[PATCH v2 0/1] Encapsulate the populate and alloc as one atomic

[Hubert Mazur <hmazur@google.com>]

Hello,
thanks for the review of the v1 patchset. I tried to make v2 diff as
small as possible and without a modification of the core logic.

When a block of memory is requested from the execmem manager the
free_areas tree is traversed to find area of given size. If it is not
found then a new fragment, aligned to a PAGE_SIZE, is allocated and
added to free_areas. Afterwards, the free_areas tree is being traversed
again to fullfil the request.

The above operations of allocation and tree traversal are not atomic
hence another request may consume this newly allocated memory
block dedicated to the original request. As a result - the first
request fails to get the memory. Such occurence can be spotted on
evices running the 6.18 kernel during the paralell modules loading.

Regards
Hubert

Changes in v2:
The __execmem_cache_alloc_locked function (lockless version of
__execmem_cache_alloc) is introduced and called after
execmem_cache_add_locked from the __execmem_cache_populate_alloc
function (renamed from execmem_cache_populate). Both calls are
guarded now with a single mutex.

Changes in v1:
Allocate new memory fragment and assign it directly to the busy_areas
inside execmem_cache_populate function.

Link to v1:
https://lore.kernel.org/all/20260312131438.361746-1-hmazur@google.com/T/#t

Hubert Mazur (1):
  mm/execmem: Make the populate and alloc atomic

 mm/execmem.c | 61 +++++++++++++++++++++++++++++-----------------------
 1 file changed, 34 insertions(+), 27 deletions(-)

--
2.53.0.851.ga537e3e6e9-goog