This patch is for ready kernel only. For new release the following patch should be used. [PATCH VZ9] fs/fuse kio: Refactor pcs_mr to avoid large contiguous memory allocation ________________________________ From: Liu Kui <[email protected]> Sent: 01 August 2025 20:55 To: [email protected] <[email protected]> Cc: Alexey Kuznetsov <[email protected]>; Andrey Zaitsev <[email protected]>; Konstantin Khorenko <[email protected]>; Kui Liu <[email protected]> Subject: [PATCH VZ9] fs/fuse kio: Use kvcalloc to prevent failure in fragmented systems
Address large memory allocation failure by replacing kcalloc with kvcalloc. This allows fallback to vmalloc for large allocations, as virtually contiguous meomory is sufficient here, improving reliability in highly fragmented systems Related to: #VSTOR-112413 https://virtuozzo.atlassian.net/browse/VSTOR-112413 Signed-off-by: Liu Kui <[email protected]> --- fs/fuse/kio/pcs/pcs_mr.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/fs/fuse/kio/pcs/pcs_mr.c b/fs/fuse/kio/pcs/pcs_mr.c index cbd3b440dd1b..28e04d1b2639 100644 --- a/fs/fuse/kio/pcs/pcs_mr.c +++ b/fs/fuse/kio/pcs/pcs_mr.c @@ -17,7 +17,7 @@ void pcs_umem_release(struct pcs_umem *umem) unpin_user_pages(umem->pages, umem->npages); atomic64_sub(umem->npages, &mm_s->pinned_vm); mmdrop(mm_s); - kfree(umem->pages); + kvfree(umem->pages); kfree(umem); } @@ -48,7 +48,7 @@ struct pcs_umem *pcs_umem_get(u64 start, u64 len) mmap_read_lock(mm_s); umem->fp_addr = fp_va; - umem->pages = kcalloc(npages, sizeof(struct page *), GFP_KERNEL); + umem->pages = kvcalloc(npages, sizeof(struct page *), GFP_KERNEL); if (!umem->pages) { ret = -ENOMEM; goto out_err; -- 2.39.5 (Apple Git-154)
_______________________________________________ Devel mailing list [email protected] https://lists.openvz.org/mailman/listinfo/devel
