f2fs: expose main_blkaddr in sysfs
Subsystems (2)
Jaegeuk Kim
maintainerPATCH 1/2 f2fs: expose main_blkaddr in sysfs
Expose in /sys/fs/f2fs/<blockdev>/main_blkaddr the block address where the main area starts. This allows user mode programs to determine:
- That pinned files that are made exclusively of fully allocated 2MB segments will never be unpinned by the file system.
- Where the main area starts. This is required by programs that want to verify if a file is made exclusively of 2MB f2fs segments, the alignment boundary for segments starts at this address. Testing for 2MB alignment relative to the start of the device is incorrect, because for some filesystems main_blkaddr is not at a 2MB boundary relative to the start of the device.
The entry will be used when validating reliable pinning file feature proposed by "f2fs: support aligned pinned file".
--- Documentation/ABI/testing/sysfs-fs-f2fs | 6 ++++++ Documentation/filesystems/f2fs.txt | 3 +++ fs/f2fs/sysfs.c | 2 ++ 3 files changed, 11 insertions(+)
Documentation/ABI/testing/sysfs-fs-f2fs +6 -0
@@ -31,6 +31,12 @@ Contact: "Jaegeuk Kim" <[email protected]> Description: Controls the issue rate of segment discard commands. +What: /sys/fs/f2fs/<disk>/max_blkaddr +Date: November 2019 +Contact: "Ramon Pantin" <[email protected]> +Description: + Shows first block address of MAIN area. + What: /sys/fs/f2fs/<disk>/ipu_policy Date: November 2013 Contact: "Jaegeuk Kim" <[email protected]>
Documentation/filesystems/f2fs.txt +3 -0
@@ -297,6 +297,9 @@ Files in /sys/fs/f2fs/<devname> reclaim the prefree segments to free segments. By default, 5% over total # of segments. + main_blkaddr This value gives the first block address of + MAIN area in the partition. + max_small_discards This parameter controls the number of discard commands that consist small blocks less than 2MB. The candidates to be discarded are cached until
fs/f2fs/sysfs.c +2 -0
@@ -445,6 +445,7 @@ F2FS_RW_ATTR(GC_THREAD, f2fs_gc_kthread, gc_no_gc_sleep_time, no_gc_sleep_time);F2FS_RW_ATTR (F2FS_SBI , f2fs_sb_info, gc_idle, gc_mode);F2FS_RW_ATTR (F2FS_SBI , f2fs_sb_info, gc_urgent, gc_mode);F2FS_RW_ATTR (SM_INFO , f2fs_sm_info, reclaim_segments, rec_prefree_segments); +F2FS_RW_ATTR (SM_INFO , f2fs_sm_info, main_blkaddr, main_blkaddr);F2FS_RW_ATTR (DCC_INFO , discard_cmd_control, max_small_discards, max_discards);F2FS_RW_ATTR (DCC_INFO , discard_cmd_control, discard_granularity, discard_granularity);F2FS_RW_ATTR (RESERVED_BLOCKS , f2fs_sb_info, reserved_blocks, reserved_blocks);
@@ -512,6 +513,7 @@ static struct attribute *f2fs_attrs[] = {ATTR_LIST (gc_idle),ATTR_LIST (gc_urgent),ATTR_LIST (reclaim_segments), +ATTR_LIST (main_blkaddr),ATTR_LIST (max_small_discards),ATTR_LIST (discard_granularity), ATTR_LIST(batched_trim_sections),
Signature · 1 line
2.19.0.605.g01d371f741-goog
Jaegeuk Kim
maintainerPATCH 2/2 f2fs: stop GC when the victim becomes fully valid
We must stop GC, once the segment becomes fully valid. Otherwise, it can produce another dirty segments by moving valid blocks in the segment partially.
Ramon hit no free segment panic sometimes and saw this case happens when validating reliable file pinning feature.
--- fs/f2fs/gc.c | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-)
fs/f2fs/gc.c +8 -2
@@ -1012,8 +1012,14 @@ static int gc_data_segment(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,block_t start_bidx;nid_t nid =le32_to_cpu (entry->nid); -/* stop BG_GC if there is not enough free sections. */ -if (gc_type ==BG_GC &&has_not_enough_free_secs (sbi,0 ,0 )) +/* +* stop BG_GC if there is not enough free sections. +* Or, stop GC if the segment becomes fully valid caused by +* race condition along with SSR block allocation. +*/ +if ((gc_type ==BG_GC &&has_not_enough_free_secs (sbi,0 ,0 )) || +get_valid_blocks (sbi, segno, false) == + sbi->blocks_per_seg)return submitted;if (check_valid_map (sbi, segno, off) ==0 )
Signature · 1 line
2.19.0.605.g01d371f741-goog