Pull zonefs fixes from Damien Le Moal: "Two fixes in here: - Revert the initial decision to silently ignore IOCB_NOWAIT for asynchronous direct IOs to sequential zone files. Instead, return an error to the user to signal that the feature is not supported (from Christoph) - A fix to zonefs Kconfig to select FS_IOMAP to avoid build failures if no other file system already selected this option (from Johannes)" * tag 'zonefs-5.6-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/dlemoal/zonefs: zonefs: select FS_IOMAP zonefs: fix IOCB_NOWAIT handling
Unified · Split
| @@ -2,6 +2,7 @@ config ZONEFS_FS · 1 unmodified line | ||
| 2 | 2 | tristate "zonefs filesystem support" |
| 3 | 3 | depends on BLOCK |
| 4 | 4 | depends on BLK_DEV_ZONED |
| 5 | + select FS_IOMAP | |
| 5 | 6 | help |
| 6 | 7 | zonefs is a simple file system which exposes zones of a zoned block |
| 7 | 8 | device (e.g. host-managed or host-aware SMR disk drives) as files. |
| @@ -601,13 +601,13 @@ static ssize_t zonefs_file_dio_write(struct kiocb *iocb, struct iov_iter *from) · 600 unmodified lines | ||
| 601 | 601 | ssize_t ret; |
| 602 | 602 | |
| 603 | 603 | /* |
| 604 | − * For async direct IOs to sequential zone files, ignore IOCB_NOWAIT | |
| 604 | + * For async direct IOs to sequential zone files, refuse IOCB_NOWAIT | |
| 605 | 605 | * as this can cause write reordering (e.g. the first aio gets EAGAIN |
| 606 | 606 | * on the inode lock but the second goes through but is now unaligned). |
| 607 | 607 | */ |
| 608 | − if (zi->i_ztype == ZONEFS_ZTYPE_SEQ && !is_sync_kiocb(iocb) | |
| 609 | − && (iocb->ki_flags & IOCB_NOWAIT)) | |
| 610 | − iocb->ki_flags &= ~IOCB_NOWAIT; | |
| 608 | + if (zi->i_ztype == ZONEFS_ZTYPE_SEQ && !is_sync_kiocb(iocb) && | |
| 609 | + (iocb->ki_flags & IOCB_NOWAIT)) | |
| 610 | + return -EOPNOTSUPP; | |
| 611 | 611 | |
| 612 | 612 | if (iocb->ki_flags & IOCB_NOWAIT) { |
| 613 | 613 | if (!inode_trylock(inode)) |