ext4: allow to change s_last_trim_minblks via sysfs
authorLukas Czerner <lczerner@redhat.com>
Wed, 3 Nov 2021 14:51:22 +0000 (15:51 +0100)
committerTheodore Ts'o <tytso@mit.edu>
Mon, 10 Jan 2022 18:25:55 +0000 (13:25 -0500)
Ext4 has an optimization mechanism for batched disacrd (FITRIM) that
should help speed up subsequent calls of FITRIM ioctl by skipping the
groups that were previously trimmed. However because the FITRIM allows
to set the minimum size of an extent to trim, ext4 stores the last
minimum extent size and only avoids trimming the group if it was
previously trimmed with minimum extent size equal to, or smaller than
the current call.

There is currently no way to bypass the optimization without
umount/mount cycle. This becomes a problem when the file system is
live migrated to a different storage, because the optimization will
prevent possibly useful discard calls to the storage.

Fix it by exporting the s_last_trim_minblks via sysfs interface which
will allow us to set the minimum size to the number of blocks larger
than subsequent FITRIM call, effectively bypassing the optimization.

By setting the s_last_trim_minblks to ULONG_MAX the optimization will be
effectively cleared regardless of the previous state, or file system
configuration.

For example:
getconf ULONG_MAX > /sys/fs/ext4/dm-1/last_trim_minblks

Signed-off-by: Lukas Czerner <lczerner@redhat.com>
Reported-by: Laurent GUERBY <laurent@guerby.net>
Reviewed-by: Andreas Dilger <adilger@dilger.ca>
Link: https://lore.kernel.org/r/20211103145122.17338-2-lczerner@redhat.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
fs/ext4/sysfs.c

index 2a4ae3d105716e015c0dff52859518db0aba1613..f61e65ae27d8d78311a188fb3265ddceae7dc43a 100644 (file)
@@ -245,6 +245,7 @@ EXT4_ATTR(last_error_time, 0444, last_error_time);
 EXT4_ATTR(journal_task, 0444, journal_task);
 EXT4_RW_ATTR_SBI_UI(mb_prefetch, s_mb_prefetch);
 EXT4_RW_ATTR_SBI_UI(mb_prefetch_limit, s_mb_prefetch_limit);
+EXT4_RW_ATTR_SBI_UL(last_trim_minblks, s_last_trim_minblks);
 
 static unsigned int old_bump_val = 128;
 EXT4_ATTR_PTR(max_writeback_mb_bump, 0444, pointer_ui, &old_bump_val);
@@ -295,6 +296,7 @@ static struct attribute *ext4_attrs[] = {
 #endif
        ATTR_LIST(mb_prefetch),
        ATTR_LIST(mb_prefetch_limit),
+       ATTR_LIST(last_trim_minblks),
        NULL,
 };
 ATTRIBUTE_GROUPS(ext4);