kernel/tools/testing/selftests/cgroup/test_memcontrol.c, branch linux-6.2.y

selftests: cgroup: make sure reclaim target memcg is unprotected

2022-12-12T02:12:19Z

Make sure that we ignore protection of a memcg that is the target of memcg reclaim. Link: https://lkml.kernel.org/r/20221202031512.1365483-4-yosryahmed@google.com Signed-off-by: Yosry Ahmed Reviewed-by: Roman Gushchin Cc: Chris Down Cc: David Rientjes Cc: Johannes Weiner Cc: Matthew Wilcox Cc: Michal Hocko Cc: Muchun Song Cc: Shakeel Butt Cc: Tejun Heo Cc: Vasily Averin Cc: Vlastimil Babka Cc: Yu Zhao Signed-off-by: Andrew Morton

selftests: cgroup: refactor proactive reclaim code to reclaim_until()

2022-12-12T02:12:19Z

Refactor the code that drives writing to memory.reclaim (retrying, error handling, etc) from test_memcg_reclaim() to a helper called reclaim_until(), which proactively reclaims from a memcg until its usage reaches a certain value. While we are at it, refactor and simplify the reclaim loop. This will be used in a following patch in another test. Link: https://lkml.kernel.org/r/20221202031512.1365483-3-yosryahmed@google.com Signed-off-by: Yosry Ahmed Suggested-by: Roman Gushchin Reviewed-by: Roman Gushchin Cc: Chris Down Cc: David Rientjes Cc: Johannes Weiner Cc: Matthew Wilcox Cc: Michal Hocko Cc: Muchun Song Cc: Shakeel Butt Cc: Tejun Heo Cc: Vasily Averin Cc: Vlastimil Babka Cc: Yu Zhao Signed-off-by: Andrew Morton

selftests: memcg: factor out common parts of memory.{low,min} tests

2022-05-27T16:33:45Z

The memory protection test setup and runtime is almost equal for memory.low and memory.min cases. It makes modification of the common parts prone to mistakes, since the protections are similar not only in setup but also in principle, factor the common part out. Past exceptions between the tests: - missing memory.min is fine (kept), - test_memcg_low protected orphaned pagecache (adapted like test_memcg_min and we keep the processes of protected memory running). The evaluation in two tests is different (OOM of allocator vs low events of protégés), this is kept different. Link: https://lkml.kernel.org/r/20220518161859.21565-6-mkoutny@suse.com Signed-off-by: Michal Koutný Acked-by: Roman Gushchin CC: Johannes Weiner Cc: Michal Hocko Cc: Shakeel Butt Cc: Richard Palethorpe Cc: David Vernet Signed-off-by: Andrew Morton

selftests: memcg: remove protection from top level memcg

2022-05-27T16:33:45Z

The reclaim is triggered by memory limit in a subtree, therefore the testcase does not need configured protection against external reclaim. Also, correct respective comments. Link: https://lkml.kernel.org/r/20220518161859.21565-5-mkoutny@suse.com Signed-off-by: Michal Koutný Acked-by: Roman Gushchin Cc: David Vernet Cc: Johannes Weiner Cc: Michal Hocko Cc: Richard Palethorpe Cc: Shakeel Butt Signed-off-by: Andrew Morton

selftests: memcg: adjust expected reclaim values of protected cgroups

2022-05-27T16:33:45Z

The numbers are not easy to derive in a closed form (certainly mere protections ratios do not apply), therefore use a simulation to obtain expected numbers. Link: https://lkml.kernel.org/r/20220518161859.21565-4-mkoutny@suse.com Signed-off-by: Michal Koutný Acked-by: Roman Gushchin Cc: David Vernet Cc: Johannes Weiner Cc: Michal Hocko Cc: Richard Palethorpe Cc: Shakeel Butt Signed-off-by: Andrew Morton

selftests: memcg: expect no low events in unprotected sibling

2022-05-27T16:33:45Z

This is effectively a revert of commit cdc69458a5f3 ("cgroup: account for memory_recursiveprot in test_memcg_low()"). The case test_memcg_low will fail with memory_recursiveprot until resolved in reclaim code. However, this patch preserves the existing helpers and variables for later uses. Link: https://lkml.kernel.org/r/20220518161859.21565-3-mkoutny@suse.com Signed-off-by: Michal Koutný Reviewed-by: David Vernet Cc: Johannes Weiner Cc: Michal Hocko Cc: Richard Palethorpe Cc: Roman Gushchin Cc: Shakeel Butt Signed-off-by: Andrew Morton

selftests: memcg: fix compilation

2022-05-27T16:33:44Z

Patch series "memcontrol selftests fixups", v2. Flushing the patches to make memcontrol selftests check the events behavior we had consensus about (test_memcg_low fails). (test_memcg_reclaim, test_memcg_swap_max fail for me now but it's present even before the refactoring.) The two bigger changes are: - adjustment of the protected values to make tests succeed with the given tolerance, - both test_memcg_low and test_memcg_min check protection of memory in populated cgroups (actually as per Documentation/admin-guide/cgroup-v2.rst memory.min should not apply to empty cgroups, which is not the case currently. Therefore I unified tests with the populated case in order to to bring more broken tests). This patch (of 5): This fixes mis-applied changes from commit 72b1e03aa725 ("cgroup: account for memory_localevents in test_memcg_oom_group_leaf_events()"). Link: https://lkml.kernel.org/r/20220518161859.21565-1-mkoutny@suse.com Link: https://lkml.kernel.org/r/20220518161859.21565-2-mkoutny@suse.com Signed-off-by: Michal Koutný Reviewed-by: David Vernet Acked-by: Roman Gushchin Cc: Johannes Weiner Cc: Michal Hocko Cc: Richard Palethorpe Cc: Shakeel Butt Signed-off-by: Andrew Morton

cgroup: fix an error handling path in alloc_pagecache_max_30M()

2022-05-25T17:47:46Z

If the first goto is taken, 'fd' is not opened yet (and is un-initialized). So a direct return is safer. Link: https://lkml.kernel.org/r/628312312eb40e0e39463a2c06415fde5295c716.1653229120.git.christophe.jaillet@wanadoo.fr Fixes: c1a31a2f7a9c ("cgroup: fix racy check in alloc_pagecache_max_30M() helper function") Signed-off-by: Christophe JAILLET Reviewed-by: Andrew Morton Cc: Dan Carpenter Cc: Johannes Weiner Cc: Michal Hocko Cc: Roman Gushchin Cc: Shakeel Butt Cc: Muchun Song Cc: Tejun Heo Cc: Zefan Li Cc: Shuah Khan Cc: David Vernet Signed-off-by: Andrew Morton

cgroup: fix racy check in alloc_pagecache_max_30M() helper function

2022-05-13T14:20:12Z

alloc_pagecache_max_30M() in the cgroup memcg tests performs a 50MB pagecache allocation, which it expects to be capped at 30MB due to the calling process having a memory.high setting of 30MB. After the allocation, the function contains a check that verifies that MB(29) < memory.current <= MB(30). This check can actually fail non-deterministically. The testcases that use this function are test_memcg_high() and test_memcg_max(), which set memory.min and memory.max to 30MB respectively for the cgroup under test. The allocation can slightly exceed this number in both cases, and for memory.max, the process performing the allocation will not have the OOM killer invoked as it's performing a pagecache allocation. This patchset therefore updates the above check to instead use the verify_close() helper function. Link: https://lkml.kernel.org/r/20220423155619.3669555-6-void@manifault.com Signed-off-by: David Vernet Acked-by: Roman Gushchin Cc: Johannes Weiner Cc: Michal Hocko Cc: Shakeel Butt Cc: Tejun Heo Signed-off-by: Andrew Morton

cgroup: remove racy check in test_memcg_sock()

2022-05-13T14:20:12Z

test_memcg_sock() in the cgroup memcg tests, verifies expected memory accounting for sockets. The test forks a process which functions as a TCP server, and sends large buffers back and forth between itself (as the TCP client) and the forked TCP server. While doing so, it verifies that memory.current and memory.stat.sock look correct. There is currently a check in tcp_client() which asserts memory.current >= memory.stat.sock. This check is racy, as between memory.current and memory.stat.sock being queried, a packet could come in which causes mem_cgroup_charge_skmem() to be invoked. This could cause memory.stat.sock to exceed memory.current. Reversing the order of querying doesn't address the problem either, as memory may be reclaimed between the two calls. Instead, this patch just removes that assertion altogether, and instead relies on the values_close() check that follows to validate the expected accounting. Link: https://lkml.kernel.org/r/20220423155619.3669555-5-void@manifault.com Signed-off-by: David Vernet Acked-by: Roman Gushchin Cc: Johannes Weiner Cc: Michal Hocko Cc: Shakeel Butt Cc: Tejun Heo Signed-off-by: Andrew Morton