mm/slab: introduce alloc_flags and slab_alloc_context
Get series with b4
b4 am [email protected] change-id 20260601-slab_alloc_flags-25c782b0c57cVersions newest first
- v3 Latest15 patches 23 repliesIn review
- v216 patches 60 repliesSuperseded
- v1 RFC15 patches 10 repliesSuperseded
Changes in v3 from the cover letter
- Applied R-b tags from Harry, Hao, Suren (thanks!)
- Former Patch 1 "mm/slab: do not limit zeroing to orig_size when only
red zoning is enabled" fast tracked as a fix to slab-for-7.2 PR.
- Patch 1: refactor kasan_init handling (Harry).
- Constify struct slab_alloc_context usage eveywhere (Suren)
- Rename SLAB_ALLOC_TRYLOCK to SLAB_ALLOC_NOLOCK (Suren, Alexei)
- Reorder patches 5 and 6 (formerly 6 7) (Suren)
- Move trynode_flags refactoring from 7 to 6 to avoid bisection
hazard.
- In Patch 14, support temporarily both __GFP_NO_OBJ_EXT and SLAB_ALLOC_NO_RECURSE to prevent obj_ext -> sheaves -> obj_ext
recursion (Sashiko)
- Expand OBJCGS_CLEAR_MASK to allow kmalloc_nolock() warnings
(Hao Li, Shengming Hu).
- Link to v2: https://patch.msgid.link/[email protected]
Patches in v3 No applies recorded yet
Reviews trailers per patch
All 15 patches have a Reviewed-by or Acked-by.
| Patch | Harry Yoo (Oracle) | Suren Baghdasaryan | Hao Li | Marco Elver |
|---|---|---|---|---|
| 1/15 | R-b | R-b | R-b | R-b |
| 2/15 | R-b | R-b | R-b | |
| 3/15 | R-b | R-b | R-b | |
| 4/15 | R-b | R-b | R-b | |
| 5/15 | R-b | R-b | R-b | |
| 6/15 | R-b | R-b | R-b | |
| 7/15 | R-b | R-b | R-b | |
| 8/15 | R-b | R-b | R-b | |
| 9/15 | R-b | R-b | R-b | |
| 10/15 | R-b | R-b | R-b | |
| 11/15 | R-b | R-b | R-b | |
| 12/15 | R-b | R-b | R-b | |
| 13/15 | R-b | R-b | R-b | |
| 14/15 | R-b | R-b | R-b | |
| 15/15 | R-b | R-b | R-b |
Trailers given in replies are filled; those carried in the patches from earlier versions are outlined with dashes. Scroll the table sideways for all 4 reviewers.
Cover letter [PATCH v3 00/15] mm/slab: introduce alloc_flags and slab_alloc_context
This series is now in slab/for-next, based on the slab-for-7.2 tag that was sent as first PR to Linus. Posting new version due to many accumulated changes, for final rounds of review. The plan is to send a second slab PR with this early next week, if nothing explodes.
Git: https://git.kernel.org/pub/scm/linux/kernel/git/vbabka/linux.git/log/?h=b4/slab_alloc_flags
The slab implementation currently relies on gfp flags to convey some context information internally:
- The absence of both __GFP_RECLAIM flags is interpreted as "cannot spin on locks", and intended to be used by kmalloc_nolock(). But false positives are possible e.g. during early boot where gfp_allowed_mask clears __GFP_RECLAIM from all allocations. This leads to unnecessary allocation failures and workarounds such as fd3634312a04 ("debugobject:
Make it work with deferred page initialization - again").
- __GFP_NO_OBJ_EXT exists and takes up valuable bit in the gfp flags space, only to prevent recursive kmalloc() allocations for obj_ext arrays and sheaves.
The page allocator uses its internal alloc_flags to convey various context information, including ALLOC_TRYLOCK (meaning "cannot spin"). This series copies that concept for the slab allocator, with its own slab-specific internal flags:
- SLAB_ALLOC_DEFAULT - no extra flags (the value is 0), but explicit
- SLAB_ALLOC_NOLOCK - do not spin on locks (used by kmalloc_nolock())
- SLAB_ALLOC_NEW_SLAB - replacing existing 'bool new_slab' parameter for allocating obj_ext arrays
- SLAB_ALLOC_NO_RECURSE - replacing usage of __GFP_NO_OBJ_EXT
To reduce the amount of parameters in various internal functions, we additionally introduce slab_alloc_context (also inspired by page allocator's alloc_context) for passing a number of existing arguments and the new alloc_flags:
/* Structure holding extra parameters for slab allocations */
struct slab_alloc_context {
unsigned long caller_addr;
size_t orig_size;
unsigned int alloc_flags;
struct list_lru *lru;
};This also replaces the existing struct partial_context.
The last necessary piece is kmalloc_flags() which can take the alloc_flags in addition to gfp flags and is intended for the recursive allocations of sheaves and obj_ext arrays, so that both SLAB_ALLOC_NOLOCK and SLAB_ALLOC_NO_RECURSE can be communicated. Internally it decides between kmalloc_nolock() and normal kmalloc() depending on SLAB_ALLOC_NOLOCK.
The rest of the series is gradually expanding the usage of both alloc_flags and slab_alloc_context as necessary, with bits of refactoring. Then, __GFP_NO_OBJ_EXT is removed completely.
Note that some usage of gfpflags_allow_spinning() relying on absence of __GFP_RECLAIM remains outside of slab (and page allocator) in memcg, page_owner and stackdepot code. These can thus yield false-positive decisions that spinning is not allowed, but should not result in important allocations failing anymore.
---
Changes in v3:
- Applied R-b tags from Harry, Hao, Suren (thanks!)
- Former Patch 1 "mm/slab: do not limit zeroing to orig_size when only
red zoning is enabled" fast tracked as a fix to slab-for-7.2 PR.
- Patch 1: refactor kasan_init handling (Harry).
- Constify struct slab_alloc_context usage eveywhere (Suren)
- Rename SLAB_ALLOC_TRYLOCK to SLAB_ALLOC_NOLOCK (Suren, Alexei)
- Reorder patches 5 and 6 (formerly 6 7) (Suren)
- Move trynode_flags refactoring from 7 to 6 to avoid bisection
hazard.
- In Patch 14, support temporarily both __GFP_NO_OBJ_EXT and SLAB_ALLOC_NO_RECURSE to prevent obj_ext -> sheaves -> obj_ext
recursion (Sashiko)
- Expand OBJCGS_CLEAR_MASK to allow kmalloc_nolock() warnings
(Hao Li, Shengming Hu).
- Link to v2: https://patch.msgid.link/[email protected]
Changes in v2:
- Due to Sashiko review, drop the idea of zeroing orig_size
unconditionally, as it can break krealloc(). Thanks to that found a pre-existing bug fixed by the new Patch 1. The kfence zeroing related cleanup is implemented differently in Patch 2.
- Prevent nested kmalloc_nolock warnings due to added gfp flags
(Sashiko)
- Fix a pre-existing issue with opportunistic slab allocation from the
target node only effectively dropping __GFP_NOMEMALLOC and __GFP_RECLAIM. (Sashiko)
- Move kmalloc_flags() definitions to mm/slab.h (per Harry).
- Link to v1: https://patch.msgid.link/[email protected]
---
Vlastimil Babka (SUSE) (15):
mm/slab: do not init any kfence objects on allocation mm/slab: stop inlining __slab_alloc_node()
mm/slab: introduce slab_alloc_context
mm/slab: introduce alloc_flags and SLAB_ALLOC_NOLOCK mm/slab: replace struct partial_context with slab_alloc_context mm/slab: add alloc_flags to slab_alloc_context
mm/slab: pass alloc_flags to new slab allocation
mm/slab: pass alloc_flags through slab_post_alloc_hook() chain mm/slab: replace slab_alloc_node() parameters with slab_alloc_context mm/slab: allow kmem_cache_alloc_bulk() with any gfp flags mm/slab: pass slab_alloc_context to __do_kmalloc_node() mm/slab: allow __GFP_NOMEMALLOC and __GFP_NOWARN for kmalloc_nolock() mm/slab: introduce kmalloc_flags()
mm/slab: remove __GFP_NO_OBJ_EXT usage from alloc_slab_obj_exts() mm/slab: replace __GFP_NO_OBJ_EXT with SLAB_ALLOC_NO_RECURSE for sheaves
include/linux/slab.h | 5 +- mm/kfence/core.c | 2 +- mm/memcontrol.c | 5 +- mm/slab.h | 29 ++- mm/slub.c | 488 +++++++++++++++++++++++++++++++-------------------- 5 files changed, 329 insertions(+), 200 deletions(-)
---
base-commit: dfdfd58cce1c3f5df8733b64595448996c08e424
change-id: 20260601-slab_alloc_flags-25c782b0c57c
Best regards,
Signature · 1 line
Vlastimil Babka (SUSE) <[email protected]>