Skip to content

Conversation

@KFilipek
Copy link
Contributor

Description

Checklist

  • Code compiles without errors locally
  • All tests pass locally
  • CI workflows execute properly
  • CI workflows, not executed per PR (e.g. Nightly), execute properly
  • New tests added, especially if they will fail without my changes
  • Added/extended example(s) to cover this functionality
  • Extended the README/documentation
  • All newly added source files have a license
  • All newly added source files are referenced in CMake files
  • Logger (with debug/info/... messages) is used
  • All API changes are reflected in docs and def/map files, and are tested

PatKamin and others added 30 commits March 6, 2025 12:17
For tag checkout it's no surprise; for main we print full version
during 'Configure' step.
[CI] Fix failing address sanitizer
…an_error_handling_path_of_umf_ba_alloc

Add missing `unlock()` in an error handling path of `umf_ba_alloc()`
Extend compatibility tests with GPU runners
Enable MT bench for disjoint_pool in fixed provider
…verity

check for alloc fails in disjoint pool init
Run Compatibility GPU CI jobs only on upstream,
since forks do not have the required HW.

Signed-off-by: Lukasz Dorau <[email protected]>
It fixes ThreadSanitizer data race in find_predecessor()
vs critnib_insert() and critnib_remove().

Signed-off-by: Lukasz Dorau <[email protected]>
Check valgrind log files. Do not print an error message like:

ls: cannot access './examples/umf_example_*.log': No such file or directory

when only the tests log files are present for example.

Signed-off-by: Lukasz Dorau <[email protected]>
…CI_jobs_only_on_upstream

Run Compatibility GPU CI jobs only on upstream
Use atomics in find_successor() like in find_predecessor().

Ref: oneapi-src#1175

Signed-off-by: Lukasz Dorau <[email protected]>
Multilevel maps are needed to support the case
when one memory pool acts as a memory provider
for another memory pool (nested memory pooling).

Signed-off-by: Lukasz Dorau <[email protected]>
lukaszstolarczuk and others added 27 commits March 12, 2025 13:18
and move the step gathering info to the end (in multi numa CI).
…ITIALIZED_error_in_cu2umf_result

Handle `CUDA_ERROR_DEINITIALIZED` error in `cu2umf_result()`
Fix segfault in cu_memory_provider_get_last_native_error()
when it is called after a CUDA device was destroyed.

Signed-off-by: Lukasz Dorau <[email protected]>
[CI] Enable SLES in multi numa workflow
…ory_provider_get_last_native_error

Fix segfault in `cu_memory_provider_get_last_native_error()`
…ults

use current ctx and dev by default in CUDA prov
…est_with_Fixed_provider

Enable jemalloc pool test with Fixed provider
Install libnuma-dev in the proxy lib workflow,
because it is required.

It has worked correctly so far most probably,
because libnuma-dev has been installed
as a dependency of libhwloc-dev.
It failed when libhwloc-dev was not installed.

Signed-off-by: Lukasz Dorau <[email protected]>
…_the_proxy_lib_workflow

Install libnuma-dev in the proxy lib workflow
Suppress false-positive in IPC tests under helgrind
Remove old SHM files /tmp/umf_file_provider_*
(at the beginning of the test)
left from the previous runs, because of crashes
of the ipc_file_prov test.

Signed-off-by: Lukasz Dorau <[email protected]>
…eft_from_the_previous_runs_because_of_crashes

Remove old SHM files left from the previous runs because of crashes
fix jemalloc benchmark with fixedprovider
increase number of threads in benchmarks
@KFilipek KFilipek marked this pull request as ready for review March 18, 2025 13:28
@KFilipek KFilipek requested a review from a team as a code owner March 18, 2025 13:28
@KFilipek KFilipek closed this Mar 18, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

7 participants