From 0da15f9194bfd181f26207c9daac420e6a3f9609 Mon Sep 17 00:00:00 2001 From: Mark Johnston Date: Fri, 21 Jan 2022 14:54:05 -0500 Subject: [PATCH] Fix handling of errors from dmu_write_uio_dbuf() on FreeBSD FreeBSD's implementation of zfs_uio_fault_move() returns EFAULT when a page fault occurs while copying data in or out of user buffers. The VFS treats such errors specially and will retry the I/O operation (which may have made some partial progress). When the FreeBSD and Linux implementations of zfs_write() were merged, the handling of errors from dmu_write_uio_dbuf() changed such that EFAULT is not handled as a partial write. For example, when appending to a file, the z_size field of the znode is not updated after a partial write resulting in EFAULT. Restore the old handling of errors from dmu_write_uio_dbuf() to fix this. This should have no impact on Linux, which has special handling for EFAULT already. Reviewed-by: Andriy Gapon Reviewed-by: Ryan Moeller Signed-off-by: Mark Johnston Closes #12964 --- module/zfs/zfs_vnops.c | 15 +++++++++++---- 1 file changed, 11 insertions(+), 4 deletions(-) diff --git a/module/zfs/zfs_vnops.c b/module/zfs/zfs_vnops.c index 170e392abe..54749810d4 100644 --- a/module/zfs/zfs_vnops.c +++ b/module/zfs/zfs_vnops.c @@ -323,7 +323,7 @@ out: int zfs_write(znode_t *zp, zfs_uio_t *uio, int ioflag, cred_t *cr) { - int error = 0; + int error = 0, error1; ssize_t start_resid = zfs_uio_resid(uio); /* @@ -561,7 +561,11 @@ zfs_write(znode_t *zp, zfs_uio_t *uio, int ioflag, cred_t *cr) continue; } #endif - if (error != 0) { + /* + * On FreeBSD, EFAULT should be propagated back to the + * VFS, which will handle faulting and will retry. + */ + if (error != 0 && error != EFAULT) { dmu_tx_commit(tx); break; } @@ -645,7 +649,7 @@ zfs_write(znode_t *zp, zfs_uio_t *uio, int ioflag, cred_t *cr) while ((end_size = zp->z_size) < zfs_uio_offset(uio)) { (void) atomic_cas_64(&zp->z_size, end_size, zfs_uio_offset(uio)); - ASSERT(error == 0); + ASSERT(error == 0 || error == EFAULT); } /* * If we are replaying and eof is non zero then force @@ -655,7 +659,10 @@ zfs_write(znode_t *zp, zfs_uio_t *uio, int ioflag, cred_t *cr) if (zfsvfs->z_replay && zfsvfs->z_replay_eof != 0) zp->z_size = zfsvfs->z_replay_eof; - error = sa_bulk_update(zp->z_sa_hdl, bulk, count, tx); + error1 = sa_bulk_update(zp->z_sa_hdl, bulk, count, tx); + if (error1 != 0) + /* Avoid clobbering EFAULT. */ + error = error1; zfs_log_write(zilog, tx, TX_WRITE, zp, woff, tx_bytes, ioflag, NULL, NULL);