KMC_KVMEM disrupts kv_alloc() memory alignment expectations

On kernels with KASAN enabled the following failure can be observed as
soon as the zfs module is loaded:

  VERIFY(IS_P2ALIGNED(ptr, PAGE_SIZE)) failed
  PANIC at spl-kmem-cache.c:228:kv_alloc()

The problem is kmalloc() has never guaranteed aligned allocations; this
requirement resulted in zfsonlinux/spl@8b45dda which removed all
kmalloc() usage in kv_alloc().

Until a GFP_ALIGNED flag (or equivalent functionality) is provided by
the kernel this commit partially reverts 66955885 and 6d948c35 to
prevent k(v)malloc() allocations in kv_alloc().

Reviewed-by: Kjeld Schouten <kjeld@schouten-lebbing.nl>
Reviewed-by: Michael Niewöhner <foss@mniewoehner.de>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: loli10K <ezomori.nozomu@gmail.com>
Closes #9813
This commit is contained in:
loli10K 2020-01-14 18:09:59 +01:00 committed by Brian Behlendorf
parent 68a192e4b7
commit 7e2da7786e
1 changed files with 2 additions and 20 deletions

View File

@ -202,26 +202,8 @@ kv_alloc(spl_kmem_cache_t *skc, int size, int flags)
if (skc->skc_flags & KMC_KMEM) {
ASSERT(ISP2(size));
ptr = (void *)__get_free_pages(lflags, get_order(size));
} else if (skc->skc_flags & KMC_KVMEM) {
ptr = spl_kvmalloc(size, lflags);
} else {
/*
* GFP_KERNEL allocations can safely use kvmalloc which may
* improve performance by avoiding a) high latency caused by
* vmalloc's on-access allocation, b) performance loss due to
* MMU memory address mapping and c) vmalloc locking overhead.
* This has the side-effect that the slab statistics will
* incorrectly report this as a vmem allocation, but that is
* purely cosmetic.
*
* For non-GFP_KERNEL allocations we stick to __vmalloc.
*/
if ((lflags & GFP_KERNEL) == GFP_KERNEL) {
ptr = spl_kvmalloc(size, lflags);
} else {
ptr = __vmalloc(size, lflags | __GFP_HIGHMEM,
PAGE_KERNEL);
}
ptr = __vmalloc(size, lflags | __GFP_HIGHMEM, PAGE_KERNEL);
}
/* Resulting allocated memory will be page aligned */
@ -249,7 +231,7 @@ kv_free(spl_kmem_cache_t *skc, void *ptr, int size)
ASSERT(ISP2(size));
free_pages((unsigned long)ptr, get_order(size));
} else {
spl_kmem_free_impl(ptr, size);
vfree(ptr);
}
}