From 895cb689d310e49cc21040f565d63dd34cc60ab3 Mon Sep 17 00:00:00 2001 From: oromenahar Date: Tue, 8 Aug 2023 18:37:06 +0200 Subject: [PATCH] zfs_clone_range should return a descriptive error codes Return the more descriptive error codes instead of `EXDEV` when the parameters don't match the requirements of the clone function. Updated the comments in `brt.c` accordingly. The first three errors are just invalid parameters, which zfs can not handle. The fourth error indicates that the block which should be cloned is created and cloned or modified in the same transaction group (`txg`). Reviewed-by: Brian Behlendorf Reviewed-by: Rob Norris Signed-off-by: Kay Pedersen Closes #15148 --- module/os/freebsd/zfs/zfs_vnops_os.c | 2 +- module/os/linux/zfs/zpl_file_range.c | 4 ++-- module/zfs/brt.c | 6 +++--- module/zfs/zfs_vnops.c | 13 +++++++------ 4 files changed, 13 insertions(+), 12 deletions(-) diff --git a/module/os/freebsd/zfs/zfs_vnops_os.c b/module/os/freebsd/zfs/zfs_vnops_os.c index 45cf6fdfc4..513accf18a 100644 --- a/module/os/freebsd/zfs/zfs_vnops_os.c +++ b/module/os/freebsd/zfs/zfs_vnops_os.c @@ -6290,7 +6290,7 @@ zfs_freebsd_copy_file_range(struct vop_copy_file_range_args *ap) error = zfs_clone_range(VTOZ(invp), ap->a_inoffp, VTOZ(outvp), ap->a_outoffp, &len, ap->a_outcred); - if (error == EXDEV || error == EOPNOTSUPP) + if (error == EXDEV || error == EINVAL || error == EOPNOTSUPP) goto bad_locked_fallback; *ap->a_lenp = (size_t)len; out_locked: diff --git a/module/os/linux/zfs/zpl_file_range.c b/module/os/linux/zfs/zpl_file_range.c index 72384b638b..43ba9a4982 100644 --- a/module/os/linux/zfs/zpl_file_range.c +++ b/module/os/linux/zfs/zpl_file_range.c @@ -103,7 +103,7 @@ zpl_copy_file_range(struct file *src_file, loff_t src_off, * Since Linux 5.3 the filesystem driver is responsible for executing * an appropriate fallback, and a generic fallback function is provided. */ - if (ret == -EOPNOTSUPP || ret == -EXDEV) + if (ret == -EOPNOTSUPP || ret == -EINVAL || ret == -EXDEV) ret = generic_copy_file_range(src_file, src_off, dst_file, dst_off, len, flags); #else @@ -111,7 +111,7 @@ zpl_copy_file_range(struct file *src_file, loff_t src_off, * Before Linux 5.3 the filesystem has to return -EOPNOTSUPP to signal * to the kernel that it should fallback to a content copy. */ - if (ret == -EXDEV) + if (ret == -EINVAL || ret == -EXDEV) ret = -EOPNOTSUPP; #endif /* HAVE_VFS_GENERIC_COPY_FILE_RANGE */ diff --git a/module/zfs/brt.c b/module/zfs/brt.c index e8218fb268..ddd8eefe60 100644 --- a/module/zfs/brt.c +++ b/module/zfs/brt.c @@ -174,7 +174,7 @@ * size_t len, unsigned int flags); * * Even though offsets and length represent bytes, they have to be - * block-aligned or we will return the EXDEV error so the upper layer can + * block-aligned or we will return an error so the upper layer can * fallback to the generic mechanism that will just copy the data. * Using copy_file_range(2) will call OS-independent zfs_clone_range() function. * This function was implemented based on zfs_write(), but instead of writing @@ -192,9 +192,9 @@ * Some special cases to consider and how we address them: * - The block we want to clone may have been created within the same * transaction group that we are trying to clone. Such block has no BP - * allocated yet, so cannot be immediately cloned. We return EXDEV. + * allocated yet, so cannot be immediately cloned. We return EAGAIN. * - The block we want to clone may have been modified within the same - * transaction group. We return EXDEV. + * transaction group. We return EAGAIN. * - A block may be cloned multiple times during one transaction group (that's * why pending list is actually a tree and not an append-only list - this * way we can figure out faster if this block is cloned for the first time diff --git a/module/zfs/zfs_vnops.c b/module/zfs/zfs_vnops.c index 8ffcf37134..a3d54c6e67 100644 --- a/module/zfs/zfs_vnops.c +++ b/module/zfs/zfs_vnops.c @@ -1028,6 +1028,10 @@ zfs_exit_two(zfsvfs_t *zfsvfs1, zfsvfs_t *zfsvfs2, const char *tag) * * On success, the function return the number of bytes copied in *lenp. * Note, it doesn't return how much bytes are left to be copied. + * On errors which are caused by any file system limitations or + * brt limitations `EINVAL` is returned. In the most cases a user + * requested bad parameters, it could be possible to clone the file but + * some parameters don't match the requirements. */ int zfs_clone_range(znode_t *inzp, uint64_t *inoffp, znode_t *outzp, @@ -1171,7 +1175,7 @@ zfs_clone_range(znode_t *inzp, uint64_t *inoffp, znode_t *outzp, * We cannot clone into files with different block size. */ if (inblksz != outzp->z_blksz && outzp->z_size > inblksz) { - error = SET_ERROR(EXDEV); + error = SET_ERROR(EINVAL); goto unlock; } @@ -1179,7 +1183,7 @@ zfs_clone_range(znode_t *inzp, uint64_t *inoffp, znode_t *outzp, * Offsets and len must be at block boundries. */ if ((inoff % inblksz) != 0 || (outoff % inblksz) != 0) { - error = SET_ERROR(EXDEV); + error = SET_ERROR(EINVAL); goto unlock; } /* @@ -1187,7 +1191,7 @@ zfs_clone_range(znode_t *inzp, uint64_t *inoffp, znode_t *outzp, */ if ((len % inblksz) != 0 && (len < inzp->z_size - inoff || len < outzp->z_size - outoff)) { - error = SET_ERROR(EXDEV); + error = SET_ERROR(EINVAL); goto unlock; } @@ -1246,9 +1250,6 @@ zfs_clone_range(znode_t *inzp, uint64_t *inoffp, znode_t *outzp, * in the current transaction group. Return an error, * so the caller can fallback to just copying the data. */ - if (error == EAGAIN) { - error = SET_ERROR(EXDEV); - } break; } /*