diff options
| author | Matthew Ahrens <[email protected]> | 2014-09-13 15:40:05 +0200 |
|---|---|---|
| committer | Brian Behlendorf <[email protected]> | 2014-10-23 15:30:32 -0700 |
| commit | 9635861742c6ac334877f944e17948aa11c2e57a (patch) | |
| tree | 2fe4744c568f26163707363d6c6c3cb22eb627ae /module/zfs/metaslab.c | |
| parent | b02fe35d3743c92e26f4158938e94093a2309a92 (diff) | |
Illumos 5164-5165 - space map fixes
5164 space_map_max_blksz causes panic, does not work
5165 zdb fails assertion when run on pool with recently-enabled
space map_histogram feature
Reviewed by: Christopher Siden <[email protected]>
Reviewed by: George Wilson <[email protected]>
Reviewed by: Saso Kiselkov <[email protected]>
Approved by: Dan McDonald <[email protected]>
References:
https://www.illumos.org/issues/5164
https://www.illumos.org/issues/5165
https://github.com/illumos/illumos-gate/commit/b1be289
Porting Notes:
The metaslab_fragmentation() hunk was dropped from this patch
because it was already resolved by commit 8b0a084.
The comment modified in metaslab.c was updated to use the correct
variable name, space_map_blksz. The upstream commit incorrectly
used space_map_blksize.
Ported by: Turbo Fredriksson <[email protected]>
Signed-off-by: Brian Behlendorf <[email protected]>
Closes #2697
Diffstat (limited to 'module/zfs/metaslab.c')
| -rw-r--r-- | module/zfs/metaslab.c | 14 |
1 files changed, 10 insertions, 4 deletions
diff --git a/module/zfs/metaslab.c b/module/zfs/metaslab.c index d41a570c5..76410ec41 100644 --- a/module/zfs/metaslab.c +++ b/module/zfs/metaslab.c @@ -67,7 +67,7 @@ int zfs_condense_pct = 200; /* * Condensing a metaslab is not guaranteed to actually reduce the amount of * space used on disk. In particular, a space map uses data in increments of - * MAX(1 << ashift, SPACE_MAP_INITIAL_BLOCKSIZE), so a metaslab might use the + * MAX(1 << ashift, space_map_blksz), so a metaslab might use the * same number of blocks after condensing. Since the goal of condensing is to * reduce the number of IOPs required to read the space map, we only want to * condense when we can be sure we will reduce the number of blocks used by the @@ -1864,6 +1864,15 @@ metaslab_sync(metaslab_t *msp, uint64_t txg) mutex_enter(&msp->ms_lock); + /* + * Note: metaslab_condense() clears the space_map's histogram. + * Therefore we muse verify and remove this histogram before + * condensing. + */ + metaslab_group_histogram_verify(mg); + metaslab_class_histogram_verify(mg->mg_class); + metaslab_group_histogram_remove(mg, msp); + if (msp->ms_loaded && spa_sync_pass(spa) == 1 && metaslab_should_condense(msp)) { metaslab_condense(msp, txg, tx); @@ -1872,9 +1881,6 @@ metaslab_sync(metaslab_t *msp, uint64_t txg) space_map_write(msp->ms_sm, *freetree, SM_FREE, tx); } - metaslab_group_histogram_verify(mg); - metaslab_class_histogram_verify(mg->mg_class); - metaslab_group_histogram_remove(mg, msp); if (msp->ms_loaded) { /* * When the space map is loaded, we have an accruate |