| Message ID | AANLkTimPwhK1NggZAkCzB8Q29Ogf5SswH1kBbH8WAjX7@mail.gmail.com |
|---|---|
| State | Accepted, archived |
| Headers | show |
On Fri, Jan 14, 2011 at 02:05:15PM -0800, Aditya Kali wrote: > Currently ext4_mb_load_buddy is called for every group, irrespective > of whether the group info is already in memory, while reading > /proc/fs/ext4/<partition>/mb_groups proc file. For the purpose of > mb_groups proc file, it is unnecessary to load the file group info > from disk if it was loaded in past. These calls to ext4_mb_load_buddy > make reading the mb_groups proc file expensive. > Also, the locks around ext4_get_group_info are not required. > This patch modifies the code to call ext4_mb_load_buddy only if the > group info had never been loaded into memory in past. It also removes > the mb group locking around ext4_get_group_info call. > > Signed-off-by: Aditya Kali <adityakali@google.com> I realized that this had dropped through the cracks. Sorry! I've queued this for the next merge window. - Ted -- To unsubscribe from this list: send the line "unsubscribe linux-ext4" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c index 851f49b..5738082 100644 --- a/fs/ext4/mballoc.c +++ b/fs/ext4/mballoc.c @@ -2171,8 +2171,9 @@ static int ext4_mb_seq_groups_show(struct seq_file *seq, void *v) struct super_block *sb = seq->private; ext4_group_t group = (ext4_group_t) ((unsigned long) v); int i; - int err; + int err, buddy_loaded = 0; struct ext4_buddy e4b; + struct ext4_group_info *grinfo; struct sg { struct ext4_group_info info;
Currently ext4_mb_load_buddy is called for every group, irrespective of whether the group info is already in memory, while reading /proc/fs/ext4/<partition>/mb_groups proc file. For the purpose of mb_groups proc file, it is unnecessary to load the file group info from disk if it was loaded in past. These calls to ext4_mb_load_buddy make reading the mb_groups proc file expensive. Also, the locks around ext4_get_group_info are not required. This patch modifies the code to call ext4_mb_load_buddy only if the group info had never been loaded into memory in past. It also removes the mb group locking around ext4_get_group_info call. Signed-off-by: Aditya Kali <adityakali@google.com> --- fs/ext4/mballoc.c | 23 +++++++++++++++-------- 1 files changed, 15 insertions(+), 8 deletions(-) ext4_grpblk_t counters[16]; @@ -2189,15 +2190,21 @@ static int ext4_mb_seq_groups_show(struct seq_file *seq, void *v) i = (sb->s_blocksize_bits + 2) * sizeof(sg.info.bb_counters[0]) + sizeof(struct ext4_group_info); - err = ext4_mb_load_buddy(sb, group, &e4b); - if (err) { - seq_printf(seq, "#%-5u: I/O error\n", group); - return 0; + grinfo = ext4_get_group_info(sb, group); + /* Load the group info in memory only if not already loaded. */ + if (unlikely(EXT4_MB_GRP_NEED_INIT(grinfo))) { + err = ext4_mb_load_buddy(sb, group, &e4b); + if (err) { + seq_printf(seq, "#%-5u: I/O error\n", group); + return 0; + } + buddy_loaded = 1; } - ext4_lock_group(sb, group); + memcpy(&sg, ext4_get_group_info(sb, group), i); - ext4_unlock_group(sb, group); - ext4_mb_unload_buddy(&e4b); + + if (buddy_loaded) + ext4_mb_unload_buddy(&e4b); seq_printf(seq, "#%-5u: %-5u %-5u %-5u [", group, sg.info.bb_free, sg.info.bb_fragments, sg.info.bb_first_free);