From: Jing Zhang <[email protected]>
Date: Wed Mar 25 23:15:06 2010
The function, ext4_mb_discard_group_preallocations(), works alone as
hard as possible without correct understanding its caller's good
thinking.
And now try to relieve it in simple way.
But the variable, free, may be replaced with freed.
Cc: Theodore Ts'o <[email protected]>
Cc: Andreas Dilger <[email protected]>
Cc: Dave Kleikamp <[email protected]>
Cc: "Aneesh Kumar K. V" <[email protected]>
Signed-off-by: Jing Zhang <[email protected]>
---
--- linux-2.6.32/fs/ext4/mballoc.c 2009-12-03 11:51:22.000000000 +0800
+++ ext4_mm_leak/mballoc-11-1.c 2010-03-25 23:06:00.000000000 +0800
@@ -3633,7 +3633,7 @@ repeat:
spin_lock(&pa->pa_lock);
if (atomic_read(&pa->pa_count)) {
spin_unlock(&pa->pa_lock);
- busy = 1;
+ busy += 1;
continue;
}
if (pa->pa_deleted) {
@@ -3654,7 +3654,7 @@ repeat:
}
/* if we still need more blocks and some PAs were used, try again */
- if (free < needed && busy) {
+ if (free < needed && needed <= free + busy) {
busy = 0;
ext4_unlock_group(sb, group);
/*
From: Jing Zhang <[email protected]>
Date: Wed Mar 25 23:15:06 2010
The function, ext4_mb_discard_group_preallocations(), works alone as
hard as possible without correct understanding its caller's good thinking.
With more understanding the work, the caller should be responsible for
the mission.
And the callee just does the correct, then feeds back, like chip.
Then both get relieved.
Cc: Theodore Ts'o <[email protected]>
Cc: Andreas Dilger <[email protected]>
Cc: Dave Kleikamp <[email protected]>
Cc: "Aneesh Kumar K. V" <[email protected]>
Signed-off-by: Jing Zhang <[email protected]>
---
--- linux-2.6.32/fs/ext4/mballoc.c 2009-12-03 11:51:22.000000000 +0800
+++ ext4_mm_leak/mballoc-11-2.c 2010-03-25 23:56:24.000000000 +0800
@@ -3595,8 +3595,6 @@ ext4_mb_discard_group_preallocations(str
struct ext4_allocation_context *ac;
struct list_head list;
struct ext4_buddy e4b;
- int err;
- int busy = 0;
int free = 0;
mb_debug(1, "discard preallocation for group %u\n", group);
@@ -3611,29 +3609,25 @@ ext4_mb_discard_group_preallocations(str
return 0;
}
- err = ext4_mb_load_buddy(sb, group, &e4b);
- if (err) {
+ if (ext4_mb_load_buddy(sb, group, &e4b)) {
ext4_error(sb, __func__, "Error in loading buddy "
"information for %u", group);
put_bh(bitmap_bh);
return 0;
}
- if (needed == 0)
- needed = EXT4_BLOCKS_PER_GROUP(sb) + 1;
INIT_LIST_HEAD(&list);
ac = kmem_cache_alloc(ext4_ac_cachep, GFP_NOFS);
if (ac)
ac->ac_sb = sb;
-repeat:
+
ext4_lock_group(sb, group);
list_for_each_entry_safe(pa, tmp,
&grp->bb_prealloc_list, pa_group_list) {
spin_lock(&pa->pa_lock);
if (atomic_read(&pa->pa_count)) {
spin_unlock(&pa->pa_lock);
- busy = 1;
continue;
}
if (pa->pa_deleted) {
@@ -3653,23 +3647,6 @@ repeat:
list_add(&pa->u.pa_tmp_list, &list);
}
- /* if we still need more blocks and some PAs were used, try again */
- if (free < needed && busy) {
- busy = 0;
- ext4_unlock_group(sb, group);
- /*
- * Yield the CPU here so that we don't get soft lockup
- * in non preempt case.
- */
- yield();
- goto repeat;
- }
-
- /* found anything to free? */
- if (list_empty(&list)) {
- BUG_ON(free != 0);
- goto out;
- }
/* now free all selected PAs */
list_for_each_entry_safe(pa, tmp, &list, u.pa_tmp_list) {
@@ -3688,7 +3665,7 @@ repeat:
call_rcu(&(pa)->u.pa_rcu, ext4_mb_pa_callback);
}
-out:
+
ext4_unlock_group(sb, group);
if (ac)
kmem_cache_free(ext4_ac_cachep, ac);
@@ -4183,14 +4160,24 @@ static int ext4_mb_discard_preallocation
ext4_group_t i, ngroups = ext4_get_groups_count(sb);
int ret;
int freed = 0;
+ int tried = 0;
trace_ext4_mb_discard_preallocations(sb, needed);
+again:
for (i = 0; i < ngroups && needed > 0; i++) {
ret = ext4_mb_discard_group_preallocations(sb, i, needed);
freed += ret;
needed -= ret;
}
+ if (needed > 0)
+ /* log it */
+ mb_debug(1, "%d PAs undiscard\n", needed);
+ if (! freed && ! tried++)
+ /* try to avoid -ENOSPC */
+ goto again;
+ if (! freed)
+ mb_debug(1, "discard PAs failed\n");
return freed;
}