diff options
| author | Konstantin Belousov <kib@FreeBSD.org> | 2026-04-28 02:34:49 +0000 |
|---|---|---|
| committer | Konstantin Belousov <kib@FreeBSD.org> | 2026-05-01 01:06:42 +0000 |
| commit | 839d3266d8c6f6471cb92a3c0ae32eb16d117427 (patch) | |
| tree | a556e741c278a2f77596ce3e6da79e20770c5b1e | |
| parent | 8f46ba065143d9d89968a20b23844287d54c04a2 (diff) | |
uipc_shm.c: make large page allocation interruptible
in cases there is no page pressure or when the user lost patience
waiting for very large allocation. Other case is already handled by
vm_wait_intr().
Reported by: "Lizzie from Eden Emulator project"
Reviewed by: adrian, markj
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
Differential revision: https://reviews.freebsd.org/D56725
| -rw-r--r-- | sys/kern/uipc_shm.c | 8 |
1 files changed, 8 insertions, 0 deletions
diff --git a/sys/kern/uipc_shm.c b/sys/kern/uipc_shm.c index 0ad5be2e8d71..197409d97f0c 100644 --- a/sys/kern/uipc_shm.c +++ b/sys/kern/uipc_shm.c @@ -829,12 +829,20 @@ shm_dotruncate_largepage(struct shmfd *shmfd, off_t length, void *rl_cookie) * object lock might allowed mapping of them. */ while (object->size < newobjsz) { + error = sig_intr(); + if (error != 0) + return (error); m = vm_page_alloc_contig(object, object->size, aflags, pagesizes[psind] / PAGE_SIZE, 0, ~0, pagesizes[psind], 0, VM_MEMATTR_DEFAULT); if (m == NULL) { VM_OBJECT_WUNLOCK(object); + error = sig_intr(); + if (error != 0) { + VM_OBJECT_WLOCK(object); + return (error); + } if (shmfd->shm_lp_alloc_policy == SHM_LARGEPAGE_ALLOC_NOWAIT || (shmfd->shm_lp_alloc_policy == |