The file was modified
doc/hwloc.doxy (diff) The file was modified
contrib/completion/bash/hwloc (diff)
Commit
682f24c1bd4ad3b8982503223ff27738cae7964e
by Brice.Goglin contrib/dist/publish_doc: display home page links We often have to update home page links to examples/upgradeAPI pages, have the doc install script list them. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
contrib/dist/publish_doc (diff)
Commit
ac6a1381e99510792393b7ba927f4caa69ea663a
by Brice.Goglin linux: add support for generic-initiator ACPI HMAT information Starting with Linux 5.10, the kernel may expose Generic Initiators (could be a FPGA/GPU) that have faster access to some memory targets than usual CPU initiators. nodes' access0 sysfs directory now contains the fastest of GI and CPU initiators. access1 contains the fastest of CPU initiators only (just like access0 in previous kernels). Hence we now try access1 to avoid GI if any, or fallback to access0 otherwise (earlier kernels). Older hwloc releases without knowledge of GI/access1 will work fine as long as the hardware doesn't expose a GI (unlikely in the near future). If the hardware exposes some GIs, we may fail to find the CPU-locality of some memory targets. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-linux.c (diff) The file was modified
NEWS (diff)
Commit
9f1aafbde3a9a9e7ad24e18cb88c3d4896fadaae
by bgoglin Move PGI compiler check before GNU * PGI compiler has a GNU compatability mode so `__GNUC__` is defined by `pgcc`. So first check for `__PGI` before checking for GNU. Signed-off-by: Joshua Hursey <jhursey@us.ibm.com> The file was modified
config/hwloc_check_vendor.m4 (diff) The file was modified
hwloc/topology-linux.c (diff)
Commit
f2c3562842248bf55290dfa76411af0f053ddfc3
by Brice.Goglin linux: add support for generic-initiator ACPI HMAT performance information Commit ac6a1381e99510792393b7ba927f4caa69ea663a was about locality, this one is for latency/bandwidth memattrs. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-linux.c (diff)
Commit
cf159165ff706ad54d92e3bf02fc099542584861
by Brice.Goglin tests/linux: add a sysfs test with complex initiators/targets There's one node per pair of core and one per 4-core package. All nodes are either cpu-only (0-3) or memory-only (5-10) Hence we have to hide CPU-only nodes and find the locality of memory nodes using sysfs target/initiator files. There's also a generic initiator (node4) that has faster access to memory node5/6/7 than its local CPU. It doesn't appear in our topology but forces us to use access1 instead of access0. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/linux/Makefile.am (diff) The file was added tests/hwloc/linux/fakememinitiators-1np2c+1npp+gi.output The file was added tests/hwloc/linux/fakememinitiators-1np2c+1npp+gi.tar.bz2
Commit
d18c44a3d6b1b8163d980d50f33718d2bd68f045
by Brice.Goglin tests/linux: remove fakememinitiators-1npc+1npc Its name was wrong (1npc+1npc instead of 1npc+1npp), and the new fakememinitiators-1np2c+1npp+gi added earlier covers more features and corner cases. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was removed tests/hwloc/linux/fakememinitiators-1npc+1npc.output The file was modified
tests/hwloc/linux/Makefile.am (diff) The file was removed tests/hwloc/linux/fakememinitiators-1npc+1npc.tar.bz2
Commit
606f988d074878d33a9359a0317ea494aaef6a3b
by Brice.Goglin contrib/android: fix the VERSION That version is only used internally, e.g. it appears in generated XMLs. The apk version comes from the android-specific files. Signed-off-by: Valentin Hoyet <valentin.hoyet@inria.fr> The file was modified
contrib/android/include/hwloc/autogen/config.h (diff) The file was modified
contrib/android/include/hwloc/autogen/config.h (diff) The file was modified
contrib/android/AndroidApp/lstopo/build.gradle (diff) The file was modified
contrib/android/include/private/autogen/config.h (diff)
Commit
093ecec7b5d10d362bb0dab1c151913ee526e6b4
by Brice.Goglin contrib/git-hooks: add a script for checking version numbers in pre-commit hooks Android and Windows have their own custom config.h files with hardwired version numbers. We always forget to update them (that's why contrib/windows/test-windows-version.sh fails in CI after releases, but checking before commit/push would be better). Android also requires to update the apk version in the gradle config. We add a check-versions.sh script (which verifies both windows and android version numbers) and may be installed in a git pre-commit hook. Signed-off-by: Valentin Hoyet <valentin.hoyet@inria.fr> Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was added contrib/git-hooks/check-versions.sh
Commit
a3a68b489f2d1627380ae3c30cf4e75d7a4eefca
by Brice.Goglin contrib/git-hooks: add a pre-commit hook template It calls check-versions.sh and update-my-copyright.pl It must be installed in .git/hooks/pre-commit (and/or in ~/.git_template/hooks/pre-commit for future clones). Not sure we can enforce this installation, but we don't want anyway since people may want to add more checks there. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> signed-off-by: Valentin Hoyet <valentin.hoyet@inria.fr> The file was added contrib/git-hooks/pre-commit The file was modified
hwloc/bitmap.c (diff) The file was modified
tests/hwloc/hwloc_bitmap.c (diff) The file was modified
contrib/ci.inria.fr/job-1-visualstudio.bat (diff) The file was modified
contrib/ci.inria.fr/job-3-cygwin.sh (diff) The file was modified
utils/lstopo/lstopo-cairo.c (diff)
Commit
35e3824d47bc7316eacc7ea65d66da6d605f72a8
by Brice.Goglin contrib/android: update to apk 1.5.1 - Add link to the discovery debug/log file (instead of having to dig into android storage manually) - Add about page with links to website, github, CI - Random small improvements Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
contrib/android/AndroidApp/lstopo/src/main/AndroidManifest.xml (diff) The file was added contrib/android/AndroidApp/lstopo/src/main/res/layout/about.xml The file was modified
contrib/android/AndroidApp/gradle/wrapper/gradle-wrapper.properties (diff) The file was modified
contrib/android/AndroidApp/lstopo/src/main/java/com/hwloc/lstopo/Lstopo.java (diff) The file was modified
contrib/android/AndroidApp/lstopo/src/main/res/values/strings.xml (diff) The file was added contrib/android/AndroidApp/lstopo/src/main/java/com/hwloc/lstopo/About.java The file was modified
contrib/android/AndroidApp/lstopo/src/main/res/menu/activity_main_menu_drawer.xml (diff) The file was modified
contrib/android/AndroidApp/lstopo/build.gradle (diff) The file was added contrib/android/AndroidApp/lstopo/src/main/androidManifest.txt The file was modified
contrib/android/AndroidApp/lstopo/src/main/java/com/hwloc/lstopo/MainActivity.java (diff) The file was modified
contrib/android/AndroidApp/lstopo/src/main/cpp/lib.c (diff) The file was modified
contrib/android/AndroidApp/lstopo/src/main/java/com/hwloc/lstopo/Lstopo.java (diff) The file was modified
contrib/git-hooks/check-versions.sh (diff)
Commit
a10b9e7b3afc2de3487354b6a86f49492538e1c3
by samuel.thibault lstopo-ascii: Use last colors instead of first colors People would be used to defining their own colors, typically starting from color 16. When we know max_value from ncurses, lstopo should rather define colors from last colors, to avoid clashing with user-defined colors. Fixes #388 The file was modified
utils/lstopo/lstopo-ascii.c (diff) The file was modified
include/hwloc/memattrs.h (diff) The file was modified
include/hwloc/memattrs.h (diff)
Commit
4c2e69957227369647c063ad39460fc143089f64
by Brice.Goglin linux: factorize some fd/path reading functions back There's no strong need to have read_fd_as_cpumask/list() separated from read_path_as_cpumask/list(), merge them back. Also it avoids using open() in some cases (no fsroot) and hwloc_open() in others (fsroot). Also it will simplify support for readfile() (if it ever gets merged) since we now open/read/close in the same function, but there's still some work to do for reading cpumask/cpulist with readfile(). Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-linux.c (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
hwloc/topology-linux.c (diff)
Commit
806a8c2a8eeea73c8a18d1a4da265a6c9e5ae247
by Brice.Goglin tests/diff: fix the test when the first NUMA node is CPU less NUMA differences may actually appear before PU differences if the first NUMA node is CPU less. Handle both orders in the assertions. Thanks to James D. Trotter for reporting the issue. Closes #374. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/hwloc_topology_diff.c (diff) The file was modified
hwloc/topology.c (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
include/private/private.h (diff)
Commit
3a1dbc303138063814760bbc27ff07d3d86e8a0d
by Brice.Goglin API: add hwloc/cpukinds.h for exposing difference between cores Backends may register cpusets with optional info attributes describing contained PUs. The core will agregate these info into individual "kinds" of PUs with same attributes. Each PU may be part of a single kind. Backends may also force an efficiency value that the core will use to rank kinds. We'll later add heuristics for ranking kinds based on info attributes. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
include/Makefile.am (diff) The file was modified
doc/doxygen-config.cfg.in (diff) The file was modified
NEWS (diff) The file was modified
hwloc/Makefile.am (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
include/hwloc.h (diff) The file was modified
include/private/private.h (diff) The file was added hwloc/cpukinds.c The file was modified
tests/hwloc/hwloc_topology_abi.c (diff) The file was modified
contrib/windows/libhwloc.vcxproj.filters (diff) The file was modified
contrib/android/AndroidApp/lstopo/src/main/cpp/lib.c (diff) The file was added include/hwloc/cpukinds.h The file was modified
doc/Makefile.am (diff) The file was modified
contrib/windows/libhwloc.vcxproj (diff) The file was modified
hwloc/topology.c (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
hwloc/hwloc2.dtd (diff) The file was modified
hwloc/topology-xml.c (diff) The file was added tests/hwloc/xml/fakecpukinds.xml The file was modified
tests/hwloc/xml/Makefile.am (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff) The file was modified
utils/lstopo/lstopo-text.c (diff) The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
NEWS (diff) The file was modified
contrib/completion/bash/hwloc (diff) The file was modified
utils/lstopo/lstopo.h (diff) The file was modified
hwloc/topology-x86.c (diff) The file was modified
include/hwloc/cpukinds.h (diff) The file was modified
NEWS (diff)
Commit
aef06ffb4a22c23e0abb552f00c40d1505e8dbcd
by Brice.Goglin linux: read cpufreq/cpuinfo_max_freq and base_frequency to create cpukinds cpuinfo_max_freq is available in all cpufreq drivers, but it's not totally clear it matches efficiency. base_frequency is likely more reliable, but only available in the intel_pstate cpufreq driver. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
include/hwloc/cpukinds.h (diff) The file was modified
tests/hwloc/linux/2arm-2c.output (diff) The file was modified
tests/hwloc/linux/8ia64-2s2c2t.output (diff) The file was modified
tests/hwloc/linux/32amd64-4s2n4c-cgroup2.xml.output (diff) The file was modified
NEWS (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
tests/hwloc/linux/32em64t-2n8c+1mic.output (diff) The file was modified
tests/hwloc/linux/8em64t-4c2t.output (diff) The file was modified
tests/hwloc/linux/32amd64-4s2n4c-cgroup2.output (diff) The file was modified
tests/hwloc/linux/128ia64-17n4s2c.output (diff) The file was modified
tests/hwloc/linux/48amd64-4d2n6c-sparse.output (diff)
Commit
82eb4209cbd74b241c8a050c36a92bef01069c8a
by Brice.Goglin windows: use RtlGetVersion() for reliable version checks without manifest GetVersion() works fine inside hwloc now that we build with a manifest. However, external users will need a manifest too when building their program so that their hwloc enables Windows10+ code properly (even if they use a libhwloc that was built with a manifest). Not easy. Use the RtlGetversion() symbol to get the real Windows version. This is sort of ugly but many people have used this for a while, hence Microsoft will likely have a hard time breaking it. If the symbol is ever missing, we'll fallback to manifest+GetVersion. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/ports/include/windows/windows.h (diff) The file was modified
hwloc/topology-windows.c (diff) The file was modified
hwloc/topology-windows.c (diff) The file was modified
NEWS (diff) The file was modified
include/hwloc/cpukinds.h (diff)
Commit
6e4731e81775718864365f37dd7347fecf82a931
by Brice.Goglin cpukinds: add some ranking/efficiency heuristics We first look at the forced efficiency if any (provided by Windows 10 only for now). Otherwise, we use the coretype (IntelAtom or IntelCore) and frequencies to define efficiencies (base frequency if available, max frequency otherwise). On ARM, we only use frequencies since there is no information about core type. It's not clear how things will evolve in the future, so we have several other heuristics that may be enbaled through the HWLOC_CPUKINDS_RANKING env var (coretype+frequency/coretype/frequency/frequency_max/frequency_base/ forced_efficiency/no_forced_efficiency/none). Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
doc/hwloc.doxy (diff) The file was modified
tests/hwloc/linux/8ia64-2s2c2t.output (diff) The file was modified
hwloc/cpukinds.c (diff) The file was modified
tests/hwloc/linux/128ia64-17n4s2c.output (diff) The file was modified
tests/hwloc/linux/32amd64-4s2n4c-cgroup2.output (diff) The file was modified
include/hwloc/cpukinds.h (diff) The file was modified
tests/hwloc/linux/48amd64-4d2n6c-sparse.output (diff) The file was modified
tests/hwloc/linux/2arm-2c.output (diff) The file was modified
tests/hwloc/linux/8em64t-4c2t.output (diff) The file was modified
hwloc/cpukinds.c (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
include/hwloc/cpukinds.h (diff) The file was removed tests/hwloc/linux/fakeheteronuma.env The file was removed tests/hwloc/linux/fakeheteronuma.tar.bz2 The file was added tests/hwloc/linux/fakeheterocpunuma.output The file was added tests/hwloc/linux/fakeheterocpunuma.env The file was added tests/hwloc/linux/fakeheterocpunuma.options The file was removed tests/hwloc/linux/fakeheteronuma.output The file was modified
tests/hwloc/linux/Makefile.am (diff) The file was removed tests/hwloc/linux/fakeheteronuma.options The file was added tests/hwloc/linux/fakeheterocpunuma.tar.bz2
Commit
d14aea6e6c12d5801044bbfa70a5d7d45fe16296
by Brice.Goglin tests/x86: add a test with fake cpukind information Modified from my laptop by adding 0x1a CPUID leaf exposing Intel Atom/Core types, and by setting the hybrid feature (bit 15 of cpuid7.edx). Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was added tests/hwloc/x86/Intel-CPUID.1A-1p2co2t.tar.bz2 The file was modified
tests/hwloc/x86/Makefile.am (diff) The file was added tests/hwloc/x86/Intel-CPUID.1A-1p2co2t.env The file was added tests/hwloc/x86/Intel-CPUID.1A-1p2co2t.output The file was modified
utils/hwloc/hwloc-calc.1in (diff) The file was modified
NEWS (diff) The file was modified
utils/hwloc/hwloc-calc.c (diff) The file was modified
contrib/completion/bash/hwloc (diff) The file was modified
utils/hwloc/test-hwloc-calc.sh.in (diff) The file was modified
utils/hwloc/test-hwloc-calc.output (diff) The file was modified
utils/hwloc/hwloc-info.c (diff) The file was modified
NEWS (diff) The file was modified
utils/hwloc/test-hwloc-info.sh.in (diff) The file was modified
utils/hwloc/test-hwloc-info.output (diff) The file was modified
hwloc/cpukinds.c (diff) The file was modified
doc/Makefile.am (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
include/hwloc/cpukinds.h (diff) The file was modified
tests/hwloc/Makefile.am (diff) The file was added tests/hwloc/cpukinds.c The file was modified
utils/hwloc/hwloc-annotate.c (diff) The file was modified
utils/hwloc/hwloc-annotate.1in (diff) The file was modified
utils/hwloc/test-hwloc-annotate.output (diff) The file was modified
utils/hwloc/test-hwloc-annotate.sh.in (diff)
Commit
f2d852fbc3e1d060a9a757734dad0acf2f137790
by Brice.Goglin API: add a cpukind_efficiency bit in the discovery support array Only set on Windows for now. Forced efficiency are only available on Windows, or when given by the user. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/hwloc-info.c (diff) The file was modified
tests/hwloc/xml/fakecpukinds.xml (diff) The file was modified
include/hwloc.h (diff) The file was modified
tests/hwloc/hwloc_topology_abi.c (diff) The file was modified
hwloc/topology-windows.c (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
include/hwloc/cpukinds.h (diff) The file was modified
utils/hwloc/test-hwloc-info.output (diff) The file was modified
utils/lstopo/lstopo.h (diff) The file was modified
utils/lstopo/lstopo.c (diff)
Commit
a55ae9eae6d12fcec27a230c8af4c1865aea9c1a
by Brice.Goglin lstopo/draw: change the format of PU boxes for CPU kinds When there are multiple kinds, use dot/dash lines with higher thickness for PUs not of the first kind. Bold font is also used for kinds with odd indexes. Kind#0 gets normal boxes Kind#1 gets boxes with small dashes and bold font Kind#2 gets larger dashes, thicker lines, and non-bold font etc The thickness isn't the same in all backends, depends on the actual rendering. Windows doesn't support dash/dot if thickness>1, hence we don't change the thickness at all there. All this may be disabled with --no-cpukinds, or toggled with 'k' in the interactive outputs. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> Thanks to Clement Foyer for the tikz part. Signed-off-by: Clement Foyer <cfoyer@hpe.com> The file was modified
utils/lstopo/lstopo-tikz.c (diff) The file was modified
contrib/completion/bash/hwloc (diff) The file was modified
contrib/android/AndroidApp/lstopo/src/main/java/com/hwloc/lstopo/MainActivity.java (diff) The file was modified
NEWS (diff) The file was modified
utils/lstopo/lstopo-android.c (diff) The file was modified
contrib/android/AndroidApp/lstopo/src/main/cpp/lib.c (diff) The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
contrib/android/AndroidApp/lstopo/src/main/java/com/hwloc/lstopo/Lstopo.java (diff) The file was modified
utils/lstopo/lstopo.h (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff) The file was modified
utils/lstopo/lstopo-cairo.c (diff) The file was modified
tests/hwloc/ports/include/windows/windows.h (diff) The file was modified
utils/lstopo/lstopo-fig.c (diff) The file was modified
utils/lstopo/lstopo-svg.c (diff) The file was modified
utils/lstopo/lstopo-windows.c (diff) The file was modified
hwloc/topology-linux.c (diff)
Commit
51795fba9d9d3dbbb473389c2822e60bceb755ef
by Brice.Goglin linux/cpukinds: sort by frequency before registering Lower frequency likely means lower efficiency. Hence register frequency-based cpukinds in that order for consistency with cases where we have enough info for ranking. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/test-hwloc-calc.output (diff) The file was modified
tests/hwloc/linux/fakeheterocpunuma.output (diff) The file was modified
utils/hwloc/test-hwloc-info.output (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
include/hwloc.h (diff) The file was modified
contrib/ci.inria.fr/browse_jenkins_logs.sh (diff) The file was removed contrib/android/AndroidApp/lstopo/src/main/androidManifest.txt The file was added contrib/android/AndroidApp/lstopo/src/main/androidManifest.xml The file was added contrib/android/assets/lstopo.svg The file was added contrib/android/assets/lstopo.png The file was modified
contrib/android/AndroidApp/lstopo/src/main/java/com/hwloc/lstopo/MainActivity.java (diff)
Commit
04a4977a2e076883fc5a94edbdb571bea2bc824e
by Brice.Goglin linux: use Linux 5.11 sysfs cpu types to detect Intel Atom/Core types It's not clear it will be in 5.11 or not, and whether the sysfs files/directories will be different. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-linux.c (diff) The file was modified
NEWS (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
utils/hwloc/test-hwloc-info.output (diff) The file was modified
tests/hwloc/linux/fakeheterocpunuma.output (diff) The file was modified
tests/hwloc/linux/fakeheterocpunuma.tar.bz2 (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
contrib/android/include/hwloc/autogen/config.h (diff) The file was modified
contrib/android/AndroidApp/lstopo/build.gradle (diff) The file was modified
contrib/windows/hwloc_config.h (diff) The file was modified
VERSION (diff) The file was modified
hwloc/cpukinds.c (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/topology-solaris.c (diff) The file was modified
include/hwloc/distances.h (diff) The file was modified
hwloc/topology-freebsd.c (diff)
Commit
d1ef509505d776a8251bcafaa00e888b106100a1
by Brice.Goglin autogen: fix MacOS X Big Sur libtool.m4 bug by patching configure This libtool patch isn't applied upstream yet but it is already used by several projects to fix build failures on the latest MacOS X (Big Sur). Upstream patch at https://lists.gnu.org/archive/html/libtool-patches/2020-06/msg00001.html applies to libtool.m4 but we have to apply it to configure after autoconf used the (buggy) libtool.m4 instead. See also open-mpi/ompi#8218. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
autogen.sh (diff) The file was modified
Makefile.am (diff) The file was added config/libtool-big-sur-fixup.patch The file was modified
utils/hwloc/misc.h (diff)
Commit
27e970774885927504a2e77be591a531ce1a7d9b
by Brice.Goglin utils/misc: improve the printing of distance matrices Scale the column width to the max of what we have to show in the matrix, don't limit values to 5 digits, etc. Also print OSDevices by name instead of indexes. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/xml/64intel64-3g2n+2n-irregulargroups+pci.console.output (diff) The file was modified
tests/hwloc/xml/16amd64-4distances.console.output (diff) The file was modified
utils/hwloc/misc.h (diff)
Commit
dc262b08504fd4d20722c0fb2a59e1355837f553
by Brice.Goglin core: fortify bridge downstream attribute use All our bridges are currently downstream to PCI, but it's not impossible it changes in the future, make the code a bit more safe. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-linux.c (diff) The file was modified
hwloc/topology-pci.c (diff) The file was modified
hwloc/traversal.c (diff) The file was modified
hwloc/pci-common.c (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
tests/hwloc/hwloc_type_sscanf.c (diff) The file was modified
contrib/git-hooks/check-versions.sh (diff) The file was modified
NEWS (diff) The file was modified
hwloc/topology-linux.c (diff)
Commit
e7bf9720fade2c1924cd55a90ece0590a46c0246
by Brice.Goglin distances: fix the import of large heterogeneous distance matrices Properly import the list of heterogeneous indexes when it is split on multiple lines. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-xml.c (diff) The file was modified
tests/hwloc/xml/fakeheterodistances.xml (diff) The file was modified
include/hwloc/plugins.h (diff) The file was modified
contrib/ci.inria.fr/job-3-sonarscanner.sh (diff) The file was modified
contrib/ci.inria.fr/job-3-sonarscanner.sh (diff)
Commit
83a16c20a2ceb5f4cba76397ba5d58f91a315f9d
by bgoglin configure: add --with-cuda=<dir> Specifies where CUDA is installed to avoid having to pass C_INCLUDE_PATH and LIBRARY_PATH at build time. LD_LIBRARY_PATH is still required at runtime. The CUDA directory is also used for looking for OpenCL and NVML subcomponents. Thanks to Andrea Bocci for the suggestion. Refs #392. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/Makefile.am (diff) The file was modified
config/hwloc.m4 (diff) The file was modified
config/hwloc_internal.m4 (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
NEWS (diff)
Commit
93692657413a8cfe2f8e963e7fdfa5ce14dd6965
by Brice.Goglin hwloc/opencl.h: fix AMD OpenCL device locality when PCI bus/device >= 128 cl_device_topology_amd stores the PCI bus/device/function as cl_char (signed). bus/device can be >= 128 and those don't cast correctly to unsigned int. Applications that call the inline function hwloc_opencl_get_device_pci_busid() directly must be recompiled to get this fix. Thanks to Edgar Leon for reporting the issue (some AMD OpenCL devices were not inserted in the PCI parent while RSMI devices were fine). Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
include/hwloc/opencl.h (diff)
Commit
706e0fbfe7bb8a535cfb1afd8f06fc4fd80c5f2b
by Brice.Goglin ci.inria.fr/sonarscanner: update exclusions Fix tests/ports exclusion Add XMLs (we already verify them with xmllint and lstopo) Add netloc_draw (we don't want to test all this HTML/JS code) Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
contrib/ci.inria.fr/job-3-sonarscanner.sh (diff)
Commit
7f793d74d07d1bf3284d0625f941a2401f412c94
by bgoglin Fix GCC checks to use the vendor name * The PGI compiler is identified as a GCC compiler (so `$GCC == yes`) This can cause problems if the user accidentally adds `--enable-picky` or if the `$GCC` variable is used to check if it is safe to add GCC specific compiler flags. * Instead of using `$GCC` we can use the vendor string returned by `_HWLOC_C_COMPILER_VENDOR`. Signed-off-by: Joshua Hursey <jhursey@us.ibm.com> The file was modified
config/hwloc.m4 (diff)
Commit
db423bc93bf10114015ba652ce36d35cf1c5207f
by Brice.Goglin contrib/android: remove unused duplicate android manifest Wasn't correctly fixed in a55490d4f65bfc97033b49a5ae237943e28d5cc7 Thanks to Eric Schnetter for reporting the issue, it caused a filename conflict on case-insensitive filesystem (MacOS). Closes #441 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was removed contrib/android/AndroidApp/lstopo/src/main/androidManifest.xml The file was modified
hwloc/topology-x86.c (diff)
Commit
45aaeb2010f32543cb3dd66fdfa784fad2597497
by Brice.Goglin x86: fix full discovery enabling when FreeBSD/NetBSD backends only created PUs Backends such as FreeBSD and NetBSD only create PUs but their priority is higher than x86. Hence x86 has to detect whether only PUs were created before adding everything else. This requires to reconnect levels earlier. Thanks to Thibault Payet and Yuri Victorovich for the report. Closes #442 Downstream FreeBSD bug report https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=252643 HPX issue about requiring Core objects https://github.com/STEllAR-GROUP/hpx/issues/5115 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-x86.c (diff) The file was modified
hwloc/topology-x86.c (diff) The file was modified
doc/Makefile.am (diff)
Commit
a3753c64dfa33753767349131c279f01983098e7
by Brice.Goglin configure: pass -Werror when checking compiler options clang warns about unknown options (while gcc errors-out) so pass -Werror when checking for options at configure time. Thanks to Jeff Squyres for the report. Refs #447 By the way, rename _HWLOC_CHECK_GCC_OPTION() to _HWLOC_CHECK_CC_OPTION() since this isn't GCC specific. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
config/hwloc_check_vendor.m4 (diff) The file was modified
config/hwloc.m4 (diff) The file was modified
autogen.sh (diff)
Commit
73c568866fcf99aa85d8f4343c3cab4d13c52765
by Brice.Goglin configure: detect when a patched libtool was used Some libtool are now available with a Big Sur patch. Detect both versions of the patch before trying to patch. Thanks to Jeff Squyres for the report. Closes #447 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
autogen.sh (diff)
Commit
eca3838533f8d58a3a087001107963e77f60b48d
by Brice.Goglin core: mark the topology as modified when we replace an object When inserting a custom group after load(), it may get merged and replace an existing group. This requires a reconnect. So mark the topology as modified so that levels are rebuilt, etc. Refs #446 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology.c (diff)
Commit
663a65969b394403d7c46802c6f67ba3568da4e3
by Brice.Goglin core: fix inserting of custom groups replacing an existing ones When inserting a group identical to an existing one, the new one may "replace" the old one (depends on their kinds). If so, we need to rebuild the level, fill its cpusets, recompute its depth and total_memory, etc. This happened when lstopo --top inserted a group (kind 0 for now) identical to a NUMA group (kind higher). Thanks to Scott Hicks for reporting the issue and providing lots of debugging information. Closes #446 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology.c (diff) The file was modified
tests/hwloc/hwloc_groups.c (diff)
Commit
4ff8c5ecc5863b89c437e400518bca6ff1ec2ffe
by Brice.Goglin API: clarify the documentation of alloc/insert_group() Detail the behavior of merging with respect to dont_merge and kinds. Allow to use high kinds for discarding whenever possible, etc. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
include/hwloc.h (diff) The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
hwloc/topology.c (diff) The file was modified
utils/hwloc/hwloc-ps.1in (diff)
Commit
5cf8a51e564e19b096aa895f0040fef45500ff8e
by Brice.Goglin tests/shmem: cleanup the enabling of specific distance checks Add a custom info attr instead on relaying on the existing SyntheticDescription (which will exist in some upcoming XML topology tests). Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/shmem.c (diff) The file was modified
tests/hwloc/shmem.c (diff) The file was modified
tests/hwloc/wrapper.sh.in (diff)
Commit
b518dcee527805d002d71a59cf03b373a91b18b7
by bgoglin Fix check for IBM xl compilers for v13.1 and later. By default newer xlc compilers only define __ibmxl__ now. https://www.ibm.com/support/knowledgecenter/en/SSXVZZ_13.1.6/com.ibm.xlcpp1316.lelinux.doc/compiler_ref/xlmacros.html Signed-off-by: Austen Lauria <awlauria@us.ibm.com> The file was modified
config/hwloc_check_vendor.m4 (diff)
Commit
b09f6539794db92d476ee117d1827488ebd21b89
by bgoglin Fix configury where most compilers will get mislabeled as 'gnu'. Most compilers define the __GNU__ macro, so put it at the bottom as a catch-all. Tested with gcc, xlc, pgi and clang. Signed-off-by: Austen Lauria <awlauria@us.ibm.com> The file was modified
config/hwloc_check_vendor.m4 (diff) The file was modified
utils/hwloc/hwloc-gather-cpuid.c (diff) The file was modified
config/hwloc.m4 (diff) The file was modified
hwloc/cpukinds.c (diff) The file was modified
tests/hwloc/x86+linux/test-topology.sh.in (diff)
Commit
1a97da7a02b6e851229082720ec4bd7b684962c2
by Brice.Goglin tests/x86+linux: add a test combining x86 CoreType and linux cpufreq Based on Intel Lakefield. Regression tests for commit 9cd818b7715dc48d0610e05cdafc30e7625af229 x86 must be called after linux because linux cannot annotate cpufreqs yet. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was added tests/hwloc/x86+linux/5intel64-hybrid-lakefield.options The file was added tests/hwloc/x86+linux/5intel64-hybrid-lakefield.tar.bz2 The file was added tests/hwloc/x86+linux/5intel64-hybrid-lakefield.output The file was added tests/hwloc/x86+linux/5intel64-hybrid-lakefield.env The file was modified
tests/hwloc/x86+linux/Makefile.am (diff) The file was modified
tests/hwloc/x86+linux/Makefile.am (diff)
Commit
20d69af930de0557c44bf936e09e6fea1dda6ef4
by Brice.Goglin linux: allow annotating cpukinds when PUs were discovered by another backend (x86) Will be useful when x86 must run first because the CPUs isn't correctly exposed by Linux. This behavior is now tested in 64amd64-4p2n4ca2co which now gets cpukinds frequency info. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/x86+linux/64amd64-4p2n4ca2co.topoextnuma.output (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
tests/hwloc/x86+linux/64amd64-4p2n4ca2co.output (diff)
Commit
7bcc273efd50536961ba16d474efca4ae163229b
by bgoglin linux: fix support for CPU0 being hotplugged Linux kernel with CONFIG_BOOTPARAM_HOTPLUG_CPU0 as compile time allow offlining CPU0. The Linux early discovery used CPU0 to detect whether we should use /sys/bus/cpu or /sys/devices/system/cpu, and whether new core/package_cpus are available. If this fails, read the first online CPU number at look at it. The overhead for the normal case (CPU0 online) is 0. Thanks to Garrett Clay for the report. Fixes #450 For hwloc 2.5, we should moderniez this code: /sys/devices/system/cpu is unlikely going away because many files were added in there since /sys/bus/cpu was added. Kernels without sysfs cpu topology files are also too old now. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-linux.c (diff) The file was modified
NEWS (diff) The file was modified
utils/hwloc/hwloc-ps.c (diff) The file was modified
utils/hwloc/hwloc-ps.1in (diff)
Commit
cbb6b960f9c1d5c2cc205e3a3df78454bd551d13
by Brice.Goglin hwloc-ps: don't apply --pid-cmd to processes that will be filtered out Split the calling of the user-provided command out of the main common-ps functions and let the caller calls it after filtering interesting processes. Otherwise hwloc-ps --pid-cmd takes several seconds on all platforms since we have to query hundreds of (uninteresting because not bound) processes. Thanks to Valentin Hoyet for the report. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/hwloc-ps.c (diff) The file was modified
utils/hwloc/common-ps.h (diff) The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
utils/hwloc/common-ps.c (diff) The file was modified
hwloc/topology-cuda.c (diff) The file was modified
hwloc/topology-pci.c (diff) The file was modified
hwloc/topology-nvml.c (diff) The file was modified
hwloc/topology-opencl.c (diff) The file was modified
hwloc/topology-rsmi.c (diff)
Commit
5fd60f712f4fe52b07ee54e738ee76288a5c7111
by Brice.Goglin utils/common-ps: fix popen/pclose for Windows Broken in cbb6b960f9c1d5c2cc205e3a3df78454bd551d13 Not strictly needed since hwloc-ps isn't built on Windows and lstopo --ps does nothing there, but the build is easy to fix. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/common-ps.c (diff) The file was modified
config/hwloc.m4 (diff) The file was modified
NEWS (diff)
Commit
9298b7769f936be35b0233f1f6e301d5dcde0026
by Brice.Goglin utils/common-ps: refix popen/pclose for Windows/Cygwin Completes commit 5fd60f712f4fe52b07ee54e738ee76288a5c7111, which tried to fix cbb6b960f9c1d5c2cc205e3a3df78454bd551d13 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/common-ps.c (diff) The file was modified
config/hwloc_internal.m4 (diff) The file was modified
config/hwloc.m4 (diff) The file was modified
config/hwloc_internal.m4 (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/Makefile.am (diff) The file was modified
config/hwloc.m4 (diff) The file was modified
tests/hwloc/Makefile.am (diff) The file was modified
config/hwloc.m4 (diff) The file was modified
config/hwloc_internal.m4 (diff) The file was modified
tests/hwloc/Makefile.am (diff)
Commit
a5695d47251ef80d391e2d18688d5af3ac4488d6
by Brice.Goglin configure: use CPPFLAGS/CFLAGS/LIBS/LDFLAGS consistently Some of these variables aren't defined for some backends, depending on whether we detect using pkg-config or not. Don't propagate these differences everywhere, always use CPPFLAGS/CFLAGS/LIBS/LDFLAGS even if they are empty for now. Applies to global lib build variables, plugin variables, and tests variables. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
config/hwloc.m4 (diff) The file was modified
tests/hwloc/Makefile.am (diff) The file was modified
hwloc/Makefile.am (diff) The file was modified
NEWS (diff) The file was modified
config/hwloc_internal.m4 (diff) The file was modified
config/hwloc.m4 (diff) The file was modified
NEWS (diff) The file was modified
doc/hwloc.doxy (diff)
Commit
f95e4b8ecb38819ad067b9b49053d4f7a66771f3
by Brice.Goglin cuda: don't warn if no device were found cuda returns an error when there are no devices. Unfortunately, nvmlInit() returns "unknown error" (999) in the same case. I am not sure we can filter-out this error in all 999 cases. Thanks to Aurelien Bouteiller for the report. Refs #453 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/ports/include/cuda/cuda_runtime_api.h (diff) The file was modified
hwloc/topology-cuda.c (diff)
Commit
6295cb21c2663097e69efb161359ad4f55a74a06
by Brice.Goglin API: add hwloc/windows.h for querying windows processor groups Windows processor groups are not a physical hierarchical level (although it may be exposed by hwloc if it's different from NUMA nodes, packages, etc) but it has an impact of application because binding must remain inside a single group. Expose those groups explicitly so that applications know what their binding masks should be restricted to. Thanks to Ivan Kochin for the suggestion. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
NEWS (diff) The file was modified
doc/hwloc.doxy (diff) The file was added include/hwloc/windows.h The file was modified
contrib/windows/lstopo-no-graphics.vcxproj.filters (diff) The file was modified
contrib/windows/lstopo.vcxproj (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
doc/Makefile.am (diff) The file was modified
contrib/windows/lstopo.vcxproj.filters (diff) The file was modified
contrib/windows/libhwloc.vcxproj (diff) The file was modified
include/Makefile.am (diff) The file was modified
contrib/windows/lstopo-no-graphics.vcxproj (diff) The file was modified
doc/doxygen-config.cfg.in (diff) The file was modified
contrib/windows/libhwloc.vcxproj.filters (diff) The file was modified
hwloc/topology-windows.c (diff) The file was modified
contrib/windows/lstopo-win.vcxproj.filters (diff) The file was modified
contrib/windows/lstopo-win.vcxproj (diff) The file was modified
tests/hwloc/rename/main.c (diff) The file was added tests/hwloc/windows_processor_groups.c The file was modified
tests/hwloc/Makefile.am (diff) The file was modified
NEWS (diff) The file was modified
utils/lstopo/lstopo.h (diff) The file was modified
utils/lstopo/lstopo-text.c (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff) The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/topology-windows.c (diff) The file was modified
hwloc/topology-windows.c (diff)
Commit
eada3d251b13531a45efee8bff978d974efe7a3f
by Brice.Goglin lstopo: fix --windows-processor-groups name confusion --processor-groups was only used during development, real/final name is --windows-processor-groups Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
contrib/ci.inria.fr/job-3-cygwin.sh (diff) The file was modified
contrib/ci.inria.fr/job-3-mingw.sh (diff) The file was modified
contrib/ci.inria.fr/job-1-visualstudio.bat (diff)
Commit
835dfbe577fcd7fa5e2b86f0af5b0051803d591d
by Brice.Goglin core: don't display "less critical" error messages by default Failure to initialize CUDA isn't critical enough to pollute external users' stacks with hwloc warnings. Set HWLOC_HIDE_ERRORS to 1 by default to that those aren't displayed by default anymore. Set it back to 0 in lstopo to get them there. Invalid hardware topology, invalid XML input, missing 32bits PCI domains, etc. are much more important, display them unless HWLOC_HIDE_ERRORS is set to 2. Thanks to Aurelien Bouteiller in #453. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-xml.c (diff) The file was modified
include/hwloc/plugins.h (diff) The file was modified
hwloc/cpukinds.c (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/topology-pci.c (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
hwloc/topology.c (diff) The file was modified
hwloc/pci-common.c (diff) The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
hwloc/components.c (diff) The file was modified
hwloc/topology-aix.c (diff) The file was modified
hwloc/topology-solaris.c (diff) The file was modified
hwloc/cpukinds.c (diff) The file was modified
hwloc/pci-common.c (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
hwloc/topology.c (diff) The file was modified
NEWS (diff) The file was modified
hwloc/topology-rsmi.c (diff) The file was modified
utils/hwloc/hwloc-ps.c (diff) The file was modified
NEWS (diff) The file was modified
contrib/completion/bash/hwloc (diff) The file was modified
utils/hwloc/hwloc-ps.1in (diff) The file was modified
contrib/completion/bash/hwloc (diff)
Commit
5958eff9c63e2243b23edb1215b16ac5c23b602d
by Brice.Goglin contrib/completion: add hwloc-ps --short-name Forgotten in ce8ea0212129ba093e50588754d7d28a864e33bf By the way, uniformize the order here and in usage. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
contrib/completion/bash/hwloc (diff) The file was modified
utils/hwloc/hwloc-ps.c (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/topology-nvml.c (diff) The file was modified
hwloc/topology-gl.c (diff) The file was modified
hwloc/topology-rsmi.c (diff)
Commit
3c8ed197d9a017ca5399007861981b60032e7ca6
by Brice.Goglin API: add hwloc_get_obj_with_same_locality() May convert normal/memory objects with same cpu/nodesets or I/O object (PCI/OS in the same PCI). Will be useful for transforming upcoming GPU Link distances which will use rsmi/nvml objects by default. HWLOC_API_VERSION gets bumped to 0x20500 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
NEWS (diff) The file was modified
include/hwloc.h (diff) The file was modified
include/hwloc/helper.h (diff) The file was added tests/hwloc/hwloc_get_obj_with_same_locality.c The file was modified
include/hwloc/distances.h (diff) The file was modified
hwloc/traversal.c (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
tests/hwloc/Makefile.am (diff) The file was modified
include/hwloc/distances.h (diff) The file was modified
utils/hwloc/misc.h (diff)
Commit
0a0cec6222c8410b369b318cdf261ae0973c6b98
by Brice.Goglin doxy: cite DRM card0/renderD128 devices in OS device examples Those are filtered out by default in recent releases, but it's better to keep them documented. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
doc/hwloc.doxy (diff)
Commit
fbfff1594d06559228685f0e45a0bf3786cb53ce
by Brice.Goglin OpenCL: the OpenCL device type is a bitset CL_DEVICE_TYPE may return DEFAULT|CPU|GPU Ignore DEFAULT, ignore CPU-only, and only keep the first one of GPU|ACCELERATOR|CUSTOM. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-opencl.c (diff) The file was modified
tests/hwloc/ports/include/opencl/CL/cl.h (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
NEWS (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff) The file was modified
utils/hwloc/hwloc-gather-topology.in (diff)
Commit
e17cb9634e5c0fa149a512a7176c800867469a5e
by bgoglin tests/linux: add a testcase with NEC Vector Engine Based on a real output from RWTH Aachen, pruned to make it smaller, and modified to put the 8th VE in NUMA partitioned mode. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was added tests/hwloc/linux/32intel64-2p8co2t+8ve.output The file was added tests/hwloc/linux/32intel64-2p8co2t+8ve.tar.bz2 The file was modified
tests/hwloc/linux/Makefile.am (diff) The file was modified
include/hwloc/cuda.h (diff) The file was modified
include/hwloc/cudart.h (diff) The file was modified
include/hwloc/gl.h (diff) The file was modified
include/hwloc/nvml.h (diff) The file was modified
include/hwloc/opencl.h (diff) The file was modified
include/hwloc/openfabrics-verbs.h (diff) The file was modified
include/hwloc/rsmi.h (diff) The file was modified
doc/hwloc.doxy (diff)
Commit
185b4b73326ae4d4f382725b0b42f98354baea92
by Brice.Goglin Add a "levelzero" backend "Level Zero" is the lowlevel API for managing oneAPI devices. We create "ze%d" devices for these GPUs (mostly Intel for now), because "ze" is the prefix for function names in the levelzero API. CPPFLAGS should point to ze_api.h that is currently installed under $prefix/include/level_zero by the oneAPI compute runtime. This code is for the L0 1.0 API, it won't work with previous releases (e.g. 0.91). We try to enable the required "Sysman" L0 sub-API but it may be too late. More to come about this. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
config/hwloc_internal.m4 (diff) The file was added tests/hwloc/ports/include/levelzero/zes_api.h The file was modified
doc/hwloc.doxy (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
configure.ac (diff) The file was modified
NEWS (diff) The file was modified
contrib/ci.inria.fr/job-3-sonarscanner.sh (diff) The file was modified
config/hwloc.m4 (diff) The file was added tests/hwloc/ports/include/levelzero/ze_api.h The file was modified
tests/hwloc/ports/Makefile.am (diff) The file was modified
include/private/internal-components.h (diff) The file was modified
hwloc/Makefile.am (diff) The file was added hwloc/topology-levelzero.c The file was modified
config/hwloc_check_attributes.m4 (diff)
Commit
3108bf38640a9fea1f06e7c81106d166da4d5d66
by Brice.Goglin levelzero: define ZES_ENABLE_SYSMAN=1 in the library constructor and Windows DllMain The variable must be defined before anybody else enables L0, so try to define as early as possible. This is in the main hwloc library and not in the levelzero component because the latter might be loaded as a plugin (i.e. later, and too late). Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology.c (diff) The file was modified
hwloc/topology-levelzero.c (diff) The file was modified
doc/doxygen-config.cfg.in (diff) The file was modified
tests/hwloc/rename/Makefile.am (diff) The file was modified
NEWS (diff) The file was modified
doc/Makefile.am (diff) The file was modified
include/Makefile.am (diff) The file was modified
tests/hwloc/rename/main.c (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
doc/hwloc.doxy (diff) The file was added include/hwloc/levelzero.h The file was modified
tests/hwloc/Makefile.am (diff) The file was added tests/hwloc/levelzero.c
Commit
00b1d2660fa8c56f42d9c9e5b839b13d21e674bc
by Brice.Goglin configure: add support for level-zero.pc My upstream PR was merged a couple hours ago, let's support it, even if we don't know when it'll be included in official packages and/or whether le .pc might be renamed in the meantime. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
config/hwloc.m4 (diff)
Commit
9914081175c9aa177656474599424328b07bad11
by Brice.Goglin tests/rename: pass rsmi/nvml/cuda/opencl/levelzero CPPFLAGS If these backends are detected, tests/rename will also check their interoperability headers, which requires their CPPFLAGS. It never broke the CI because those dependencies are not available on the CI slaves (except OpenCL) and we don't use anything installed outside of the standard directories. We actually use both CFLAGS and CPPFLAGS because pkg-config doesn't distinguish them. openfabrics doesn't get CPPFLAGS because our build doesn't define them (only needed for tests) and it's always installed in standard paths anyway. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/rename/Makefile.am (diff)
Commit
12659255eb90394a78b3ba53a7e236d67fbf73f9
by bgoglin Make -ltermcap the last option for ncurses On macOS there seems to be a symlink from /usr/lib/libtermcap.dylib to libncurses.5.4.dylib, which is the first option hwloc currently considers for detecting ncurses. If you install ncurses with brew or spack, you don't get this symlink, so hwloc will always link to the system libncurses instead of the brew/spack version. By considering libncurses first this should be fixed. Signed-off-by: Harmen Stoppels <me@harmenstoppels.nl> The file was modified
config/hwloc_internal.m4 (diff) The file was modified
hwloc/distances.c (diff)
Commit
22b3688aa5207d8ecc68d00a615d13320eb7ac91
by Brice.Goglin distances: add hwloc_distances_transform() Modifies a distances structure obtained with _get(). Does not propage changes into the core distances info yet. Current transformations are: * REMOVE_NULL to remove objects that were replaced by NULL in the objs array * LINKS to replace bandwidth matrices with number of links More to come with GPU matrices. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
include/hwloc/rename.h (diff) The file was modified
doc/Makefile.am (diff) The file was modified
hwloc/distances.c (diff) The file was modified
include/hwloc/distances.h (diff) The file was modified
NEWS (diff) The file was modified
tests/hwloc/hwloc_distances.c (diff) The file was modified
include/hwloc/distances.h (diff) The file was modified
hwloc/distances.c (diff)
Commit
cd98115cf6cfe4c1630faae9d48cb1c6b5a5e9d6
by Brice.Goglin distances: rework internals of adding distances Split the process into create/values/commit so that we can reduce the crazy number of arguments in public functions and allow future additions to the API. Nothing changes outside of distances.c yet. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-xml.c (diff) The file was modified
hwloc/distances.c (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
include/private/private.h (diff)
Commit
fc0e312aadd63c894c4f851a4eebbfb5fae5d87f
by Brice.Goglin plugins.h: export low-level distances add_create/add_values/add_commit functions to plugins These functions are more flexible and will allow future additions of parameters such as info attributes attached to distances. For now this is only the plugin API because GPU backends will need them soon. The public API for users will come in other commits. add_values_by_index() isn't exported since it's unlikely to be used outside of the XML component, which isn't plugin'able. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
include/private/private.h (diff) The file was modified
include/hwloc/plugins.h (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
hwloc/distances.c (diff)
Commit
ad7d08c26ae77f83a988bc8569fadb08fc04deff
by Brice.Goglin distances.h: replace add() with add_create(), add_values(), add_commit() This is the public API for end users following the plugin API added in previous commit. It lets users specify the distance name, and it will be extendable later if we ever need to add different things to distances. hwloc_distances_add() is now deprecated. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/distances.c (diff) The file was modified
include/hwloc/plugins.h (diff) The file was modified
include/hwloc/distances.h (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
doc/Makefile.am (diff) The file was modified
NEWS (diff) The file was modified
include/hwloc/deprecated.h (diff) The file was modified
include/hwloc/distances.h (diff) The file was modified
doc/Makefile.am (diff) The file was modified
tests/hwloc/hwloc_topology_dup.c (diff) The file was modified
tests/hwloc/hwloc_distances.c (diff) The file was modified
tests/hwloc/hwloc_groups.c (diff) The file was modified
tests/hwloc/hwloc_topology_restrict.c (diff) The file was modified
tests/hwloc/shmem.c (diff) The file was modified
utils/hwloc/hwloc-annotate.c (diff) The file was modified
utils/hwloc/test-hwloc-annotate.sh.in (diff) The file was modified
utils/hwloc/hwloc-annotate.1in (diff) The file was modified
utils/hwloc/test-hwloc-annotate.output (diff) The file was modified
utils/hwloc/hwloc-annotate.c (diff) The file was modified
utils/hwloc/hwloc-annotate.c (diff) The file was modified
utils/hwloc/hwloc-annotate.c (diff) The file was modified
utils/hwloc/hwloc-annotate.1in (diff) The file was modified
NEWS (diff) The file was modified
utils/hwloc/test-hwloc-annotate.sh.in (diff)
Commit
32c994b7f3822ffe143423886840a632f88e07ca
by Brice.Goglin pci: keep PCI-to-non-PCI bridges until the core filters them out NVLink endpoints can land on custom bridges near the CPU, but the current code ignores those bridges by default. Now keep all pci-to-non-pci bridges in the PCI backends (those are very uncommon anyway) and let the core filter them out later if they have no child just like PCI bridges. This adds a PCISlot to some fake Linux tests because an IDE controller gets it from an ISABridge that doesn't get ignored early anymore (their PCI BDF is identical except the F). Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/linux/2pa-pcidomain32bits.output (diff) The file was modified
include/hwloc/plugins.h (diff) The file was modified
tests/hwloc/linux/2pa-pcidomain32bits-disabled.output (diff) The file was modified
hwloc/topology.c (diff) The file was modified
include/hwloc/plugins.h (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
include/private/private.h (diff) The file was modified
include/hwloc/plugins.h (diff)
Commit
fa8fe41ca71e9d06dc90bd0328ed7aacd71bf305
by Brice.Goglin nvml: add a NVLinkBandwidth matrix Objects are GPU "nvml" OS devices for GPUs, Package for POWER OpenCAPI/NVLink CPU ports, and a PCI device for NVSwitch. We assume that NVLink peers of PCI class 0x0680 and vendor IBM are NVLink ports on the CPU. That's true for POWER8/9 but no reply from IBM about the future or some sort of standardization of these OpenCAPI ports. We also assume that NVLink peers of PCI class 0x0680 and vendor NVIDIA are NVSwitches. NVIDIA doesn't provide any way to distinguish peer types. POWER8 doesn't have correct PCI locality for these OpenCAPI ports, hence we use the GPU locality instead, unless HWLOC_NVML_USE_OPENCAPI_LOCALITY is set in the environment. We will need some high level functions to summarize this matrix because it's not obvious on DGX platforms with 16 GPUs and a NVSwitch: * there are two independant set of connected GPUs/Switch, inter-switch links are not reported * there are 6 hardware switches on each side, connected to one of the 6 links of the gpus Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-nvml.c (diff) The file was modified
tests/hwloc/ports/include/nvml/nvml.h (diff) The file was modified
NEWS (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
include/hwloc/distances.h (diff) The file was modified
hwloc/topology.c (diff) The file was modified
hwloc/topology-nvml.c (diff) The file was modified
NEWS (diff) The file was modified
include/hwloc/distances.h (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/topology-rsmi.c (diff) The file was modified
doc/hwloc.doxy (diff)
Commit
5aee2044ab6a8f2a486ef449ba55806d75800f76
by Brice.Goglin distances: allow to add distances with the heterogeneous kind It'll be recomputed internally, but users may try to pass it back when transforming distances and re-adding them. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/distances.c (diff) The file was modified
include/hwloc/distances.h (diff) The file was modified
utils/hwloc/hwloc-bind.1in (diff) The file was modified
utils/hwloc/hwloc-ps.1in (diff) The file was modified
utils/hwloc/hwloc-calc.1in (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff) The file was modified
contrib/completion/bash/hwloc (diff) The file was modified
utils/hwloc/hwloc-gather-topology.1in (diff)
Commit
64ced880959e43d546a387149057ce7e0b54a536
by Brice.Goglin hwloc-ps: --pid-cmd now supports "mpirank" and "env=<name>" "mpirank" will look for widespread MPI-rank-related envvars to show the MPI rank. "env=<name>" lets you customize the actual envvar to query in each process. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/common-ps.c (diff) The file was modified
NEWS (diff) The file was modified
utils/hwloc/hwloc-ps.1in (diff) The file was modified
utils/hwloc/hwloc-diff.c (diff) The file was modified
utils/hwloc/hwloc-dump-hwdata.c (diff) The file was modified
utils/hwloc/hwloc-gather-topology.in (diff) The file was modified
utils/hwloc/hwloc-calc.c (diff) The file was modified
utils/hwloc/hwloc-patch.c (diff) The file was modified
utils/hwloc/hwloc-annotate.c (diff) The file was modified
utils/hwloc/hwloc-distrib.c (diff) The file was modified
utils/hwloc/hwloc-compress-dir.in (diff) The file was modified
utils/hwloc/hwloc-gather-cpuid.c (diff) The file was modified
utils/hwloc/hwloc-bind.c (diff) The file was modified
contrib/completion/bash/hwloc (diff) The file was modified
utils/hwloc/hwloc-annotate.1in (diff) The file was modified
utils/hwloc/hwloc-annotate.c (diff) The file was modified
utils/hwloc/hwloc-ps.1in (diff) The file was modified
contrib/completion/bash/hwloc (diff) The file was modified
utils/hwloc/hwloc-ps.c (diff) The file was modified
utils/hwloc/hwloc-dump-hwdata.1in (diff) The file was modified
utils/hwloc/hwloc-dump-hwdata.c (diff) The file was modified
utils/hwloc/hwloc-gather-cpuid.1in (diff) The file was modified
utils/hwloc/hwloc-gather-cpuid.c (diff) The file was modified
contrib/completion/bash/hwloc (diff) The file was modified
utils/hwloc/hwloc-gather-topology.1in (diff) The file was modified
utils/hwloc/hwloc-gather-topology.in (diff) The file was modified
contrib/completion/bash/hwloc (diff) The file was modified
utils/hwloc/hwloc-compress-dir.in (diff) The file was modified
contrib/completion/bash/hwloc (diff) The file was modified
utils/hwloc/hwloc-compress-dir.1in (diff) The file was modified
utils/hwloc/hwloc-calc.c (diff) The file was modified
utils/hwloc/hwloc-diff.c (diff) The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
utils/hwloc/hwloc-calc.c (diff) The file was modified
utils/hwloc/hwloc-bind.c (diff) The file was modified
utils/hwloc/hwloc-ps.c (diff) The file was modified
utils/hwloc/hwloc-annotate.c (diff) The file was modified
utils/hwloc/hwloc-distrib.c (diff) The file was modified
utils/hwloc/hwloc-info.c (diff) The file was modified
utils/hwloc/hwloc-patch.c (diff) The file was modified
utils/hwloc/hwloc-compress-dir.in (diff) The file was modified
utils/hwloc/hwloc-dump-hwdata.c (diff) The file was modified
utils/hwloc/hwloc-gather-topology.in (diff)
Commit
0114c2b0b3e39265e0829eebfff87ac9f4412fe9
by Brice.Goglin linux: fix support for NUMA node0 being offline Just like we didn't support offline CPU#0 until commit 7bcc273efd50536961ba16d474efca4ae163229b, we need to support node0 being offline as well. It's not clear whether it's a new Linux feature or not, this was reported on a POWER LPAR VM. The symptoms are different here because we got no NUMA nodes at all, hence the core hwloc added a default machine-wide node. But this node got marked disallowed by Linux cgroups. Hence load() failed with "Topology does not contain any NUMA node, aborting!" We opportunistically assume node0 is online to avoid the overhead in the vast majority of cases. If node0 is missing, we parse "online" to find the first node. Thanks to Jirka Hladky for the report. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-linux.c (diff)
Commit
03bddd0037c62116d6171e4bf1fb72c48a57d825
by Brice.Goglin tests/linux: add a testcase with CPU#0 and NUMA node#0 offline Test for commits 7bcc273efd50536961ba16d474efca4ae163229b and 0114c2b0b3e39265e0829eebfff87ac9f4412fe9. Generated from a dual-12-core haswell platform, after marking cores 0-4,21-23 offline and NUMA node 0 offline, and updating sysfs cpu/cache/node bitmap accordingly. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was added tests/hwloc/linux/offline-cpu0-node0.output The file was modified
tests/hwloc/linux/Makefile.am (diff) The file was added tests/hwloc/linux/offline-cpu0-node0.tar.bz2
Commit
46b12176937a6532dfea40970db49334b7d8d9b9
by Brice.Goglin contrib/git-hooks: add a pre-commit hook checking options in manpage/completion This script is intended to run in pre-commit (see contrib/git-hooks/pre-commit for an example). It looks for command-line options in git-added usage lines in utils (C or shell) and verifies that they are also in the manpage and completions. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> Signed-off-by: Brice Goglin <Valentin.Hoyet@inria.fr> The file was added contrib/git-hooks/check-manpages-completions.pl The file was modified
contrib/git-hooks/pre-commit (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff)
Commit
aaf857630c0b8bab95252f0f4fdce1fea5ae996d
by Brice.Goglin doc: add support for building the HTML doxygen on readthedocs.io If branch/tag support appears to work well, this may replace the doc on www.open-mpi.org in the future. The readthedocs doxygen config is based on the default config but it disables non-HTML builds and it enables some things that weren't possible earlier (SHORT_NAMES required because tarballs don't like very long filenames, GENERATE_TREEVIEW didn't look good in the OMPI website layout). Signed-off-by: Brice Goglin <Valentin.Hoyet@inria.fr> The file was added doc/readthedocs/readthedocs.io.sh The file was added doc/readthedocs/conf.py The file was added doc/readthedocs/index.rst The file was modified
tests/hwloc/ports/include/levelzero/ze_api.h (diff) The file was modified
hwloc/topology-levelzero.c (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
contrib/git-hooks/check-manpages-completions.pl (diff) The file was modified
hwloc/hwloc2.dtd (diff) The file was modified
utils/hwloc/hwloc-annotate.1in (diff) The file was modified
utils/hwloc/hwloc-annotate.c (diff)
Commit
bb203b711405faa7273ad38dd7eaa078903db820
by Brice.Goglin distances: add MERGE_SWITCH_PORTS and TRANSITIVE_CLOSURE transformations Those are currently specific to NVIDIA GPUs connected through a NVSwitch. The NVLinkBandwidth exposes bandwidth from GPUs to 6 individual NVSwitch ports. MERGE_SWITCH_PORTS will combine all switch ports into a single object. We loose the PCI locality (one half of the switch is on each socket) but it doesn't matetr much. TRANSITIVE_CLOSURE gives inter-GPU bandwidth across the NVSwitch. The XGMIBandwidth is not affected for now because XGMI traffic cannot be routed through GPUs or switches yet. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
include/hwloc/distances.h (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
doc/Makefile.am (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/distances.c (diff) The file was modified
utils/hwloc/hwloc-annotate.c (diff) The file was modified
utils/hwloc/hwloc-annotate.1in (diff) The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
utils/lstopo/lstopo-text.c (diff) The file was modified
contrib/completion/bash/hwloc (diff) The file was modified
utils/lstopo/lstopo.h (diff) The file was modified
NEWS (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff)
Commit
12df5292d8dcce7510a56f6c550c7c7ee737f7c5
by Brice.Goglin tests/xml: add power8 distances between NVIDIA GPUs and Packages Restricted horizontally (only 2 PU per core, and 4 cores par packages) Restricted vertically (removed caches). Removed all I/Os except GPUs. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was added tests/hwloc/xml/power8gpudistances.xml The file was modified
tests/hwloc/xml/Makefile.am (diff) The file was modified
utils/hwloc/hwloc-annotate.c (diff) The file was modified
utils/hwloc/test-hwloc-annotate.sh.in (diff) The file was modified
utils/hwloc/Makefile.am (diff) The file was added utils/hwloc/test-hwloc-annotate.output2 The file was modified
NEWS (diff)
Commit
11df5ed4972714e8ef2aa5f03a1aec22b1ea4359
by Brice.Goglin levelzero: fix device name numbering variable k is used as iterator now for CQs, use dedicated variable for device numbering. (broken in 3a8c9fb5b1307ff1d30650f810797a391dfb6140) Signed-off-by: Peyton, Jonathan L <jonathan.l.peyton@intel.com> The file was modified
hwloc/topology-levelzero.c (diff) The file was modified
hwloc/cpukinds.c (diff)
Commit
cb0ed0353ae5c08f566c8acf7ec6c66d9455e7be
by Brice.Goglin cpukinds: remove arch-specific ranking strategy selection The default coretype+frequency now works on both ARM (only frequency) and Intel (either coretype or frequency if available). Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/cpukinds.c (diff) The file was modified
utils/lstopo/lstopo.desktop (diff) The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
utils/hwloc/hwloc-bind.c (diff)
Commit
3487fc68dd4003600dbf6d390956cb6bd47b1638
by Brice.Goglin levelzero: prefix headers include path with level_zero/ It looks like the recommended way to include L0 headers is "level_zero/ze_api.h" instead of "ze_api.h". It's not clear yet if level-zero.pc will remain and if it will point to level_zero/ze_api.h or ze_api.h. Hence we use libze_loader.pc instead because that one existed in older L0 releases and points to level_zero/ze_api.h, and it's unlikely to change. If level-zero.pc ever appears in a L0 release, we'll update the code to look at it first. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-levelzero.c (diff) The file was modified
config/hwloc.m4 (diff) The file was modified
tests/hwloc/ports/Makefile.am (diff) The file was added tests/hwloc/ports/include/levelzero/level_zero/zes_api.h The file was removed tests/hwloc/ports/include/levelzero/ze_api.h The file was removed tests/hwloc/ports/include/levelzero/zes_api.h The file was added tests/hwloc/ports/include/levelzero/level_zero/ze_api.h The file was modified
include/hwloc/levelzero.h (diff) The file was modified
tests/hwloc/levelzero.c (diff) The file was modified
utils/lstopo/lstopo-text.c (diff) The file was modified
utils/lstopo/lstopo-text.c (diff) The file was modified
utils/hwloc/hwloc-calc.c (diff)
Commit
97e37e5463bf7bb155e517d8f9d1137c2014e8f7
by Brice.Goglin utils/info: don't display unknown memattr initiator OS index we have the logical index before it, just remove the OS index instead of printing 4294967295. Refs #468 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/hwloc-info.c (diff)
Commit
da4e290b9147b9d1077fc8a3da184470becb18b3
by Brice.Goglin lstopo/memattrs: cleanup object printing * show -1 instead of 4294967295 when os_index is unknown * show the specific OSDev subtype in initiators * show the object name if any * factorize all this between target and initiator kinds Refs #468 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/lstopo/lstopo-text.c (diff) The file was modified
utils/hwloc/hwloc-calc.h (diff)
Commit
551c1bcc9c4b1ee48d4ae807c7ac8e13d6e3df8b
by bgoglin API: add HWLOC_TOPOLOGY_FLAG_RESTRICT_TO_{CPU,MEM}BINDING These flags apply a restrict during hwloc_topology_load() and directly ignore excluded PUs or NUMAs. The main goal is to avoid rebinding the x86 discovery to all PUs since it breaks the process affinity on Windows (see #366). The THISSYSTEM flag is required as well since binding is needed. It only matters for PU but people may want to restrict to NUMA for consistency, hence we provide both flags even if NUMA discovery doesn't break Windows binding (and CPU and memory binding are basically identical on Windows). The implementation could be improved to filter excluded PUs/NUMAs early in the backend, but that would require lots of changes and testing. For now, we only update the x86 backend to avoid breaking the binding, and the core filters things out at the end of the discovery. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology.c (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
utils/hwloc/misc.h (diff) The file was modified
hwloc/topology-x86.c (diff) The file was modified
NEWS (diff) The file was modified
doc/Makefile.am (diff) The file was modified
include/hwloc.h (diff)
Commit
5a7d127b222f52f8ed9475f8dc17ee9061eb1601
by bgoglin API: add HWLOC_TOPOLOGY_FLAG_DONT_CHANGE_BINDING This completely disables discovery steps that modify thread or process binding, i.e. the entire x86 backend for now. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
doc/Makefile.am (diff) The file was modified
hwloc/topology-x86.c (diff) The file was modified
include/hwloc.h (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
utils/hwloc/misc.h (diff) The file was modified
NEWS (diff) The file was modified
hwloc/topology.c (diff) The file was modified
contrib/windows/libhwloc.vcxproj.filters (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
include/hwloc/distances.h (diff) The file was modified
config/hwloc.m4 (diff) The file was modified
config/hwloc_internal.m4 (diff) The file was modified
config/hwloc_check_visibility.m4 (diff) The file was modified
config/hwloc.m4 (diff) The file was modified
configure.ac (diff) The file was modified
configure.ac (diff) The file was modified
config/hwloc.m4 (diff) The file was modified
config/hwloc_check_attributes.m4 (diff) The file was modified
config/hwloc.m4 (diff) The file was modified
configure.ac (diff) The file was modified
contrib/windows/hwloc_config.h (diff) The file was modified
contrib/android/AndroidApp/lstopo/build.gradle (diff) The file was modified
contrib/android/include/hwloc/autogen/config.h (diff) The file was modified
VERSION (diff) The file was modified
tests/hwloc/shmem.c (diff) The file was modified
tests/hwloc/hwloc_get_obj_with_same_locality.c (diff)
Commit
c51886bdf45b2a607e471c116484f8d5665ab507
by Brice.Goglin windows: warn if we find processor groups on a 32bits windows Processor groups are documented as not working there, some CPUs will be NUMA less, caches may intersect, etc. https://docs.microsoft.com/en-us/windows/win32/procthread/processor-groups "This functionality is not available on 32-bit versions of Windows." Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-windows.c (diff) The file was modified
include/hwloc.h (diff) The file was modified
hwloc/topology-rsmi.c (diff) The file was modified
NEWS (diff) The file was modified
utils/lstopo/lstopo-tikz.c (diff) The file was modified
utils/lstopo/lstopo.h (diff) The file was modified
utils/lstopo/lstopo-svg.c (diff) The file was modified
utils/lstopo/lstopo-cairo.c (diff) The file was modified
utils/lstopo/lstopo-fig.c (diff) The file was modified
utils/lstopo/lstopo-android.c (diff) The file was modified
utils/lstopo/lstopo-windows.c (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
include/hwloc/cpukinds.h (diff) The file was modified
utils/hwloc/hwloc-calc.1in (diff) The file was modified
include/hwloc/cpukinds.h (diff)
Commit
1def2f1ac0ea12210b1324edffdba0ff31001084
by bgoglin darwin: add cpukinds support using IOKit cpus information We read the 'cluster-type' ('E' for energy and 'P' for performance) and the 'compatible' string (either "apple,icestorm;ARM,v8" or "apple,firestorm;ARM,v8" on M1 processor) to build two cpukinds. Thanks to Michael Hirsch and Francois Ozog for the help. Closes: #454 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/Makefile.am (diff) The file was modified
config/hwloc.m4 (diff) The file was added tests/hwloc/ports/include/darwin/CoreFoundation/CoreFoundation.h The file was modified
NEWS (diff) The file was modified
include/hwloc/cpukinds.h (diff) The file was modified
hwloc/topology-darwin.c (diff) The file was added tests/hwloc/ports/include/darwin/IOKit/IOKitLib.h The file was modified
doc/hwloc.doxy (diff) The file was modified
tests/hwloc/ports/Makefile.am (diff) The file was modified
hwloc/topology-linux.c (diff)
Commit
82ff12253cd29ae8324d9fb07e5bd064bd42cdab
by Brice.Goglin linux: use the optional "cpu_capacity" sysfs cpu file to rank cpukinds This file is used (at least on ARM) platforms to rank core performance in the kernel scheduler. It's not strictly guaranteed to be rank perfectly (because throttling, etc that may slow down some cores) but it should be fine in the vast majority of cases. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
include/hwloc/cpukinds.h (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
NEWS (diff) The file was modified
doc/hwloc.doxy (diff)
Commit
6ed7e9def5ac70186a0932a315a8092dda7f03b4
by Brice.Goglin linux/cpukinds: remove sysfs cpu type support The kernel will rather use the "cpu_capacity" file, cpu types support has never been merged. This basically reverts 04a4977a2e076883fc5a94edbdb571bea2bc824e Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
tests/hwloc/linux/fakeheterocpunuma.output (diff) The file was modified
utils/hwloc/test-hwloc-info.output (diff)
Commit
e688667906516b77ef3e675b89596cc2df6675c2
by Brice.Goglin tests/linux: update fakeheterocpunuma to use cpu_capacity instead of cpu_types This reverts ee7b0dc8b5c736a607858da9ef2d369cfeb23b91 and adds capacities. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/linux/fakeheterocpunuma.output (diff) The file was modified
utils/hwloc/test-hwloc-info.output (diff) The file was modified
tests/hwloc/linux/fakeheterocpunuma.tar.bz2 (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
tests/hwloc/linux/40intel64-2g2n4c+pci.env (diff)
Commit
330a846de4f96e14583565dbf3e4cd5bc4dc3636
by Brice.Goglin tests/gather: ignore missing DAXDevice NUMA info attr Those attributes require dax sysfs files, but we don't gather those I/O files in this test. (enabling I/O in this test would likely raise many corner cases that aren't worth fixing) Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/linux/gather/test-gather-topology.sh.in (diff)
Commit
01e54c4e2302d4776f6cbfa7863b7c772fdd7c77
by Brice.Goglin lstopo/x11: fix autoresizing when window is maximized If the window is maximized, pressing + or - to change the scale raises a ConfigureNotify event (as if the window had been resized) that causes the scaling to go back to fit the window. Detect whether the window is maximized during each ConfigureNotify and don't ask for a scale update when we're maximized. If the window is maximized in a single dimension, keep the current behavior because zoom in/out will unmaximize it. Thanks to Clement Foyer for the report. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/lstopo/lstopo-cairo.c (diff) The file was modified
utils/hwloc/hwloc-calc.1in (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
utils/hwloc/hwloc-calc.1in (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff) The file was modified
hwloc/topology-levelzero.c (diff) The file was modified
hwloc/dolib.c (diff) The file was modified
hwloc/pci-common.c (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff)
Commit
de93efd669eb6f77b671f3d67d37b6b339529cc4
by Brice.Goglin configure: look for cuda in /usr/local/cuda by default And add some configure notice depending on whether we're using a cuda-$version.pc, or cuda custom path, or this default. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
config/hwloc.m4 (diff) The file was modified
config/hwloc.m4 (diff)
Commit
46f16178e1d22afc5706c9360e9c7c9e350d2849
by Brice.Goglin configure: cleanup pthread_mutex_lock checks Move it above together with non-backend-specific things, and clear the cache instead of working around it. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
config/hwloc.m4 (diff) The file was modified
config/hwloc.m4 (diff)
Commit
e50c371a56780b35cb99846a93e54b9beff68df7
by Brice.Goglin configure/rsmi: allow to specify the ROCm install path Use --with-rocm=<dir> first. Or use /opt/rocm-<version> if version specified in --with-rocm-version=<version> or ROCM_VERSION envvar. Or use /opt/rocm if it exists. Or assume ROCm is installed in standard directories. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
config/hwloc_internal.m4 (diff) The file was modified
NEWS (diff) The file was modified
config/hwloc.m4 (diff)
Commit
dafbfac07744d12f4e0325cf968bbcdcb49e1f1e
by Brice.Goglin utils/gather-cpuid: bring back pconfig It's not clear what happened to pconfig in the x86 manual. The doc was clear when first added. Now most of it moved to a specific section without keeping all details about the CPUID leaf. One may read this as if the subleaf value is the entire EAX but everybody, including Linux, still uses bits 0:11 as in the old documentation. So just revert 8e42ed34984ea36c2f6cd1dacd891e2865bdfaa8 and assume the (new) doc needs to be clarified. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/hwloc-gather-cpuid.c (diff) The file was modified
contrib/ci.inria.fr/job-3-android.sh (diff) The file was modified
contrib/android/AndroidApp/gradlew (diff)
Commit
3919415b85f06357e12061e4467314e47e7fa886
by Brice.Goglin contrib: cleanup/factorize scripts fot checking windows/android versions contrib/git-hooks/check-versions.sh tested both android and windows. contrib/windows/test-windows-version.sh.in only tested windows but it had verbose messages, and it was only called in ci.inria.fr/job-0-tarball.sh. Have a single windows check script under contrib/windows and an Android one under contrib/android. Both are called by the git hooks with --quiet, and by ci.inria.fr/job-0-tarball.sh without --quiet. No need to make them "make check" compatible. No need to generate anything from configure. They have a parameter to run outside of the git root directory but it's actually not required here. contrib/android/check-versions.sh isn't distributed since contrib/android isn't. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
contrib/git-hooks/check-versions.sh (diff) The file was modified
contrib/ci.inria.fr/job-0-tarball.sh (diff) The file was modified
Makefile.am (diff) The file was removed contrib/windows/test-windows-version.sh.in The file was added contrib/android/check-versions.sh The file was modified
config/hwloc_internal.m4 (diff) The file was modified
.gitignore (diff) The file was modified
contrib/windows/Makefile.am (diff) The file was added contrib/windows/check-versions.sh
Commit
47c745526cc6f048266a65ae1bd31e4de36cd055
by Brice.Goglin android: don't ignore the greek Android releases are often based on hwloc snapshots, make greek such as "a1" appear in the android version and add an optional snapshot number. The greek was already in the release posted last friday (bypassing contrib/android/check-versions.sh) but not the snapshot ("2" means ready for next version). Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
contrib/android/AndroidApp/lstopo/build.gradle (diff) The file was modified
contrib/android/check-versions.sh (diff) The file was modified
contrib/android/include/hwloc/autogen/config.h (diff)
Commit
6abf03dd20caa2e3d0231ac0fcebf43d6acdd294
by Brice.Goglin linux: use MPOL_PREFERRED_MANY when available MPOL_PREFERRED is less strict than MPOL_BIND since it falls back to other node if the given one is full. But it works only with a single node. Other (non-first physical index) are just ignored. MPOL_PREFERRED_MANY in 5.15 fixes this. Use MPOL_PREFERRED_MANY by default for membind. If it fails, try MPOL_PREFERRED. If it works, assume MPOL_PREFERRED_MANY isn't supported in the current kernel. Closes #236 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-linux.c (diff) The file was modified
NEWS (diff)
Commit
bbbb2671208974719162ab8b182fb3c70827e8fb
by bgoglin linux: expose cache os_index when available x86 and arm (with PPTT ACPI table) have a "hardware" cache ID that is exposed in sysfs. Expose that ID as the os_index. It is useful for "resctrl" since you need to know which cache to configure (for partitioning, monitoring, etc). From the resctrl documentation (section 20.3.3) at https://www.kernel.org/doc/html/latest/x86/resctrl.html "[...] instead of using “socket” or “core” to define the set of logical cpus sharing a resource we use a “Cache ID”. At a given cache level this will be a unique number across the whole system (but it isn’t guaranteed to be a contiguous sequence, there may be gaps). To find the ID for each logical CPU look in /sys/devices/system/cpu/cpu*/cache/index*/id" These IDs appear in the "schemata" file used for resctrl config in sysfs: $ cat /sys/fs/resctrl/schemata L3:0=ffff;1=ffff;2=ffff;3=ffff;4=ffff;5=ffff;6=ffff;7=ffff;16=ffff;17=ffff;18=ffff;19=ffff;20=ffff;21=ffff;22=ffff;23=ffff ^ ^ ^ here and here etc Thanks to Clement Foyer for the suggestion. Closes #343 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/linux/offline-cpu0-node0.output (diff) The file was modified
NEWS (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
tests/hwloc/x86+linux/5intel64-hybrid-lakefield.output (diff) The file was modified
tests/hwloc/linux/32intel64-2p8co2t+8ve.output (diff) The file was modified
tests/hwloc/linux/fakecpuid1f-64intel64-2p4d2n2c2t.output (diff)
Commit
78216defd38b7e5e0c348fcd12277a018d6d77fd
by bgoglin x86: add a comment to remember why we don't expose cache os_index Short story: It's hard to make sure x86 cache IDs are compatible with resctrl on Linux. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-x86.c (diff) The file was modified
include/hwloc.h (diff) The file was modified
include/hwloc/linux.h (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff) The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
utils/lstopo/lstopo.h (diff) The file was modified
utils/lstopo/lstopo-ascii.c (diff)
Commit
42929f8a304f32f11fc3d4795015a0a0610b1dc1
by Brice.Goglin lstopo/draw: add flag for backends returning approximative textwidth The core will change some layout if it requires precise text width for scaling/placing things. tikz, (native)svg, fig and android get this flag. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/lstopo/lstopo-tikz.c (diff) The file was modified
utils/lstopo/lstopo-svg.c (diff) The file was modified
utils/lstopo/lstopo-fig.c (diff) The file was modified
utils/lstopo/lstopo.h (diff) The file was modified
utils/lstopo/lstopo-android.c (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff) The file was modified
utils/lstopo/lstopo.h (diff) The file was modified
utils/lstopo/lstopo.h (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff) The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
utils/lstopo/lstopo.h (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff) The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff)
Commit
3b4343733486943809f202d220307e5dde594b9e
by Brice.Goglin lstopo/draw: place I/O and Misc separately from CPU by default The placement algorithm doesn't work well with objects of very different sizes (e.g. CPU cores are small and all the same, PCI is often big, Misc is very small). I/O and Misc may now be placed on the right (by default, or below) CPU children without considering all them at the same time. By default --children-order is now memory:above,io:right,misc:right. The old behavior is memory:above,plain. io:below and misc:below are also accepted. plain puts everything not specified with CPU children. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/lstopo/lstopo.h (diff) The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
NEWS (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff)
Commit
9088f2714e2789d2a6233c6647f76181e1c16ea9
by Brice.Goglin lstopo/draw: allow to specify horiz/vert/rect layout for right/below io/misc io:right:vert means I/O children are placed vertically on the right of CPUs. misc:below:horiz means Misc children are placed horizontally below CPUs. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff) The file was modified
utils/lstopo/lstopo.h (diff) The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff) The file was modified
doc/hwloc.doxy (diff)
Commit
520e59b2d03127e852966b9467e220bb90684b48
by Brice.Goglin rsmi: add a "XGMIHops" matrix RSMI reports the same bandwidth between directly-connected GPUs and XGMI-routed ones (hardware routing). I don't know yet if we should remove non-direct BW and force users to apply the transitive closure (like they do for NVSwitch). Add this hops matrix in the meantime. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/topology-rsmi.c (diff) The file was modified
include/hwloc/distances.h (diff) The file was modified
hwloc/topology-rsmi.c (diff)
Commit
fdcba6a6b34e13366bcc73f0096206006c30927d
by Brice.Goglin configure: use ROCm config for detecting OpenCL AMD's OpenCL is installed the ROCm directory, use the ROCm detection result to improve the automatic OpenCL detection. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
config/hwloc.m4 (diff)
Commit
8ff9e5e50960f2876f958c41873ede6656923220
by Brice.Goglin hwloc-calc.1: add an example for resctrl bitmap strings The "resctrl" Linux subsystem is one of the (rare?) case where users have to give bitmask to sysfs files. hwloc doesn't have a way to generate those bitmask. They must be split in non-empty 32bits pieces separated by commas, without leading 0x. Provide a sed command for conversion in hwloc-calc.1 examples. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/hwloc-calc.1in (diff) The file was modified
NEWS (diff)
Commit
1bd7c0d9dee476c1ba3ed5736171a277c10706b2
by Brice.Goglin linux: add "bxi0" OFED OS devices for Atos/Bull BXI HCAs It's not OFED, but it's close enough not to deserve a new MISC osdev subtype. We only have a BXIUUID info attribute for now. The PCI device is ignored by default because its class is 0000 (unclassified), whitelisting it isn't very convenient yet. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
NEWS (diff) The file was modified
include/hwloc.h (diff) The file was modified
utils/hwloc/hwloc-gather-topology.in (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/topology-linux.c (diff)
Commit
12342fd4191641efdc8c492f095613a9b7b9ee5c
by Brice.Goglin core: fix a corner case of reconnecting levels during the merging of identical level Merging "keep structure" requires normal levels to be reconnected. Special levels aren't needed, but they may need reconnecting after merging (if some parent/child special children list are merged). Instead of reconnecting everything before hwloc_filter_levels_keep_structure() and adding a reconnect of special levels after, integrate the required reconnect calls inside hwloc_filter_levels_keep_structure(). This fixes a corner case that should unlikely occur in practice. If we ever attach Memory (or IO or Misc) to 2 different objects that are merged by "keep structure" (eg Package and Group), the special level might need reordering after merging. These merged objects have same sets, hence special children should have been attached to the same one from the beginning anyway. Except if sets are different because one was inserted by allowed_cpuset and the other by normal cpuset. Or if sysfs files were tweaked and not properly restricted to the final remaining topology sets. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology.c (diff) The file was modified
hwloc/topology.c (diff)
Commit
1b9a5431f80be1776277326de9c1a4f16463dc2a
by Brice.Goglin tests/linux: update outputs We ignore indentation differences to avoid multiple commits in multiple files for useless changes. Let's update all files once because distances output changed a while ago. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/linux/2amd64-2n.output (diff) The file was modified
tests/hwloc/linux/nvidiagpunumanodes.output (diff) The file was modified
tests/hwloc/linux/40intel64-4n10c+pci-conflicts.output (diff) The file was modified
tests/hwloc/linux/256ppc-8n8s4t.output (diff) The file was modified
tests/hwloc/linux/256ia64-64n2s2c.output (diff) The file was modified
tests/hwloc/linux/48amd64-4d2n6c-sparse.output (diff) The file was modified
tests/hwloc/linux/16amd64-8n2c-cpusets.output (diff) The file was modified
tests/hwloc/linux/64amd64-4s2n4ca2co.output (diff) The file was modified
tests/hwloc/linux/8ia64-2n2s2c.output (diff) The file was modified
tests/hwloc/linux/128ia64-17n4s2c.output (diff) The file was modified
tests/hwloc/linux/16amd64-8n2c-cpusets_noadmin.output (diff) The file was modified
tests/hwloc/linux/16amd64-4n4c-cgroup-distance-merge.output (diff) The file was modified
tests/hwloc/linux/16ia64-8n2s.output (diff) The file was modified
tests/hwloc/linux/96em64t-4n4d3ca2co.output (diff) The file was modified
tests/hwloc/linux/fakememinitiators-1np2c+1npp+gi.output (diff) The file was modified
tests/hwloc/linux/fakecpuid1f-64intel64-2p4d2n2c2t.output (diff) The file was modified
tests/hwloc/linux/8amd64-4n2c.output (diff) The file was modified
tests/hwloc/linux/32amd64-4s2n4c-cgroup2.output (diff) The file was modified
tests/hwloc/linux/16amd64-8n2c.output (diff) The file was modified
tests/hwloc/linux/nvidiagpunumanodes.kept.output (diff)
Commit
ad9019216bb84b4e50698381fc67420422d9ba5e
by Brice.Goglin tests/linux: cleanup fake PCI sysfs files for nvidia GPUs with NUMA nodes Specify the 0300 class instead of keeping it uninitialized (0000) which means unclassified like Atos/Bull BXI HCAs. And fix local_cpus to match the PUs that we kept in syfs. Thus also fix cpumap in real NUMA nodes to avoid having NUMA attached to different groups and being reordered once groups are merged (see commit 12342fd4191641efdc8c492f095613a9b7b9ee5c) Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/linux/nvidiagpunumanodes.kept.output (diff) The file was modified
tests/hwloc/linux/nvidiagpunumanodes.output (diff) The file was modified
tests/hwloc/linux/nvidiagpunumanodes.tar.bz2 (diff)
Commit
1a68aec09eae0f6f137b4bb4ea3ae271286bf71a
by Brice.Goglin pci: consider unclassified devices as important Required so that Atos/Bull BXI HCAs aren't ignored by default. Hopefully this change won't cause many new PCI devices to appear in the default hwloc PCI hierarchy because most devices have a valid class. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
include/hwloc/plugins.h (diff) The file was modified
utils/hwloc/hwloc-gather-cpuid.c (diff)
Commit
7e35a3de81b2b6951c08b9d4f8c36e76522e60d7
by Brice.Goglin x86: AMD Family 19h (Zen3) only reports 1 node per proc Update our warning accordingly. It's not clear whether only early Fam 17h models could report 2 or 4 as well. They might have disabled it because it didn't match the BIOS NPS configuration. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-x86.c (diff) The file was modified
hwloc/topology-x86.c (diff)
Commit
7a2808e69032f48f7cf886636dde20e26b962348
by Brice.Goglin x86: update the AMD L3 APIC ID sharing quirk for Family 19h (Zen3) CPU models with less cores still have missing APIC IDs part of each L3, which means we cannot divide by the real number of cores to find the cache ID. We previously hardwired the number of cores from Zen1/2 (up to 4 HT cores per L3), let's do something clean that works for Zen3 (up to 8). And it's actually similar to what the kernel does in cacheinfo_amd_init_llc_id(). The Hygon quirk might need the same cleanup if they release a Zen3 clone in the future. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-x86.c (diff) The file was modified
utils/hwloc/hwloc-annotate.c (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
utils/hwloc/hwloc-annotate.c (diff) The file was modified
utils/hwloc/hwloc-annotate.1in (diff) The file was modified
NEWS (diff) The file was modified
utils/hwloc/hwloc-annotate.c (diff) The file was modified
NEWS (diff) The file was modified
utils/hwloc/hwloc-annotate.1in (diff)
Commit
0a233a5872242da1a049070ef72373e45830a8e4
by Brice.Goglin utils: add a test of building a custom heterogeneous topology First package has 4 SMT "Big" cores with 200 GB of DRAM. Second package has 8 "Little" cores (no HT) with 100GB of DRAM and 10GB of HBM. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
config/hwloc_internal.m4 (diff) The file was modified
utils/hwloc/Makefile.am (diff) The file was added utils/hwloc/test-build-custom-topology.sh.in The file was added utils/hwloc/test-build-custom-topology.output The file was modified
doc/hwloc.doxy (diff) The file was modified
NEWS (diff) The file was modified
utils/hwloc/test-build-custom-topology.output (diff) The file was modified
utils/hwloc/test-build-custom-topology.sh.in (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/topology-rsmi.c (diff) The file was modified
NEWS (diff) The file was modified
tests/hwloc/ports/include/rsmi/rocm_smi/rocm_smi.h (diff) The file was modified
NEWS (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
doc/hwloc.doxy (diff)
Commit
1d9b10b60bda998a99e718ade1856b75206bd3f5
by Brice.Goglin doxy+test-build-custom-topology: use the modern annotate command-line layout Put location between -- so that we can apply operations to multiple locations. We could even remove "none" locations when unneeded but that's not clearly documented yet. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
doc/hwloc.doxy (diff) The file was modified
utils/hwloc/test-build-custom-topology.sh.in (diff) The file was modified
contrib/windows/hwloc_config.h (diff) The file was modified
contrib/android/include/hwloc/autogen/config.h (diff) The file was modified
VERSION (diff) The file was modified
contrib/android/AndroidApp/lstopo/build.gradle (diff) The file was modified
doc/hwloc.doxy (diff)
Commit
887b930cf27573c08c84a9d5f493dc21c97b480e
by Brice.Goglin dolib: don't use _snprintf when snprintf is available For some reason, loading MSVC vars on the cygwin ci.inria.fr slave (required for CMake to find MSVC stuff) doesn't like _snprintf() anymore. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/Makefile.am (diff) The file was modified
hwloc/dolib.c (diff)
Commit
9a03e30be1af9eee00055173cd36b8bb8087e45b
by Brice.Goglin core: fix a #if vs #ifdef for constructor attribute support Unnoticed until now because these lines were added recently, way after all compilers added support for this attribute... until somebody tried autoconf with MSVC. Thanks to Alexander Neumann for the report. Closes #485 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology.c (diff) The file was modified
hwloc/Makefile.am (diff)
Commit
9bafca186b2ebdbddba9d565db3c7a9359589c41
by Brice.Goglin private/misc.h: use HWLOC_HAVE_ATTRIBUTE_UNUSED instead of __HWLOC_HAVE_ATTRIBUTE_UNUSED The latter is defined in public headers depending on whether private headers are available or not. The formet is the official internal thing. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
include/private/misc.h (diff) The file was modified
hwloc/memattrs.c (diff) The file was modified
hwloc/cpukinds.c (diff)
Commit
f9f10c7aad05f66601026a2056fbc58126a171e1
by Brice.Goglin contrib: add CMakeLists.txt under contrib/windows-cmake Inspired from https://github.com/microsoft/vcpkg/tree/master/ports/hwloc (MIT license, OK for BSD-3). See the README for details. Closes #88 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
doc/hwloc.doxy (diff) The file was modified
NEWS (diff) The file was added contrib/windows-cmake/CMakeLists.txt The file was modified
Makefile.am (diff) The file was added contrib/windows-cmake/README The file was modified
contrib/git-hooks/check-versions.sh (diff) The file was added contrib/windows-cmake/check-versions.sh The file was modified
contrib/ci.inria.fr/Jenkinsfile (diff) The file was added contrib/ci.inria.fr/job-1-wincmake.bat The file was modified
doc/hwloc.doxy (diff) The file was modified
.gitignore (diff) The file was modified
configure.ac (diff)
Commit
cc8ff199ed771f4fb6d363ee33ce227db12fd51d
by bgoglin build/windows: directly call MSLIB instead of going through our dolib dolib was needed in old msys environment where mixing Unix and windows path didn't work well. We now use -option instead of /option for MSLIB options to avoid MSYS2 path rewriting: Passing /def:.libs/libhwloc.def /out:.libs/libhwloc.lib would fail with LINK : fatal error LNK1104: cannot open file 'C:\msys64\def;.libs\libhwloc.lib' Sidenote: MSYS2 path rewriting can also be avoided by setting the env var MSYS2_ARG_CONV_EXCL="/machine;/def;/out;/name" Side-sidenote: MSYS_NO_PATHCONV=1 is often cited as an option but it's Git-bash specific and we don't want to disable path rewriting entirely. Thanks to Alexander Neumann for the help. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/Makefile.am (diff) The file was modified
configure.ac (diff) The file was modified
contrib/ci.inria.fr/job-3-mingw.bat (diff) The file was modified
contrib/ci.inria.fr/job-3-mingw.sh (diff)
Commit
ccc25d97569469b0b96f16811e5eab20ab2c1669
by bgoglin build: remove dolib.c By the way, dolib.c should never have been built in the MSVC solution (it wasn't meant to be included in libhwloc). Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/Makefile.am (diff) The file was removed hwloc/dolib.c The file was modified
contrib/windows/libhwloc.vcxproj (diff) The file was modified
contrib/windows/libhwloc.vcxproj.filters (diff) The file was modified
HACKING (diff) The file was modified
configure.ac (diff) The file was modified
contrib/android/AndroidApp/lstopo/build.gradle (diff) The file was modified
contrib/android/AndroidApp/lstopo/src/main/cpp/CMakeLists.txt (diff)
Commit
7cf69ea1969c9ea3ce6e88eefa8ba367863ce03e
by Brice.Goglin contrib/android: cleanup version code management Previous official android released had version codes ABI*100xx+151 xx was bumped at each release, was 14 in 2.6.0rc1-1-1.5.1 151 was somehow related to the core android hwloc code, currently 1.5.1 This gave 80263 for the current max ABI(8). Starting with next release, we will do ABI*10000 + 264 where 264 increases at the release gives x0264 for each ABI, which is still higher than previous release valued. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
contrib/android/AndroidApp/lstopo/build.gradle (diff) The file was modified
contrib/ci.inria.fr/job-0-tarball.sh (diff) The file was modified
contrib/android/check-versions.sh (diff) The file was modified
contrib/windows/check-versions.sh (diff) The file was modified
contrib/windows/hwloc_config.h (diff) The file was modified
contrib/windows-cmake/check-versions.sh (diff) The file was added contrib/update-versions.sh
Commit
62a924ea0d4d1eaf7bc7d780169c6fc389b22a3f
by Brice.Goglin contrib/android: update/fix gradle-wrapper As explained by the fdroid bot in https://gitlab.com/fdroid/fdroiddata/-/merge_requests/9983 we declared a 4.10.1 but actually has 4.6. Also we were missing its sha256sum Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
contrib/android/AndroidApp/gradle/wrapper/gradle-wrapper.jar (diff) The file was modified
contrib/android/AndroidApp/gradle/wrapper/gradle-wrapper.properties (diff) The file was modified
utils/hwloc/misc.h (diff) The file was modified
utils/lstopo/lstopo-tikz.c (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff) The file was modified
hwloc/topology-xml.c (diff)
Commit
8b06b7f1a486b985c4aeb9b304cebfad2bf60464
by Brice.Goglin cpukinds: fix the sorting of cpukinds by ranking_value Returning the difference of uint64_t from the qsort compare function is wrong, and it actually ignores the high bits here. Compare those uint64_t values properly. Thanks to the MSVC warning, that one was useful. Fortunately, those high bits aren't used in current ranking heuristics (frequencies are in MHz, Linux capacity is currently ~100 or ~1000, x86 coretype is 0x20 or 0x40 <<20, Windows and MacOS forced efficiencies are 0 and 1). Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/cpukinds.c (diff)
Commit
2770bfe27d7ac920c98f14ce3ae41ff8e5cbfe7c
by Brice.Goglin xml/export/v1: fix distances export vs v1 memory group insert When exporting some memory children to v1, we have to insert intermediate Groups in some cases. Those groups did not have their parent pointer initialized because it wasn't needed... except when deciding whether a distance matrix should be exported or not: only machine-wide matrices are exported (matrices attached to an object whose parent is NULL). This caused distance matrices to be exported multiple times (in the root and in some subgroups) and warnings on import (duplicates were ignored). Thanks to Ralph Castain for the report. Ref #494. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-xml.c (diff)
Commit
bf1f6bb2db8891e3340ebb5e7246bdbd7fac05da
by Brice.Goglin API: clarify the documentation of hwloc_alloc_membind_policy() And remove an obsolete reference to a _nodeset() variant. Thanks to Philippe Swartvagher. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
include/hwloc.h (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
doc/hwloc.doxy (diff)
Commit
492396e381faf48d302a78b9eb1544422fa85ae6
by Brice.Goglin doxygen: remove some obsolete config options to fix warnings TCL_SUBST and MSCGEN_PATH were empty. COLS_IN_ALPHA_INDEX and PERL_PATH were set to their default value. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
doc/doxygen.cfg (diff)
Commit
a53e2f578dc0f6672dbf8082d8e147062ff09d14
by Brice.Goglin doxygen: a4wide support has been dropped PAPER_TYPE=a4wide was removed in doxygen 1.7.2 (in 2010) because a4wide was being removed from texlive. doxygen has been using a4 since then. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
doc/doxygen.cfg (diff) The file was modified
doc/Makefile.am (diff)
Commit
0595531fdc8659f0bc4f5440a21dc52642d5b073
by Brice.Goglin linux: don't look at packages if we're not the first PU of the core "not first PU of core" always means "not first PU of die" on Linux. Before Die support was added, this worked fine. Now it works fine if the kernel exposes Dies. But missing Die info in old kernels meant a package is parsed once per core, because no PU was marked as first of a die. Fix this. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-linux.c (diff)
Commit
01969735dee4998f3f9d05207dafa45fa66c4c37
by Brice.Goglin linux: add "Cluster" groups for Linux 5.16+ cluster topology Expose "Cluster" group objects on Linux kernel 5.16+ for CPUs that share some internal cache or bus. This can be equivalent to the L2 Cache level on some platforms (e.g. x86) or a specific level between L2 and L3 on others (e.g. ARM Kungpeng 920). Based on a patch by Jonathan Cameron <Jonathan.Cameron@huawei.com> These allow us to represent groups of cores where they are grouped because of shared elements other than caches. Typical example would be a heirarchical coherency fabric with a cluster sharing a single bridge to a wider coherency network on chip. In ACPI this is represented in PPTT as a processor hierarchy node structure above the level representing a CPU (possibly with multiple threads) and below the hierarchy node structure that represents either a die, or the level at which the last level cache is shared. An example of this is the Kunpeng920 SoC which has 4 CPU cores per cluster and up to 8 of those clusters share an L3 cache. Similar concepts exist in device tree. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-linux.c (diff) The file was modified
include/private/private.h (diff) The file was modified
NEWS (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
NEWS (diff) The file was added tests/hwloc/linux/128arm-2pa2n8cluster4co.output The file was added tests/hwloc/linux/128arm-2pa2n8cluster4co.tar.bz2 The file was modified
tests/hwloc/linux/Makefile.am (diff) The file was modified
hwloc/topology-windows.c (diff)
Commit
2213d4f7ddebe051d8229c1553a660e0b6aaf56d
by Brice.Goglin windows: move some common definitions to private/windows.h Those are useful in both topology-windows.c and lstopo-windows.c when running make check under ports, but also soon for native builds when we'll always redefine windows structures/unions unconditionally. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-windows.c (diff) The file was modified
include/Makefile.am (diff) The file was modified
utils/lstopo/lstopo-windows.c (diff) The file was added include/private/windows.h The file was modified
tests/hwloc/ports/include/windows/windows.h (diff)
Commit
db730cf4e4304b4f21df8e71d914cdb298997fb8
by Brice.Goglin windows: cleanup the management of windows query structures PROCESSOR_RELATIONSHIP changed between Windows versions. Some old MSVC/Windows version may miss the EfficiencyClass field that we want to always use at compile time. That's why we used a hack to access that field without naming it. Cleanup this by always using our redefinition of PROCESSOR_RELATIONSHIP that contains that field instead of using the Windows/MSVC-provided versions. By the way, the field is guaranteed to be 0 on old Windows versions at runtime. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-windows.c (diff)
Commit
0185d003629f21511b335c0193ba1e092593abb0
by Brice.Goglin windows: support multiple processor Groups per NUMA The previous Windows API could not describe NUMA nodes with PUs from more than a single Processor Group (64 PUs max). Starting with Windows 10 build 20348 and Windows 11, the NUMA affinity can be described in more than one group affinity structure. Refs #465 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-windows.c (diff) The file was modified
NEWS (diff)
Commit
5858e76744e4a97fe406f76d2bf81b8945bb29f5
by Brice.Goglin windows: support multiple processor Groups per cache The previous Windows API could not describe caches with PUs from more than a single Processor Group (64 PUs max). This was extended for NUMA nodes in Windows 10 build 20348 and Windows 11 (see previous commit), and approximately at the same time for caches. Refs #465 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-windows.c (diff) The file was modified
NEWS (diff) The file was modified
contrib/windows-cmake/CMakeLists.txt (diff) The file was added contrib/windows-cmake/private_config.h.in The file was modified
contrib/windows-cmake/CMakeLists.txt (diff) The file was modified
contrib/windows-cmake/CMakeLists.txt (diff) The file was modified
contrib/windows-cmake/CMakeLists.txt (diff)
Commit
a5809731f6411a8eded8e1b00b74956cd8dd9da9
by scivision windows-cmake: hwloc object for deduplicated internal logic Naming targets with a "lib" prefix is unconventional for CMake and causes unexpected file naming. Instead, we set CMAKE_*_LIBRARY_PREFIX to name libhwloc.* consistently across platforms Signed-off-by: Michael Hirsch <michael@scivision.dev> The file was modified
contrib/windows-cmake/CMakeLists.txt (diff) The file was modified
tests/hwloc/Makefile.am (diff) The file was modified
contrib/windows-cmake/CMakeLists.txt (diff) The file was added tests/hwloc/CMakeLists.txt The file was modified
contrib/ci.inria.fr/job-1-wincmake.bat (diff) The file was modified
contrib/windows-cmake/CMakeLists.txt (diff) The file was modified
contrib/windows-cmake/CMakeLists.txt (diff) The file was modified
contrib/windows-cmake/private_config.h.in (diff) The file was modified
contrib/windows-cmake/CMakeLists.txt (diff) The file was added contrib/windows-cmake/static-components.h.in The file was modified
contrib/windows-cmake/CMakeLists.txt (diff) The file was modified
contrib/windows-cmake/private_config.h.in (diff) The file was modified
contrib/windows-cmake/static-components.h.in (diff)
Commit
11c07152fc7508b3d71b7e02621dff51f873f071
by scivision windows-cmake: do not force lib filename prefix on MSVC Signed-off-by: Michael Hirsch <michael@scivision.dev> Co-authored-by: Alexander Neumann <30894796+Neumann-A@users.noreply.github.com> The file was modified
contrib/windows-cmake/CMakeLists.txt (diff) The file was modified
contrib/windows-cmake/CMakeLists.txt (diff) The file was modified
contrib/windows-cmake/CMakeLists.txt (diff) The file was modified
doc/hwloc.doxy (diff)
Commit
ffdcaa9e555a7aeda0e76d9cecf6fe4b6c94b6a6
by Brice.Goglin hwloc-calc.1: improve the resctrl bitmask conversion Commit 8ff9e5e50960f2876f958c41873ede6656923220 added a regexp conversion that doesn't handle consecutive empty masks ",,,". Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/hwloc-calc.1in (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff)
Commit
67b08aacd7f60f8f2b28f80f141a82d12af7f21e
by Brice.Goglin lstopo: properly skip --top-color argument after processing We were getting "Cannot infer output type for file `none' without any extension, using default output." which means exporting to a file would fail. Hopefully we rarely export --top to a file. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff)
Commit
38cbbe655a9bad7c390207e325b721b664e029ed
by Brice.Goglin lstopo/draw: cleanup line color management Line color (for bridges) was always set to black in callers. Also it was not correctly implemented in the xfig backend (it was changing the fill color instead of the pen color). Box border color was hardwired to black i the backends, do the same for bridge lines. It's unlikely that we'll ever need to change those. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/lstopo/lstopo-ascii.c (diff) The file was modified
utils/lstopo/lstopo-tikz.c (diff) The file was modified
utils/lstopo/lstopo-cairo.c (diff) The file was modified
utils/lstopo/lstopo-android.c (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff) The file was modified
utils/lstopo/lstopo-windows.c (diff) The file was modified
utils/lstopo/lstopo-svg.c (diff) The file was modified
utils/lstopo/lstopo.h (diff) The file was modified
utils/lstopo/lstopo-fig.c (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff) The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
utils/lstopo/lstopo.h (diff) The file was modified
NEWS (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff) The file was modified
utils/lstopo/lstopo.h (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff) The file was modified
contrib/completion/bash/hwloc (diff) The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
NEWS (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff) The file was modified
contrib/completion/bash/hwloc (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff) The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
contrib/completion/bash/hwloc (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff) The file was modified
utils/lstopo/lstopo.h (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff) The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff) The file was modified
utils/lstopo/lstopo.h (diff) The file was modified
NEWS (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff) The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff) The file was modified
utils/lstopo/lstopo.h (diff)
Commit
1fe1d2fdc007b325bdef0409f0abc0d057a70342
by Brice.Goglin windows: don't create groups for windows processor groups by default As discussed in #497, Windows may now create processor groups that strictly intersect with NUMAs or packages. This causes a big warning when hwloc tries to insert the corresponding group objects in the topology. Fortunately, these groups aren't as useful as before in Windows 11 and WS2022. So disable them by default. We still have the Windows specific API to query them, and group objs may be reenabled by setting HWLOC_WINDOWS_PROCESSOR_GROUP_OBJS=1 in the environment. Closes #497. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
NEWS (diff) The file was modified
hwloc/topology-windows.c (diff) The file was modified
doc/hwloc.doxy (diff)
Commit
62582b3259a8415687fb0915f24cfa45dfb0abf6
by Brice.Goglin NEWS: clarify the NEWS entry about the new windows API Even if the online doc talks about some features appearing in Windows 10 build 20348, Microsoft says the doc will be fixed and it's only Windows 11 and WS2012. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
NEWS (diff) The file was modified
NEWS (diff) The file was modified
hwloc/topology-windows.c (diff)
Commit
ad1d888a4b851c7fca9c74ecc1781a7f06af3df2
by Brice.Goglin windows: fix the source string when inserting processor group objects Can be either "GetLogicalProcessorInformation" or "GetLogicalProcessorInformationEx". Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-windows.c (diff)
Commit
4682d2c2751ba7ba76eb3b76060c700e13d0a5a5
by Brice.Goglin windows: remove support for the old non-"Ex" GetLogicalProcessorInformation() Still detect it at runtime whether GetLogicalProcessorInformationEx() is available but don't fallback to the old GetLogicalProcessorInformation() anymore. GetLogicalProcessorInformationEx() is available since Windows7/WS2008R2. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
contrib/android/include/private/autogen/config.h (diff) The file was modified
config/hwloc.m4 (diff) The file was modified
contrib/windows/private_config.h (diff) The file was modified
hwloc/topology-windows.c (diff)
Commit
67644a5fb05bf3128591bd213245c4bc3fb6ef81
by Brice.Goglin lstopo/windows: fix the initial window size by using the right fonts The window initial dimensions are computed by drawing once in an invisible window. We need to use the right font(s) to get the correct dimensions that will match the final drawing in the window. Instead of creating a temporary font without specifying its non-boldness (0 means FW_DONTCARE), define both normal and bold fonts as usual during the initial step. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/lstopo/lstopo-windows.c (diff) The file was modified
contrib/windows-cmake/private_config.h.in (diff) The file was modified
NEWS (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff)
Commit
584541f65209add4dbd60cdff53710b94f400451
by Brice.Goglin levelzero: don't skip devices if sysman is missing We won't have all attributes and we won't have the locality, but some attributes are still useful to some apps. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-levelzero.c (diff) The file was modified
hwloc/topology-levelzero.c (diff)
Commit
ac8f22469ab570531d168b04692ff636a28e083e
by Brice.Goglin levelzero: update the management of vendor/device/brand/... strings Recent implementations rather return "unknown" than "Unknown", and the Model name seems correct, at least for Intel GPUs now. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-levelzero.c (diff) The file was modified
hwloc/topology-levelzero.c (diff) The file was modified
tests/hwloc/ports/include/levelzero/level_zero/ze_api.h (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
NEWS (diff) The file was modified
hwloc/topology-levelzero.c (diff) The file was modified
NEWS (diff) The file was modified
tests/hwloc/ports/include/levelzero/level_zero/ze_api.h (diff) The file was modified
tests/hwloc/ports/include/levelzero/level_zero/zes_api.h (diff) The file was modified
hwloc/topology-levelzero.c (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/topology-levelzero.c (diff) The file was modified
hwloc/topology-levelzero.c (diff)
Commit
e9505f5878cd74aedfcad4ca2722dc293d7254e8
by Brice.Goglin windows: update comments about new topology features in recent versions Even if the doc talks about Windows 20348, Microsoft says that the new API for describing NUMA nodes and caches spanning multiple processor groups is only in Windows 11 and Server 2022. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-windows.c (diff) The file was modified
hwloc/topology-levelzero.c (diff)
Commit
3bd29845f1dad88c90a9b2d76f15a5ffdec2ee59
by Brice.Goglin levelzero: query memory and expose LevelZeroHBM/DDR/MemorySize info attrs Both "core API" and "sysman api" may report useful info, but drivers and implementations don't work in all cases on all hardware yet. For now we use sysman if enabled, otherwise coreapi. HWLOC_L0_COREAPI_MEMORY may be set to 0 or 1 to force sysman or core API. We'll revisit this later once L0 bugs are fixed. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/ports/include/levelzero/level_zero/ze_api.h (diff) The file was modified
tests/hwloc/ports/include/levelzero/level_zero/zes_api.h (diff) The file was modified
hwloc/topology-levelzero.c (diff) The file was modified
NEWS (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/topology-levelzero.c (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
doc/hwloc.doxy (diff)
Commit
46a7d94f3ab57ba56c9a8124dbf73ed147fbd155
by Brice.Goglin linux/cpukinds: ignore max frequencies is base frequency is known Some Intel processors have Turbo Boost Max 3.0 which allows some cores to have a slightly higher max turboboost frequency than others. For instance i9-10920x gives 4.6GHz max to 8 cores, 4.7 to 2, and 4.8 to 2. When the Linux backend sees this, it creates 3 cpukinds because max frequencies are slightly different, while the base frequency is the same. For now, we just ignore max frequencies is base frequencies are available (which is the case on Intel CPUs with not-too-old Linux kernels). We'll improve things later. This disabling may be reverted with HWLOC_CPUKINDS_MAXFREQ=1. maxfreq is forced back in tests/linux/fakeheterocpunuma but disabled in tests/x86+linux/5intel64-hybrid-lakefield. Refs #502. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
doc/hwloc.doxy (diff) The file was modified
tests/hwloc/x86+linux/5intel64-hybrid-lakefield.output (diff) The file was modified
NEWS (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
tests/hwloc/linux/fakeheterocpunuma.env (diff) The file was modified
NEWS (diff) The file was modified
contrib/android/include/hwloc/autogen/config.h (diff) The file was modified
contrib/android/AndroidApp/lstopo/build.gradle (diff) The file was modified
contrib/windows-cmake/CMakeLists.txt (diff) The file was modified
VERSION (diff) The file was modified
contrib/windows/hwloc_config.h (diff) The file was modified
autogen.sh (diff) The file was modified
autogen.sh (diff) The file was modified
contrib/android/AndroidApp/lstopo/build.gradle (diff)
Commit
4b7540f34bc10d7489da30ff7335a48ea3771954
by Brice.Goglin linux/cpukinds: adjust max frequencies instead of ignoring them If some cores have the same base frequency but different max freqs (e.g. because of Intel Turbo Boost Max 3.0) adjust those max freqs down to report the same for those cores. By default, this is only enabled for differences < 10% but it may be configured with HWLOC_CPUKINDS_MAXFREQ=adjust=X (X in percents) or disabled (=1). Ignoring (default behavior in 2.7) is still possible with =0. Closes #502. This is the real fix instead of the work-around in 2.7.0 in commit 46a7d94f3ab57ba56c9a8124dbf73ed147fbd155 . Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/x86+linux/5intel64-hybrid-lakefield.output (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
NEWS (diff) The file was modified
hwloc/topology-linux.c (diff)
Commit
82d202791bb12ae5cb2bc416dc9dd5e9975aef31
by Brice.Goglin levelzero: ignore DDR memory if the device is integrated The Core API reports "DDR" memory for both integrated GPUs and discrete with GDDR. The former is likely the host DRAM, so ignore it by default (can be unignored with HWLOC_L0_COREAPI_MEMORY=2). Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-levelzero.c (diff) The file was modified
tests/hwloc/ports/include/levelzero/level_zero/ze_api.h (diff)
Commit
d194199ce528e89d87286e42b5919dbed149c4c8
by Brice.Goglin contrib/android: bump compileSdkVersion to 30 as well We bumpded targetSdkVersion to 30 in c62e5c8bb7870ae211f41e8a8a2d4a10e73daa59 and lint errors out if compileSdkVersion is lower. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
contrib/android/AndroidApp/lstopo/build.gradle (diff)
Commit
41b09946a98774db047a03805b6291ef255a8bd1
by Brice.Goglin xml: support id="objXYZ" instead of gp_index="XYZ" We'll likely switch to the former in a future major release because it simplifies XML-based manipulation outside of hwloc. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-xml.c (diff) The file was modified
README (diff) The file was modified
doc/main.doxy (diff) The file was modified
hwloc/topology-darwin.c (diff) The file was added tests/hwloc/ports/include/darwin/Availability.h The file was modified
tests/hwloc/ports/Makefile.am (diff) The file was modified
tests/hwloc/ports/include/darwin/IOKit/IOKitLib.h (diff) The file was modified
utils/hwloc/common-ps.c (diff) The file was modified
config/hwloc_internal.m4 (diff) The file was modified
hwloc/topology.c (diff)
Commit
c90ad95636b0f1530e16a502b14a1353f6c01797
by Brice.Goglin API: add topology flag NO_DISTANCES To reduce the time+space overhead when doing basic things. Linux still loads distances internally for better NUMA topology organization. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/misc.h (diff) The file was modified
hwloc/topology.c (diff) The file was modified
hwloc/topology-nvml.c (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
hwloc/topology-freebsd.c (diff) The file was modified
hwloc/topology-rsmi.c (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
hwloc/topology-solaris.c (diff) The file was modified
include/hwloc.h (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
hwloc/topology.c (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
include/hwloc.h (diff) The file was modified
utils/hwloc/misc.h (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
hwloc/topology-darwin.c (diff) The file was modified
include/hwloc.h (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
utils/hwloc/misc.h (diff) The file was modified
hwloc/topology-x86.c (diff) The file was modified
hwloc/topology.c (diff) The file was modified
NEWS (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
include/hwloc.h (diff)
Commit
7c594ccce2536121911d76270c7900a1bf81ba41
by Brice.Goglin contrib/ci.inria.fr: never disable make check in coverity and sonar-scanner jobs NO_CHECK would make some issues disappear just because we disabled make check in a custom job config. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
contrib/ci.inria.fr/job-3-sonarscanner.sh (diff) The file was modified
contrib/ci.inria.fr/job-3-coverity.sh (diff) The file was modified
hwloc/topology-darwin.c (diff)
Commit
25d69491cdf0475026a474b0200fb5253ec07bce
by Brice.Goglin darwin: split to a generic cpukinds query and register So that we may query from other sources (eg sysctl soon), look at cpukinds during the rest of the discovery, and register them the end. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-darwin.c (diff) The file was modified
hwloc/topology-darwin.c (diff)
Commit
565c241497519841758649def21ebf472e38ca5a
by Brice.Goglin darwin: optionally get cpukinds from new sysctl hw.perflevels too Use the existing IOKit code by default since it provides info about core types, and the precise list of core indexes, while sysctl only gives the number of cores per kind. If we found nothing in IOKit or HWLOC_DARWIN_CPUKINDS_FROM_SYSCTL=1 in the env, use sysctl instead. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/topology-darwin.c (diff)
Commit
f7c9aa86c2f8ce8717ce4381357d3b98518d14ca
by Brice.Goglin darwin: detect heterogeneous caches from sysctl perflevels When we get cpukinds from IOKit, we have to match them with perflevels (P=0 and E=1 for now) to find out which cpukinds correspond to sysctl perflevel caches. If we can't match (may occur if future core types are different from E and P), we fallback to sysctl perflevels. Fixes #507 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-darwin.c (diff) The file was modified
NEWS (diff) The file was modified
contrib/windows-cmake/CMakeLists.txt (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/topology-levelzero.c (diff) The file was modified
hwloc/topology-levelzero.c (diff) The file was modified
tests/hwloc/ports/include/levelzero/level_zero/ze_api.h (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/topology-levelzero.c (diff)
Commit
fe363de1647013c190ac3e9d25bb1d7d3fbe574d
by Brice.Goglin core+levelzero: Set ZES_ENABLE_SYSMAN via setenv instead of putenv Setting `ZES_ENABLE_SYSMAN` via `putenv` placed a constant string in the environ array which cannot be touched. If the user is manipulating that environ array then touching this envar will result in a segv. - Instead of using `putenv` use `setenv` which will put a copy of the constant string in the `environ` array allowing the end user to manipulate that array as needed. - Note that I could not find a `setenv` function for windows so I left a comment and did not touch that code. Signed-off-by: Joshua Hursey <jhursey@us.ibm.com> Also change the putenv() inside topology-levelzero.c for consistency and update some comments. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-levelzero.c (diff) The file was modified
hwloc/topology.c (diff)
Commit
9f1fefe75eea888057b49d910f1077b302778013
by Brice.Goglin darwin: read cachelinesize before creating hybrid perflevel caches Otherwise we get linesize=0 for all these caches. Thanks to Ralph Castain for the report. Closes #513 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-darwin.c (diff)
Commit
e7214879cdc7998ff57a99eb902378cc5684fff6
by Brice.Goglin core: fix setenv(ZES_ENABLE_SYSMAN) on mingw MinGW has gcc constructor but no setenv(), use putenv() there too. Fixes fe363de1647013c190ac3e9d25bb1d7d3fbe574d Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology.c (diff) The file was modified
hwloc/topology.c (diff) The file was modified
hwloc/topology-levelzero.c (diff) The file was modified
include/hwloc.h (diff)
Commit
748fd86e065ebfa32892a1fcc96dd33e809219cd
by jhursey Strengthen the NVML check by adding a runtime check * The Nvidia ML library is provided by a stub which can fool the linker in thinking that the actual library is on the system even when it is not. This can cause a runtime failure on such systems. - Strengthen the `nvidia-ml` library check by adding a runtime check that the resulting program will run not just compile and link. Signed-off-by: Joshua Hursey <jhursey@us.ibm.com> The file was modified
configure.ac (diff) The file was modified
config/hwloc.m4 (diff) The file was modified
include/hwloc/memattrs.h (diff)
Commit
8fb90fb27bf941e26b8381e39bfb9d24a9d53f50
by Brice.Goglin configure: add HWLOC_FOO_COMPONENT_BUILTIN macros to public headers So that external users know if CUDA/NVML/... support is built-in (since they may bring additional lib dependencies). Refs #515 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
include/hwloc/autogen/config.h.in (diff) The file was modified
config/hwloc.m4 (diff) The file was modified
configure.ac (diff) The file was modified
config/hwloc.m4 (diff) The file was modified
config/hwloc_internal.m4 (diff)
Commit
7a558845f9835103b5e8bbdfbe8e9fb1181fb1c9
by Brice.Goglin configure+doxy: document that --with-cuda=/nonexisting will disable everything installed by CUDA Alternative to --disable-cuda --disable-nvml --disable-opencl for NVIDIA GPUs. It works because all of them start by looking at the common NVIDIA detection which looks at --with-cuda. Refs #515 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
config/hwloc_internal.m4 (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
config/hwloc.m4 (diff) The file was modified
configure.ac (diff) The file was modified
configure.ac (diff) The file was modified
config/hwloc.m4 (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/topology-synthetic.c (diff) The file was modified
utils/hwloc/hwloc-annotate.c (diff) The file was modified
utils/hwloc/hwloc-annotate.1in (diff) The file was modified
utils/hwloc/test-build-custom-topology.sh.in (diff) The file was modified
tests/hwloc/hwloc_synthetic.c (diff) The file was modified
utils/hwloc/test-build-custom-topology.output (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff) The file was modified
include/hwloc.h (diff) The file was modified
include/hwloc.h (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff)
Commit
514d63fab6e7ae60e0fae4a4f224b02b53a60afb
by Brice.Goglin core: don't enable L0 Sysman if L0 is disabled If the levelzero backend isn't enabled, don't force enable Sysman since it may disturb people using levelzero outside of hwloc. In theory, it's still possible to build hwloc without levelzero enabled, but load an external levelzero component as a plugin at runtime. That's unlikely to happen, ignore this case, they may have to manually set ZES_ENABLE_SYSMAN=1. Thanks to Jonathan Peyton for the suggestion. Closes #521 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology.c (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
tests/hwloc/memattrs.c (diff) The file was modified
hwloc/memattrs.c (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
include/hwloc/memattrs.h (diff)
Commit
51665ffa29cb87e46d95a858b9c43cba64a48aaf
by Brice.Goglin memattrs: add Read/Write Bandwidth/Latency attributes These could be manually added in the past, now they have a dedicated ID and name. It looks like newest Intel Platform expose separed bandwidth values for read and write, so expose them. The old Bandwidth/Latency values are still exposed as "average" values. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/lstopo/test-lstopo.output (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
NEWS (diff) The file was modified
include/hwloc/memattrs.h (diff) The file was modified
hwloc/memattrs.c (diff) The file was modified
tests/hwloc/memattrs.c (diff) The file was modified
NEWS (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
tests/hwloc/linux/fakememinitiators-1np2c+1npp+gi.output (diff) The file was modified
utils/lstopo/lstopo-text.c (diff) The file was modified
tests/hwloc/linux/fakememinitiators-1np2c+1npp+gi.output (diff) The file was modified
utils/lstopo/lstopo-text.c (diff) The file was modified
utils/lstopo/test-lstopo.output (diff) The file was modified
tests/hwloc/x86/Makefile.am (diff) The file was added tests/hwloc/x86/AMD-19h-Zen3-2xEpyc-7763.tar.bz2 The file was added tests/hwloc/x86/AMD-19h-Zen3-2xEpyc-7763.output
Commit
ff102fdfa95d911a4a1eac33c6cd80cdfe30445d
by Brice.Goglin x86: workaround VMs reporting invalid core/thread info Check some CPUID outputs before dividing since some VMs do not report coherent values. Crostini / Chrome M99 on Acer C933 Chromebook gets a division by zero error because: CPUID leaf 0x1 returns register EDX with bit HTT=28 unset. According to the Intel x86 manual, this means: "A value of 0 for HTT indicates there is only a single logical processor in the package and software should assume only a single APIC ID is reserved." This seems wrong on a quad-core Celeron 4100 processor. Moreover, CPUID leaf 0x4 with subleaf 0 returns a valid first level of cache which says (in EAX bits 26-31)that there are 4 cores in the physical package. This is correct, and contradicts CPUID leaf 0x1 above. Thanks to Peter Bense for the report. Closes #525 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-x86.c (diff) The file was modified
NEWS (diff)
Commit
b8574c213343dd50c07f353648c9ceeb9a488fb0
by Brice.Goglin configure: add --disable-readme Updating the top-level README can fail in some corner cases when "mixing" srcdir and builddir builds, e.g. in #526. This could be properly fixed by: * explicitly depending on $(DOX_HTML_DIR)/index.html so that make can take it either from srcdir or builddir. * creating $(DOX_HTML_DIR) first since it doesn't exist in the builddir when using srcdir doxygen output. I'd rather not touch that code unless strictly required. Closes #526. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
NEWS (diff) The file was modified
config/hwloc_internal.m4 (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff)
Commit
80e8e5b328cae3361f192fc33f7297e8a71db8a4
by Brice.Goglin utils: allow loading fsroot/cpuid topologies from a tarball If -i/--input is given a file ending with .tar.bz2 or .tar.gz (or --input-format/--if "archive" is given), use archivemount to fusemount the tarball instead of requiring unpacking it before load. We enter the mountpoint, 'umount -l' and 'rmdir' it immediately so that we don't leak a mountpoint if lstopo crashes. test-topology.sh under tests/hwloc/{linux,x86} still unpacks since archivemount is rarely installed. Specific archivemount tests will be committed separately. This optional feature is mostly convenient to avoid having millions of small files in lots of fsroot topologies on disk in case there's a need to test lstopo manually. I have 2 millions of them, and my backup tool doesn't like that, replacing them with tarballs will cleanup all this. Closes #531 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
NEWS (diff) The file was modified
utils/hwloc/hwloc-calc.1in (diff) The file was modified
config/hwloc_internal.m4 (diff) The file was modified
utils/hwloc/hwloc-info.1in (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff) The file was modified
utils/hwloc/hwloc-distrib.1in (diff) The file was modified
utils/hwloc/misc.h (diff) The file was modified
utils/lstopo/test-lstopo.sh.in (diff) The file was modified
config/hwloc.m4 (diff)
Commit
b294ce85c9223d48d55b3b2ac67a3521539e29f0
by Brice.Goglin distances: fix clang warning about NULL pointer substraction in container_offset warning: performing pointer subtraction with a null pointer has undefined behavior [-Wnull-pointer-subtraction] Substracting NULL is really required anymore since NULL is always 0 since C99 anyway. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/distances.c (diff) The file was modified
utils/hwloc/hwloc-gather-cpuid.c (diff) The file was modified
utils/hwloc/hwloc-gather-cpuid.c (diff) The file was modified
utils/hwloc/hwloc-gather-cpuid.c (diff) The file was modified
utils/hwloc/hwloc-gather-cpuid.c (diff)
Commit
8223863f4f6ac66844840715a9b061827674d305
by Brice.Goglin lstopo: add --os/logical-index-prefix To replace " P#" and " L#" with a custom prefix. Useful for avoiding "P" and "L" in figures in papers when we don't want to talk about logical-vs-os indexes. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/lstopo/lstopo-text.c (diff) The file was modified
NEWS (diff) The file was modified
contrib/completion/bash/hwloc (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff) The file was modified
utils/lstopo/lstopo.h (diff) The file was modified
utils/lstopo/lstopo.c (diff)
Commit
59d1bd61afe687dceaa54496dbf65e788061078a
by Brice.Goglin levelzero: use the core PCI prop extension if available When available, it gives the PCI locality without requiring Sysman. If not, we fallback to Sysman as usual. Seems to be available since L0 1.3 spec, but it's not clear when it's actually implemented (runtime version and hardware model). Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
config/hwloc.m4 (diff) The file was modified
hwloc/topology-levelzero.c (diff) The file was modified
tests/hwloc/ports/include/levelzero/level_zero/ze_api.h (diff)
Commit
2e8fedad73e8a2550af94cae5b17570791699579
by Brice.Goglin pci: rework locality quirks Haven't been used for a while, will be back soon. Apply quirks early in the pci locality discovery instead of only fixing buggy cpusets later. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/pci-common.c (diff) The file was modified
include/private/private.h (diff) The file was modified
hwloc/topology.c (diff)
Commit
71d27c1636a29bb0721567ece1c41c181891e301
by Brice.Goglin pci: add a locality quirk for the HPE Cray EX235a This server is used in Frontier@ORNL and several other supercomputers. It has a complex PCI locality that cannot easily be exposed in ACPI tables. It's specific/rare/important enough to justify a quirk in hwloc. This may be disabled by setting HWLOC_PCI_LOCALITY= (empty) in the environment. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/pci-common.c (diff) The file was modified
NEWS (diff)
Commit
2d262de4ba844576875586be17c48785d724c4c1
by Brice.Goglin pci: add a fake pci locality quirk for testing It attaches all PCI hierarchies to the last PU. Enabled with HWLOC_PCI_LOCALITY_FAKE_QUIRK=1 (unless HWLOC_PCI_LOCALITY is set in the env) Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/pci-common.c (diff) The file was modified
NEWS (diff) The file was modified
hwloc/topology-linux.c (diff)
Commit
c413eb14fe9744ecfa84d58c285dba909da617f2
by Brice.Goglin linux/io: remove support for /sys/class/dax DAX devices moved to /sys/bus/dax soon after their support were added in Linux, distributions disabled it and now it's removed from latest kernels. Also /sys/class/dax misses many features, such as kmem driver for exposing as NUMA nodes, and hmem driver for special-purpose memory. So don't bother supporting it. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-linux.c (diff)
Commit
b4b352d0ca08ee6ba9ff83c8ab85f4f3ba3ea2ef
by Brice.Goglin linux/io: cleanup dax/non-dax devtype management Clarify the difference between "class" and "bus" devices, which explains why we follow the "device" symlink for the former and not the latter. And then clarify why we look at the parent for DAX bus devices. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-linux.c (diff)
Commit
48673bc74b8435631325448224508cbf6d5b51ae
by Brice.Goglin linux/io: no need for sysfs local_cpus for OSdev locality Most OSdev have a numa_node attribute in their sysfs hierarchy, but local_cpus is only available PCI sysfs devices. OSdev that are related to a PCIdev are attached below that PCIdev. We don't need to look at numa_node or local_cpus here. The locality of the PCIdev was obtained earlier by looking at local_cpus in hwloc_linux_backend_get_pci_busid_cpuset(). numa_node is also available (since 2007) but not used. OSdev that are NOT related to a PCIdev have no relevant local_cpus to look at, hence this code was never used. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-linux.c (diff)
Commit
81b0bf3487698d6fbaa4f90f344712dc1a993e1b
by Brice.Goglin linux/io: rework/fix numa_node attribute in sysfs For class devices (from /sys/class/foo), the actual device is pointed by the "device" symlink, there's usually a numa_node attribute in there. For bus devices (from /sys/bus/foo/devices), numa_node can be in the device itself (hmem DAX devices in some Suse 5.3 kernel), or in its parent (old DAX devices in pre-5.5 kernels), or in both (newer kernels). The previous code only supported the parent case. We now use the device itself first for bus devices (enough in the vast majority of cases), and then the parent if a flag was given (for DAX on some old kernels). Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-linux.c (diff) The file was modified
utils/hwloc/hwloc-gather-topology.in (diff) The file was modified
tests/hwloc/linux/32em64t-2n8c+1mic.tar.bz2 (diff) The file was modified
tests/hwloc/linux/fakememinitiators-1np2c+1npp+gi.tar.bz2 (diff)
Commit
161dbb2c18c5fbf5faceffeb1b0e11134cfdac15
by Brice.Goglin linux: fix and factorize the checking of whether a DAX device is exposed as NUMA node Even if target_node points an existing NUMA node, it doesn't mean that specific DAX was added to that node (for instance, it could be a subpart of it when memmap/efi_fake_addr boot parameters were used to change memory region attributes). Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-linux.c (diff)
Commit
ed52753dbec61ac3c0257b5c89d24e2caa4c9192
by Brice.Goglin linux: add DAXParent and DAXType info attr Added to both DAX OS devices or their corresponding NUMA nodes. DAXParent is a string describing the sysfs hierarchy going to the parent device (contains "hmem" for soft-reserved specific-purpose memory and "ndbus" for NVDIMMs). DAXType is either "SPM" or "NVM" for now. We'll use bandwidth later to detect when SPM is actually HBM. The "ndbus" subsystem driving these DAX devices is for nvdimms only right now, but I wouldn't be surprised if non-nvdimm hardware ended up there in the future too, hence the name "NVM" instead of "NVDIMM". Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/linux/32em64t-2n8c+1mic.output (diff) The file was modified
tests/hwloc/linux/fakememinitiators-1np2c+1npp+gi.output (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
NEWS (diff) The file was modified
hwloc/topology-linux.c (diff)
Commit
2bb8dad3346a3a45e59885f1c341401c975c6c28
by Brice.Goglin linux/block: replace "NVDIMM" subtype with "NVM" or "SPM" to match DAX attributes A DAX file may come from a HBM node marked as specific-purpose (SPM). Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
tests/hwloc/linux/32em64t-2n8c+1mic.output (diff)
Commit
a82267bcf4d0c220529eff886760b432c7392cf8
by Brice.Goglin memattrs: heuristics to set NUMA node subtype to DRAM/HBM/SPM/NVM Internally, we classify NUMA nodes by tier: 1) "UNKNOWN" (usually DRAM) 2) "SPM" (UEFI name "Specific-Purpose Memory") is what Linux exposes as "Soft-Reserved" DAX, usually HBM but could be something else. 3) "NVM" (NVDIMMs already detected on Linux through dax/kmem) 4) "GPU" (NVIDIA-only, already detected on Linux, and exposed with subtype "GPUMemory") If (2) has 2x higher bandwidth than (1), (2) becomes HBM and (1) become DRAM. If HWLOC_MEMTIERS_GUESS=spm_is_hbm is set in the environment, we don't even look a the bandwidth. In the end, we set DRAM/HBM/NVM to NUMA node subtypes. We keep SPM if we couldn't guess that SPM was HBM. DRAM isn't set unless there's anything else in the system. The heuristics is applied at the end of the topology, even when loading from XML. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/linux/fakememinitiators-1np2c+1npp+gi.output (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/memattrs.c (diff) The file was modified
include/private/private.h (diff) The file was modified
hwloc/topology.c (diff) The file was modified
NEWS (diff) The file was modified
include/hwloc/rename.h (diff) The file was added tests/hwloc/memtiers.c The file was modified
tests/hwloc/Makefile.am (diff)
Commit
526cdf1b7fc87f1f4a3b1e14d675139d65342681
by Brice.Goglin tests/linux: add a complex test case with lots of heterogeneous memories Package0 has DRAM+HBM+NVM but HBM is exposed as DAX (SPM instead of HBM since DAX have no BW info). Package1 has DRAM+2xHBM+NVM but one HBM is as DAX (SPM instead of HBM). Package2 has HBM+2xNVM but one NVM is as DAX (NVM). The case was generated with: qemu-system-x86_64 -accel kvm \ -machine pc,nvdimm=on,hmat=on \ -drive if=pflash,format=raw,file=$FILES/OVMF.fd \ -drive media=disk,format=qcow2,file=$FILES/efi.qcow2 \ -smp 6 \ -m 6G,slots=4,maxmem=8G \ -object memory-backend-ram,size=3G,id=ram0 \ -object memory-backend-ram,size=1G,id=ram1 \ -object memory-backend-ram,size=512M,id=ram2 \ -object memory-backend-ram,size=512M,id=ram3 \ -object memory-backend-ram,size=512M,id=ram4 \ -object memory-backend-ram,size=512M,id=ram5 \ -numa node,nodeid=0,memdev=ram0,cpus=0-1 \ -numa node,nodeid=1,memdev=ram1,cpus=2-3 \ -numa node,nodeid=2,memdev=ram2,initiator=0 \ -numa node,nodeid=3,memdev=ram3,initiator=1 \ -numa node,nodeid=4,memdev=ram4,initiator=1 \ -numa node,nodeid=5,memdev=ram5,cpus=4-5 \ -numa node,nodeid=6,initiator=0 \ -numa node,nodeid=7,initiator=1 \ -numa node,nodeid=8,initiator=5 \ -numa node,nodeid=9,initiator=5 \ -object memory-backend-file,id=nvdimm1,share=on,mem-path=/tmp/nvdimm1.img,size=512M \ -device nvdimm,id=nvdimm1,memdev=nvdimm1,unarmed=off,node=6 \ -object memory-backend-file,id=nvdimm2,share=on,mem-path=/tmp/nvdimm2.img,size=512M \ -device nvdimm,id=nvdimm2,memdev=nvdimm2,unarmed=off,node=7 \ -object memory-backend-file,id=nvdimm3,share=on,mem-path=/tmp/nvdimm3.img,size=512M \ -device nvdimm,id=nvdimm3,memdev=nvdimm3,unarmed=off,node=8 \ -object memory-backend-file,id=nvdimm4,share=on,mem-path=/tmp/nvdimm4.img,size=512M \ -device nvdimm,id=nvdimm4,memdev=nvdimm4,unarmed=off,node=9 Booted with 5.18 with efi_fake_mem=2G@5G:0x40000 to mark PXM 2-5 as SPM. Then all NVDIMM namespaces (5.0 to 8.0) are converted to devdax with ndctl disable-namespace namespaceX.0 ndctl create-namespace -f -e namespaceX.0 -t pmem --mode=devdax Then some DAX (1.0 3.0 5.0 7.0 and 8.0) are exposed as NUMA nodes with daxctl reconfigure-device --mode=system-ram daxX.0 Then hwloc-gather-topology and tweak things: * Remove dax4.0 since it doesn't really exist (it appears because we used a single 2G efi_fake_mem= instead of 4 consecutive ones, but it actually covers dax0.0-dax3.0). * Fix the locality of some DAX (in /sys/bus/dax/devices/*/{,..}/numa_node) that Qemu doesn't set correctly because we didn't give the HMAT and SLIT tables (and initiator= doesn't work very well). * Then set R/W bandwidth of all nodes (100 for NVM, 1000 for DRAM, 10000 for HBM) in /sys/devices/system/node/node?/access1/initiators/*bandwidth (possible with Qemu but requires loooooots of useless values on the command-line to fill the matrix). Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was added tests/hwloc/linux/fakeheteromemtiers.output The file was modified
tests/hwloc/linux/Makefile.am (diff) The file was added tests/hwloc/linux/fakeheteromemtiers.tar.bz2 The file was modified
NEWS (diff) The file was modified
hwloc/pci-common.c (diff) The file was modified
hwloc/topology-cuda.c (diff) The file was modified
hwloc/pci-common.c (diff) The file was modified
include/private/windows.h (diff) The file was modified
contrib/android/AndroidApp/lstopo/build.gradle (diff) The file was modified
VERSION (diff) The file was modified
contrib/android/include/hwloc/autogen/config.h (diff) The file was modified
contrib/windows/hwloc_config.h (diff) The file was modified
contrib/windows-cmake/CMakeLists.txt (diff) The file was removed tests/hwloc/linux/1alpha.tar.bz2 The file was modified
tests/hwloc/linux/Makefile.am (diff) The file was removed tests/hwloc/linux/1alpha.output
Commit
cf4a652fde50009c8c2a48fd06ce8017ad6480ff
by Brice.Goglin tests/linux: cleanup missing cpus in nvidiagpunumanodes Some cpus were removed from /sys/devices/system/cpu to reduce the size of the topology, but the corresponding symlinks in /sys/bus/cpu/devices were still there. This causes these cpu to appear as offline when discovering through /sys/bus/cpu/devices (while they don't exist at all), and to not appear at all when discovering through /sys/devices/system/cpu. Remove the broken symlinks so that discovery from both directory sees the same (reduced) cpuset. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/linux/nvidiagpunumanodes.output (diff) The file was modified
tests/hwloc/linux/nvidiagpunumanodes.kept.output (diff) The file was modified
tests/hwloc/linux/nvidiagpunumanodes.tar.bz2 (diff)
Commit
2f0815709ef83fe3fe278d2702c8cca52cc77870
by Brice.Goglin linux: improve the readdir'ing of /sys/devices/system/{cpu,node} Explicitly ignore entries starting with {cpu,node} but not followed by a digit. At least cpufreq and cpuidle exist. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-linux.c (diff)
Commit
9db28f4394a3e160d581213e0b087745168d99c8
by Brice.Goglin linux: abort if sysfs cpu topology is missing We dropped /proc/cpuinfo-based topology 3.5 years ago in 56e01f01c31f2afd2094a716c0b94a9a463137bd. The fallback for missing CPU topo was only useful for kernels with NUMA sysfs topology but no CPU sysfs topology (before 2.6.16), no need to keep supporting these. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-linux.c (diff)
Commit
b588277a948351e6b2c881de664505bd5397946a
by Brice.Goglin linux: don't support both /sys/bus/cpu/devices and /sys/devices/system/cpu The former will never replace the latter. /sys/devices/system/cpu contains more, so just always use that one. /sys/devices/ is where all devices are, and /sys/bus/*/devices contains symlinks to "similar" stuff in /sys/devices/... Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-linux.c (diff) The file was modified
hwloc/topology-linux.c (diff)
Commit
c241441da9f554fff7bd55ada9143c2c1ad6f16f
by Brice.Goglin linux: don't check whether numa topology is properly available in sysfs Remove the now useless code that checked NUMA support in /sys/devices/system/node and /sys/bus/node/devices. Either NUMA is enabled in the kernel (since early 2.6 kernels) and /sys/devices/system/node exists and we assume it works, or ignore NUMA discovery. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-linux.c (diff) The file was modified
hwloc/topology-linux.c (diff)
Commit
62648f1d7c9ceda62d3b2752b1e43e919dd9bff2
by Brice.Goglin utils/calc: don't segfault if IO is given to -N or -I Walk up parents until we find a normal one before comparing sets. Thanks to Clement Foyer for the report. Closes #537. However -N gpu or -I gpu still doesn't work because it actually counts osdev ignoring the gpu subtype. This would require more work. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/hwloc-calc.c (diff)
Commit
4965e5bd4e5cd663f7f7ab82fcd797ec0c00b28f
by Brice.Goglin utils/calc: allow -H with NUMA nodes There is no code change except explicitly allowing NUMA instead of rejecting it like other special levels (I/O and Misc). However NUMA may cause multiple redundant outputs as documented in the manpage: "on a platform with both DRAM and HBM memory on a package, the first core will be considered both as first core of first NUMA node (DRAM) and as first core of second NUMA node (HBM)." Refs #539 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/hwloc-calc.c (diff) The file was modified
utils/hwloc/hwloc-calc.1in (diff) The file was modified
hwloc/pci-common.c (diff) The file was modified
utils/hwloc/test-hwloc-calc.output (diff) The file was modified
utils/hwloc/test-hwloc-calc.sh.in (diff)
Commit
7833d5200c8372753f6796095e4a46bdad9c6964
by Brice.Goglin utils/calc: allow to retrieve object attributes when scanning type strings for 3.0, we should provide a public API to convert a string into type, depth, subtype (string), and these attributes. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/hwloc-calc.c (diff) The file was modified
utils/hwloc/hwloc-calc.c (diff) The file was modified
utils/hwloc/hwloc-calc.c (diff) The file was modified
utils/hwloc/hwloc-calc.1in (diff) The file was modified
hwloc/topology-levelzero.c (diff) The file was added codemeta.json
Commit
69fbc208ee373cc042aa803325e65febd0bd17ae
by Brice.Goglin pci+linux: make 32bits domain warnings non-critical This issue seems to happen more frequently, don't report it unless all errors are enabled. This warning means that some PCI devices might be missing. That's not critical enough to warn users (most of them know nothing about hwloc). For the record, CUDA init warnings were demoted the same a couple releases ago. Refs #354 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-linux.c (diff) The file was modified
hwloc/topology-pci.c (diff) The file was modified
hwloc/topology.c (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
hwloc/memattrs.c (diff) The file was modified
hwloc/pci-common.c (diff) The file was modified
hwloc/topology-solaris.c (diff) The file was modified
hwloc/topology-aix.c (diff) The file was modified
hwloc/cpukinds.c (diff) The file was modified
hwloc/topology-darwin.c (diff) The file was modified
hwloc/topology-rsmi.c (diff) The file was modified
hwloc/topology-pci.c (diff) The file was modified
hwloc/topology-cuda.c (diff) The file was modified
hwloc/topology-windows.c (diff) The file was modified
hwloc/topology.c (diff) The file was modified
hwloc/topology-nvml.c (diff) The file was modified
hwloc/components.c (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
hwloc/topology-opencl.c (diff) The file was modified
hwloc/topology-levelzero.c (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
include/hwloc/plugins.h (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
contrib/completion/bash/hwloc (diff) The file was modified
codemeta.json (diff) The file was modified
include/hwloc/bitmap.h (diff) The file was modified
include/hwloc/memattrs.h (diff) The file was modified
doc/hwloc.doxy (diff) The file was added doc/examples/memory-attributes.c The file was modified
doc/examples/Makefile.am (diff)
Commit
d316fa749474b5ba0a2d557989af5fb9e0b556de
by Brice.Goglin memattrs.h: cleanup doxygen formatting of the enum of memory attributes Escape double-quotes in comments (they break the parser and the remaining enum item doc is ignored in manpages). Hide initializer values. Uniformize the source formatting to keep it easy to read even if the manpage formatting isn't perfect (looks like doxygen/manpage wants only two paragraph per enum item, but we'd have to change lots of other places, and it would make the source harder to read). Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
include/hwloc/memattrs.h (diff) The file was modified
doc/examples/memory-attributes.c (diff)
Commit
352348a1d861614f67553dd5fd6ad1ef1d4ebca5
by Brice.Goglin nvml/bw: add nvlink versions 3&4 Possible version values aren't documented in NVML API doc, but wikipedia and other sources say 50GB/s per link for v3 (Ampere) and v4 (Hopper). Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-nvml.c (diff)
Commit
0de22a593ff280cca67a01724a01bcf81d4490fc
by Brice.Goglin tests/backends: fix build with ClangCI 15 compiler frontend for MSVC hwloc_backends.c needs some open/read/close/mktemp redirection to _foo. Copy what we did in pci-common.c and add mktemp. Thanks to Mario Emmenlauer for the report. Fixes #546 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/hwloc_backends.c (diff) The file was modified
contrib/completion/bash/hwloc (diff)
Commit
2993fba464749d3a17e08fa2e402f41144675f13
by Brice.Goglin lstopo: if stdout is redirected, use console mode Except if tcgetpgrp(STDIN) fails, we're likely launched by a window manager outside of a terminal in this case. Closes #503 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff) The file was modified
config/hwloc_internal.m4 (diff) The file was modified
utils/lstopo/lstopo.c (diff)
Commit
1a6426878487b986d1a258c584811b35fc23d549
by Brice.Goglin lstopo: only show interactive help if stdout is a tty Otherwise things like .xsession-errors (depending on the window manager) might get flooded with those messages. Ignored on Windows since I have now idea where those messages go when the app doesn't run from a terminal. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
NEWS (diff) The file was modified
contrib/completion/bash/hwloc (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff) The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
hwloc/topology-cuda.c (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff) The file was modified
hwloc/topology-cuda.c (diff) The file was modified
include/hwloc/bitmap.h (diff) The file was modified
include/hwloc/deprecated.h (diff) The file was modified
include/hwloc.h (diff) The file was modified
hwloc/diff.c (diff) The file was modified
include/hwloc/plugins.h (diff) The file was modified
hwloc/pci-common.c (diff) The file was modified
utils/hwloc/hwloc-gather-topology.in (diff) The file was modified
tests/hwloc/ports/include/levelzero/level_zero/zes_api.h (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
NEWS (diff) The file was modified
hwloc/topology-levelzero.c (diff) The file was modified
include/hwloc/distances.h (diff) The file was modified
NEWS (diff)
Commit
795bc1db852b38ce826633703c5cd6b92ca53146
by Brice.Goglin linux+lstopo/draw: expose CXL memory device (Type 3) size Linux support for exposing this memory as NUMA node isn't totally ready yet (at least for RAM), but we already have device ram/pmem size in sysfs. This is a Block device just like DAX for now, we'll work on a better way to expose all these memory things in 3.0. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-linux.c (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff) The file was modified
NEWS (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
contrib/android/AndroidApp/lstopo/build.gradle (diff) The file was modified
contrib/windows-cmake/CMakeLists.txt (diff) The file was modified
contrib/windows/hwloc_config.h (diff) The file was modified
VERSION (diff) The file was modified
contrib/android/include/hwloc/autogen/config.h (diff) The file was modified
hwloc/topology-levelzero.c (diff) The file was modified
VERSION (diff) The file was modified
contrib/android/AndroidApp/lstopo/build.gradle (diff) The file was modified
contrib/android/include/hwloc/autogen/config.h (diff) The file was modified
contrib/windows-cmake/CMakeLists.txt (diff) The file was modified
contrib/windows/hwloc_config.h (diff) The file was modified
hwloc/topology-pci.c (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
include/hwloc.h (diff) The file was removed tests/hwloc/linux/2pa-pcidomain32bits-disabled.output The file was modified
hwloc/topology-linux.c (diff) The file was modified
tests/hwloc/hwloc_topology_abi.c (diff) The file was modified
config/hwloc_internal.m4 (diff) The file was modified
include/hwloc/autogen/config.h.in (diff) The file was removed tests/hwloc/linux/2pa-pcidomain32bits-disabled.source The file was modified
tests/hwloc/linux/Makefile.am (diff) The file was modified
NEWS (diff) The file was modified
include/private/private.h (diff) The file was modified
config/hwloc.m4 (diff)
Commit
704c3166208e6d517a172a2c6f3c397b18520eab
by Brice.Goglin pci: add the prog-if (register level programming interface) to pci device attributes It's the 3rd byte of the class, useful in some corner cases. It'll be assumed 0 when importing from pre-3.x XMLs. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/test-hwloc-annotate.output (diff) The file was modified
tests/hwloc/linux/40intel64-2g2n4c+pci.output (diff) The file was modified
hwloc/topology-nvml.c (diff) The file was modified
tests/hwloc/xml/96em64t-4n4d3ca2co-pci.xml (diff) The file was modified
utils/hwloc/test-hwloc-compress-dir.output.tar.gz (diff) The file was modified
utils/hwloc/test-hwloc-compress-dir.input.tar.gz (diff) The file was modified
tests/hwloc/linux/32em64t-2n8c+1mic.output (diff) The file was modified
tests/hwloc/xml/16intel64-manyVFs.xml (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
tests/hwloc/xml/32em64t-2n8c2t-pci-wholeio.xml (diff) The file was modified
utils/hwloc/test-hwloc-annotate.output2 (diff) The file was modified
utils/hwloc/test-hwloc-diffpatch.input2 (diff) The file was modified
tests/hwloc/xml/28intel64-2p2g7c-CoD.nogroups.v1tov2.xml (diff) The file was modified
tests/hwloc/xml/28intel64-2p2g7c-CoDgroups.v1tov2.xml (diff) The file was modified
hwloc/topology-pci.c (diff) The file was modified
tests/hwloc/xml/32em64t-2n8c2t-pci-normalio.xml (diff) The file was modified
NEWS (diff) The file was modified
utils/hwloc/test-hwloc-diffpatch.input1 (diff) The file was modified
tests/hwloc/xml/power8gpudistances.xml (diff) The file was modified
tests/hwloc/xml/64intel64-3g2n+2n-irregulargroups+pci.xml (diff) The file was modified
include/hwloc.h (diff) The file was modified
tests/hwloc/xml/8em64t-2p2ca2co-nonodesets.v1tov2.xml (diff) The file was modified
tests/hwloc/xml/192em64t-24n8c2t.xml (diff) The file was modified
tests/hwloc/xml/192em64t-12gr2n8c2t.xml (diff) The file was modified
tests/hwloc/xml/24em64t-2n6c2t-pci.xml (diff) The file was modified
hwloc/topology-linux.c (diff)
Commit
c0c5151c2d82c2add1c93458bfa438fcac3ca028
by Brice.Goglin rename.h: replace hwloc_ with HWLOC_ when renaming capital names Thanks to Samuel K. Gutierrez for reporting this issue a long time ago. This shouldn't break anything since people are supposed to call non-renamed symbols and let rename.h rename them. In case it manages to break something, we're only changing this in a major ABI-breaking release. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
doc/hwloc.doxy (diff) The file was modified
include/hwloc/rename.h (diff)
Commit
5a8c01a3c449adbba1ff88640a276076a15e90b7
by Brice.Goglin BGQ: remove support It's more than 10 years old, IBM abandoned the BlueGene line in 2015, and most BGQ installations are gone now. No need to support this in future hwloc releases. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/bind.c (diff) The file was removed hwloc/topology-bgq.c The file was modified
hwloc/Makefile.am (diff) The file was modified
include/hwloc/plugins.h (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
include/private/private.h (diff) The file was modified
contrib/windows/private_config.h (diff) The file was modified
tests/hwloc/ports/Makefile.am (diff) The file was modified
contrib/ci.inria.fr/job-3-sonarscanner.sh (diff) The file was modified
include/private/internal-components.h (diff) The file was modified
config/hwloc_internal.m4 (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
NEWS (diff) The file was modified
hwloc/misc.c (diff) The file was removed tests/hwloc/ports/include/bgq/spi/include/kernel/process.h The file was modified
contrib/android/include/private/autogen/config.h (diff) The file was modified
codemeta.json (diff) The file was removed tests/hwloc/ports/include/bgq/spi/include/kernel/location.h The file was modified
config/hwloc.m4 (diff) The file was modified
contrib/windows-cmake/private_config.h.in (diff) The file was modified
tests/hwloc/hwloc_backends.c (diff) The file was modified
hwloc/topology.c (diff) The file was modified
hwloc/shmem.c (diff) The file was modified
tests/hwloc/hwloc_topology_abi.c (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
hwloc/topology-windows.c (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
hwloc/components.c (diff) The file was modified
hwloc/distances.c (diff) The file was modified
hwloc/memattrs.c (diff) The file was modified
hwloc/bind.c (diff) The file was modified
hwloc/diff.c (diff) The file was modified
hwloc/topology-synthetic.c (diff) The file was modified
include/private/private.h (diff) The file was modified
include/private/private.h (diff) The file was modified
hwloc/topology.c (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
hwloc/components.c (diff)
Commit
91399fbe8d05af5f664b30c43e6b92c76334a514
by Brice.Goglin deprecated.h: remove functions deprecated since 2.0 hwloc_topology_insert_misc_object_by_parent() hwloc_obj_cpuset_snprintf() hwloc_obj_type_sscanf() hwloc_cpuset_to_nodeset_strict() hwloc_cpuset_from_nodeset_strict() Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
NEWS (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
include/hwloc.h (diff) The file was modified
include/hwloc/deprecated.h (diff) The file was modified
NEWS (diff) The file was modified
hwloc/pci-common.c (diff) The file was modified
include/hwloc/helper.h (diff)
Commit
ee87ef7bd086692fc640240fd29245b6507cb5c4
by Brice.Goglin API: add hwloc_get_type_depth_with_attr() Generalized hwloc_get_type_depth() that may disambiguate HWLOC_TYPE_DEPTH_MULTIPLE by using optional attributes. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/traversal.c (diff) The file was modified
utils/hwloc/hwloc-calc.c (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
doc/Makefile.am (diff) The file was modified
NEWS (diff) The file was modified
include/hwloc.h (diff) The file was modified
tests/hwloc/hwloc_type_sscanf.c (diff) The file was modified
utils/hwloc/hwloc-calc.h (diff) The file was modified
utils/hwloc/hwloc-info.c (diff) The file was modified
utils/hwloc/hwloc-distrib.c (diff) The file was modified
NEWS (diff) The file was modified
include/hwloc.h (diff) The file was modified
doc/Makefile.am (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
include/hwloc/deprecated.h (diff)
Commit
8fd1dfc0b0cefd6251f1efe63c40ce46a7885b58
by Brice.Goglin API: replace hwloc_obj_type/attr_snprintf() verbose attribute with flags 1 keeps the same meaning as in 2.x (HWLOC_OBJ_SNPRINTF_FLAG_OLD_VERBOSE). No other flags yet. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/traversal.c (diff) The file was modified
hwloc/topology.c (diff) The file was modified
include/hwloc.h (diff) The file was modified
utils/lstopo/lstopo-android.c (diff) The file was modified
NEWS (diff) The file was modified
hwloc/topology-synthetic.c (diff) The file was modified
utils/lstopo/lstopo-text.c (diff) The file was modified
utils/hwloc/hwloc-ps.c (diff) The file was modified
tests/hwloc/hwloc_type_sscanf.c (diff) The file was modified
utils/hwloc/hwloc-info.c (diff) The file was modified
utils/hwloc/hwloc-calc.c (diff) The file was modified
utils/hwloc/misc.h (diff) The file was modified
utils/hwloc/test-parsing-flags.sh.in (diff)
Commit
f2f4f06f4a8e2275a9588c5b31730072c79b729b
by Brice.Goglin misc: replace hwloc_memory_size_printf_unit/value() with hwloc_memory_size_snprintf() Move it to private/private.h since it uses hwloc_snprintf() and it's now inline instead of a macro. And use obj_snprintf flags instead of only a verbose parameter. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/lstopo/lstopo-draw.c (diff) The file was modified
hwloc/traversal.c (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
include/private/private.h (diff) The file was modified
include/private/misc.h (diff) The file was modified
utils/lstopo/lstopo-text.c (diff)
Commit
7d2aa6a3a5cf58477309f78ec652229f216e0f8c
by Brice.Goglin obj_attr_snprintf: switch to KiB/MiB/GiB/TiB Only the displayed unit changes from KB/MB/... to KiB/MiB/... Values were already divided by 1024 instead of 1000. Fixes #520 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
include/private/private.h (diff) The file was modified
NEWS (diff) The file was modified
utils/lstopo/test-lstopo.output (diff) The file was modified
tests/hwloc/xml/16em64t-4s2c2t.console.output (diff) The file was modified
tests/hwloc/linux/4em64t-2c2t-ignore-reorder.output (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff)
Commit
aeee9e368afd53322705792e7554b34b0370c112
by Brice.Goglin obj_snprintf(): add flags to tune type names and units Allow to switch to raw sizes without units, or actual KB/MB/GB/... Use them instead of the old verbose flag for most tools and tests, except the main lstopo where more tuning is coming. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-synthetic.c (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
tests/hwloc/hwloc_type_sscanf.c (diff) The file was modified
utils/hwloc/test-parsing-flags.sh.in (diff) The file was modified
hwloc/topology.c (diff) The file was modified
utils/hwloc/hwloc-info.c (diff) The file was modified
include/hwloc.h (diff) The file was modified
utils/hwloc/hwloc-calc.c (diff) The file was modified
utils/hwloc/misc.h (diff) The file was modified
utils/hwloc/hwloc-ps.c (diff) The file was modified
hwloc/traversal.c (diff) The file was modified
utils/lstopo/lstopo-android.c (diff) The file was modified
include/private/private.h (diff)
Commit
a077be8b9e77e5973cf8145a3d5db45a6fcf7e68
by Brice.Goglin lstopo: add --osf/--obj-snprintf-flags option to tune obj printing By default, no flags are given (short names, GiB units and no other attributes). --osf/--obj-snprintf-flags passes flags to hwloc_obj_type/attr_snprintf(). Flags are also affected by existing -v --verbose option: If verbose (-v), longer names are used (even in graphical window mode now). If verbose (-v) and textual, more attributes are also displayed, and we use MiB/GiB sizes instead of the old broken KiB-but-KB sizes. Refs #520 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/x86+linux/5intel64-hybrid-lakefield.output (diff) The file was modified
tests/hwloc/linux/16em64t-4s2c2t.output (diff) The file was modified
tests/hwloc/linux/nvidiagpunumanodes.output (diff) The file was modified
contrib/completion/bash/hwloc (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff) The file was modified
tests/hwloc/linux/96em64t-4n4d3ca2co.output (diff) The file was modified
tests/hwloc/linux/2s390-2c.output (diff) The file was modified
tests/hwloc/linux/2amd64-2n.output (diff) The file was modified
tests/hwloc/linux/256ia64-64n2s2c.output (diff) The file was modified
tests/hwloc/linux/4ia64-4s.output (diff) The file was modified
tests/hwloc/linux/2pa-pcidomain32bits.output (diff) The file was modified
utils/lstopo/lstopo.h (diff) The file was modified
tests/hwloc/linux/128ia64-17n4s2c.output (diff) The file was modified
tests/hwloc/linux/16amd64-8n2c-cpusets_noadmin.output (diff) The file was modified
tests/hwloc/linux/16em64t-4s2c2t_merge.output (diff) The file was modified
tests/hwloc/linux/32amd64-4s2n4c-cgroup2.output (diff) The file was modified
tests/hwloc/linux/64intel64-fakeKNL-A2A-cache.output (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff) The file was modified
tests/hwloc/xml/16-2gr2gr2n2c+misc.console.output (diff) The file was modified
tests/hwloc/xml/16intel64-manyVFs.console.output (diff) The file was modified
tests/hwloc/linux/20s390-2g6s4c.output (diff) The file was modified
tests/hwloc/linux/4fake-4gr1nu1pu.output (diff) The file was modified
tests/hwloc/linux/8em64t-2s2ca2c.output (diff) The file was modified
tests/hwloc/linux/16em64t-4s2ca2c-cpusetreorder.output (diff) The file was modified
tests/hwloc/linux/16amd64-8n2c-cpusets.output (diff) The file was modified
tests/hwloc/linux/16amd64-8n2c.output (diff) The file was modified
tests/hwloc/linux/16em64t-4s2c2t-offlines.output (diff) The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
NEWS (diff) The file was modified
tests/hwloc/linux/16amd64-4n4c-cgroup-distance-merge.output (diff) The file was modified
tests/hwloc/linux/8ia64-2s2c2t.output (diff) The file was modified
tests/hwloc/xml/64intel64-3g2n+2n-irregulargroups+pci.console.output (diff) The file was modified
tests/hwloc/linux/2arm-2c.output (diff) The file was modified
tests/hwloc/xml/16amd64-4distances.console.output (diff) The file was modified
tests/hwloc/linux/8amd64-4n2c.output (diff) The file was modified
tests/hwloc/linux/16em64t-4s2ca2c-cpusetreorder-merge.output (diff) The file was modified
utils/hwloc/misc.h (diff) The file was modified
tests/hwloc/linux/16em64t-4s2c2t_ncaches.output (diff) The file was modified
tests/hwloc/linux/8em64t-4c2t.output (diff) The file was modified
tests/hwloc/linux/fakememinitiators-1np2c+1npp+gi.output (diff) The file was modified
tests/hwloc/linux/fakecpuid1f-64intel64-2p4d2n2c2t.output (diff) The file was modified
tests/hwloc/linux/16ia64-8n2s.output (diff) The file was modified
tests/hwloc/linux/8ia64-2n2s2c.output (diff) The file was modified
tests/hwloc/linux/offline-cpu0-node0.output (diff) The file was modified
tests/hwloc/linux/32intel64-2p8co2t+8ve.output (diff) The file was modified
utils/lstopo/test-lstopo.output (diff) The file was modified
tests/hwloc/linux/8em64t-2s2ca2c-buggynuma.output (diff) The file was modified
tests/hwloc/xml/16intel64-manyVFs.console.nocollapse.output (diff) The file was modified
tests/hwloc/linux/48amd64-4d2n6c-sparse.output (diff) The file was modified
tests/hwloc/linux/64intel64-fakeKNL-SNC4-hybrid.output (diff) The file was modified
utils/lstopo/lstopo-text.c (diff) The file was modified
tests/hwloc/linux/64fake-4n2s2ca2c2t.output (diff) The file was modified
tests/hwloc/linux/256ppc-8n8s4t.output (diff) The file was modified
tests/hwloc/linux/fakeheteromemtiers.output (diff) The file was modified
tests/hwloc/linux/64amd64-4s2n4ca2co.output (diff) The file was modified
tests/hwloc/linux/64intel64-fakeKNL-SNC4-hybrid-msc.output (diff) The file was modified
tests/hwloc/linux/128arm-2pa2n8cluster4co.output (diff) The file was modified
tests/hwloc/linux/nvidiagpunumanodes.kept.output (diff) The file was modified
tests/hwloc/linux/40intel64-4n10c+pci-conflicts.output (diff) The file was modified
tests/hwloc/linux/8em64t-2s4c-heterogeneous.output (diff) The file was modified
tests/hwloc/linux/8ia64-4s2c.output (diff) The file was modified
NEWS (diff)
Commit
41ecc23af8528911bea23e7b598648dd83be8965
by Brice.Goglin hwloc-gather-topology: only unalias ls 'unalias -a ls' unaliases everything, while we only wanted 'ls' in ce2ab33a408205cd3ec51f8aa845af5a2ddffd18. 'unalias -a' is bashism anyway. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/hwloc-gather-topology.in (diff) The file was modified
contrib/android/AndroidApp/lstopo/build.gradle (diff)
Commit
c255f14c9aafa781ea192ad79d4c9b22b8375c69
by Brice.Goglin contrib/android: update build parameters for Android 12 Main activity with intent-filter required "android:exported" parameter, and it must be "true" because this activity may be called by other apps or the launcher. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
contrib/android/AndroidApp/lstopo/src/main/AndroidManifest.xml (diff) The file was modified
contrib/android/AndroidApp/lstopo/build.gradle (diff) The file was modified
NEWS (diff) The file was modified
contrib/windows/check-versions.sh (diff) The file was modified
contrib/android/check-versions.sh (diff) The file was modified
contrib/windows-cmake/check-versions.sh (diff) The file was modified
contrib/android/check-versions.sh (diff)
Commit
accb7020c814610ea381438c199843514fe6c9b4
by Brice.Goglin contrib/dist/make_dist_tarball: update android/windows VERSION too HWLOC_VERSION* macros were X.Yrc1 instead of X.Y in the MSVC builds. Windows CMAKE builds don't include suffixes such as rc1, but we still update the hwloc version there for consistency (and in case X.Y was changed too). Android isn't updated since it isn't distributed in tarballs. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
contrib/dist/make_dist_tarball (diff) The file was modified
doc/hwloc.doxy (diff)
Commit
14a76592bd79f398ec63870048576d6e9763fffa
by Brice.Goglin test-gather-topology: skip if /sys isn't available Support for old Linux kernels without /sys support was removed in 2.9 but it can still happen in containerized builds. Skip the Linux gather-topology test in that case. Thanks to Ludovic Courtes for the patch. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/linux/gather/test-gather-topology.sh.in (diff) The file was modified
utils/hwloc/hwloc-gather-cpuid.c (diff)
Commit
66ace095b9347395c0e12920837dce83ab140015
by Brice.Goglin utils/gather-cpuid: always stop ecx loops after 256 iterations In case we ever meet a buggy CPUID implementation. subleaf 256 is way too much for most leaves anyway. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/hwloc-gather-cpuid.c (diff)
Commit
ba0bac0c6ea2f28db8377de39d34e414f6fec807
by Brice.Goglin x86: limit the number of cache and topology subleaves An invalid subleaf is supposed to break these loops. but add a max limit (16 for caches and 32 for topology levels) in case we ever meet a broken CPUID implementation. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-x86.c (diff)
Commit
51f3c81e8cf874952047b4fb4553501253e3bc0a
by Brice.Goglin lstopo: improve the error message in case of unrecognized option When an option is unrecognized, it gets used as the output filename. The real filename, if any, then throws an error saying it's an unrecognized option. Clarify that it couldn't be used as an output filename because another one was already used. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff)
Commit
43c0045e71de5d0f1b85d345cad3ab649529ebcc
by Brice.Goglin lstopo: remove .format from -.format special output early Keep "-" as soon as possible instead of handling "-.format" or even "/dev/stdout" later in the core or in backends. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
utils/lstopo/lstopo.c (diff)
Commit
096c89dd51c03d01d32a15da4e4b1bc023a9e45e
by Brice.Goglin lstopo: show a message about the export filename and format Only shown if exporting to a file, that file is not stdout, stdout is a terminal, and the output format is not window. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
include/hwloc/deprecated.h (diff) The file was modified
utils/hwloc/hwloc-distrib.1in (diff) The file was modified
utils/hwloc/test-hwloc-calc.sh.in (diff) The file was modified
utils/hwloc/hwloc-calc.c (diff) The file was modified
utils/hwloc/misc.h (diff) The file was modified
hwloc/traversal.c (diff)
Commit
501620266a809841a13842c58cabb297f7d1d154
by Brice.Goglin tests: filter-out the upcoming "id" xml attribute It's the future replacement of "gp_index". Just like "gp_index", it may change for (internal) reasons without actually breaking things. Hence, ignore "id" just like we ignore "gp_index" in test outputs. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/lstopo/test-lstopo.sh.in (diff) The file was modified
tests/hwloc/linux/test-topology.sh.in (diff) The file was modified
tests/hwloc/x86/test-topology.sh.in (diff) The file was modified
tests/hwloc/x86+linux/test-topology.sh.in (diff)
Commit
d1029c34976d2519e440a2b21a2200c172f8a28f
by Brice.Goglin xml: add a v2 export flag and enable it by default for now It doesn't do anything yet and it's force-enabled internally for now so that tests outputs don't change yet. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
include/hwloc/rename.h (diff) The file was modified
include/hwloc/export.h (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
utils/hwloc/misc.h (diff)
Commit
5e004cc95bad103f828da9513d477941944bad5c
by Brice.Goglin xml: add support for 3.0 xml format For now, it only changes the XML header to <topology version="3.0"> We still use hwloc2.dtd, we'll fork it later into a hwloc3.dtd if needed. 2.0 is still enabled internally for now, but 2.0 and 3.0 import/export is supported. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
NEWS (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
hwloc/topology-xml-libxml.c (diff) The file was modified
hwloc/topology-xml-nolibxml.c (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff)
Commit
261082bf1f95124949904c1d5af39d42be610957
by Brice.Goglin xml: export id="objXXX" in XML 3.0 Having id="objXXX" in the XML makes external users much easier since they can directly identify objects in CSS, etc. gp_index="XXX" is still exported for now for backward compatibility with 2.x before 2.8, even if they are redundant. It'll be removed in 4.0. id="objXXX" is not exported if v2 is forced. Test outputs aren't updated yet since v2 is still forced by default. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-xml.c (diff) The file was modified
hwloc/hwloc2.dtd (diff)
Commit
45e0f28845c11543997de9813c87e180d6967270
by Brice.Goglin xml: stop forcing v2 export by default v1tov2 XML tests remain v2, others aren't forced to v2 anymore. This reverts the temporary v2 forcing in a previous commit. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/xml/16amd64-8n2c-cpusets.xml (diff) The file was modified
tests/hwloc/xml/192em64t-12gr2n8c2t.xml (diff) The file was modified
utils/hwloc/test-hwloc-diffpatch.input2 (diff) The file was modified
tests/hwloc/xml/16amd64-4distances.xml (diff) The file was modified
tests/hwloc/xml/16intel64-manyVFs.xml (diff) The file was modified
tests/hwloc/x86/Intel-SandyBridge-2xXeon-E5-2650.output (diff) The file was modified
tests/hwloc/x86/AMD-K10-MagnyCours-2xOpteron-6164HE.output (diff) The file was modified
tests/hwloc/linux/2i386-2c-nohugepage.xml.output (diff) The file was modified
tests/hwloc/linux/16em64t-4s2c2t.xml.output (diff) The file was added tests/hwloc/xml/8intel64-fakeKNL-A2A-hybrid.rootattachednumas.v1tov2.options The file was modified
tests/hwloc/xml/8intel64-4n2t-memattrs.xml (diff) The file was modified
tests/hwloc/x86/AMD-15h-Piledriver-4xOpteron-6348.output (diff) The file was modified
tests/hwloc/xml/power8gpudistances.xml (diff) The file was modified
tests/hwloc/x86/AMD-19h-Zen3-2xEpyc-7763.output (diff) The file was modified
tests/hwloc/xml/32em64t-2n8c2t-pci-wholeio.xml (diff) The file was modified
utils/hwloc/test-hwloc-compress-dir.output.tar.gz (diff) The file was modified
tests/hwloc/x86/Hygon-Dhyana-32cores.output (diff) The file was modified
tests/hwloc/x86/Intel-CascadeLake-2xXeon6230.output (diff) The file was modified
tests/hwloc/x86/AMD-15h-Bulldozer-4xOpteron-6272.output (diff) The file was modified
tests/hwloc/x86/Intel-CPUID.1F-Qemu-2p3d3c2t.output (diff) The file was modified
tests/hwloc/x86/Intel-Broadwell-2xXeon-E5-2650Lv4.output (diff) The file was modified
tests/hwloc/x86/Zhaoxin-CentaurHauls-ZXD-4600.output (diff) The file was modified
tests/hwloc/x86/AMD-K10-Istanbul-8xOpteron-8439SE.output (diff) The file was modified
tests/hwloc/x86/Intel-CPUID.1A-1p2co2t.output (diff) The file was modified
tests/hwloc/x86/Intel-Westmere-2xXeon-X5650.output (diff) The file was modified
utils/lstopo/test-lstopo.output (diff) The file was modified
tests/hwloc/linux/40intel64-2g2n4c+pci.output (diff) The file was modified
tests/hwloc/xml/Makefile.am (diff) The file was modified
tests/hwloc/xml/8intel64-fakeKNL-A2A-hybrid.rootattachednumas.xml (diff) The file was modified
tests/hwloc/xml/28intel64-2p2g7c-CoD.nogroups.v1tov2.options (diff) The file was added tests/hwloc/xml/16amd64-4distances.v1tov2.options The file was modified
tests/hwloc/xml/96em64t-4n4d3ca2co-pci.xml (diff) The file was modified
tests/hwloc/x86/Intel-IvyBridge-12xXeon-E5-4620v2.output (diff) The file was modified
tests/hwloc/x86/Intel-KnightsCorner-XeonPhi-SE10P.output (diff) The file was modified
tests/hwloc/xml/32em64t-2n8c2t-pci-noio.xml (diff) The file was modified
tests/hwloc/xml/64intel64-3g2n+2n-irregulargroups+pci.xml (diff) The file was modified
tests/hwloc/xml/fakecpukinds.xml (diff) The file was modified
tests/hwloc/x86/Intel-Nehalem-2xXeon-X5550.output (diff) The file was added tests/hwloc/xml/64intel64-fakeKNL-SNC4-hybrid.v1tov2.options The file was added tests/hwloc/xml/28intel64-2p2g7c-CoDgroups.v1tov2.options The file was modified
tests/hwloc/x86/Intel-Penryn-4xXeon-X7460.output (diff) The file was modified
tests/hwloc/xml/24em64t-2n6c2t-pci.xml (diff) The file was modified
utils/hwloc/test-hwloc-annotate.output2 (diff) The file was modified
tests/hwloc/linux/32amd64-4s2n4c-cgroup2.xml.output (diff) The file was modified
tests/hwloc/x86/AMD-17h-Zen-2xEpyc-7451.output (diff) The file was modified
tests/hwloc/xml/16em64t-4s2c2t-offlines.xml (diff) The file was modified
tests/hwloc/x86/AMD-K8-SantaRosa-2xOpteron-2218.output (diff) The file was added tests/hwloc/xml/8ia64-2n2s2c+1n.v1tov2.options The file was modified
tests/hwloc/x86+linux/64amd64-4p2n4ca2co.topoextnuma.output (diff) The file was modified
tests/hwloc/linux/fakeheterocpunuma.output (diff) The file was modified
tests/hwloc/x86/Intel-Core-2xXeon-E5345.output (diff) The file was modified
tests/hwloc/xml/192em64t-24n8c2t.xml (diff) The file was modified
tests/hwloc/linux/2i386-2t-hugepagesizecount.xml.output (diff) The file was modified
tests/hwloc/linux/16em64t-4s2c2t-offlines.xml.output (diff) The file was modified
tests/hwloc/x86/AMD-K8-SledgeHammer-2xOpteron-250.output (diff) The file was modified
tests/hwloc/xml/64intel64-fakeKNL-SNC4-hybrid.xml (diff) The file was modified
utils/hwloc/test-hwloc-annotate.output (diff) The file was modified
tests/hwloc/xml/fakeheterodistances.xml (diff) The file was modified
tests/hwloc/linux/32em64t-2n8c+1mic.output (diff) The file was modified
tests/hwloc/linux/16amd64-8n2c-cpusets.xml.output (diff) The file was modified
tests/hwloc/x86/Zhaoxin-Shanghai-KaiSheng-ZXC+-FC1081.output (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
tests/hwloc/xml/32em64t-2n8c2t-pci-normalio.xml (diff) The file was modified
tests/hwloc/x86/Intel-Skylake-2xXeon6140.output (diff) The file was modified
tests/hwloc/x86+linux/64amd64-4p2n4ca2co.output (diff) The file was modified
tests/hwloc/xml/16-2gr2gr2n2c+misc.xml (diff) The file was modified
utils/hwloc/test-hwloc-annotate.input (diff) The file was modified
tests/hwloc/x86/Intel-Haswell-2xXeon-E5-2680v3.output (diff) The file was modified
utils/hwloc/test-hwloc-diffpatch.input1 (diff) The file was modified
utils/hwloc/test-build-custom-topology.output (diff) The file was added tests/hwloc/xml/2intel64-1n2c-numaroot.v1tov2.options The file was modified
utils/hwloc/test-hwloc-compress-dir.input.tar.gz (diff) The file was modified
tests/hwloc/xml/16em64t-4s2c2t.xml (diff) The file was modified
tests/hwloc/linux/allowed/test1.output (diff) The file was added tests/hwloc/xml/8em64t-2p2ca2co-nonodesets.v1tov2.options The file was modified
tests/hwloc/x86/Intel-KnightsLanding-XeonPhi-7210.output (diff) The file was modified
doc/images/devel09-pci.xml (diff) The file was modified
doc/images/hagrid.xml (diff) The file was modified
doc/images/emmett.xml (diff) The file was modified
doc/images/dudley.xml (diff) The file was modified
doc/images/hagrid.png (diff) The file was modified
doc/images/ppc64-with-smt.png (diff) The file was modified
doc/images/ppc64-without-smt.png (diff) The file was modified
doc/images/ppc64-full-with-smt.png (diff) The file was modified
doc/images/devel09-pci.png (diff) The file was modified
doc/images/emmett.png (diff) The file was modified
doc/images/HACKING (diff) The file was modified
doc/images/dudley.png (diff) The file was removed tests/hwloc/xml/8em64t-2p2ca2co-nonodesets.v1tov2.source The file was removed tests/hwloc/xml/16amd64-4distances.v1tov2.source The file was removed tests/hwloc/xml/8ia64-2n2s2c+1n.v1tov2.xml The file was removed tests/hwloc/xml/28intel64-2p2g7c-CoD.nogroups.v1tov2.source The file was removed tests/hwloc/xml/28intel64-2p2g7c-CoDgroups.v1tov2.xml The file was removed tests/hwloc/xml/28intel64-2p2g7c-CoDgroups.v1.xml The file was removed tests/hwloc/xml/2intel64-1n2c-numaroot.v1tov2.source The file was removed tests/hwloc/xml/8intel64-fakeKNL-A2A-hybrid.rootattachednumas.v1tov2.source The file was removed tests/hwloc/xml/2intel64-1n2c-numaroot.v1tov2.xml The file was removed tests/hwloc/xml/8em64t-2p2ca2co-nonodesets.v1tov2.xml The file was modified
tests/hwloc/xml/Makefile.am (diff) The file was removed tests/hwloc/xml/8em64t-2p2ca2co-nonodesets.v1.xml The file was removed tests/hwloc/xml/16amd64-4distances.v1.xml The file was removed tests/hwloc/xml/64intel64-fakeKNL-SNC4-hybrid.v1tov2.source The file was removed tests/hwloc/xml/16amd64-4distances.v1tov2.xml The file was removed tests/hwloc/xml/28intel64-2p2g7c-CoD.nogroups.v1tov2.xml The file was removed tests/hwloc/xml/2intel64-1n2c-numaroot.v1.xml The file was removed tests/hwloc/xml/28intel64-2p2g7c-CoDgroups.v1tov2.source The file was removed tests/hwloc/xml/64intel64-fakeKNL-SNC4-hybrid.v1tov2.xml The file was removed tests/hwloc/xml/28intel64-2p2g7c-CoD.nogroups.v1tov2.options The file was removed tests/hwloc/xml/8ia64-2n2s2c+1n.v1.xml The file was removed tests/hwloc/xml/8intel64-fakeKNL-A2A-hybrid.rootattachednumas.v1tov2.xml The file was removed tests/hwloc/xml/8ia64-2n2s2c+1n.v1tov2.source The file was modified
tests/netloc/data/plafrim.txz (diff) The file was modified
tests/netloc/data/plafrim2.txz (diff) The file was modified
tests/netloc/data/avakas.txz (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
NEWS (diff) The file was modified
include/private/xml.h (diff) The file was modified
utils/lstopo/test-lstopo.output (diff) The file was modified
utils/lstopo/test-lstopo.sh.in (diff) The file was modified
tests/hwloc/xml/Makefile.am (diff) The file was removed tests/hwloc/xml/16amd64-4distances.v2tov1.xml The file was removed tests/hwloc/xml/8intel64-fakeKNL-A2A-hybrid.rootattachednumas.v2tov1.options The file was removed tests/hwloc/xml/16amd64-4distances.v2tov1.source The file was removed tests/hwloc/xml/16amd64-4distances.v2tov1.options The file was removed tests/hwloc/xml/64intel64-fakeKNL-SNC4-hybrid.v2tov1.source The file was removed tests/hwloc/xml/64intel64-fakeKNL-SNC4-hybrid.v2tov1.xml The file was removed tests/hwloc/xml/64intel64-fakeKNL-SNC4-hybrid.v2tov1.options The file was removed tests/hwloc/xml/8intel64-fakeKNL-A2A-hybrid.rootattachednumas.v2tov1.source The file was removed tests/hwloc/xml/8intel64-fakeKNL-A2A-hybrid.rootattachednumas.v2tov1.xml The file was modified
hwloc/topology-xml-nolibxml.c (diff) The file was modified
include/hwloc/export.h (diff) The file was modified
include/private/xml.h (diff) The file was modified
NEWS (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
hwloc/topology-xml-libxml.c (diff) The file was modified
utils/hwloc/misc.h (diff) The file was removed hwloc/hwloc.dtd The file was modified
hwloc/topology-xml-libxml.c (diff) The file was modified
hwloc/Makefile.am (diff) The file was modified
hwloc/topology-xml.c (diff) The file was removed tests/hwloc/xml/8intel64-fakeKNL-A2A-hybrid.rootattachednumas.v1tov2.options The file was removed tests/hwloc/xml/64intel64-fakeKNL-SNC4-hybrid.v1tov2.options The file was removed tests/hwloc/xml/8em64t-2p2ca2co-nonodesets.v1tov2.options The file was removed tests/hwloc/xml/2intel64-1n2c-numaroot.v1tov2.options The file was removed tests/hwloc/xml/8ia64-2n2s2c+1n.v1tov2.options The file was removed tests/hwloc/xml/16amd64-4distances.v1tov2.options The file was removed tests/hwloc/xml/28intel64-2p2g7c-CoDgroups.v1tov2.options
Commit
aa0ef16c41279cdfd1a6f5d276cfdb2d46ee88bd
by Brice.Goglin darwin: don't forget to filter perflevel caches Forgotten in f7c9aa86c2f8ce8717ce4381357d3b98518d14ca Thanks to Michel Lesoinne for the report. Closes #564 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-darwin.c (diff)
Commit
cb82437df9335e0b036f5634f04ff17d5b06d363
by Brice.Goglin API: rename Block OS devices into Storage We've had more than Blocks in there, and Storage is more explicit, while Block came from the Linux specific block device type. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/lstopo/lstopo-draw.c (diff) The file was modified
NEWS (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
include/hwloc.h (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
tests/hwloc/linux/40intel64-4n10c+pci-conflicts.output (diff) The file was modified
tests/hwloc/linux/2pa-pcidomain32bits.output (diff) The file was modified
include/hwloc/deprecated.h (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
hwloc/traversal.c (diff) The file was modified
tests/hwloc/linux/fakeheteromemtiers.output (diff) The file was modified
tests/hwloc/xml/64intel64-3g2n+2n-irregulargroups+pci.console.output (diff) The file was modified
doc/Makefile.am (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
include/hwloc.h (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
hwloc/traversal.c (diff) The file was modified
tests/hwloc/linux/32em64t-2n8c+1mic.output (diff) The file was modified
tests/hwloc/linux/fakeheteromemtiers.output (diff) The file was modified
NEWS (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
tests/hwloc/xml/Makefile.am (diff) The file was added tests/hwloc/xml/cxlmem+dax.v2tov3.xml The file was added tests/hwloc/xml/cxlmem+dax.v2tov3.source The file was added tests/hwloc/xml/cxlmem+dax.v3tov2.options The file was added tests/hwloc/xml/cxlmem+dax.v3tov2.xml The file was added tests/hwloc/xml/cxlmem+dax.v3tov2.source
Commit
ca7f6459a5db7c099b2ac62516bdddf1010bac6e
by Brice.Goglin linux: force an ending \0 in our readlink readlink() doesn't add an ending \0. Add it in our wrapper to avoid surprises during future development/debug (this behavior was discovered while playing with CXL symlinks). This doesn't really fix anything yet because current callers: 1. (dax bus device symlink) were guaranteed to end before the end of the symlink. 2. (generic device symlink) manually added the ending \0 (without checking for buffer overflow, very unlikely since sysfs symlinks are much shorter than 256) Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-linux.c (diff)
Commit
4463e2e4b1cb8ba7ac4f5bde1abde5c8fbf202c4
by Brice.Goglin linux/dax: don't reuse block-specific sysfs parsing code There's not many attributes to get for DAX devices, don't bother factorizing with config flags, etc. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-linux.c (diff) The file was modified
hwloc/topology-linux.c (diff)
Commit
0780a3b9002c3567f3a8041a62f0d81bbddfd027
by bgoglin tests/backends.c: fix a failure when /sys isn't available on Linux on non-x86 Since 2.9, Linux discovery aborts if /sys isn't available. Usually it will fallback to the x86 backend... except on non-x86 platforms obviously where it would fallback to "noos" instead. The hwloc_backends.c test verified that the "Backend" info attribute was the same in the XML-loaded topology and the original topology, but it did not handle the case where that info attribute did not exist at all, which is the case when only "noos" is used. This is first reported by Simon South on https://issues.guix.gnu.org/61493 Thanks to Ludovic Courtes for forwarding to me. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/hwloc_backends.c (diff) The file was modified
utils/hwloc/hwloc-gather-cpuid.c (diff) The file was modified
utils/hwloc/hwloc-gather-cpuid.c (diff) The file was modified
utils/hwloc/hwloc-gather-cpuid.c (diff) The file was modified
.github/issue_template.md (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/topology-linux.c (diff)
Commit
194ae573f20c2e205540f501224b9debccb09026
by Brice.Goglin configure: update to latest ROCm include/lib paths OpenCL and ROCm SMI lib (and others we don't used) previously had their libs and headers under /opt/rocm.../{opencl,rocm_smi,...}/{lib/include} Since ROCm 5.2 (maybe earlier), they are in /opt/rocm.../{lib/include} and some warnings appear when using the old paths. Only use the old paths if the new ones aren't available. Closes #571 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
config/hwloc.m4 (diff)
Commit
d0e32feed5c98f032b493ad5354b58f9841d5ad8
by Brice.Goglin tests/xml: rewrite the framework Replace the old .source + .options + .env optional files with a single mandatory .test file that contains the source and target XML filenames, optional options, and optional environment variables. Now we iterate over *.test instead of *.xml and *.console outputs (with some .xml that could also be input). Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was added tests/hwloc/xml/fakeheterodistances.test The file was added tests/hwloc/xml/192em64t-12gr2n8c2t.test The file was added tests/hwloc/xml/16-2gr2gr2n2c+misc.console.test The file was removed tests/hwloc/xml/16-2gr2gr2n2c+misc.console.output The file was added tests/hwloc/xml/16intel64-manyVFs.console The file was added tests/hwloc/xml/cxlmem+dax.v2.xml The file was removed tests/hwloc/xml/cxlmem+dax.v3tov2.xml The file was added tests/hwloc/xml/16-2gr2gr2n2c+misc.console The file was removed tests/hwloc/xml/32em64t-2n8c2t-pci-noio.options The file was added tests/hwloc/xml/32em64t-2n8c2t-pci-noio.test The file was removed tests/hwloc/xml/64intel64-3g2n+2n-irregulargroups+pci.console.options The file was removed tests/hwloc/xml/32em64t-2n8c2t-pci-wholeio.options The file was modified
tests/hwloc/xml/test-topology.sh.in (diff) The file was removed tests/hwloc/xml/32em64t-2n8c2t-pci-noio.source The file was added tests/hwloc/xml/16em64t-4s2c2t-offlines.test The file was added tests/hwloc/xml/power8gpudistances.test The file was removed tests/hwloc/xml/16intel64-manyVFs.console.source The file was removed tests/hwloc/xml/16amd64-4distances.console.source The file was removed tests/hwloc/xml/cxlmem+dax.v3tov2.source The file was added tests/hwloc/xml/16amd64-4distances.console The file was removed tests/hwloc/xml/64intel64-3g2n+2n-irregulargroups+pci.console.env The file was added tests/hwloc/xml/24em64t-2n6c2t-pci.test The file was added tests/hwloc/xml/32em64t-2n8c2t-pci-wholeio.test The file was added tests/hwloc/xml/32em64t-2n8c2t-pci-normalio.test The file was added tests/hwloc/xml/16intel64-manyVFs.console.test The file was removed tests/hwloc/xml/cxlmem+dax.v3tov2.options The file was added tests/hwloc/xml/16intel64-manyVFs.nocollapse.console The file was added tests/hwloc/xml/8intel64-fakeKNL-A2A-hybrid.rootattachednumas.test The file was modified
tests/hwloc/xml/Makefile.am (diff) The file was added tests/hwloc/xml/cxlmem+dax.v2tov3.test The file was removed tests/hwloc/xml/16intel64-manyVFs.console.output The file was added tests/hwloc/xml/16intel64-manyVFs.nocollapse.console.test The file was removed tests/hwloc/xml/64intel64-3g2n+2n-irregulargroups+pci.console.output The file was removed tests/hwloc/xml/16intel64-manyVFs.console.options The file was removed tests/hwloc/xml/16amd64-4distances.console.env The file was removed tests/hwloc/xml/cxlmem+dax.v2tov3.xml The file was added tests/hwloc/xml/64intel64-3g2n+2n-irregulargroups+pci.disallowed.console.test The file was added tests/hwloc/xml/64intel64-3g2n+2n-irregulargroups+pci.disallowed.test The file was removed tests/hwloc/xml/16em64t-4s2c2t.console.output The file was removed tests/hwloc/xml/16intel64-manyVFs.console.nocollapse.options The file was added tests/hwloc/xml/16intel64-manyVFs.test The file was added tests/hwloc/xml/16amd64-4distances.test The file was added tests/hwloc/xml/192em64t-24n8c2t.test The file was added tests/hwloc/xml/16amd64-4distances.console.test The file was removed tests/hwloc/xml/16em64t-4s2c2t.console.source The file was removed tests/hwloc/xml/32em64t-2n8c2t-pci-normalio.source The file was removed tests/hwloc/xml/64intel64-3g2n+2n-irregulargroups+pci.console.source The file was added tests/hwloc/xml/16amd64-8n2c-cpusets.test The file was removed tests/hwloc/xml/16em64t-4s2c2t.console.options The file was added tests/hwloc/xml/cxlmem+dax.v3.xml The file was added tests/hwloc/xml/fakecpukinds.test The file was added tests/hwloc/xml/96em64t-4n4d3ca2co-pci.test The file was added tests/hwloc/xml/cxlmem+dax.v3tov2.test The file was removed tests/hwloc/xml/16intel64-manyVFs.console.nocollapse.output The file was added tests/hwloc/xml/16em64t-4s2c2t.test The file was removed tests/hwloc/xml/cxlmem+dax.v2tov3.source The file was removed tests/hwloc/xml/16-2gr2gr2n2c+misc.console.source The file was removed tests/hwloc/xml/16amd64-4distances.console.output The file was added tests/hwloc/xml/16em64t-4s2c2t.console The file was removed tests/hwloc/xml/16amd64-4distances.console.options The file was added tests/hwloc/xml/8intel64-4n2t-memattrs.test The file was removed tests/hwloc/xml/64intel64-3g2n+2n-irregulargroups+pci.options The file was removed tests/hwloc/xml/16intel64-manyVFs.console.nocollapse.source The file was added tests/hwloc/xml/64intel64-3g2n+2n-irregulargroups+pci.console The file was removed tests/hwloc/xml/16-2gr2gr2n2c+misc.console.options The file was added tests/hwloc/xml/16-2gr2gr2n2c+misc.test The file was added tests/hwloc/xml/16em64t-4s2c2t.console.test The file was added tests/hwloc/xml/64intel64-fakeKNL-SNC4-hybrid.test
Commit
5a51da1c145accb24f5596e7f838b1789d713d9e
by Brice.Goglin tests/xml: remove unneeded environment variables v1.x exported floating point numbers for distances, hence failing if the current locale didn't match the locale of the expected output. v2+ export integers instead, no locale difference there. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/xml/16amd64-4distances.console.test (diff) The file was modified
tests/hwloc/xml/64intel64-3g2n+2n-irregulargroups+pci.disallowed.console.test (diff)
Commit
4c0aa9d1ac5a3eeda6b7aceeacb4f65c8901563f
by Brice.Goglin tests/xml: remove 8intel64-fakeKNL-A2A-hybrid.rootattachednumas This test was explicitly designed to test the v1->v2 xml import where KNL needed special care. Doesn't matter anymore in v3. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was removed tests/hwloc/xml/8intel64-fakeKNL-A2A-hybrid.rootattachednumas.test The file was removed tests/hwloc/xml/8intel64-fakeKNL-A2A-hybrid.rootattachednumas.xml The file was modified
tests/hwloc/xml/Makefile.am (diff) The file was modified
tests/hwloc/linux/Makefile.am (diff) The file was modified
tests/hwloc/x86/Makefile.am (diff) The file was modified
tests/hwloc/xml/Makefile.am (diff) The file was modified
tests/hwloc/linux/allowed/Makefile.am (diff) The file was modified
tests/hwloc/x86+linux/Makefile.am (diff) The file was modified
tests/hwloc/linux/allowed/test-topology.sh.in (diff) The file was modified
tests/hwloc/x86+linux/test-topology.sh.in (diff) The file was modified
tests/hwloc/linux/test-topology.sh.in (diff)
Commit
bdc440633d205e03460cfe4bd3a82faa99b52e70
by bgoglin Update .gitignore to include cross-compile libtool artifacts When configured with a non-native --host value, the libtool file will be prefixed by the host, which the current .gitignore misses. Signed-off-by: Nicholas Corgan <nicholas.corgan@skylarkwireless.com> The file was modified
.gitignore (diff) The file was removed tests/hwloc/xml/64intel64-3g2n+2n-irregulargroups+pci.disallowed.test The file was modified
tests/hwloc/xml/Makefile.am (diff) The file was added tests/hwloc/xml/irregulargroups-disallowed.console The file was added tests/hwloc/xml/irregulargroups-disallowed.console.test The file was added tests/hwloc/xml/irregulargroups-disallowed.test The file was removed tests/hwloc/xml/64intel64-3g2n+2n-irregulargroups+pci.console The file was removed tests/hwloc/xml/64intel64-3g2n+2n-irregulargroups+pci.disallowed.console.test The file was added tests/hwloc/xml/irregulargroups-disallowed.xml The file was removed tests/hwloc/xml/64intel64-3g2n+2n-irregulargroups+pci.xml
Commit
d66768c32d16cdbf5b62a7e0f0d069250ed8a72d
by Brice.Goglin tests/xml: improve the format of .test files Prefix lines with source/target/options/env: instead of requiring a specific line order. foo.test should contain: - source: input XML filename (must be in same directory as foo.test) - target: expected output filename (must be in same directory as foo.test, ends with either .console or .xml) - options: lstopo options, if any - env: environment variable lines to be sourced, if any This also allows comments to describe the test in other lines. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/xml/16em64t-4s2c2t.test (diff) The file was modified
tests/hwloc/xml/16intel64-manyVFs.console.test (diff) The file was modified
tests/hwloc/xml/32em64t-2n8c2t-pci-normalio.test (diff) The file was modified
tests/hwloc/xml/16-2gr2gr2n2c+misc.console.test (diff) The file was modified
tests/hwloc/xml/16-2gr2gr2n2c+misc.test (diff) The file was modified
tests/hwloc/xml/8intel64-4n2t-memattrs.test (diff) The file was modified
tests/hwloc/xml/test-topology.sh.in (diff) The file was modified
tests/hwloc/xml/16em64t-4s2c2t.console.test (diff) The file was modified
tests/hwloc/xml/32em64t-2n8c2t-pci-wholeio.test (diff) The file was modified
tests/hwloc/xml/fakeheterodistances.test (diff) The file was modified
tests/hwloc/xml/16amd64-8n2c-cpusets.test (diff) The file was modified
tests/hwloc/xml/16em64t-4s2c2t-offlines.test (diff) The file was modified
tests/hwloc/xml/16intel64-manyVFs.nocollapse.console.test (diff) The file was modified
tests/hwloc/xml/cxlmem+dax.v2tov3.test (diff) The file was modified
tests/hwloc/xml/16amd64-4distances.console.test (diff) The file was modified
tests/hwloc/xml/16amd64-4distances.test (diff) The file was modified
tests/hwloc/xml/cxlmem+dax.v3tov2.test (diff) The file was modified
tests/hwloc/xml/64intel64-fakeKNL-SNC4-hybrid.test (diff) The file was modified
tests/hwloc/xml/fakecpukinds.test (diff) The file was modified
tests/hwloc/xml/192em64t-12gr2n8c2t.test (diff) The file was modified
tests/hwloc/xml/16intel64-manyVFs.test (diff) The file was modified
tests/hwloc/xml/192em64t-24n8c2t.test (diff) The file was modified
tests/hwloc/xml/irregulargroups-disallowed.test (diff) The file was modified
tests/hwloc/xml/96em64t-4n4d3ca2co-pci.test (diff) The file was modified
tests/hwloc/xml/24em64t-2n6c2t-pci.test (diff) The file was modified
tests/hwloc/xml/irregulargroups-disallowed.console.test (diff) The file was modified
tests/hwloc/xml/32em64t-2n8c2t-pci-noio.test (diff) The file was modified
tests/hwloc/xml/power8gpudistances.test (diff)
Commit
d6fda2d87f980ac57ae61a00865eb487f6eea8dc
by Brice.Goglin tests/linux: remove some of the many tests that aren't useful anymore * 4em64t-2c2t-ignore-reorder Output is almost empty and there's no actual reordering anymore. * console output of 32amd64-4s2n4c-cgroup2.output XML output is enough (and better) for testing Cgroup v2 support. * 2s390-2c very basic S/390, we have a much bigger one with "books". * 4ia64-4s * 8ia64-4s2c Old ia64 that are less interesting than others (with multiple sockets, NUMA and/or SMT). ia64 is pretty much dead anyway, we still have too many after removing those. * 8em64t-4c2t * 8amd64-4n2c * 2amd64-2n Old and basic x86_64 outputs, we have many others and much better ones. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/linux/Makefile.am (diff) The file was removed tests/hwloc/linux/8amd64-4n2c.output The file was removed tests/hwloc/linux/2amd64-2n.output The file was removed tests/hwloc/linux/32amd64-4s2n4c-cgroup2.output The file was removed tests/hwloc/linux/4ia64-4s.output The file was removed tests/hwloc/linux/8em64t-4c2t.output The file was removed tests/hwloc/linux/2s390-2c.output The file was removed tests/hwloc/linux/2amd64-2n.tar.bz2 The file was removed tests/hwloc/linux/8ia64-4s2c.output The file was removed tests/hwloc/linux/8amd64-4n2c.tar.bz2 The file was removed tests/hwloc/linux/4em64t-2c2t-ignore-reorder.options The file was removed tests/hwloc/linux/4em64t-2c2t-ignore-reorder.tar.bz2 The file was removed tests/hwloc/linux/8ia64-4s2c.tar.bz2 The file was removed tests/hwloc/linux/2s390-2c.tar.bz2 The file was removed tests/hwloc/linux/4ia64-4s.tar.bz2 The file was removed tests/hwloc/linux/8em64t-4c2t.tar.bz2 The file was removed tests/hwloc/linux/4em64t-2c2t-ignore-reorder.output
Commit
9ce01f32c226651c28897d4fd642815fa352f44f
by Brice.Goglin tests/linux: rewrite the framework with .test files Replace the old .source + .options + .exclude + .env optional files with a single mandatory .test file that contains everything. foo.test should contain: - source: tarball filename (must be in same directory as foo.test) - target: expected output filename (must be in same directory as foo.test, ends with either .console or .xml) - options: lstopo options, if any - exclude: exclude file to be passed to tar --exclude-from=, if any - env: environment variable lines to be sourced, if any Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was added tests/hwloc/linux/96em64t-4no4pa3ca2co.console The file was removed tests/hwloc/linux/16em64t-4s2ca2c-cpusetreorder.output The file was removed tests/hwloc/linux/16em64t-4s2c2t_merge.options The file was added tests/hwloc/linux/16ia64-8n2s.console The file was added tests/hwloc/linux/16em64t-4s2c2t.test The file was removed tests/hwloc/linux/16em64t-4s2c2t-offlines.xml.options The file was removed tests/hwloc/linux/16em64t-4s2c2t.xml.output The file was removed tests/hwloc/linux/2i386-2t-hugepagesizecount.xml.options The file was added tests/hwloc/linux/32intel64-2p8co2t+8ve.test The file was removed tests/hwloc/linux/40intel64-4n10c+pci-conflicts.output The file was added tests/hwloc/linux/128arm-2pa2n8cluster4co.console The file was removed tests/hwloc/linux/8em64t-2s2ca2c.output The file was added tests/hwloc/linux/20s390-2g6s4c.console The file was added tests/hwloc/linux/8ia64-2n2s2c.test The file was removed tests/hwloc/linux/16em64t-4s2ca2c-cpusetreorder-merge.source The file was added tests/hwloc/linux/nvidiagpunumanodes.test The file was removed tests/hwloc/linux/16amd64-8n2c-cpusets_noadmin.output The file was added tests/hwloc/linux/32em64t-2n8c+dax+nvme+mic+dimms.test The file was added tests/hwloc/linux/8em64t-2s2ca2c.test The file was added tests/hwloc/linux/8em64t-2s4c-asymcaches.console The file was added tests/hwloc/linux/2i386-2c-nohugepage.test The file was added tests/hwloc/linux/16amd64-8n2c-cpusets.test The file was added tests/hwloc/linux/64intel64-fakeKNL-SNC4-hybrid.test The file was added tests/hwloc/linux/fakememinitiators-1np2c+1npp+gi.test The file was added tests/hwloc/linux/256ia64-64n2s2c.test The file was added tests/hwloc/linux/64intel64-fakeKNL-SNC4-hybrid-msc.console The file was modified
utils/lstopo/test-lstopo.sh.in (diff) The file was removed tests/hwloc/linux/16em64t-4s2c2t_ncaches.options The file was added tests/hwloc/linux/40intel64-2g2n4c+pcilocality.test The file was added tests/hwloc/linux/nvidiagpunumanodes-kept.console The file was removed tests/hwloc/linux/16em64t-4s2c2t.xml.options The file was removed tests/hwloc/linux/fakeheterocpunuma.options The file was added tests/hwloc/linux/8ia64-2s2c2t.console The file was added tests/hwloc/linux/8em64t-2s2ca2c-buggynuma.test The file was removed tests/hwloc/linux/256ppc-8n8s4t.output The file was added tests/hwloc/linux/offline-cpu0-node0.console The file was added tests/hwloc/linux/48amd64-4pa2n6c-sparse.test The file was removed tests/hwloc/linux/fakeheterocpunuma.env The file was removed tests/hwloc/linux/48amd64-4d2n6c-sparse.tar.bz2 The file was added tests/hwloc/linux/128ia64-17n4s2c.test The file was removed tests/hwloc/linux/64amd64-4s2n4ca2co.output The file was removed tests/hwloc/linux/16em64t-4s2c2t-offlines.xml.source The file was added tests/hwloc/linux/16amd64-4n4c-cgroup-distance-merge.test The file was removed tests/hwloc/linux/8em64t-2s2ca2c-buggynuma.output The file was removed tests/hwloc/linux/32amd64-4s2n4c-cgroup2.xml.source The file was removed tests/hwloc/linux/fakeheteromemtiers.output The file was added tests/hwloc/linux/fakeheteromemtiers.console The file was added tests/hwloc/linux/fakecpuid1f-64intel64-2p4d2n2c2t.console The file was removed tests/hwloc/linux/96em64t-4n4d3ca2co.tar.bz2 The file was added tests/hwloc/linux/fakeheterocpunuma.xml The file was added tests/hwloc/linux/2i386-2c-nohugepage.xml The file was added tests/hwloc/linux/40intel64-2g2n4c+pcilocality.xml The file was removed tests/hwloc/linux/16amd64-8n2c-cpusets.xml.options The file was added tests/hwloc/linux/64intel64-fakeKNL-SNC4-hybrid.console The file was removed tests/hwloc/linux/40intel64-2g2n4c+pci.env The file was added tests/hwloc/linux/48amd64-4pa2n6c-sparse.tar.bz2 The file was added tests/hwloc/linux/64fake-4n2s2ca2c2t.console The file was removed tests/hwloc/linux/8ia64-2s2c2t.output The file was removed tests/hwloc/linux/16amd64-8n2c-cpusets_noadmin.source The file was added tests/hwloc/linux/16em64t-4s2c2t-offlines.console The file was removed tests/hwloc/linux/2ps3-2t.output The file was removed tests/hwloc/linux/16amd64-4n4c-cgroup-distance-merge.output The file was removed tests/hwloc/linux/32em64t-2n8c+1mic.tar.bz2 The file was removed tests/hwloc/linux/40intel64-2g2n4c+pci.tar.bz2 The file was removed tests/hwloc/linux/fakecpuid1f-64intel64-2p4d2n2c2t.output The file was added tests/hwloc/linux/16em64t-4s2c2t-nocaches.test The file was removed tests/hwloc/linux/2i386-2t-hugepagesizecount.xml.source The file was removed tests/hwloc/linux/64intel64-fakeKNL-SNC4-hybrid-msc.env The file was removed tests/hwloc/linux/32em64t-2n8c+1mic.output The file was removed tests/hwloc/linux/16em64t-4s2c2t_merge.source The file was removed tests/hwloc/linux/16em64t-4s2c2t_ncaches.source The file was removed tests/hwloc/linux/offline-cpu0-node0.output The file was added tests/hwloc/linux/8ia64-2n2s2c.console The file was added tests/hwloc/linux/32amd64-4s2n4c-cgroup2.xml The file was added tests/hwloc/linux/96em64t-4no4pa3ca2co.test The file was removed tests/hwloc/linux/16em64t-4s2c2t_ncaches.output The file was added tests/hwloc/linux/64intel64-fakeKNL-A2A-cache.console The file was modified
utils/hwloc/test-hwloc-info.sh.in (diff) The file was modified
tests/hwloc/linux/test-topology.sh.in (diff) The file was added tests/hwloc/linux/2pa-pcidomain32bits.console The file was added tests/hwloc/linux/32intel64-2p8co2t+8ve.console The file was added tests/hwloc/linux/8em64t-2s2ca2c-buggynuma.console The file was added tests/hwloc/linux/32em64t-2n8c+dax+nvme+mic+dimms.xml The file was added tests/hwloc/linux/8ia64-2s2c2t.test The file was added tests/hwloc/linux/256ia64-64n2s2c.console The file was removed tests/hwloc/linux/64fake-4n2s2ca2c2t.output The file was added tests/hwloc/linux/16em64t-4s2ca2c-cpusetreorder.console The file was added tests/hwloc/linux/40intel64-2g2n4c+pcilocality.tar.bz2 The file was added tests/hwloc/linux/16ia64-8n2s.test The file was added tests/hwloc/linux/32em64t-2n8c+dax+nvme+mic+dimms.tar.bz2 The file was removed tests/hwloc/linux/40intel64-2g2n4c+pci.options The file was added tests/hwloc/linux/64amd64-4s2n4ca2co.console The file was added tests/hwloc/linux/2pa-pcidomain32bits.test The file was added tests/hwloc/linux/16em64t-4s2c2t-nocaches.console The file was added tests/hwloc/linux/128arm-2pa2n8cluster4co.test The file was added tests/hwloc/linux/fakememinitiators-1np2c+1npp+gi.console The file was removed tests/hwloc/linux/32em64t-2n8c+1mic.options The file was removed tests/hwloc/linux/32amd64-4s2n4c-cgroup2.xml.output The file was added tests/hwloc/linux/64intel64-fakeKNL-A2A-cache.test The file was added tests/hwloc/linux/96em64t-4no4pa3ca2co.tar.bz2 The file was removed tests/hwloc/linux/16em64t-4s2c2t.xml.source The file was removed tests/hwloc/linux/2i386-2c-nohugepage.xml.source The file was removed tests/hwloc/linux/40intel64-4n10c+pci-conflicts.env The file was removed tests/hwloc/linux/4fake-4gr1nu1pu.output The file was added tests/hwloc/linux/16em64t-4s2ca2c-cpusetreorder.test The file was added tests/hwloc/linux/20s390-2g6s4c.test The file was added tests/hwloc/linux/2i386-2t-hugepagesizecount.xml The file was added tests/hwloc/linux/nvidiagpunumanodes.console The file was removed tests/hwloc/linux/16em64t-4s2c2t-offlines.output The file was added tests/hwloc/linux/fakeheterocpunuma.test The file was removed tests/hwloc/linux/2i386-2c-nohugepage.xml.options The file was added tests/hwloc/linux/64intel64-fakeKNL-SNC4-hybrid-msc.test The file was added tests/hwloc/linux/nvidiagpunumanodes-kept.test The file was removed tests/hwloc/linux/nvidiagpunumanodes.kept.env The file was added tests/hwloc/linux/16em64t-4s2c2t-offlines-xml.xml The file was removed tests/hwloc/linux/96em64t-4n4d3ca2co.output The file was removed tests/hwloc/linux/48amd64-4d2n6c-sparse.output The file was removed tests/hwloc/linux/64intel64-fakeKNL-A2A-cache.output The file was removed tests/hwloc/linux/16em64t-4s2c2t-offlines.xml.output The file was added tests/hwloc/linux/256ppc-8n8s4t.test The file was removed tests/hwloc/linux/2i386-2t-hugepagesizecount.xml.output The file was modified
utils/hwloc/test-hwloc-calc.sh.in (diff) The file was added tests/hwloc/linux/16amd64-8n2c-cpusets-disallowed.console The file was removed tests/hwloc/linux/2pa-pcidomain32bits.output The file was removed tests/hwloc/linux/128arm-2pa2n8cluster4co.output The file was removed tests/hwloc/linux/64intel64-fakeKNL-SNC4-hybrid-msc.output The file was added tests/hwloc/linux/8em64t-2s4c-asymcaches.tar.bz2 The file was added tests/hwloc/linux/16em64t-4s2c2t-merge.console The file was added tests/hwloc/linux/2ps3-2t.test The file was added tests/hwloc/linux/16amd64-8n2c.console The file was removed tests/hwloc/linux/16ia64-8n2s.output The file was removed tests/hwloc/linux/16amd64-8n2c-cpusets.output The file was removed tests/hwloc/linux/nvidiagpunumanodes.kept.source The file was added tests/hwloc/linux/128ia64-17n4s2c.console The file was added tests/hwloc/linux/16em64t-4s2c2t-offlines.test The file was added tests/hwloc/linux/16amd64-8n2c-cpusets-xml.xml The file was removed tests/hwloc/linux/16em64t-4s2c2t_merge.output The file was added tests/hwloc/linux/8em64t-2s2ca2c.console The file was added tests/hwloc/linux/16amd64-8n2c-cpusets-disallowed.test The file was removed tests/hwloc/linux/fakememinitiators-1np2c+1npp+gi.output The file was removed tests/hwloc/linux/16em64t-4s2ca2c-cpusetreorder-merge.options The file was added tests/hwloc/linux/4fake-4gr1nu1pu.console The file was modified
tests/hwloc/linux/Makefile.am (diff) The file was removed tests/hwloc/linux/64intel64-fakeKNL-SNC4-hybrid.output The file was removed tests/hwloc/linux/16amd64-8n2c-cpusets.xml.source The file was removed tests/hwloc/linux/8em64t-2s4c-heterogeneous.tar.bz2 The file was removed tests/hwloc/linux/64intel64-fakeKNL-SNC4-hybrid-msc.source The file was removed tests/hwloc/linux/32intel64-2p8co2t+8ve.output The file was removed tests/hwloc/linux/2arm-2c.output The file was removed tests/hwloc/linux/40intel64-2g2n4c+pci.output The file was added tests/hwloc/linux/16em64t-4s2c2t-merge.test The file was removed tests/hwloc/linux/8ia64-2n2s2c.output The file was added tests/hwloc/linux/offline-cpu0-node0.test The file was removed tests/hwloc/linux/nvidiagpunumanodes.kept.output The file was added tests/hwloc/linux/32amd64-4s2n4c-cgroup2.test The file was removed tests/hwloc/linux/20s390-2g6s4c.output The file was removed tests/hwloc/linux/128ia64-17n4s2c.output The file was removed tests/hwloc/linux/2i386-2c-nohugepage.xml.output The file was removed tests/hwloc/linux/16amd64-8n2c-cpusets_noadmin.options The file was removed tests/hwloc/linux/16amd64-8n2c-cpusets.xml.output The file was added tests/hwloc/linux/16em64t-4s2c2t.console The file was added tests/hwloc/linux/2i386-2t-hugepagesizecount.test The file was removed tests/hwloc/linux/16em64t-4s2ca2c-cpusetreorder-merge.output The file was added tests/hwloc/linux/2arm-2c.test The file was added tests/hwloc/linux/16amd64-8n2c-cpusets-xml.test The file was added tests/hwloc/linux/2ps3-2t.console The file was removed tests/hwloc/linux/fakeheterocpunuma.output The file was added tests/hwloc/linux/fakecpuid1f-64intel64-2p4d2n2c2t.test The file was removed tests/hwloc/linux/8em64t-2s4c-heterogeneous.output The file was removed tests/hwloc/linux/nvidiagpunumanodes.output The file was added tests/hwloc/linux/16amd64-4n4c-cgroup-distance-merge.console The file was added tests/hwloc/linux/fakeheteromemtiers.test The file was added tests/hwloc/linux/16amd64-8n2c-cpusets.console The file was added tests/hwloc/linux/64fake-4n2s2ca2c2t.test The file was removed tests/hwloc/linux/16amd64-4n4c-cgroup-distance-merge.options The file was removed tests/hwloc/linux/32amd64-4s2n4c-cgroup2.xml.options The file was removed tests/hwloc/linux/16em64t-4s2c2t.output The file was added tests/hwloc/linux/40intel64-4n10c+pci-conflicts.test The file was added tests/hwloc/linux/16em64t-4s2c2t-offlines-xml.test The file was added tests/hwloc/linux/16amd64-8n2c.test The file was added tests/hwloc/linux/16em64t-4s2c2t-xml.xml The file was added tests/hwloc/linux/48amd64-4pa2n6c-sparse.console The file was added tests/hwloc/linux/16em64t-4s2ca2c-cpusetreorder-merge.console The file was added tests/hwloc/linux/2arm-2c.console The file was removed tests/hwloc/linux/16amd64-8n2c.output The file was added tests/hwloc/linux/8em64t-2s4c-asymcaches.test The file was removed tests/hwloc/linux/256ia64-64n2s2c.output The file was added tests/hwloc/linux/64amd64-4s2n4ca2co.test The file was added tests/hwloc/linux/16em64t-4s2ca2c-cpusetreorder-merge.test The file was added tests/hwloc/linux/256ppc-8n8s4t.console The file was added tests/hwloc/linux/4fake-4gr1nu1pu.test The file was added tests/hwloc/linux/16em64t-4s2c2t-xml.test The file was added tests/hwloc/linux/40intel64-4n10c+pci-conflicts.console
Commit
a34aa2003ee133fe7d7b0c9ba70ced3d5db34992
by Brice.Goglin tests/linux/allowed: rewrite the framework with .test files Replace the old .synthetic + .fsroot + .options + .env optional files with a single mandatory .test file that contains everything. foo.test should contain: - synthetic: synthetic topology description - fsroot: tarball filename for applying local restrictions (must be in same directory as foo.test) - target: expected output filename (must be in same directory as foo.test, ends with either .console or .xml) - options: lstopo options, if any - env: environment variable lines to be sourced, if any By the way, remove gp_index/id just like we do in other tests. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was removed tests/hwloc/linux/allowed/test1.output The file was modified
tests/hwloc/linux/allowed/Makefile.am (diff) The file was modified
tests/hwloc/linux/allowed/test-topology.sh.in (diff) The file was added tests/hwloc/linux/allowed/test1.test The file was removed tests/hwloc/linux/allowed/test1.options The file was removed tests/hwloc/linux/allowed/test1.synthetic The file was added tests/hwloc/linux/allowed/test1.xml
Commit
3f53f9cd8946d22918170250490b1dc05af146e3
by Brice.Goglin tests/x86+linux: rewrite the framework with .test files Replace the old .source + .options + .env optional files with a single mandatory .test file that contains everything. foo.test should contain: - source: tarball filename (must be in same directory as foo.test) - target: expected output filename (must be in same directory as foo.test, ends with either .console or .xml) - options: lstopo options, if any - env: environment variable lines to be sourced, if any Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was removed tests/hwloc/x86+linux/64amd64-4p2n4ca2co.topoextnuma.env The file was removed tests/hwloc/x86+linux/5intel64-hybrid-lakefield.output The file was removed tests/hwloc/x86+linux/5intel64-hybrid-lakefield.env The file was removed tests/hwloc/x86+linux/5intel64-hybrid-lakefield.options The file was added tests/hwloc/x86+linux/64amd64-4p2n4ca2co+topoextnuma.test The file was removed tests/hwloc/x86+linux/64amd64-4p2n4ca2co.topoextnuma.output The file was removed tests/hwloc/x86+linux/64amd64-4p2n4ca2co.output The file was modified
tests/hwloc/x86+linux/Makefile.am (diff) The file was modified
tests/hwloc/x86+linux/test-topology.sh.in (diff) The file was added tests/hwloc/x86+linux/5intel64-hybrid-lakefield.console The file was added tests/hwloc/x86+linux/64amd64-4p2n4ca2co.test The file was added tests/hwloc/x86+linux/5intel64-hybrid-lakefield.test The file was added tests/hwloc/x86+linux/64amd64-4p2n4ca2co.xml The file was removed tests/hwloc/x86+linux/64amd64-4p2n4ca2co.topoextnuma.source
Commit
1216179792ed636de28b2b791e1f58c36ad05ed9
by Brice.Goglin tests/x86: rewrite the framework with .test files Replace the old .options + .env optional files with a single mandatory .test file that contains everything. foo.test should contain: - source: tarball filename (must be in same directory as foo.test) - target: expected output filename (must be in same directory as foo.test, ends with either .console or .xml) - options: lstopo options, if any - env: environment variable lines to be sourced, if any Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was removed tests/hwloc/x86/AMD-K8-SantaRosa-2xOpteron-2218.output The file was removed tests/hwloc/x86/Intel-IvyBridge-12xXeon-E5-4620v2.output The file was added tests/hwloc/x86/Intel-Nehalem-2xXeon-X5550.xml The file was removed tests/hwloc/x86/AMD-15h-Bulldozer-4xOpteron-6272.output The file was removed tests/hwloc/x86/AMD-K10-MagnyCours-2xOpteron-6164HE.output The file was added tests/hwloc/x86/Intel-SandyBridge-2xXeon-E5-2650.test The file was added tests/hwloc/x86/Intel-Skylake-2xXeon6140.xml The file was modified
tests/hwloc/x86/test-topology.sh.in (diff) The file was added tests/hwloc/x86/Intel-KnightsCorner-XeonPhi-SE10P.test The file was added tests/hwloc/x86/Intel-Penryn-4xXeon-X7460.xml The file was added tests/hwloc/x86/AMD-K8-SantaRosa-2xOpteron-2218.test The file was modified
utils/lstopo/test-lstopo.sh.in (diff) The file was added tests/hwloc/x86/Intel-KnightsCorner-XeonPhi-SE10P.xml The file was removed tests/hwloc/x86/Intel-CascadeLake-2xXeon6230.output The file was removed tests/hwloc/x86/Intel-Skylake-2xXeon6140.output The file was removed tests/hwloc/x86/Intel-Broadwell-2xXeon-E5-2650Lv4.output The file was added tests/hwloc/x86/Intel-Broadwell-2xXeon-E5-2650Lv4.xml The file was added tests/hwloc/x86/AMD-K8-SledgeHammer-2xOpteron-250.test The file was removed tests/hwloc/x86/Intel-Nehalem-2xXeon-X5550.output The file was added tests/hwloc/x86/Intel-CascadeLake-2xXeon6230.xml The file was added tests/hwloc/x86/AMD-K10-MagnyCours-2xOpteron-6164HE.xml The file was added tests/hwloc/x86/AMD-K10-MagnyCours-2xOpteron-6164HE.test The file was added tests/hwloc/x86/Hygon-Dhyana-32cores.test The file was added tests/hwloc/x86/AMD-19h-Zen3-2xEpyc-7763.xml The file was removed tests/hwloc/x86/Intel-Westmere-2xXeon-X5650.output The file was added tests/hwloc/x86/AMD-15h-Piledriver-4xOpteron-6348.xml The file was removed tests/hwloc/x86/Intel-KnightsCorner-XeonPhi-SE10P.output The file was removed tests/hwloc/x86/AMD-17h-Zen-2xEpyc-7451.output The file was added tests/hwloc/x86/Intel-CPUID.1F-Qemu-2p3d3c2t.test The file was removed tests/hwloc/x86/Hygon-Dhyana-32cores.output The file was added tests/hwloc/x86/AMD-K8-SantaRosa-2xOpteron-2218.xml The file was added tests/hwloc/x86/Zhaoxin-Shanghai-KaiSheng-ZXC+-FC1081.test The file was added tests/hwloc/x86/Intel-IvyBridge-12xXeon-E5-4620v2.xml The file was removed tests/hwloc/x86/Intel-Penryn-4xXeon-X7460.output The file was removed tests/hwloc/x86/AMD-K10-Istanbul-8xOpteron-8439SE.output The file was added tests/hwloc/x86/Intel-CascadeLake-2xXeon6230.test The file was removed tests/hwloc/x86/Intel-CPUID.1F-Qemu-2p3d3c2t.output The file was added tests/hwloc/x86/Intel-Haswell-2xXeon-E5-2680v3.test The file was added tests/hwloc/x86/Intel-KnightsLanding-XeonPhi-7210.xml The file was removed tests/hwloc/x86/Intel-SandyBridge-2xXeon-E5-2650.output The file was added tests/hwloc/x86/Intel-Skylake-2xXeon6140.test The file was added tests/hwloc/x86/AMD-17h-Zen-2xEpyc-7451.test The file was added tests/hwloc/x86/Zhaoxin-Shanghai-KaiSheng-ZXC+-FC1081.xml The file was removed tests/hwloc/x86/Zhaoxin-CentaurHauls-ZXD-4600.output The file was removed tests/hwloc/x86/AMD-15h-Piledriver-4xOpteron-6348.output The file was added tests/hwloc/x86/AMD-K8-SledgeHammer-2xOpteron-250.xml The file was added tests/hwloc/x86/AMD-15h-Piledriver-4xOpteron-6348.test The file was added tests/hwloc/x86/AMD-19h-Zen3-2xEpyc-7763.test The file was added tests/hwloc/x86/Intel-IvyBridge-12xXeon-E5-4620v2.test The file was removed tests/hwloc/x86/Zhaoxin-Shanghai-KaiSheng-ZXC+-FC1081.output The file was removed tests/hwloc/x86/AMD-19h-Zen3-2xEpyc-7763.output The file was added tests/hwloc/x86/Intel-CPUID.1A-1p2co2t.test The file was modified
tests/hwloc/x86/Makefile.am (diff) The file was removed tests/hwloc/x86/AMD-K8-SledgeHammer-2xOpteron-250.output The file was removed tests/hwloc/x86/Intel-Haswell-2xXeon-E5-2680v3.output The file was added tests/hwloc/x86/Intel-Westmere-2xXeon-X5650.xml The file was added tests/hwloc/x86/AMD-15h-Bulldozer-4xOpteron-6272.xml The file was added tests/hwloc/x86/Intel-Broadwell-2xXeon-E5-2650Lv4.test The file was added tests/hwloc/x86/Intel-KnightsLanding-XeonPhi-7210.test The file was added tests/hwloc/x86/AMD-K10-Istanbul-8xOpteron-8439SE.test The file was removed tests/hwloc/x86/Intel-CPUID.1A-1p2co2t.output The file was added tests/hwloc/x86/Intel-Westmere-2xXeon-X5650.test The file was added tests/hwloc/x86/AMD-15h-Bulldozer-4xOpteron-6272.test The file was added tests/hwloc/x86/Intel-Core-2xXeon-E5345.xml The file was added tests/hwloc/x86/AMD-K10-Istanbul-8xOpteron-8439SE.xml The file was added tests/hwloc/x86/Intel-Haswell-2xXeon-E5-2680v3.xml The file was added tests/hwloc/x86/Intel-Penryn-4xXeon-X7460.test The file was added tests/hwloc/x86/Zhaoxin-CentaurHauls-ZXD-4600.xml The file was added tests/hwloc/x86/Intel-SandyBridge-2xXeon-E5-2650.xml The file was removed tests/hwloc/x86/Intel-Core-2xXeon-E5345.output The file was removed tests/hwloc/x86/Intel-KnightsLanding-XeonPhi-7210.output The file was added tests/hwloc/x86/Hygon-Dhyana-32cores.xml The file was added tests/hwloc/x86/Intel-Nehalem-2xXeon-X5550.test The file was added tests/hwloc/x86/Intel-CPUID.1A-1p2co2t.xml The file was added tests/hwloc/x86/AMD-17h-Zen-2xEpyc-7451.xml The file was added tests/hwloc/x86/Intel-CPUID.1F-Qemu-2p3d3c2t.xml The file was added tests/hwloc/x86/Intel-Core-2xXeon-E5345.test The file was added tests/hwloc/x86/Zhaoxin-CentaurHauls-ZXD-4600.test
Commit
01fee40216706d41b4487d76655f554bb217e6fd
by Brice.Goglin nvml: fix nvlink 3/4 bandwidth in our distances matrix 50GB/s is the v3/v4 bidrectional bandwidth. unidirectional is still 25GB/s per link as in v2. Closes #574. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-nvml.c (diff)
Commit
22a193101bbeff92f6371d71eca6ad87d3b1a7d9
by Brice.Goglin restrict: fix when doing nothing except because of REMOVE_CPULESS If we restrict to the whole topology cpuset, we usually do nothing ... except if REMOVE_CPULESS is passed: some NUMA nodes with empty cpusets will get removed. We had a buggy assertion because of this, raised in #575. Moreover, fix the way we mark the object has "modified". With the current code, the parent wouldn't be considered modified, hence we wouldn't recurse in children to actually remove those NUMA nodes. And that caused the nodeset/complete_nodeset to remain untouched while allowed_nodeset was reduced in the caller. Also fix the corresponding case when restricting by nodeset and removing memory-less CPUs. Thanks to Mark Grondona for the report. Closes #574 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology.c (diff) The file was modified
tests/hwloc/hwloc_topology_restrict.c (diff)
Commit
f2724bbf4f068ed45edd40ac1f3f4f981fbd2693
by Brice.Goglin tests/restrict: add some redundant restrict() Apply the same restrict twice, or once with and once without REMOVE_CPULESS. Regression tests for #575 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/hwloc_topology_restrict.c (diff)
Commit
c54e0292f3c2c110c5aa9841e737a2b98ec683c8
by Brice.Goglin tests/{linux,xml,x86,x86+linux,linux/allowed}: allow to run tests in an optional extra subdirectory If extra/extra-tests.sh exists, call it with our TEST_LOG_COMPILER as parameter. Useful for additional testing that we can't or don't want to put in the repo (confidential, huge files, ...). Mostly useful for x86 and linux but supporting others is easy. Contrary to #191, this does not touch automake stuff at all, everything is managed by the outside subdirectory script (which basically iterates over tests and pass them to out TEST_LOG_COMPILER). Closes #191 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/linux/allowed/Makefile.am (diff) The file was modified
tests/hwloc/x86+linux/Makefile.am (diff) The file was modified
tests/hwloc/linux/Makefile.am (diff) The file was modified
tests/hwloc/xml/Makefile.am (diff) The file was modified
tests/hwloc/x86/Makefile.am (diff) The file was modified
NEWS (diff)
Commit
050f532f1b61cb313102b2de2cfeabc567d5311e
by Brice.Goglin synthetic/export/memory: replace a buggy assertion with some warnings We currently cannot export complex memory hierarchies with memory-side caches, caches are ignored, as well as some non-first NUMA nods below them. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-synthetic.c (diff) The file was modified
hwloc/topology-synthetic.c (diff) The file was modified
utils/hwloc/misc.h (diff) The file was modified
utils/hwloc/misc.h (diff) The file was modified
config/hwloc_internal.m4 (diff) The file was modified
utils/lstopo/test-lstopo.sh.in (diff)
Commit
af43952957b0e686a6e31187e61b5dd43db37796
by Brice.Goglin utils/archivemount: disable on !linux At least, the code doesn't build on Windows. And the tool will likely only be used on Linux for loading Linux sysfs or cpuid tarballs. Fixes commit 9e90d68bebe93c8e6ab7e633cd561868e9d1038f Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/lstopo/test-lstopo.sh.in (diff) The file was modified
utils/hwloc/misc.h (diff)
Commit
577b327ec57cf1980aac113b58f686d505fcc637
by Brice.Goglin core: fix group->total_memory after hwloc_topology_insert_group_object() It was kept as 0. The issue was never detected because our tests use synthetic topologies with empty nodes (changed in next commit). Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology.c (diff) The file was modified
tests/hwloc/hwloc_synthetic.c (diff) The file was modified
hwloc/topology-synthetic.c (diff) The file was modified
utils/hwloc/test-hwloc-info.output (diff)
Commit
02eaabce08a07e8fd3637d0198bd15aa833d7e17
by Brice.Goglin core: fix the updating of total_memory in hwloc_topology_insert_group_object() Objects might get merged, update the remaining one, not the original one. If merged with another group, there's no need to save total_memory during the merge because we'll recompute it in hwloc_topology_insert_group_object() anyway. Fixes 577b327ec57cf1980aac113b58f686d505fcc637 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology.c (diff) The file was modified
hwloc/topology-synthetic.c (diff) The file was modified
hwloc/topology-synthetic.c (diff)
Commit
24e2c76f7af718441f5e6bcd63039c300edb215b
by Brice.Goglin synthetic/import: allow memory-side caches It's just a single level, specified as an attribute to NUMAnodes pack:2 [numa(size=16GiB memorysidecache=1GiB)] pu:2 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-synthetic.c (diff)
Commit
8db8fd3aae0e2a25af4b90e6cda809e1ebf627fc
by Brice.Goglin synthetic/export: export memory-side caches Accumulate all levels in a single MS$ attribute of the NUMA node. Ignore memory side caches when exporting to v1 since the object doesn't exist there. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-synthetic.c (diff) The file was added SECURITY.md The file was modified
hwloc/topology.c (diff) The file was modified
include/hwloc.h (diff)
Commit
4519fabec95ee90a0c84b5d7db3355f237624ab1
by bgoglin hwloc/glibc-sched.h: handle bitmap realloc errors And document them in the API, as well as the other function that cannot fail. Refs #578. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
include/hwloc/glibc-sched.h (diff)
Commit
5aa875e603a42670a98c9463759d994ed1da0fbf
by bgoglin hwloc/linux-libnuma.h: handle bitmap realloc errors And document them in the API, as well as the other functions that can or cannot fail. Refs #578. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
include/hwloc/linux-libnuma.h (diff) The file was modified
include/hwloc.h (diff) The file was modified
include/hwloc/shmem.h (diff) The file was modified
include/hwloc/opencl.h (diff) The file was modified
include/hwloc/rsmi.h (diff) The file was modified
include/hwloc/openfabrics-verbs.h (diff) The file was modified
include/hwloc/levelzero.h (diff) The file was modified
include/hwloc/cudart.h (diff) The file was modified
include/hwloc/gl.h (diff) The file was modified
include/hwloc/cuda.h (diff) The file was modified
include/hwloc/nvml.h (diff) The file was modified
include/hwloc/linux.h (diff) The file was modified
include/hwloc/memattrs.h (diff) The file was modified
include/hwloc/export.h (diff) The file was modified
include/hwloc/diff.h (diff) The file was modified
include/hwloc/distances.h (diff) The file was modified
include/hwloc/helper.h (diff)
Commit
f28bcf2ac734b54c91478792fe30762d573a91e1
by Brice.Goglin hwloc.h: rewrite the doc of hwloc_get_type_depth() Move things in \return to follow changes for #578. Similar functions aren't updated because there are too many thing to duplicate. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
include/hwloc.h (diff) The file was modified
include/hwloc.h (diff) The file was modified
include/hwloc/bitmap.h (diff) The file was modified
include/hwloc.h (diff)
Commit
1f6f719f9cf32202979961016afdc9bfc9fe563f
by Brice.Goglin doxy: remove broken link to XML v1 export flag The flag was removed in 945c1dde813da1a3b3b78ce2899db6acd347bf0c. The v1->v2 upgrade guide isn't removed yet, just keep the text but remove the doxygen crossref. The guide might be replace with v2->v3 later, we'll see. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
doc/hwloc.doxy (diff) The file was modified
include/hwloc/helper.h (diff) The file was modified
include/hwloc/bitmap.h (diff) The file was modified
doc/Makefile.am (diff) The file was modified
include/hwloc.h (diff) The file was modified
utils/hwloc/misc.h (diff) The file was modified
hwloc/topology-x86.c (diff)
Commit
9a54af46b9743cd45fecd3e6aadbdc81d06f7918
by Brice.Goglin x86: use a global is_hybrid flag The Intel hybrid feature bit is reserved on AMD, so define our own flag that we'll update separately on AMD later. By the way, restrict the Intel cpukinds code to Intel now that it depends on that non-Intel-specific flag. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-x86.c (diff)
Commit
57328e6c146d104be6b63dd7585937173f5a1037
by Brice.Goglin x86: remove the explicit apic_number in leaves 0xb/0x0f It's the number of logical procs in the next level, but the type of that level isn't known in advance anyway (contrary to what some comments were sort of saying, which could confuse future changes). And we don't use that number anyway, just keep it in debugging messages. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-x86.c (diff) The file was modified
utils/hwloc/hwloc-gather-cpuid.c (diff) The file was modified
hwloc/topology-x86.c (diff) The file was modified
hwloc/topology-x86.c (diff)
Commit
20c794500c810686b0c0099b1e099c7461ff9f6b
by Brice.Goglin x86: update the loop break when reading topology cpuid leaves 0x1f/0x0b AMD supports 0x0b but they don't specify the same break condition. Things worked fine in practice but a new AMD variant is coming, so don't assume any implicit compatibility with Intel anymore. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-x86.c (diff)
Commit
ae6f3cc63493190daee19ff062182980b3ebcd73
by Brice.Goglin x86: add support for AMD 0x80000026 CPUID leaf This new leaf is similar to Intel 0x0b (already supported on AMD) and Intel 0x1f but it comes with AMD-specific levels (CCX="Complex" and CCD=Die which doesn't have the same value as Intel die level). This is supported on Zen4 (at least Genoa EPYCs) where there's currently a single CCX per CCD, hence we get Dies (but no "Complex" groups). Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
include/private/private.h (diff) The file was modified
hwloc/topology-x86.c (diff) The file was modified
doc/hwloc.doxy (diff) The file was added tests/hwloc/x86/AMD-19h-Zen4-2xEpyc-9654.xml The file was modified
tests/hwloc/x86/Makefile.am (diff) The file was added tests/hwloc/x86/AMD-19h-Zen4-2xEpyc-9654.tar.bz2 The file was added tests/hwloc/x86/AMD-19h-Zen4-2xEpyc-9654.test
Commit
21787a3027cbb092d748b5b40e1ef04cb334881c
by Brice.Goglin linux: detect fake numa in the Linux kernel cmdline fake numa is a development hack for changing the NUMA layout but some of its options (uniform splitting) are going to be used in production, so detect it (without using it yet). Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-linux.c (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
doc/hwloc.doxy (diff)
Commit
43105769f55d88b64e1aa580573ed732395616cb
by Brice.Goglin linux/numa: disable memory attributes and memory-side cache in fake numa HMAT information isn't updated when fake numa is enabled, so ignore it for now. In the uniform splitting fake numa case, we could update HMAT info since node #K is replaced with nodes #KN-#(K+1)N-1. In theory, the Linux kernel should fix HMAT info to match the fakenuma config, but it looks like it's not going to happen soon, we'll see later if we need/want to fix it in hwloc. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-linux.c (diff)
Commit
45ac2a28107a0f717f7ac726ac6c376beb8b660b
by Brice.Goglin lstopo/draw: cleanup horiz/vert layout management Add a ratio for the memory:above box but it's not used yet. When rectangular layout will be allowed there, it'll be very wide. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/lstopo/lstopo-draw.c (diff)
Commit
1e89d4e1974af217db859a41838348acd1763517
by Brice.Goglin lstopo/draw: make memory:above layout configurable and use rectangular by default Useful for fake numa cases where we have 8 local nodes per socket. They'll be stacked as 2 rows of 4 instead of 1 single row of 8. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
utils/lstopo/lstopo.h (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff)
Commit
32117b9109ed0c4c905ec2e291f8e610870f3bb8
by bgoglin Fix name scheme when building with --target Building with --target may add an unexpected prefix to the binary names. This commit uses the same mecanism to take this prefix in account. Closes #560 Signed-off-by: Clément Foyer <clement.foyer@univ-reims.fr> The file was modified
utils/netloc/infiniband/Makefile.am (diff) The file was modified
utils/hwloc/Makefile.am (diff) The file was modified
utils/lstopo/Makefile.am (diff) The file was modified
hwloc/topology-linux.c (diff)
Commit
aa26f297b5240425a970d21ecbb3a2a70fca0b95
by Brice.Goglin linux/dax: add CXLDevice info attribute(s) in DAX and NUMA nodes There can be multiple entries if the region is interleaved. Might be better to merge into a single info attr? We'll see. This uses "memregion" identifiers (regionX) to match dax devices and CXL devices. The corresponding Linux code (CXL volatile regions, etc) is planned for Linux 6.3. Refs #554 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/topology-linux.c (diff)
Commit
806fa7be16f3db32420a830d63ee7c8ec7f7b66f
by Brice.Goglin xml/import: allow (and ignore for now) info children in more xml tags This commit is mostly for backport in v2.x since it will simplify xml compatibility between 3.0 export and 2.x import. Allow info tags inside all tags where it could be useful one day (topology, page_type, distances2*, memattr). We already had them in object and cpukind. It's not clear yet whether we'll use info in more objects in v3 (distances and topology have been proposed but not decided yet), but ignoring them is easy to make (current) v3 XML importable in v2. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-xml.c (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
doc/hwloc.doxy (diff)
Commit
fa7b02feb44719c5f709d1b3311ae905e54a1795
by Brice.Goglin doxygen: update the config to 1.9.4 The only new value that is changed from default is NUM_PROC_THREADS=0 so that it uses all available cores instead of only one (we should ask doxygen devs to use hwloc to do so). Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
doc/doxygen.cfg (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
README (diff) The file was modified
doc/netloc.doxy (diff) The file was modified
doc/main.doxy (diff) The file was modified
utils/netloc/infiniband/netloc_ib_extract_dats.c (diff)
Commit
a6763631dca5973eb5ea5fc8b17ad4127cb48dcf
by bgoglin utils/archivemount: Restore working directory * After loading the topology in the directory mounted with archivemount, we restore the original working directory. * Add hwloc_utils_disable_input_format() to free the file descriptor. Solves #577 Signed-off-by: Clément Foyer <clement.foyer@univ-reims.fr> The file was modified
utils/hwloc/hwloc-info.c (diff) The file was modified
utils/hwloc/misc.h (diff) The file was modified
utils/hwloc/hwloc-calc.c (diff) The file was modified
utils/hwloc/hwloc-distrib.c (diff) The file was modified
utils/lstopo/lstopo.c (diff)
Commit
927fe558960207a8e5fae47ea556497c914e4e8f
by Brice.Goglin netloc: remove Doxy files still need some updates since main.doxy only includes hwloc.doxy now. The hwloc/ source directory isn't renamed to src/ (as in 1.x), not strictly required. tests/hwloc/ doesn't go back to tests either, could be nice since there's nothing else in tests/. utils/ still contains hwloc/ and lstopo/, we're fine there. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was removed tests/netloc/data/plafrim2.txz The file was removed netloc/edge.c The file was removed utils/netloc/infiniband/netloc_ib_gather_raw.in The file was removed netlocscotch.pc.in The file was removed doc/images/netloc_design.png The file was modified
VERSION (diff) The file was modified
include/Makefile.am (diff) The file was removed netloc/support.c The file was removed doc/netloc.doxy The file was removed utils/netloc/mpi/Makefile.am The file was modified
NEWS (diff) The file was removed doc/images/netloc_draw.png The file was removed tests/netloc/tests.sh.in The file was removed utils/netloc/infiniband/netloc_ib_extract_dats.c The file was modified
utils/Makefile.am (diff) The file was removed tests/netloc/data/tests_draw.txt The file was removed utils/netloc/draw/Makefile.am The file was removed utils/netloc/draw/netloc_draw_to_json.c The file was removed include/netloc.h The file was removed netloc/Makefile.am The file was modified
contrib/android/include/private/autogen/config.h (diff) The file was modified
doc/doxygen-config.cfg.in (diff) The file was removed utils/netloc/infiniband/Makefile.am The file was removed netloc/architecture.c The file was removed netloc/mpicomm.c The file was removed utils/netloc/infiniband/README The file was modified
doc/main.doxy (diff) The file was removed utils/netloc/draw/visdist/vis.min.css The file was removed utils/netloc/scotch/Makefile.am The file was modified
config/hwloc.m4 (diff) The file was removed include/netloc/uthash.h The file was removed include/netlocscotch.h The file was removed tests/netloc/data/scotch.txz The file was removed netloc/topology.c The file was removed include/private/netloc.h The file was removed include/netloc/utarray.h The file was modified
contrib/ci.inria.fr/job-3-sonarscanner.sh (diff) The file was removed netloc/physical_link.c The file was modified
doc/hwloc.doxy (diff) The file was removed tests/netloc/data/plafrim.txz The file was removed tests/netloc/data/tests_extract.txt The file was removed utils/netloc/mpi/netloc_mpi_rank_file.c The file was removed utils/netloc/mpi/netloc_rank_order.c The file was modified
config/hwloc_internal.m4 (diff) The file was removed netloc.pc.in The file was modified
configure.ac (diff) The file was removed utils/netloc/scotch/netlocscotch_get_arch.c The file was removed config/netloc.m4 The file was removed netloc/hwloc.c The file was removed tests/netloc/data/tests_mpiscotch.txt The file was modified
.gitignore (diff) The file was removed utils/netloc/draw/visdist/vis.min.js The file was modified
README (diff) The file was removed tests/netloc/data/avakas.txz The file was removed utils/netloc/draw/netloc_draw.html The file was modified
Makefile.am (diff) The file was removed netloc/scotch.c The file was removed doc/images/netloc_design.svg The file was removed tests/netloc/data/tests_scotch.txt The file was removed utils/netloc/mpi/netloc_mpi_find_hosts.c The file was removed utils/netloc/draw/netloc_draw.js The file was modified
doc/Makefile.am (diff) The file was removed netloc/node.c The file was removed netloc/path.c The file was removed doc/README-netloc.infiniband The file was modified
contrib/update-my-copyright.pl (diff) The file was modified
tests/Makefile.am (diff) The file was removed tests/netloc/Makefile.am The file was modified
doc/Makefile.am (diff) The file was removed doc/main.doxy The file was modified
doc/doxygen-config.cfg.in (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
README (diff) The file was modified
doc/Makefile.am (diff) The file was modified
README (diff) The file was modified
utils/hwloc/hwloc-calc.1in (diff) The file was modified
utils/hwloc/hwloc-distrib.1in (diff) The file was modified
utils/hwloc/hwloc-info.1in (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff) The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff) The file was modified
utils/lstopo/test-lstopo.sh.in (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
utils/hwloc/hwloc-gather-topology.in (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/topology-x86.c (diff) The file was modified
utils/hwloc/hwloc-calc.1in (diff)
Commit
45fb910813a0eed3d173440f1e38e71457d03468
by Brice.Goglin hwloc-calc: improve --cpukind parsing Assume it's a number only if it starts with a digit. Error-out otherwise, so that things like foo:bar instead of foo=bar are rejected. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/hwloc-calc.c (diff)
Commit
8e9134bdb79e1b0d4b2ba64640ebcce533c2997b
by Brice.Goglin hwloc-calc/bind.1: more info about --no-smt We apply it at the end, hence some cases aren't easy to implement. Give example combining multiple invocations of hwloc-calc. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/hwloc-bind.1in (diff) The file was modified
utils/hwloc/hwloc-calc.1in (diff)
Commit
b149d14b98a2359beb026ed741012aa78b53c6db
by Brice.Goglin hwloc-calc/bind.1: more info about --cpukind calc applis --cpukind at the end, and bind doesn't support it. Add some examples for searching/binding without a specific cpukind by combining invocations of calc. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/hwloc-bind.1in (diff) The file was modified
utils/hwloc/hwloc-calc.1in (diff)
Commit
d01d5ffff5843704f618904ebe67e943e51e2b33
by Brice.Goglin x86: fix/cleanup the breaking-out of the topology enumeration loops Intel and AMD specify different conditions for breaking out of the loop on CPUID 0xb/0x1f and 0xb/0x80000026 leaves respectively. And Zhaoxin behaves slightly differently from Intel. We cannot factorize the break condition as we did in commit 20c794500c810686b0c0099b1e099c7461ff9f6b, just respect the actual specs depending on whether we are on AMD or not. This fixes Dies on Zhaoxin KH-4000 CPUs (that 0x1f level was ignored because it reports 0 in the number of threads in next level). These dies aren't visible in any release yet since 0x1f wasn't enabled on Zhaoxin (enabled in commit 4f89de1f902d27ebf3844632d5a7a4412ba5e450). Refs #593 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/hwloc-gather-cpuid.c (diff) The file was modified
hwloc/topology-x86.c (diff) The file was modified
hwloc/topology-levelzero.c (diff) The file was modified
tests/hwloc/levelzero.c (diff)
Commit
620a0088c661afdf5565b0218f7c656cbd1870bf
by Brice.Goglin fix prefix before error messages "hwloc/...:" instead of a couple of "[hwloc/...]" Add missing ones. verbose messages (especially in XML and synthetic backends) don't get prefixes. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-x86.c (diff) The file was modified
hwloc/topology-levelzero.c (diff) The file was modified
hwloc/topology-nvml.c (diff) The file was modified
hwloc/topology-synthetic.c (diff) The file was modified
hwloc/topology-darwin.c (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
hwloc/topology-windows.c (diff) The file was modified
hwloc/topology-levelzero.c (diff) The file was modified
hwloc/topology-levelzero.c (diff) The file was modified
hwloc/topology-pci.c (diff) The file was modified
hwloc/topology-pci.c (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/topology-levelzero.c (diff)
Commit
300fb8a551b3fd2f39f938390c82be2f5a790c78
by Brice.Goglin utils/annotate: fix [kMGT]B size units commit 4d9a9843a5616fc702575c4701a8c3cba044757f did not actually fix it, it kept them identical to the new [kMGT]iB. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/hwloc-annotate.c (diff) The file was modified
utils/hwloc/hwloc-calc.c (diff) The file was modified
utils/hwloc/hwloc-bind.c (diff) The file was modified
utils/hwloc/hwloc-ps.c (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
utils/hwloc/hwloc-calc.c (diff) The file was modified
utils/hwloc/common-ps.c (diff)
Commit
12e346d1740ce35e4a920df4adbe3e45e20444aa
by Brice.Goglin core+lstopo: don't forget to set the L3i filter when operating on all caches Fortunately, L3i doesn't exist these days (not sure it has ever existed, Itanium-2 had L2i but its L3 was unified). Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
hwloc/topology.c (diff)
Commit
690367b05affa942fa370bc2d8f4674544a5fbc1
by Brice.Goglin core: remove some redundant checks when setting filters When setting filters for multiple types, only check the topology status once, and avoid checks on hardwired type values. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology.c (diff)
Commit
f0699f2d66881f6cf17c45f4c196f89d2e4b258c
by Brice.Goglin core: allow backends to check if all cpu caches are filtered-out Useful for Linux to avoid reading multiple useless sysfs files before checking the resulting type filter. Also used for Windows and x86 since it's easy to use there. Refs #592 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-windows.c (diff) The file was modified
include/private/private.h (diff) The file was modified
hwloc/topology-x86.c (diff) The file was modified
hwloc/topology.c (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
doc/hwloc.doxy (diff)
Commit
93fbfa328e6d8611401b81febecb1e8f2e5466c2
by Brice.Goglin levelzero: use zesInit() when available When zesInit() is available (starting in specs 1.5), call it to further reduce problems with ZES_ENABLE_SYSMAN=1 not being set in the environment. The current implementation of zesInit() in the oneAPI compute runtime returns an "unsupported feature" 0x78000003 error (or return success and do nothing if ZES_ENABLE_SYSMAN=1) but hopefully things will improve quickly. Once zesInit() will be widely available/implemented, we'll consider disabling the ZES_ENABLE_SYSMAN=1 stuff if zesInit() is found at configure time, and later remove it. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/ports/include/levelzero/level_zero/zes_api.h (diff) The file was modified
hwloc/topology-levelzero.c (diff) The file was modified
tests/hwloc/levelzero.c (diff) The file was modified
config/hwloc.m4 (diff) The file was modified
tests/hwloc/ports/Makefile.am (diff) The file was modified
include/hwloc/levelzero.h (diff) The file was modified
tests/hwloc/ports/Makefile.am (diff) The file was modified
contrib/dist/publish_doc (diff) The file was modified
contrib/dist/publish_doc (diff)
Commit
49b940c3a4fa4a4e570c03884c3590e0819200a6
by bgoglin windows: Fix type of hostname_size parameter Avoids warning: passing argument 2 of 'GetComputerNameA' from incompatible pointer type [-Wincompatible-pointer-types] Signed-off-by: Evan Ramos <evan@hpccharm.com> The file was modified
hwloc/topology-windows.c (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff)
Commit
f76a7ccffa143cdceef8a4dc929b09876fac5f9e
by Brice.Goglin darwin: fix IOKit warnings about const char * being passed in arrays GCC11 emits warnings [-Wstringop-overread] when passing short C strings in io_name/string_t in the IOKit API because those types are 128/512 char arrays. It's not clear if these warnings are really valid but they are not too hard to fix (we cannot cast into an array type, so just declare a variable of that array type earlier). By the way, update the actual definition of io_name/string_t in our port test headers. Thanks to Ralph Castain for the report. Fixes #599 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-darwin.c (diff) The file was modified
tests/hwloc/ports/include/darwin/IOKit/IOKitLib.h (diff)
Commit
2c0308331d3d2f5475c1fd0271e315d59246f1fd
by Brice.Goglin contrib/dist/pubish_doc: further improve jquery.js We'll use a common jquery.js that points to the latest one. Followup to hwloc commit d94bca72ef2cdbc3f95636b37f639ebd03430643 and ompi-www commit cb8ca644a0839948874570a4aedbbaf8fa18ecca Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
contrib/dist/publish_doc (diff) The file was modified
NEWS (diff) The file was modified
utils/hwloc/hwloc-bind.1in (diff) The file was modified
include/hwloc.h (diff)
Commit
dfca160d4f741392e2e5a38357aa9a6a5243b43f
by Brice.Goglin doxy: improve the cpu_mem_bind chapter - subdivide in sections - add an introduction - talk about portability and policies - more cross-references Refs #601 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
doc/hwloc.doxy (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
README (diff) The file was modified
doc/hwloc.doxy (diff)
Commit
850a706cc3195d79b60186f87b6856c211996014
by Brice.Goglin README: update after doxy main page reorganization Skip the table of contents, keep from overview to credits. Filter-out "ttc" html lines (tooltip lines that are hidden by the CSS by default). Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
doc/README.sed (diff) The file was modified
README (diff) The file was modified
doc/Makefile.am (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/topology-linux.c (diff)
Commit
785678dc112dae4c7aa7aacc59149088bc10272a
by Brice.Goglin doxy: remove our custom doxygen.css for html docs Our css is severely outdated (for instance it misses the ttc/tooltip stuff, causing lots of things not to be hidden by default) and we didn't document much to be able to maintain it. Changes in commit 7e6bc6b297ce86915417976a59de9ef26488d4a3 could have been kept but its minor changes aren't really needed anymore. The default style is used in the official docs online, it's fine, so just use the same for all generated html docs. If we ever need changes, they should rather go in another CSS file that would override rules of the official one (using HTML_EXTRA_STYLESHEET as recommended). Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
doc/Makefile.am (diff) The file was removed doc/doxygen.css The file was modified
README (diff) The file was modified
doc/hwloc.doxy (diff)
Commit
a2e28b5a1ef3bf98139beef6327deb9693683ec8
by Brice.Goglin components: allocate the backend private_data together with the backend structure This breaks the component ABI, but it was already bumped from 7 to 8 in bf44f300e8. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-netbsd.c (diff) The file was modified
hwloc/topology-noos.c (diff) The file was modified
hwloc/topology-freebsd.c (diff) The file was modified
hwloc/topology-windows.c (diff) The file was modified
hwloc/topology-rsmi.c (diff) The file was modified
hwloc/topology-aix.c (diff) The file was modified
hwloc/topology-solaris.c (diff) The file was modified
hwloc/topology-pci.c (diff) The file was modified
hwloc/components.c (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
hwloc/topology-cuda.c (diff) The file was modified
hwloc/topology-hpux.c (diff) The file was modified
hwloc/topology-levelzero.c (diff) The file was modified
hwloc/topology-opencl.c (diff) The file was modified
hwloc/topology-gl.c (diff) The file was modified
hwloc/topology-x86.c (diff) The file was modified
hwloc/topology-nvml.c (diff) The file was modified
hwloc/topology-synthetic.c (diff) The file was modified
hwloc/topology-darwin.c (diff) The file was modified
hwloc/topology-fake.c (diff) The file was modified
include/hwloc/plugins.h (diff)
Commit
74e32c185f326f908c1e3ad3e8b2e9ad76ef4785
by Brice.Goglin cuda/levelzero/gl/nvml/opencl/rsmi: add a single Backend info attr in the root The name and subtype of these OS devices directly indicate their originating backend, no need to add a Backend=CUDA in each of them. Just keep one in the top topology info where we list other backends. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-levelzero.c (diff) The file was modified
tests/hwloc/cudart.c (diff) The file was modified
NEWS (diff) The file was modified
tests/hwloc/opencl.c (diff) The file was modified
hwloc/topology-gl.c (diff) The file was modified
tests/hwloc/cuda.c (diff) The file was modified
tests/hwloc/levelzero.c (diff) The file was modified
hwloc/topology-opencl.c (diff) The file was modified
tests/hwloc/gl.c (diff) The file was modified
tests/hwloc/nvml.c (diff) The file was modified
tests/hwloc/rsmi.c (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/topology-nvml.c (diff) The file was modified
hwloc/topology-cuda.c (diff) The file was modified
hwloc/topology-rsmi.c (diff) The file was modified
hwloc/topology-hardwired.c (diff) The file was modified
tests/hwloc/hwloc_backends.c (diff) The file was modified
hwloc/topology-noos.c (diff) The file was modified
hwloc/topology-pci.c (diff) The file was modified
hwloc/topology-freebsd.c (diff) The file was modified
tests/hwloc/linux/16amd64-8n2c-cpusets-xml.xml (diff) The file was modified
tests/hwloc/linux/16em64t-4s2ca2c-cpusetreorder.console (diff) The file was modified
tests/hwloc/linux/16amd64-8n2c-cpusets.console (diff) The file was modified
tests/hwloc/linux/16em64t-4s2ca2c-cpusetreorder-merge.console (diff) The file was modified
tests/hwloc/x86+linux/5intel64-hybrid-lakefield.console (diff) The file was modified
tests/hwloc/linux/32intel64-2p8co2t+8ve.console (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
tests/hwloc/linux/40intel64-2g2n4c+pcilocality.xml (diff) The file was modified
tests/hwloc/linux/fakecpuid1f-64intel64-2p4d2n2c2t.console (diff) The file was modified
tests/hwloc/linux/16amd64-8n2c-cpusets-disallowed.console (diff) The file was modified
tests/hwloc/linux/fakeheteromemtiers.console (diff) The file was modified
tests/hwloc/linux/nvidiagpunumanodes-kept.console (diff) The file was modified
tests/hwloc/linux/offline-cpu0-node0.console (diff) The file was modified
tests/hwloc/linux/16amd64-4n4c-cgroup-distance-merge.console (diff) The file was modified
tests/hwloc/linux/fakememinitiators-1np2c+1npp+gi.console (diff) The file was modified
tests/hwloc/linux/128arm-2pa2n8cluster4co.console (diff) The file was modified
tests/hwloc/linux/40intel64-4n10c+pci-conflicts.console (diff) The file was modified
tests/hwloc/linux/4fake-4gr1nu1pu.console (diff) The file was modified
tests/hwloc/linux/32amd64-4s2n4c-cgroup2.xml (diff) The file was modified
tests/hwloc/linux/2pa-pcidomain32bits.console (diff) The file was modified
tests/hwloc/linux/nvidiagpunumanodes.console (diff) The file was modified
tests/hwloc/linux/8em64t-2s2ca2c-buggynuma.console (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
include/private/xml.h (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
hwloc/diff.c (diff) The file was modified
include/private/xml.h (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
include/private/xml.h (diff) The file was modified
include/hwloc.h (diff) The file was modified
utils/hwloc/hwloc-calc.1in (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
include/hwloc.h (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
tests/hwloc/levelzero.c (diff) The file was modified
tests/hwloc/nvml.c (diff) The file was modified
tests/hwloc/opencl.c (diff) The file was modified
hwloc/topology.c (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
hwloc/traversal.c (diff) The file was modified
tests/hwloc/rsmi.c (diff) The file was modified
include/hwloc.h (diff) The file was modified
hwloc/diff.c (diff) The file was modified
tests/hwloc/cuda.c (diff) The file was modified
include/hwloc/inlines.h (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
utils/hwloc/hwloc-annotate.c (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff) The file was modified
hwloc/topology-x86.c (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
tests/hwloc/cudart.c (diff) The file was modified
tests/hwloc/gl.c (diff) The file was modified
NEWS (diff) The file was modified
utils/hwloc/hwloc-info.c (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
include/private/private.h (diff) The file was modified
hwloc/diff.c (diff) The file was modified
hwloc/topology-windows.c (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
hwloc/topology.c (diff) The file was modified
hwloc/cpukinds.c (diff) The file was modified
hwloc/topology-x86.c (diff) The file was modified
hwloc/topology-darwin.c (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
doc/Makefile.am (diff) The file was modified
include/hwloc/inlines.h (diff) The file was modified
include/hwloc.h (diff) The file was modified
include/hwloc/cpukinds.h (diff) The file was modified
tests/hwloc/cpukinds.c (diff) The file was modified
utils/hwloc/hwloc-info.c (diff) The file was modified
utils/lstopo/lstopo-text.c (diff) The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
hwloc/cpukinds.c (diff) The file was modified
NEWS (diff) The file was modified
utils/hwloc/hwloc-annotate.c (diff) The file was modified
utils/hwloc/hwloc-calc.c (diff) The file was modified
include/hwloc.h (diff) The file was modified
hwloc/topology-darwin.c (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
hwloc/topology.c (diff) The file was modified
utils/hwloc/hwloc-annotate.c (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
tests/hwloc/cpukinds.c (diff) The file was modified
hwloc/topology-x86.c (diff) The file was modified
include/hwloc.h (diff) The file was modified
hwloc/topology.c (diff) The file was modified
NEWS (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
doc/Makefile.am (diff) The file was modified
hwloc/topology.c (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
include/hwloc.h (diff) The file was modified
include/hwloc/inlines.h (diff) The file was modified
tests/hwloc/CMakeLists.txt (diff) The file was modified
tests/hwloc/Makefile.am (diff) The file was removed tests/hwloc/hwloc_obj_infos.c The file was added tests/hwloc/infos.c
Commit
a03b7f3fd04fe052d3fe256d78a92427fbe89557
by Brice.Goglin core: use the new internal hwloc__replace_infos() and remove add_info_nodup() And clarify that callers may always call replace, even if the info doesn't exist yet. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology.c (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
include/private/private.h (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
hwloc/topology-x86.c (diff) The file was modified
utils/hwloc/hwloc-annotate.c (diff)
Commit
d07e4ac7ff8e51bf47918f9b9bf48ee2b248c6dd
by Brice.Goglin utils/annotate: reimplement info operations using hwloc_modify_infos() It was relying on non-official/non-documentation manual changes to the info attr arrays. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/hwloc-annotate.c (diff)
Commit
6acac769c3f30c677dbc979d35515fa6265316e8
by Brice.Goglin utils/annotate: add the "cpukind#N" fake location to tweak cpukind infos And document that we may also call hwloc-annotate multiple times with same cpukind parameters but different info names and values. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/hwloc-annotate.c (diff) The file was modified
utils/hwloc/hwloc-annotate.1in (diff) The file was modified
utils/hwloc/test-hwloc-annotate.output (diff) The file was modified
utils/hwloc/test-hwloc-annotate.sh.in (diff) The file was modified
tests/hwloc/nvml.c (diff) The file was modified
tests/hwloc/gl.c (diff) The file was modified
tests/hwloc/rsmi.c (diff) The file was modified
tests/hwloc/levelzero.c (diff) The file was modified
tests/hwloc/cudart.c (diff) The file was modified
tests/hwloc/opencl.c (diff) The file was modified
tests/hwloc/cuda.c (diff) The file was modified
utils/hwloc/hwloc-annotate.1in (diff) The file was modified
hwloc/topology-rsmi.c (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
utils/hwloc/common-ps.c (diff) The file was modified
utils/hwloc/misc.h (diff) The file was modified
hwloc/components.c (diff) The file was modified
hwloc/topology-darwin.c (diff) The file was modified
hwloc/topology.c (diff) The file was modified
include/private/private.h (diff) The file was modified
doc/Makefile.am (diff) The file was modified
hwloc/diff.c (diff) The file was modified
hwloc/hwloc2.dtd (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
include/hwloc.h (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
include/hwloc/diff.h (diff) The file was modified
tests/hwloc/hwloc_topology_abi.c (diff) The file was modified
NEWS (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
utils/hwloc/hwloc-annotate.c (diff) The file was modified
utils/hwloc/test-hwloc-annotate.sh.in (diff) The file was modified
utils/hwloc/hwloc-annotate.1in (diff) The file was modified
utils/hwloc/test-hwloc-annotate.output (diff) The file was modified
utils/hwloc/test-hwloc-info.sh.in (diff) The file was modified
utils/hwloc/hwloc-info.c (diff) The file was modified
utils/hwloc/hwloc-info.1in (diff) The file was modified
utils/hwloc/test-hwloc-info.output (diff) The file was modified
utils/lstopo/lstopo-text.c (diff)
Commit
45f5e07db1dd069ab61a749ba35624567ad7b4f4
by Brice.Goglin core: move some root info attrs to topology infos Topology discovery and backend information: * Backend * SyntheticDescription * hwlocVersion * ProcessName OS information: * uname stuff (OSName, OSRelease, OSVersion, HostName, Architecture) * LinuxCgroup * WindowsBuildEnvironment Hardware info (DMI, CPU when in root because Package locality is missing, etc) remains in root. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/nvml.c (diff) The file was modified
tests/hwloc/x86/AMD-K10-MagnyCours-2xOpteron-6164HE.xml (diff) The file was modified
tests/hwloc/xml/96em64t-4n4d3ca2co-pci.xml (diff) The file was modified
tests/hwloc/linux/256ppc-8n8s4t.console (diff) The file was modified
tests/hwloc/x86/AMD-K10-Istanbul-8xOpteron-8439SE.xml (diff) The file was modified
tests/hwloc/x86/Intel-KnightsCorner-XeonPhi-SE10P.xml (diff) The file was modified
hwloc/topology-solaris.c (diff) The file was modified
tests/hwloc/linux/256ia64-64n2s2c.console (diff) The file was modified
tests/hwloc/levelzero.c (diff) The file was modified
tests/hwloc/linux/fakememinitiators-1np2c+1npp+gi.console (diff) The file was modified
tests/hwloc/xml/64intel64-fakeKNL-SNC4-hybrid.xml (diff) The file was modified
hwloc/topology-noos.c (diff) The file was modified
utils/hwloc/test-build-custom-topology.output (diff) The file was modified
tests/hwloc/linux/allowed/test1.xml (diff) The file was modified
tests/hwloc/x86/Intel-Westmere-2xXeon-X5650.xml (diff) The file was modified
hwloc/topology-windows.c (diff) The file was modified
tests/hwloc/linux/16em64t-4s2ca2c-cpusetreorder.console (diff) The file was modified
tests/hwloc/linux/8ia64-2n2s2c.console (diff) The file was modified
tests/hwloc/xml/192em64t-24n8c2t.xml (diff) The file was modified
utils/lstopo/test-lstopo.output (diff) The file was modified
tests/hwloc/xml/24em64t-2n6c2t-pci.xml (diff) The file was modified
tests/hwloc/x86/AMD-17h-Zen-2xEpyc-7451.xml (diff) The file was modified
tests/hwloc/x86/AMD-15h-Bulldozer-4xOpteron-6272.xml (diff) The file was modified
tests/hwloc/xml/16intel64-manyVFs.nocollapse.console (diff) The file was modified
tests/hwloc/xml/8intel64-4n2t-memattrs.xml (diff) The file was modified
utils/hwloc/test-hwloc-compress-dir.output.tar.gz (diff) The file was modified
tests/hwloc/x86/Intel-Skylake-2xXeon6140.xml (diff) The file was modified
tests/hwloc/xml/cxlmem+dax.v3.xml (diff) The file was modified
tests/hwloc/xml/irregulargroups-disallowed.console (diff) The file was modified
tests/hwloc/xml/32em64t-2n8c2t-pci-wholeio.xml (diff) The file was modified
hwloc/topology-nvml.c (diff) The file was modified
utils/hwloc/test-hwloc-compress-dir.input.tar.gz (diff) The file was modified
tests/hwloc/linux/nvidiagpunumanodes-kept.console (diff) The file was modified
tests/hwloc/xml/32em64t-2n8c2t-pci-normalio.xml (diff) The file was modified
tests/hwloc/linux/64fake-4n2s2ca2c2t.console (diff) The file was modified
hwloc/topology-gl.c (diff) The file was modified
tests/hwloc/x86/Intel-CPUID.1F-Qemu-2p3d3c2t.xml (diff) The file was modified
tests/hwloc/linux/16em64t-4s2c2t-nocaches.console (diff) The file was modified
tests/hwloc/xml/192em64t-12gr2n8c2t.xml (diff) The file was modified
tests/hwloc/gl.c (diff) The file was modified
hwloc/topology-rsmi.c (diff) The file was modified
hwloc/topology-aix.c (diff) The file was modified
hwloc/topology-darwin.c (diff) The file was modified
tests/hwloc/linux/2pa-pcidomain32bits.console (diff) The file was modified
tests/hwloc/cuda.c (diff) The file was modified
tests/hwloc/xml/fakeheterodistances.xml (diff) The file was modified
tests/hwloc/linux/40intel64-4n10c+pci-conflicts.console (diff) The file was modified
tests/hwloc/x86/Intel-Nehalem-2xXeon-X5550.xml (diff) The file was modified
tests/hwloc/linux/4fake-4gr1nu1pu.console (diff) The file was modified
utils/hwloc/test-hwloc-annotate.output2 (diff) The file was modified
tests/hwloc/linux/2i386-2t-hugepagesizecount.xml (diff) The file was modified
tests/hwloc/linux/2i386-2c-nohugepage.xml (diff) The file was modified
tests/hwloc/linux/32intel64-2p8co2t+8ve.console (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
tests/hwloc/linux/8em64t-2s2ca2c-buggynuma.console (diff) The file was modified
tests/hwloc/linux/16amd64-4n4c-cgroup-distance-merge.console (diff) The file was modified
hwloc/misc.c (diff) The file was modified
hwloc/topology-freebsd.c (diff) The file was modified
tests/hwloc/x86/AMD-19h-Zen3-2xEpyc-7763.xml (diff) The file was modified
tests/hwloc/linux/16em64t-4s2c2t-offlines-xml.xml (diff) The file was modified
tests/hwloc/linux/fakeheterocpunuma.xml (diff) The file was modified
utils/hwloc/test-hwloc-annotate.input (diff) The file was modified
tests/hwloc/linux/128ia64-17n4s2c.console (diff) The file was modified
tests/hwloc/linux/8em64t-2s4c-asymcaches.console (diff) The file was modified
tests/hwloc/linux/2arm-2c.console (diff) The file was modified
hwloc/topology-netbsd.c (diff) The file was modified
utils/hwloc/test-hwloc-diffpatch.input1 (diff) The file was modified
tests/hwloc/linux/64amd64-4s2n4ca2co.console (diff) The file was modified
hwloc/topology-opencl.c (diff) The file was modified
tests/hwloc/linux/64intel64-fakeKNL-SNC4-hybrid.console (diff) The file was modified
tests/hwloc/x86/Intel-CascadeLake-2xXeon6230.xml (diff) The file was modified
tests/hwloc/x86/Intel-Penryn-4xXeon-X7460.xml (diff) The file was modified
tests/hwloc/linux/64intel64-fakeKNL-A2A-cache.console (diff) The file was modified
tests/hwloc/linux/16em64t-4s2c2t.console (diff) The file was modified
utils/hwloc/test-hwloc-info.output (diff) The file was modified
tests/hwloc/x86/Intel-CPUID.1A-1p2co2t.xml (diff) The file was modified
tests/hwloc/linux/16ia64-8n2s.console (diff) The file was modified
tests/hwloc/linux/16amd64-8n2c-cpusets.console (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
tests/hwloc/linux/48amd64-4pa2n6c-sparse.console (diff) The file was modified
tests/hwloc/xml/32em64t-2n8c2t-pci-noio.xml (diff) The file was modified
tests/hwloc/linux/16em64t-4s2c2t-xml.xml (diff) The file was modified
tests/hwloc/linux/fakecpuid1f-64intel64-2p4d2n2c2t.console (diff) The file was modified
tests/hwloc/x86/AMD-15h-Piledriver-4xOpteron-6348.xml (diff) The file was modified
utils/hwloc/test-hwloc-info.sh.in (diff) The file was modified
tests/hwloc/linux/nvidiagpunumanodes.console (diff) The file was modified
tests/hwloc/xml/16intel64-manyVFs.console (diff) The file was modified
tests/hwloc/linux/16amd64-8n2c-cpusets-disallowed.console (diff) The file was modified
tests/hwloc/xml/16-2gr2gr2n2c+misc.console (diff) The file was modified
tests/hwloc/linux/16em64t-4s2c2t-offlines.console (diff) The file was modified
tests/hwloc/x86/AMD-19h-Zen4-2xEpyc-9654.xml (diff) The file was modified
tests/hwloc/linux/fakeheteromemtiers.console (diff) The file was modified
utils/hwloc/test-hwloc-annotate.output (diff) The file was modified
tests/hwloc/linux/8ia64-2s2c2t.console (diff) The file was modified
tests/hwloc/linux/32amd64-4s2n4c-cgroup2.xml (diff) The file was modified
tests/hwloc/rsmi.c (diff) The file was modified
tests/hwloc/xml/cxlmem+dax.v2.xml (diff) The file was modified
hwloc/topology.c (diff) The file was modified
tests/hwloc/xml/irregulargroups-disallowed.xml (diff) The file was modified
tests/hwloc/x86/Intel-SandyBridge-2xXeon-E5-2650.xml (diff) The file was modified
tests/hwloc/linux/offline-cpu0-node0.console (diff) The file was modified
utils/hwloc/test-hwloc-diffpatch.input2 (diff) The file was modified
hwloc/topology-synthetic.c (diff) The file was modified
NEWS (diff) The file was modified
tests/hwloc/linux/16amd64-8n2c-cpusets-xml.xml (diff) The file was modified
tests/hwloc/linux/16amd64-8n2c.console (diff) The file was modified
tests/hwloc/hwloc_backends.c (diff) The file was modified
tests/hwloc/x86/Intel-Core-2xXeon-E5345.xml (diff) The file was modified
tests/hwloc/linux/64intel64-fakeKNL-SNC4-hybrid-msc.console (diff) The file was modified
tests/hwloc/linux/32em64t-2n8c+dax+nvme+mic+dimms.xml (diff) The file was modified
tests/hwloc/linux/8em64t-2s2ca2c.console (diff) The file was modified
tests/hwloc/x86+linux/64amd64-4p2n4ca2co.xml (diff) The file was modified
tests/hwloc/xml/power8gpudistances.xml (diff) The file was modified
tests/hwloc/linux/40intel64-2g2n4c+pcilocality.xml (diff) The file was modified
tests/hwloc/x86/Zhaoxin-CentaurHauls-ZXD-4600.xml (diff) The file was modified
tests/hwloc/xml/16amd64-4distances.xml (diff) The file was modified
tests/hwloc/x86/AMD-K8-SantaRosa-2xOpteron-2218.xml (diff) The file was modified
hwloc/topology-pci.c (diff) The file was modified
hwloc/topology-hardwired.c (diff) The file was modified
tests/hwloc/x86/Hygon-Dhyana-32cores.xml (diff) The file was modified
tests/hwloc/x86/Intel-Broadwell-2xXeon-E5-2650Lv4.xml (diff) The file was modified
tests/hwloc/xml/16-2gr2gr2n2c+misc.xml (diff) The file was modified
hwloc/topology-cuda.c (diff) The file was modified
tests/hwloc/linux/20s390-2g6s4c.console (diff) The file was modified
tests/hwloc/linux/128arm-2pa2n8cluster4co.console (diff) The file was modified
tests/hwloc/x86/Intel-KnightsLanding-XeonPhi-7210.xml (diff) The file was modified
tests/hwloc/xml/fakecpukinds.xml (diff) The file was modified
tests/hwloc/linux/16em64t-4s2c2t-merge.console (diff) The file was modified
tests/hwloc/x86+linux/5intel64-hybrid-lakefield.console (diff) The file was modified
tests/hwloc/cudart.c (diff) The file was modified
tests/hwloc/xml/16amd64-4distances.console (diff) The file was modified
hwloc/topology-x86.c (diff) The file was modified
tests/hwloc/xml/16intel64-manyVFs.xml (diff) The file was modified
hwloc/topology-hpux.c (diff) The file was modified
include/hwloc.h (diff) The file was modified
tests/hwloc/linux/96em64t-4no4pa3ca2co.console (diff) The file was modified
tests/hwloc/x86/Intel-Haswell-2xXeon-E5-2680v3.xml (diff) The file was modified
hwloc/topology-levelzero.c (diff) The file was modified
tests/hwloc/x86/AMD-K8-SledgeHammer-2xOpteron-250.xml (diff) The file was modified
tests/hwloc/linux/16em64t-4s2ca2c-cpusetreorder-merge.console (diff) The file was modified
tests/hwloc/linux/2ps3-2t.console (diff) The file was modified
tests/hwloc/opencl.c (diff) The file was modified
tests/hwloc/x86/Intel-IvyBridge-12xXeon-E5-4620v2.xml (diff) The file was modified
tests/hwloc/x86/Zhaoxin-Shanghai-KaiSheng-ZXC+-FC1081.xml (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
utils/lstopo/test-lstopo.output (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
tests/hwloc/xml/cxlmem+dax.v2.xml (diff) The file was modified
tests/hwloc/Makefile.am (diff) The file was removed tests/hwloc/CMakeLists.txt The file was modified
contrib/windows-cmake/CMakeLists.txt (diff) The file was added contrib/windows-cmake/tests/CMakeLists.txt The file was modified
hwloc/shmem.c (diff)
Commit
383a7d9af42730a67cf749bd3eabe908d7cc11aa
by Brice.Goglin allow: now allowed on shmem-adopted topologies A shmem topology may now be shared between different cgroups, one only has to call hwloc_topology_allow(LOCAL) to update the list of allowed CPUs and memory after adopting the shmem topo. The source process must have built the original topology with INCLUDE_DISALLOWED. This was disabled previously because allow(LOCAL) is supposed to add "LinuxCgroup" to the info attrs, which was not possible until we added topo infos in 3.0. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
include/hwloc.h (diff) The file was modified
hwloc/topology.c (diff) The file was modified
NEWS (diff) The file was modified
include/hwloc/shmem.h (diff)
Commit
63762b9eefff7dcf582ec758c2954ede659649b8
by Brice.Goglin API: add notes about restrict/insert_group changing logical indexes Can cause confusion when using different --restrict in different command-lines. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
include/hwloc.h (diff) The file was modified
include/hwloc.h (diff) The file was modified
utils/hwloc/hwloc-info.1in (diff) The file was modified
utils/hwloc/hwloc-bind.1in (diff) The file was modified
utils/hwloc/hwloc-distrib.1in (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff) The file was modified
utils/hwloc/hwloc-calc.1in (diff) The file was modified
utils/hwloc/hwloc-bind.1in (diff) The file was modified
utils/hwloc/hwloc-calc.1in (diff) The file was modified
utils/hwloc/hwloc-distrib.1in (diff) The file was modified
utils/hwloc/hwloc-info.1in (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff) The file was modified
tests/hwloc/shmem.c (diff) The file was modified
include/hwloc.h (diff) The file was modified
hwloc/topology.c (diff) The file was modified
tests/hwloc/infos.c (diff) The file was modified
utils/hwloc/common-ps.h (diff) The file was modified
utils/hwloc/common-ps.c (diff) The file was modified
contrib/completion/bash/hwloc (diff) The file was modified
utils/hwloc/hwloc-ps.1in (diff) The file was modified
utils/hwloc/hwloc-ps.c (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff) The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff)
Commit
237ddc1550b431c41a32d07e706e956d09a66d94
by Brice.Goglin lstopo: add --misc-from Allows to give a list of Misc objects in a separate file (or stdin), will be used to bring hwloc-ps flexibility without adding many options to lstopo. Refs #606 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
contrib/completion/bash/hwloc (diff) The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff) The file was modified
contrib/completion/bash/hwloc (diff) The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff) The file was modified
utils/hwloc/hwloc-ps.c (diff) The file was modified
utils/hwloc/hwloc-ps.1in (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/topology-x86.c (diff)
Commit
db52ca9af081420d84adb8aca509bc091921c52e
by Brice.Goglin core: don't crash when clearing an already empty topology If all objects got removed because cgroups were wrong, we'd crash when trying to clear the topology because the root is already NULL. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology.c (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
doc/hwloc.doxy (diff)
Commit
abfd6136e0fa4c0d7dff9b2260e4247ef27bffa2
by Brice.Goglin linux: add more MemoryModule info attributes FormFactor=DIMM, Type=DDR4, Size=<KiB> (with support for more than 32GB DIMMs), Rank=2 Uses SMBIOS 2.6 when available. Signed-off-by: Zubiao Xiong <zubiao.xiong@memverge.com> Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-linux.c (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
tests/hwloc/linux/32em64t-2n8c+dax+nvme+mic+dimms.xml (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
include/hwloc/bitmap.h (diff)
Commit
f0eeb3a9bb8e216ffff97087192e1591ce2fd440
by Brice.Goglin utils/calc: fix searching CPU objects below heterogeneous NUMA socket:0.numa:0.core:0 looks for first core whose cpuset and nodeset are included in socket:0.numa:0. This fails on heterogeneous memory machines since this core may have 2 local NUMAs in its nodeset, hence not included in socket:0.numa:0's nodeset. Fix this by checking if the nodeset intersects instead of being included. Note that the empty bitmap is isincluded in any bitmap, while it doesn't intersect with anything, hence we have to check empty nodesets too. Thanks to Antoine Morvan for the report. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/hwloc-calc.h (diff) The file was modified
utils/hwloc/test-hwloc-calc.output (diff) The file was modified
utils/hwloc/test-hwloc-calc.sh.in (diff)
Commit
75234cb14586fc7cb2ad0fdd9f9ca136beac32bd
by Brice.Goglin utils/calc: fix searching NUMA nodes inside CPU objects when NUMA are attached higher This is the symmetric of the previous commit, but reversing NUMAs and PUs. When a NUMA node is attached at a level, and we search for that NUMA node inside children of that level (hierarchical spec), the cpuset of the NUMA node contains more than just one children, hence we wouldn't find it with isincluded(). We now use zero()||intersects() instead. Example (different from the one added in test-hwloc-calc): AMD machine with 4 L3 (4 cores each) and 1 NUMA per package (16 cores total). "hwloc-calc l3:0.numa:0" should return the first NUMA (cpuset 0xffff nodeset 0x1) but it didn't because it's not included in the L3 (cpuset 0x000f nodeset 0x1). This should not happen when the user knows the ordering of levels in the machine, but may happen when using generic scripts that don't know where NUMA is attached (L3 above NUMA on some Intel/SNC but below on AMD currently). Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/test-hwloc-calc.sh.in (diff) The file was modified
utils/hwloc/test-hwloc-calc.output (diff) The file was modified
utils/hwloc/hwloc-calc.h (diff)
Commit
bc374deed17fa7ae212050534c52698ff51c4bdc
by Brice.Goglin android: disable the platform database for now Heroku broke, the API isn't available anymore. Just open the static database webpage. By the way, rename "API" into "database". Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
contrib/android/AndroidApp/lstopo/src/main/java/com/hwloc/lstopo/MainActivity.java (diff) The file was modified
contrib/android/AndroidApp/lstopo/src/main/res/menu/activity_main_menu_drawer.xml (diff) The file was modified
contrib/android/AndroidApp/lstopo/build.gradle (diff)
Commit
e0c40cb072200dbf0bd5e3d9abd905022fb59ddd
by Brice.Goglin utils/calc: add --object-output For options which report object indexes (-I and --local-memory), this adds the type as a prefix, e.g. Core:0,Core:2 instead of 0,2. Refs #611. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/test-hwloc-calc.output (diff) The file was modified
contrib/completion/bash/hwloc (diff) The file was modified
utils/hwloc/hwloc-calc.c (diff) The file was modified
utils/hwloc/hwloc-calc.1in (diff) The file was modified
utils/hwloc/test-hwloc-calc.sh.in (diff)
Commit
257e91a99eef37185d887d86a8a11df4bb052f00
by Brice.Goglin utils/bind: warn if neither CPU nor memory binding location were given This may clarify strange error cases where locations didn't get parsed and are tried as executable name. Refs #611 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/hwloc-bind.c (diff)
Commit
95b9d6ebd487f3b6b7b381e0e8c04c1c6f923131
by Brice.Goglin utils/bind: always warn when assuming the executable starts at an unrecognized argument We support -- to start the executable since the beginning and it's documented, so people should get used to it. Refs #611. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/hwloc-bind.c (diff) The file was modified
utils/hwloc/hwloc-calc.1in (diff)
Commit
0e54376b4c503754414482cf70fdd63de01bb564
by Brice.Goglin helpers.h: fix missing MemCache in get_next_child() This function would infinitely loop between a NUMANode and its MemCache parent instead of moving to IO/Misc children. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
include/hwloc/helper.h (diff) The file was modified
include/hwloc/helper.h (diff) The file was modified
include/hwloc/helper.h (diff)
Commit
21de51486f9e58cab8388c5a8601edad2af558b5
by Brice.Goglin utils: add -q and --quiet when we have -s and --silent Some tools had -s and --silent (lstopo, gather-cpuid, info), some had -q and --quiet (calc, bind). It's never actually completely silent, so have -q --quiet in all these tools and keep the existing -s --silent for backward compatibility. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
contrib/completion/bash/hwloc (diff) The file was modified
utils/hwloc/hwloc-gather-cpuid.1in (diff) The file was modified
utils/hwloc/hwloc-info.1in (diff) The file was modified
utils/hwloc/hwloc-info.c (diff) The file was modified
utils/hwloc/hwloc-gather-cpuid.c (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff) The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
hwloc/topology-linux.c (diff)
Commit
3c0822930f34e5e8876d89f88ae9c67f12a3b38c
by Brice.Goglin tests/linux: add a testcase with memory side caches We only ever had KNL with MSC but this case doesn't use the official HMAT-based Linux sysfs discovery. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was added tests/hwloc/linux/memorysidecaches.tar.bz2 The file was added tests/hwloc/linux/memorysidecaches.test The file was added tests/hwloc/linux/memorysidecaches.console The file was modified
tests/hwloc/linux/Makefile.am (diff) The file was modified
tests/hwloc/xml/Makefile.am (diff) The file was added tests/hwloc/xml/memorysidecaches.test The file was added tests/hwloc/xml/memorysidecaches.xml The file was modified
utils/hwloc/hwloc-info.c (diff) The file was modified
utils/hwloc/hwloc-info.c (diff) The file was modified
utils/hwloc/hwloc-info.c (diff)
Commit
9d35ec037343a6bf4d8e0b40b69c50c43c89bf5f
by Brice.Goglin utils/info: add [kind=]cpu/cache/normal/memory/io/all to --ancestor and --descendants Allow to select multiple types of ancestors/descendants. Ref #612. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
contrib/completion/bash/hwloc (diff) The file was modified
utils/hwloc/hwloc-info.1in (diff) The file was modified
utils/hwloc/hwloc-info.c (diff) The file was modified
utils/hwloc/test-hwloc-info.output (diff) The file was modified
utils/hwloc/test-hwloc-info.sh.in (diff)
Commit
b67f2f462dce864d7884bb15a8f6b70a929cb23f
by Brice.Goglin utils/info: add --first to stop on first matching object Useful for finding where a NUMA is attached: $ hwloc-info --ancestor normal --first numa:1 Package:0 Refs #612. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/test-hwloc-info.sh.in (diff) The file was modified
utils/hwloc/test-hwloc-info.output (diff) The file was modified
utils/hwloc/hwloc-info.1in (diff) The file was modified
contrib/completion/bash/hwloc (diff) The file was modified
utils/hwloc/hwloc-info.c (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
include/hwloc/helper.h (diff) The file was modified
utils/hwloc/hwloc-info.1in (diff) The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
include/hwloc.h (diff) The file was modified
include/hwloc/plugins.h (diff) The file was modified
contrib/ci.inria.fr/job-0-tarball.sh (diff)
Commit
b4a99ae7bf84328cf416b8fa0248f3175f93e807
by Brice.Goglin contrib/android: fix missing text outside of boxes PCI speeds and factorization info did not appear at all because they are displayed outside of their box but drawn inside that box "view". Move them out of the box view to the main layout. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
contrib/android/AndroidApp/lstopo/src/main/java/com/hwloc/lstopo/Lstopo.java (diff) The file was modified
contrib/android/AndroidApp/lstopo/src/main/java/com/hwloc/lstopo/MainActivity.java (diff) The file was modified
utils/lstopo/lstopo-android.c (diff) The file was modified
contrib/android/AndroidApp/lstopo/src/main/cpp/lib.c (diff)
Commit
fc01868ac81fed50f019efe6ccd9294fa9d1f13b
by Brice.Goglin contrib/android: rework the drawing of lines Just use small boxes since we only draw horizontal/vertical lines. Extend these boxes a little to make sure they touch neighbor boxes and lines. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
contrib/android/AndroidApp/lstopo/src/main/java/com/hwloc/lstopo/Lstopo.java (diff) The file was modified
utils/lstopo/lstopo-android.c (diff) The file was modified
contrib/android/AndroidApp/lstopo/src/main/java/com/hwloc/lstopo/Options.java (diff) The file was modified
contrib/android/AndroidApp/lstopo/build.gradle (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
tests/hwloc/hwloc_type_sscanf.c (diff)
Commit
3b6bc32553ac4e0b1bd32fe70ca24a21739248c6
by Brice.Goglin obj_type_snprintf: show "OS[type]" or "OSDev[type]" instead of "type" by default And add a SHORT_NAMES flag to revert to the old behavior, enabled by default in lstopo. This will help disambiguating things like "memory" that could have different meaning ("memory osdev" in 3+, "memory" kind, etc). enum hwloc_obj_snprintf_flag_e gets renumbered but it wasn't released yet anyway. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/xml/irregulargroups-disallowed.console (diff) The file was modified
tests/hwloc/xml/16intel64-manyVFs.console (diff) The file was modified
tests/hwloc/linux/fakeheteromemtiers.console (diff) The file was modified
tests/hwloc/linux/2pa-pcidomain32bits.console (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
utils/hwloc/misc.h (diff) The file was modified
include/hwloc.h (diff) The file was modified
tests/hwloc/xml/16intel64-manyVFs.nocollapse.console (diff) The file was modified
hwloc/traversal.c (diff) The file was modified
tests/hwloc/linux/32intel64-2p8co2t+8ve.console (diff) The file was modified
tests/hwloc/linux/40intel64-4n10c+pci-conflicts.console (diff)
Commit
35cc3ed50833e6cc2ff72d15c4449448f7e8015f
by Brice.Goglin tests/type_sscanf: update for new osdev format Test new osdev[] and os[] prefixes, test short names. Don't enforce the result for short names since it's not guaranteed to parse back anymore. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/hwloc_type_sscanf.c (diff) The file was modified
hwloc/topology-xml.c (diff)
Commit
015b20a7184e40e03c1deb876e69735991c80de8
by Brice.Goglin osdev: change osdev.type from an enum to a bitmask Each value is now 1<<x instead of x, and some got reordered. No multiple bits set yet. Default/unknown value is 0UL instead of -1. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
include/hwloc.h (diff) The file was modified
utils/hwloc/hwloc-calc.h (diff) The file was modified
tests/hwloc/xml/32em64t-2n8c2t-pci-wholeio.xml (diff) The file was modified
tests/hwloc/xml/192em64t-12gr2n8c2t.xml (diff) The file was modified
utils/hwloc/test-hwloc-annotate.input (diff) The file was modified
tests/hwloc/hwloc_type_sscanf.c (diff) The file was modified
tests/hwloc/openfabrics-verbs.c (diff) The file was modified
tests/hwloc/xml/192em64t-24n8c2t.xml (diff) The file was modified
NEWS (diff) The file was modified
tests/hwloc/xml/power8gpudistances.xml (diff) The file was modified
tests/hwloc/xml/24em64t-2n6c2t-pci.xml (diff) The file was modified
tests/hwloc/hwloc_iodevs.c (diff) The file was modified
tests/hwloc/xml/16intel64-manyVFs.xml (diff) The file was modified
tests/hwloc/xml/32em64t-2n8c2t-pci-normalio.xml (diff) The file was modified
tests/hwloc/xml/irregulargroups-disallowed.xml (diff) The file was modified
hwloc/traversal.c (diff) The file was modified
utils/hwloc/test-hwloc-annotate.output (diff) The file was modified
utils/hwloc/test-hwloc-diffpatch.input2 (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
tests/hwloc/linux/40intel64-2g2n4c+pcilocality.xml (diff) The file was modified
tests/hwloc/xml/96em64t-4n4d3ca2co-pci.xml (diff) The file was modified
tests/hwloc/linux/32em64t-2n8c+dax+nvme+mic+dimms.xml (diff) The file was modified
utils/hwloc/hwloc-calc.c (diff) The file was modified
utils/hwloc/test-hwloc-annotate.output2 (diff) The file was modified
utils/hwloc/test-hwloc-diffpatch.input1 (diff) The file was modified
tests/hwloc/xml/cxlmem+dax.v3.xml (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
NEWS (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
include/hwloc.h (diff)
Commit
a8fe16491a776eda33b4b464005c4ba23e5dae8a
by Brice.Goglin osdev: use multiple bits in the osdev-specific type Most GPUs are also CoProcs. Most CoProcs are also GPUs. Memory may also be Storage when non-volatile. OFED is also Network. hwloc_obj_type_snprintf() only prints one type when in shortnames mode. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
include/hwloc/nvml.h (diff) The file was modified
tests/hwloc/xml/192em64t-12gr2n8c2t.xml (diff) The file was modified
include/hwloc/opencl.h (diff) The file was modified
tests/hwloc/xml/16intel64-manyVFs.nocollapse.console (diff) The file was modified
tests/hwloc/nvml.c (diff) The file was modified
tests/hwloc/xml/32em64t-2n8c2t-pci-normalio.xml (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
NEWS (diff) The file was modified
include/hwloc/cuda.h (diff) The file was modified
utils/hwloc/test-hwloc-diffpatch.input2 (diff) The file was modified
tests/hwloc/linux/32em64t-2n8c+dax+nvme+mic+dimms.xml (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
include/hwloc.h (diff) The file was modified
tests/hwloc/xml/16intel64-manyVFs.console (diff) The file was modified
utils/hwloc/test-hwloc-diffpatch.input1 (diff) The file was modified
include/hwloc/gl.h (diff) The file was modified
tests/hwloc/cudart.c (diff) The file was modified
tests/hwloc/xml/16intel64-manyVFs.xml (diff) The file was modified
doc/examples/gpu.c (diff) The file was modified
hwloc/topology-cuda.c (diff) The file was modified
include/hwloc/rsmi.h (diff) The file was modified
tests/hwloc/xml/24em64t-2n6c2t-pci.xml (diff) The file was modified
include/hwloc/cudart.h (diff) The file was modified
hwloc/topology-nvml.c (diff) The file was modified
hwloc/traversal.c (diff) The file was modified
hwloc/topology-rsmi.c (diff) The file was modified
tests/hwloc/levelzero.c (diff) The file was modified
tests/hwloc/xml/192em64t-24n8c2t.xml (diff) The file was modified
include/hwloc/openfabrics-verbs.h (diff) The file was modified
tests/hwloc/linux/40intel64-2g2n4c+pcilocality.xml (diff) The file was modified
hwloc/topology-levelzero.c (diff) The file was modified
tests/hwloc/hwloc_type_sscanf.c (diff) The file was modified
tests/hwloc/opencl.c (diff) The file was modified
tests/hwloc/xml/cxlmem+dax.v3.xml (diff) The file was modified
hwloc/topology-opencl.c (diff) The file was modified
tests/hwloc/openfabrics-verbs.c (diff) The file was modified
utils/hwloc/test-hwloc-annotate.output2 (diff) The file was modified
tests/hwloc/linux/fakeheteromemtiers.console (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff) The file was modified
include/hwloc/plugins.h (diff) The file was modified
tests/hwloc/rsmi.c (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
tests/hwloc/xml/32em64t-2n8c2t-pci-wholeio.xml (diff) The file was modified
tests/hwloc/xml/power8gpudistances.xml (diff) The file was modified
tests/hwloc/cuda.c (diff) The file was modified
utils/hwloc/test-hwloc-compress-dir.input.tar.gz (diff) The file was modified
utils/hwloc/test-hwloc-compress-dir.output.tar.gz (diff)
Commit
8b22aea3a2923e0fd6b16cb90fcb980c704850ba
by Brice.Goglin gather-topology: bring back /proc/device-tree/cpus if --dt is given May be useful for debugging what Linux exposes in sysfs, at least on ARM. This reverts commit 9f37aae9a57f9fde9300f545daa9f2243489cc0c and adds conditionals on top of the old code. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/hwloc-gather-topology.1in (diff) The file was modified
contrib/completion/bash/hwloc (diff) The file was modified
utils/hwloc/hwloc-gather-topology.in (diff) The file was modified
NEWS (diff) The file was modified
hwloc/topology.c (diff) The file was modified
hwloc/topology.c (diff) The file was modified
tests/hwloc/hwloc_groups.c (diff) The file was modified
include/hwloc.h (diff) The file was modified
include/hwloc.h (diff) The file was modified
include/hwloc.h (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
tests/hwloc/hwloc_groups.c (diff) The file was modified
doc/Makefile.am (diff) The file was modified
hwloc/topology.c (diff) The file was modified
include/hwloc.h (diff)
Commit
2e40224fa3afc6a166e230374246bcbe24d0edbb
by Brice.Goglin xml: fix statically-allocated opaque backend info array It was sized for 64bit pointers max, it's not enough on 128bits platforms such as Capability Hardware Enhanced RISC Instructions (CHERI). Use something a bit more generic but still a bit hardwired to keep things simple and statically allocatable on the stack. Thanks to Jeff Hammond for the report. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
include/private/xml.h (diff)
Commit
43ad0c9e55903ba7e2e3823106ac88d654c62ab7
by Brice.Goglin shmem: pad the end of header up to a multiple of pointer size So that the following topology is well aligned on 128 bits architectures. No change on 32/64bits architectures. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/shmem.c (diff) The file was modified
doc/hwloc.doxy (diff)
Commit
14ab4eda9c4cc39bc9960fb2a6080dca86db700a
by Brice.Goglin core: improve insertion error messages Add subtype if any, display group kind/subkind (useful for finding where the existing conflicting group came from). Also cleanup the code a bit to factorize conditionals about fields to show or not. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology.c (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/topology-linux.c (diff)
Commit
6d5e0368d441e77871a355252ca60ed6f03032f9
by Brice.Goglin API: clarify in filter doc that MemCaches are ignored by default It was specified in the object type definition, not in the filter API def. Refs #622. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
include/hwloc.h (diff)
Commit
549b84f6cab8200363fabe5255cfcefceb770ec7
by Brice.Goglin API: clarify the default filtering of dies The core keeps all dies by default. However our Linux backend applies its own KEEP_STRUCTURE because it has no way to know if the kernel exposes a real single die per package, or just a default non-existing one. x86 seems to behave properly: a recent SPR processor doesn't report a single die in CPUID topology enumeraton, hence the x86 backend doesn't have to filter any useless die out. Refs #622 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
include/hwloc.h (diff)
Commit
4bd8185ef1d4bc226bdb64d102ccaa279de94ae0
by Brice.Goglin core: don't let hwloc_topology_check() test uninitialized topologies It's documented as only valid for loaded topology. Return earlier otherwise, but still do some basic topology-independent checks. Refs #622. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology.c (diff)
Commit
e6889d3e8491a473a0cf802120e5f7fb954087ca
by Brice.Goglin API: clarify that XML input errors are reported either in set_xml*() or load() set_xml() when using libxml, load() when using our minimalistic parser. Refs #623 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
include/hwloc.h (diff) The file was modified
include/hwloc.h (diff) The file was modified
utils/hwloc/hwloc-calc.c (diff) The file was modified
utils/hwloc/hwloc-calc.h (diff) The file was modified
utils/hwloc/hwloc-calc.1in (diff)
Commit
739a1f9ad0200ade1a11a349373f5cdd37183f10
by Brice.Goglin contrib/ci.inria.fr: shorten the build path in windows cmake Our CI uses NMake CMake-generator which doesn't like very long file names. It already fails when building some branches with long names. Use only the last 9 characters of the tarball name (that's what our CI currently uses for abbreviated git hashes) instead of the full hwloc-<branch>-date.git<hash> tarball name as a build directory. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
contrib/ci.inria.fr/job-1-wincmake.bat (diff) The file was modified
contrib/ci.inria.fr/job-1-wincmake.bat (diff)
Commit
56deb3b186e8abdef2427bf78d7ec107014c3603
by Brice.Goglin utils/calc: rework parsing of levels for -I -N and -H Replace hwloc_calc_type_depth() with hwloc_calc_parse_level(). Make it more generic for future reuses in other cases, save all parsed info a new hwloc_calc_level structure. HBM and MCDRAM are allowed but only used as raw NUMA for now. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/hwloc-calc.h (diff) The file was modified
utils/hwloc/hwloc-calc.c (diff) The file was modified
utils/hwloc/hwloc-calc.h (diff) The file was modified
utils/hwloc/hwloc-info.c (diff) The file was modified
utils/hwloc/hwloc-calc.h (diff)
Commit
323bdba060eb4f5381bb54e69ccf82b173d116e2
by Brice.Goglin utils: factorize the computing of the location prefix length pci[vendor:device] needs a temporary hack because it's not handled in hwloc_calc_parse_level() yet while typelen includes it. Will be removed in next commit. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/hwloc-annotate.c (diff) The file was modified
utils/hwloc/hwloc-info.c (diff) The file was modified
utils/hwloc/hwloc-calc.h (diff)
Commit
103f1faed9327342d0e028515fbb870ad7cbcd90
by Brice.Goglin utils: move the parsing of os[] and pci[:] filters to hwloc_calc_parse_level() Don't duplicate the parsing of osdev type, just use the now official parsing of strings like "os[net]". Remove a temporary hack that was needed for parsing pci[vendor:device] now that hwloc_calc_parse_level() takes care of it. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/hwloc-info.c (diff) The file was modified
utils/hwloc/hwloc-calc.h (diff) The file was modified
utils/hwloc/hwloc-annotate.c (diff) The file was modified
utils/hwloc/hwloc-calc.h (diff) The file was modified
utils/hwloc/hwloc-calc.h (diff) The file was modified
utils/hwloc/hwloc-info.c (diff) The file was modified
utils/hwloc/hwloc-calc.h (diff) The file was modified
utils/hwloc/hwloc-calc.c (diff) The file was modified
utils/hwloc/hwloc-calc.h (diff) The file was modified
utils/hwloc/test-hwloc-info.output (diff) The file was modified
utils/hwloc/test-hwloc-calc.sh.in (diff) The file was modified
utils/hwloc/test-hwloc-info.sh.in (diff) The file was modified
utils/hwloc/test-hwloc-calc.output (diff) The file was modified
utils/hwloc/hwloc.7in (diff) The file was modified
utils/hwloc/hwloc-info.1in (diff) The file was modified
utils/hwloc/hwloc-calc.1in (diff) The file was modified
utils/lstopo/lstopo.c (diff) The file was modified
utils/lstopo/lstopo.h (diff) The file was modified
utils/lstopo/lstopo-text.c (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff) The file was modified
utils/hwloc/hwloc-bind.1in (diff) The file was modified
utils/hwloc/hwloc.7in (diff) The file was modified
utils/hwloc/hwloc-calc.h (diff) The file was modified
hwloc/topology.c (diff) The file was modified
include/hwloc.h (diff)
Commit
a1b939190938ac253d4ba27328445b61258320c8
by Brice.Goglin core: disable all distances/cpukinds/memattrs call if flags are set Don't even bother calling the distances/cpukinds/memattrs which will do not nothing. init()/destroy() calls are kept because init() doesn't know yet if flags will be set to disable distances/cpukinds/memattrs. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology.c (diff)
Commit
bc89428564311365f3f11e928a3d2106c17afb66
by Brice.Goglin memattrs: rework the heuristics to find types of memory nodes Actually create tiers internally to group nodes by type. Split tiers if bandwidth or latency difference is too high (10% by default). Don't enforce any order in types yet, just apply our easy heuristics on unknown tiers instead of enforcing them to be first. 2 slight differences: * HWLOC_MEMTIERS_GUESS=spm_is_hbm doesn't set non-HBM nodes to DRAM anymore. * HWLOC_MEMTIERS_GUESS=none still sets node subtypes when they are obvious (no need to guess anything) tests/hwloc/memtiers.c is updated, extended, and improved. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/memtiers.c (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
tests/hwloc/xml/64intel64-fakeKNL-SNC4-hybrid.xml (diff) The file was modified
hwloc/memattrs.c (diff) The file was modified
tests/hwloc/linux/64intel64-fakeKNL-SNC4-hybrid.console (diff) The file was modified
tests/hwloc/linux/64intel64-fakeKNL-SNC4-hybrid-msc.console (diff) The file was modified
tests/hwloc/memtiers.c (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/memattrs.c (diff) The file was modified
hwloc/memattrs.c (diff) The file was modified
hwloc/memattrs.c (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/memattrs.c (diff) The file was modified
tests/hwloc/memtiers.c (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
hwloc/memattrs.c (diff) The file was modified
tests/hwloc/linux/64intel64-fakeKNL-SNC4-hybrid.console (diff) The file was modified
tests/hwloc/linux/fakeheteromemtiers.console (diff) The file was modified
hwloc/memattrs.c (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
tests/hwloc/linux/fakememinitiators-1np2c+1npp+gi.console (diff) The file was modified
tests/hwloc/xml/8intel64-4n2t-memattrs.xml (diff) The file was modified
tests/hwloc/linux/nvidiagpunumanodes-kept.console (diff) The file was modified
tests/hwloc/memtiers.c (diff) The file was modified
tests/hwloc/linux/64intel64-fakeKNL-SNC4-hybrid-msc.console (diff) The file was modified
tests/hwloc/xml/64intel64-fakeKNL-SNC4-hybrid.xml (diff) The file was modified
include/hwloc.h (diff) The file was modified
doc/hwloc.doxy (diff)
Commit
c3061d1313563d473ad4bd9bdd96e16a74e40e78
by Brice.Goglin memattrs: only build memory tiers and set subtypes when not loading from XML XML already has its own subtype and tiers. If importing from <2.8 (no subtypes) or <2.10 (no tiers), one may set HWLOC_MEMTIERS_REFRESH=1 to force refresh them. May also be used if the backend set wrong subtypes for some reason. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
include/private/private.h (diff) The file was modified
hwloc/memattrs.c (diff) The file was modified
tests/hwloc/memtiers.c (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/topology.c (diff)
Commit
f15fea14daefd82fa9cd1919d0ab77a748bca878
by Brice.Goglin utils/manpages: talk much more about HBM NUMA nodes Keep references to KNL MCDRAM but make sure we have HBM nearby since it's going to be more common now. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/hwloc-bind.1in (diff) The file was modified
utils/hwloc/hwloc-calc.1in (diff) The file was modified
utils/hwloc/hwloc.7in (diff) The file was modified
utils/lstopo/lstopo-no-graphics.1in (diff) The file was modified
utils/hwloc/test-hwloc-info.sh.in (diff) The file was modified
utils/hwloc/hwloc-calc.1in (diff) The file was modified
utils/hwloc/hwloc-info.c (diff) The file was modified
utils/hwloc/test-hwloc-info.output (diff) The file was modified
utils/hwloc/hwloc-calc.h (diff) The file was modified
utils/hwloc/hwloc.7in (diff) The file was modified
doc/images/HACKING (diff) The file was modified
doc/hwloc.doxy (diff) The file was added doc/images/heteromem.xml The file was modified
include/hwloc/memattrs.h (diff) The file was added doc/images/heteromem.png The file was modified
utils/hwloc/hwloc.7in (diff) The file was modified
utils/hwloc/hwloc-gather-topology.in (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
hwloc/topology-linux.c (diff)
Commit
9ae30cc5f920d3f56a2d5e289a0898b0f7ab793d
by Brice.Goglin linux/cpukinds: separate gathering from submitting cpukind Get all sysfs frequencies and capacities in an array and then submit cpukinds. We'll add intermediate quirks. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-linux.c (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
hwloc/topology-linux.c (diff)
Commit
e0e3c7c5dc6e7cc30722ea2035e66a3ee3330a27
by Brice.Goglin linux/cpukinds: force homogeneity on NVIDIA Grace Add a quirk to uniformize basefreq (to its minimum), maxfreq and capacity (to their maximum) and enable it on NVIDIA Grace. Also add the envvar HWLOC_CPUKINDS_HOMOGENEOUS to force that quirk (=1) or avoid it (=0). Closes #634 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-linux.c (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
include/hwloc.h (diff) The file was modified
tests/hwloc/hwloc_topology_abi.c (diff) The file was modified
include/hwloc.h (diff) The file was added tests/hwloc/x86/Intel-SapphireRapids-2xXeonMax9460.test The file was added tests/hwloc/x86/Intel-SapphireRapids-2xXeonMax9460.tar.bz2 The file was added tests/hwloc/x86/Intel-SapphireRapids-2xXeonMax9460.xml The file was modified
tests/hwloc/x86/Makefile.am (diff) The file was added tests/hwloc/x86/Intel-RaptorLake-Corei7-1370P.test The file was added tests/hwloc/x86/Intel-RaptorLake-Corei7-1370P.tar.bz2 The file was modified
tests/hwloc/x86/Makefile.am (diff) The file was added tests/hwloc/x86/Intel-RaptorLake-Corei7-1370P.xml The file was added tests/hwloc/linux/20em64t-hybrid-1p6c2t+2ca4co1t.test The file was modified
tests/hwloc/linux/Makefile.am (diff) The file was added tests/hwloc/linux/20em64t-hybrid-1p6c2t+2ca4co1t.tar.bz2 The file was added tests/hwloc/linux/20em64t-hybrid-1p6c2t+2ca4co1t.console
Commit
d70f784de98288a6594f70964ee8878da74b2484
by Brice.Goglin linux: fix smbios memory Misc info attrs on big endian SMBIOS fields are apparently little-endian (although the spec isn't totally clear, that's what dmidecode assumes). So change the byte order when reading multibyte fields (only memory size and extended_size for now). The bug appeared on BE platforms when running the linux test "32em64t-2n8c+dax+nvme+mic+dimms" linux (gathered from a little-endian platform). This test is the only one where we have dimm information. Fixes abfd6136e0fa4c0d7dff9b2260e4247ef27bffa2 Closes #637 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-linux.c (diff)
Commit
2c16044e0c0a4d02d70323a80b6bb19e83383304
by Brice.Goglin linux/smbios: use endian.h htole16/32() Instead of reimplementing it ourself. Thanks to Guillaume Mercier for the suggestion. Improves d70f784de98288a6594f70964ee8878da74b2484 Refs #637 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-linux.c (diff)
Commit
cfd3a3c42fe85eb4badf2d43932d7f7d89225b86
by samuel.thibault test-gather-topology.sh: Disable test when cross-compiling In case one uses binfmt to be able to run the cross-built library & programs, the content of /proc/cpuinfo is supposed to be the guest cpu content, while cat will have gotten the host cpu content. Actually at the moment qemu erroneously gives the host cpu content, see https://gitlab.com/qemu-project/qemu/-/issues/2004 but once that will get fixed, cross-compiled gather-topology testing will fail (as expected), so disable it in that case. Refs #638 The file was modified
tests/hwloc/linux/gather/Makefile.am (diff) The file was modified
config/hwloc.m4 (diff)
Commit
ab5fbeb0432f51512a5bdf12d4430e7a6496a22a
by Brice.Goglin hwloc/import/diff/libxml: don't count the \0 in the buffer length Otherwise libxml 2.12 reports "parser error : Extra content at the end of the document" because it doesn't want the xmlReadMemory() input buffer length to include the ending \0 (as confirmed in https://gitlab.gnome.org/GNOME/libxml2/-/issues/638). Closes #639 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-xml-libxml.c (diff)
Commit
84223c7009b4798dde0a0959be5c8955ba70a56f
by Brice.Goglin API: ending \0 must be included in xml input memory buffer and length Our exporting functions include them, and importing are designed to work with those first. And commit ab5fbeb0432f51512a5bdf12d4430e7a6496a22a assumes this now in libxml import too. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
include/hwloc.h (diff) The file was modified
include/hwloc/diff.h (diff)
Commit
9eb2bf336124a0e27871b3b91be65ded6a376023
by Brice.Goglin xml/import/nolibxml: assume ending \0 is included in input buffer and length Required in the API since 84223c7009b4798dde0a0959be5c8955ba70a56f Add the ending \0 where it should already be. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-xml-nolibxml.c (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
NEWS (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff)
Commit
ff4db1ef81577771ffd084b682037d4f8536a857
by Brice.Goglin opencl.h: use the khronos PCI bus info extension when available Added in OpenCL 3.0.7. Still an extension, but looks more portable and AMD and NVIDIA ones. It brings locality for Intel GPU OpenCL devices, and works at least for NVIDIA GPUs too. Refs #337 because hopefully it works Intel FPGA too. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
include/hwloc/opencl.h (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
contrib/windows-cmake/tests/CMakeLists.txt (diff) The file was modified
hwloc/topology-levelzero.c (diff)
Commit
d081c32b2a30cf40b378814677738b1844d2e7c0
by Brice.Goglin levelzero/xelink: accumulate subdevice link bandwidths in parents Fix f207b0f9ac7b6f0651d0e5d36fc8ae8f229b4423 moved root BWs to subdevices, which means there's nothing in root anymore whenever there are subdevices. Accumulate in parents so that parent BWs are identical to what they were before the fix. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-levelzero.c (diff) The file was modified
doc/hwloc.doxy (diff)
Commit
eaecc3e331cf65954b949383ded9fdf6d1c9de96
by Brice.Goglin doxy: update the description of the I/O device figure PCI displaying changed since the text was written, it uses BDF instead vendor:device now. Also fix/improve the text a little bit. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
doc/hwloc.doxy (diff)
Commit
c38aa5dfe719b7c6993d3e6525e949cd73baff07
by Brice.Goglin doxy: remove custom HTML div sections before doxygen sectons They were only added to match the "runtime" css style, which is not used anymore. The current doxygen-generated HTML is invalid because it embeds those div declarations in <p>...</p>. Also anchors were duplicated to move them a bit up. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
doc/hwloc.doxy (diff)
Commit
cd3a1a7afa46bb8e747dd482e978955917a65feb
by Brice.Goglin doxy: add a section about building with GNU Make and CMake The GNU Make stuff is moved from the API example, and CMake is added thanks to Florent Pruvost's example at https://gitlab.inria.fr/solverstack/distrib/-/tree/master/cmake/test/hwloc Refs #565 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
doc/hwloc.doxy (diff) The file was modified
README (diff) The file was modified
include/private/cpuid-x86.h (diff)
Commit
9ff05d100442bde1f11fca058f1e8f9b334924ac
by Brice.Goglin API: improve the documentation of insert_group() Start from the easy case where we group children of the same parent, then talk about more general cases that may however fail. Thanks to HadrienG2 for the suggestion. Closes #648 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
include/hwloc.h (diff) The file was modified
include/hwloc.h (diff)
Commit
ee081625412f9a30ae5c84ba1a869029c753f807
by Brice.Goglin API: add hwloc_obj_set_subtype() hwloc_topology_insert_group() says explicitly that group->subtype may be defined by the user. However it will be freed by hwloc's internal call to free(), which may be incompatible to how the caller allocated that subtype. Now recommend the use of hwloc_obj_set_subtype() instead, it will take care of duplicating the user-given given subtype. Thanks to HadrienG2 for the report. Closes #642. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
doc/Makefile.am (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
include/hwloc.h (diff) The file was modified
hwloc/topology.c (diff) The file was modified
tests/hwloc/hwloc_groups.c (diff)
Commit
24d4281211e36bf23cc36d5b73cc2a49f25e807f
by Brice.Goglin deprecated.h: remove _nodeset() functions Those were deprecated since 2.0, not sure why I didn't remove them in commit 91399fbe8d05af5f664b30c43e6b92c76334a514 with others. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
include/hwloc/rename.h (diff) The file was modified
NEWS (diff) The file was modified
include/hwloc/deprecated.h (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
doc/hwloc.doxy (diff)
Commit
eeeda4ad2d0511a2334a02141dab6138b1e0ce47
by Brice.Goglin contrib/windows-cmake: actually enable the CUDA backend when detected Forgottin in commit 3ed566736ab06c4462c6632fece56f9622167ff8 Thanks to Moritz Kreutzer for the patch. Closes #651 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
contrib/windows-cmake/static-components.h.in (diff) The file was modified
utils/hwloc/hwloc-gather-topology.in (diff)
Commit
39fae7e3151fbd677f953f6990acd9eb2a0b9bfb
by Brice.Goglin linux: don't hide the NVIDIA GPU node on non-POWER platforms And allow to force hide it with HWLOC_KEEP_NVIDIA_GPU_NUMA_NODES=0. These nodes were hidden by default on POWER because people could use interleaved memory allocations across sockets, but they would also interleave across GPU memory. On NVIDIA Grace Hopper, interleaved allocation isn't much of an issue since there's a single CPU node (and 1 single GPU node per GPU slice). Thanks to Antoine Morvan for the report. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-linux.c (diff) The file was modified
doc/hwloc.doxy (diff)
Commit
42fea7bf6f65cea552a31bb7092432ff2320419d
by Brice.Goglin nvml: use the current PCI link speed Not the max supported by the device, since it may be in a slower/narrower slot. Just like we do in the PCI backend. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-nvml.c (diff) The file was modified
config/hwloc.m4 (diff) The file was modified
tests/hwloc/ports/include/nvml/nvml.h (diff)
Commit
ca9f08973e11e7f71b6d10dbc6a4f30a4f5f9a4f
by Brice.Goglin nvml: fix the PCI link speed for PCI Gen4+ The code was never updated Gen>3. We ignore slight encoding changes in last PCIe revisions, but we'll factorize with the common PCI code soon anyway. Thanks to Akram Sbaih for the report. Closes #653 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-nvml.c (diff)
Commit
cc260153fc2da93828e02d9b67efbce17d7d4621
by Brice.Goglin pci+nvml: factorize PCI link speed computation So that we don't forget to update NVML when adding new PCI revisions to the duplicated PCI code. Also the new NVML code doesn't handle encoding changes in last PCIe revisions. Refs #653 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-nvml.c (diff) The file was modified
include/private/misc.h (diff) The file was modified
hwloc/pci-common.c (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
utils/hwloc/hwloc-info.1in (diff)
Commit
b4ce80732bb9abd699f68229db3c489669777351
by Brice.Goglin windows: fix dll filenames when UNICODE is enabled LoadLibrary() and GetModuleHandle() failed when UNICODE is enabled because the filename was not properly encoded. Use the TEXT() macro to fix the encoding. Most dynamic function pointers would be NULL, and we'd fail to discover many features and to bind. Thanks to Martin for the report. Closes #655 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/ports/include/windows/windows.h (diff) The file was modified
hwloc/topology-windows.c (diff) The file was modified
utils/lstopo/lstopo-cairo.c (diff)
Commit
d94d7595071f640d21728184eaf6ad8add013c54
by Brice.Goglin publish_doc: update URL checks Generalize for recent outputs when grepping for section name. Generalize for the upcoming 3.x upgrade API page. Improve the grep output. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
contrib/dist/publish_doc (diff) The file was modified
doc/Makefile.am (diff)
Commit
9873cbfe08bec9ea8c73780602478d327f52733a
by Brice.Goglin core: add the KiB suffix to Size info when it's implicit All info attr except SectorSize, as documented. This does NOT break v3 import in v2 since the suffix is ignored and assumed "KiB" by default. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-levelzero.c (diff) The file was modified
hwloc/topology-opencl.c (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
hwloc/topology-cuda.c (diff) The file was modified
tests/hwloc/linux/32em64t-2n8c+dax+nvme+mic+dimms.xml (diff) The file was modified
tests/hwloc/linux/2pa-pcidomain32bits.console (diff) The file was modified
tests/hwloc/xml/cxlmem+dax.v3.xml (diff) The file was modified
tests/hwloc/xml/cxlmem+dax.v2.xml (diff) The file was modified
tests/hwloc/linux/40intel64-2g2n4c+pcilocality.xml (diff) The file was modified
hwloc/topology-linux.c (diff) The file was modified
tests/hwloc/linux/32intel64-2p8co2t+8ve.console (diff) The file was modified
tests/hwloc/linux/40intel64-4n10c+pci-conflicts.console (diff) The file was modified
tests/hwloc/linux/fakeheteromemtiers.console (diff) The file was modified
hwloc/topology-rsmi.c (diff) The file was modified
tests/hwloc/xml/power8gpudistances.xml (diff) The file was modified
tests/hwloc/xml/32em64t-2n8c2t-pci-wholeio.xml (diff) The file was modified
utils/hwloc/test-hwloc-annotate.output2 (diff) The file was modified
tests/hwloc/xml/32em64t-2n8c2t-pci-normalio.xml (diff) The file was modified
utils/lstopo/lstopo-draw.c (diff) The file was modified
NEWS (diff) The file was modified
doc/hwloc.doxy (diff)
Commit
313de564cd1ee2a2f3a1fa208316f523b3da0ca2
by Brice.Goglin utils/calc+info: rework --best-memattr to allow multiple nodes --best-memattr was very strict when selecting best nodes. The node had to be the best for the entire input CPUs. On a dual-socket machine with HBM in each socket, each HBM is the best local but not the best remote. Hence we'd report no best for the entire machine. Now we return both HBM for the entire machine by default, but may go back to the previous behavior by adding ",strict" after the attribute name. Adding ",default" also allows to return all nodes if no best was found (if no attribute values are found). Thanks to Antoine Morvan for the report. Refs #652 Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/hwloc-info.c (diff) The file was modified
utils/hwloc/misc.h (diff) The file was modified
utils/hwloc/hwloc-calc.c (diff) The file was modified
utils/hwloc/hwloc-calc.1in (diff) The file was modified
utils/hwloc/hwloc-info.1in (diff) The file was modified
utils/hwloc/hwloc-calc.c (diff) The file was modified
utils/hwloc/hwloc-info.c (diff) The file was modified
utils/hwloc/misc.h (diff) The file was modified
utils/hwloc/hwloc-bind.c (diff) The file was modified
utils/hwloc/hwloc-bind.1in (diff) The file was modified
utils/hwloc/misc.h (diff)
Commit
25aacbdd05444325c3450ba734b732b17731e018
by Brice.Goglin tests/x86: add a Zhaoxin KH-40000 Its topology enumeration CPUID leaf return 0 in max_procs in the last level, hence we cannot break-out of the enumeration using AMD's condition. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
tests/hwloc/x86/Makefile.am (diff) The file was added tests/hwloc/x86/Zhaoxin-CentaurHauls-KaiSheng-KH-40000.tar.bz2 The file was added tests/hwloc/x86/Zhaoxin-CentaurHauls-KaiSheng-KH-40000.test The file was added tests/hwloc/x86/Zhaoxin-CentaurHauls-KaiSheng-KH-40000.xml The file was modified
hwloc/topology-x86.c (diff) The file was modified
hwloc/topology-x86.c (diff)
Commit
f9df603ede427b8905a9ced5838c2cfaed188629
by Brice.Goglin tests/xml: update the Zhaoxin KH-40000 cpuid test Use a recent 2-socket output. The old one was single-socket and the BIOS was outdated, causing some CPUIDs to be wrong. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was removed tests/hwloc/x86/Zhaoxin-CentaurHauls-KaiSheng-KH-40000.test The file was added tests/hwloc/x86/Zhaoxin-2xCentaurHauls-KaiSheng-KH-40000.test The file was removed tests/hwloc/x86/Zhaoxin-CentaurHauls-KaiSheng-KH-40000.xml The file was added tests/hwloc/x86/Zhaoxin-2xCentaurHauls-KaiSheng-KH-40000.xml The file was removed tests/hwloc/x86/Zhaoxin-CentaurHauls-KaiSheng-KH-40000.tar.bz2 The file was modified
tests/hwloc/x86/Makefile.am (diff)
Commit
5d761ed922c19ca6c970ef8b0530cae1de286dfc
by Brice.Goglin x86: update topology enumeration loop break-out conditions Zhaoxin confirmed they follow the Intel specs in commercial products, so assume we can break-out of the enumeration loop once one of the Intel or AMD condition is met. Fixes #593. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-x86.c (diff) The file was added tests/hwloc/x86/Zhaoxin-2xCentaurHauls-KaiSheng-KH-40000.tar.bz2
Commit
f90b69c101f126fec938421ad5aa12d5fc17c84f
by Brice.Goglin linux: try standard cgroup/cpuset mount point before reading /proc/mounts cgroups are pretty much always mounted on /sys/fs/cgroup nowadays, both for v1 and v2. And the obsolete cpuset filesystem was often mounted on /dev/cpuset. Try those before reading /proc/mounts. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
hwloc/topology-linux.c (diff) The file was modified
utils/hwloc/hwloc-info.1in (diff) The file was modified
utils/hwloc/hwloc-info.c (diff) The file was modified
utils/hwloc/hwloc-info.c (diff)
Commit
742a3852a83b18bc07be97ae195308730970b4d3
by Brice.Goglin utils/info: "levels", "topology" and "support" may now be passed as target objects And can be combined. This will deprecate --support --topology and --objects in v3. "topology" is already used in hwloc-annotate to manipulate topology info attrs. Signed-off-by: Brice Goglin <Brice.Goglin@inria.fr> The file was modified
utils/hwloc/hwloc-info.1in (diff) The file was modified
utils/hwloc/hwloc-info.c (diff) The file was modified
utils/hwloc/test-hwloc-info.output (diff) The file was modified
utils/hwloc/test-hwloc-info.sh.in (diff) The file was modified
utils/hwloc/hwloc-info.1in (diff) The file was modified
utils/hwloc/hwloc-info.c (diff) The file was modified
NEWS (diff) The file was modified
utils/hwloc/hwloc-info.c (diff) The file was modified
utils/hwloc/hwloc-info.1in (diff) The file was modified
utils/hwloc/hwloc-info.c (diff) The file was modified
utils/hwloc/test-hwloc-info.sh.in (diff) The file was modified
contrib/completion/bash/hwloc (diff) The file was modified
utils/hwloc/test-hwloc-info.output (diff) The file was modified
hwloc/topology.c (diff) The file was modified
include/hwloc/rename.h (diff) The file was modified
include/private/private.h (diff) The file was modified
hwloc/memattrs.c (diff) The file was modified
tests/hwloc/linux/64intel64-fakeKNL-SNC4-hybrid-msc.console (diff) The file was modified
hwloc/topology-xml.c (diff) The file was modified
tests/hwloc/linux/fakeheteromemtiers.console (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
tests/hwloc/linux/nvidiagpunumanodes-kept.console (diff) The file was modified
hwloc/memattrs.c (diff) The file was modified
tests/hwloc/linux/64intel64-fakeKNL-SNC4-hybrid.console (diff) The file was modified
tests/hwloc/linux/fakememinitiators-1np2c+1npp+gi.console (diff) The file was modified
doc/hwloc.doxy (diff) The file was modified
contrib/ci.inria.fr/Jenkinsfile (diff) The file was modified
contrib/ci.inria.fr/Jenkinsfile (diff)