f2fs: avoid resetting non empty zone

[PATCH] f2fs: avoid resetting non empty zone

Posted by Sheng Yong 1 year, 6 months ago

If curseg is not the first segment in its zone, the zone is not empty,
and it should not be reset. This issue could be reproduced by:

	modprobe null_blk nr_devices=1 zoned=1 zone_max_open=6 zone_max_active=6 zone_size=1024 gb=30
	# /dev/vda is 4G
	mkfs.f2fs -m -c /dev/nullb0 /dev/vda -f

	while :; do
		mount /dev/vda /mnt/
		dd if=/dev/zero of=/mnt/file bs=4K count=11 conv=fsync status=none
		if [ $? -ne 0 ]; then
			umount /mnt
			break
		fi
		f2fs_io shutdown 1 /mnt/file
		umount /mnt
	done

And the error looks like:
[  123.169852] I/O error, dev nullb0, sector 41951232 op 0x1:(WRITE) flags 0x800 phys_seg 10 prio class 0
[  123.173070] F2FS-fs (vda): do_checkpoint failed err:-5, stop checkpoint
dd: error writing '/mnt/testfile': Input/output error

Signed-off-by: Sheng Yong <shengyong@oppo.com>
---
 fs/f2fs/segment.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
index 4c8836ded90fc..50b38cbe33401 100644
--- a/fs/f2fs/segment.c
+++ b/fs/f2fs/segment.c
@@ -5004,7 +5004,8 @@ static int fix_curseg_write_pointer(struct f2fs_sb_info *sbi, int type)
 	}
 
 	/* Allocate a new section if it's not new. */
-	if (cs->next_blkoff) {
+	if (cs->next_blkoff ||
+	    cs->segno != GET_SEG_FROM_SEC(sbi, GET_ZONE_FROM_SEC(sbi, cs_section))) {
 		unsigned int old_segno = cs->segno, old_blkoff = cs->next_blkoff;
 
 		f2fs_allocate_new_section(sbi, type, true);
-- 
2.40.1

Re: [PATCH] f2fs: avoid resetting non empty zone

Posted by Chao Yu 1 year, 6 months ago

On 2024/5/29 19:24, Sheng Yong wrote:
> If curseg is not the first segment in its zone, the zone is not empty,
> and it should not be reset. This issue could be reproduced by:
> 
> 	modprobe null_blk nr_devices=1 zoned=1 zone_max_open=6 zone_max_active=6 zone_size=1024 gb=30
> 	# /dev/vda is 4G
> 	mkfs.f2fs -m -c /dev/nullb0 /dev/vda -f
> 
> 	while :; do
> 		mount /dev/vda /mnt/
> 		dd if=/dev/zero of=/mnt/file bs=4K count=11 conv=fsync status=none
> 		if [ $? -ne 0 ]; then
> 			umount /mnt
> 			break
> 		fi
> 		f2fs_io shutdown 1 /mnt/file
> 		umount /mnt
> 	done
> 
> And the error looks like:
> [  123.169852] I/O error, dev nullb0, sector 41951232 op 0x1:(WRITE) flags 0x800 phys_seg 10 prio class 0
> [  123.173070] F2FS-fs (vda): do_checkpoint failed err:-5, stop checkpoint
> dd: error writing '/mnt/testfile': Input/output error
> 
> Signed-off-by: Sheng Yong <shengyong@oppo.com>

Reviewed-by: Chao Yu <chao@kernel.org>

Thanks,

[PATCH v2] f2fs: alloc new section if curseg is not the first seg in its zone

Posted by Sheng Yong 1 year, 6 months ago

If curseg is not the first segment in its zone, the zone is not empty.
A new section should be allocated and avoid resetting the old zone.

Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Sheng Yong <shengyong@oppo.com>
---
v2: remove and update inaccurate commit msg
---
 fs/f2fs/segment.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
index 4c8836ded90fc..50b38cbe33401 100644
--- a/fs/f2fs/segment.c
+++ b/fs/f2fs/segment.c
@@ -5004,7 +5004,8 @@ static int fix_curseg_write_pointer(struct f2fs_sb_info *sbi, int type)
 	}
 
 	/* Allocate a new section if it's not new. */
-	if (cs->next_blkoff) {
+	if (cs->next_blkoff ||
+	    cs->segno != GET_SEG_FROM_SEC(sbi, GET_ZONE_FROM_SEC(sbi, cs_section))) {
 		unsigned int old_segno = cs->segno, old_blkoff = cs->next_blkoff;
 
 		f2fs_allocate_new_section(sbi, type, true);
-- 
2.40.1