diff options
| author | Brian Behlendorf <[email protected]> | 2018-03-28 10:19:22 -0700 |
|---|---|---|
| committer | GitHub <[email protected]> | 2018-03-28 10:19:22 -0700 |
| commit | b2ab468dded24a090dba9fc8396b3d394011ce14 (patch) | |
| tree | 52ce015e474ee028851256f3b423075b7cf73f58 /module | |
| parent | 668173b5764386b47de0beb7c57ff8ce8f68f983 (diff) | |
Fix mmap / libaio deadlock
Calling uiomove() in mappedread() under the page lock can result
in a deadlock if the user space page needs to be faulted in.
Resolve the issue by dropping the page lock before the uiomove().
The inode range lock protects against concurrent updates via
zfs_read() and zfs_write().
Reviewed-by: Albert Lee <[email protected]>
Reviewed-by: Chunwei Chen <[email protected]>
Signed-off-by: Brian Behlendorf <[email protected]>
Closes #7335
Closes #7339
Diffstat (limited to 'module')
| -rw-r--r-- | module/zfs/zfs_vnops.c | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/module/zfs/zfs_vnops.c b/module/zfs/zfs_vnops.c index a2d7d7b24..14caa80e5 100644 --- a/module/zfs/zfs_vnops.c +++ b/module/zfs/zfs_vnops.c @@ -400,6 +400,7 @@ mappedread(struct inode *ip, int nbytes, uio_t *uio) pp = find_lock_page(mp, start >> PAGE_SHIFT); if (pp) { ASSERT(PageUptodate(pp)); + unlock_page(pp); pb = kmap(pp); error = uiomove(pb + off, bytes, UIO_READ, uio); @@ -409,7 +410,6 @@ mappedread(struct inode *ip, int nbytes, uio_t *uio) flush_dcache_page(pp); mark_page_accessed(pp); - unlock_page(pp); put_page(pp); } else { error = dmu_read_uio_dbuf(sa_get_db(zp->z_sa_hdl), |