d/freebsd-dev
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Vectorized fletcher_4 must be 128-bit aligned

Browse Source 
The fletcher_4_native() and fletcher_4_byteswap() functions may only
safely use the vectorized implementations when the buffer is 128-bit
aligned.  This is because both the AVX2 and SSE implementations process
four 32-bit words per iterations.  Fallback to the scalar implementation
which only processes a single 32-bit word for unaligned buffers.

Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Gvozden Neskovic <neskovic@gmail.com>
Issue #4330
This commit is contained in:
Brian Behlendorf 2016-06-28 13:31:21 -07:00
parent d1d19c7854
commit 0dab2e84fc
1 changed files with 12 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
module/zcommon/zfs_fletcher.c
View File

@ -334,7 +334,12 @@ fletcher_4_impl_get(void)
void
fletcher_4_native(const void *buf, uint64_t size, zio_cksum_t *zcp)
{
const fletcher_4_ops_t *ops = fletcher_4_impl_get();
const fletcher_4_ops_t *ops;
if (IS_P2ALIGNED(size, 4 * sizeof (uint32_t)))
ops = fletcher_4_impl_get();
else
ops = &fletcher_4_scalar_ops;
ops->init(zcp);
ops->compute(buf, size, zcp);
@ -345,7 +350,12 @@ fletcher_4_native(const void *buf, uint64_t size, zio_cksum_t *zcp)
void
fletcher_4_byteswap(const void *buf, uint64_t size, zio_cksum_t *zcp)
{
const fletcher_4_ops_t *ops = fletcher_4_impl_get();
const fletcher_4_ops_t *ops;
if (IS_P2ALIGNED(size, 4 * sizeof (uint32_t)))
ops = fletcher_4_impl_get();
else
ops = &fletcher_4_scalar_ops;
ops->init(zcp);
ops->compute_byteswap(buf, size, zcp);