2023-07-16 14:28:37

by Ritesh Harjani (IBM)

[permalink] [raw]
Subject: [PATCH] ext4: Don't use CR_BEST_AVAIL_LEN for non-regular files

Using CR_BEST_AVAIL_LEN only make sense for regular files, as for
non-regular files we never normalize the allocation request length i.e.
goal len is same as original length (ac_g_ex.fe_len == ac_o_ex.fe_len).

Hence there is no scope of trimming the goal length to make it
satisfy original request len. Thus this patch avoids using
CR_BEST_AVAIL_LEN criteria for non-regular files request.

Fixes: 33122aa930f1 ("ext4: Add allocation criteria 1.5 (CR1_5)")
Reported-by: Eric Whitney <[email protected]>
Signed-off-by: Ritesh Harjani (IBM) <[email protected]>
---
fs/ext4/mballoc.c | 13 ++++++++++++-
1 file changed, 12 insertions(+), 1 deletion(-)

diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c
index 3ab37533349f..bc004f5d3f3c 100644
--- a/fs/ext4/mballoc.c
+++ b/fs/ext4/mballoc.c
@@ -975,7 +975,18 @@ static void ext4_mb_choose_next_group_goal_fast(struct ext4_allocation_context *
*group = grp->bb_group;
ac->ac_flags |= EXT4_MB_CR_GOAL_LEN_FAST_OPTIMIZED;
} else {
- *new_cr = CR_BEST_AVAIL_LEN;
+ /*
+ * CR_BEST_AVAIL_LEN works based on the concept that we have
+ * a larger normalized goal len request which can be trimmed to
+ * a smaller goal len such that it can still satisfy original
+ * request len. However, allocation request for non-regular
+ * files never gets normalized.
+ * See function ext4_mb_normalize_request() (EXT4_MB_HINT_DATA).
+ */
+ if (ac->ac_flags & EXT4_MB_HINT_DATA)
+ *new_cr = CR_BEST_AVAIL_LEN;
+ else
+ *new_cr = CR_GOAL_LEN_SLOW;
}
}

--
2.40.1



2023-07-16 16:39:22

by Ojaswin Mujoo

[permalink] [raw]
Subject: Re: [PATCH] ext4: Don't use CR_BEST_AVAIL_LEN for non-regular files

On Sun, Jul 16, 2023 at 07:33:34PM +0530, Ritesh Harjani (IBM) wrote:
> Using CR_BEST_AVAIL_LEN only make sense for regular files, as for
> non-regular files we never normalize the allocation request length i.e.
> goal len is same as original length (ac_g_ex.fe_len == ac_o_ex.fe_len).
>
> Hence there is no scope of trimming the goal length to make it
> satisfy original request len. Thus this patch avoids using
> CR_BEST_AVAIL_LEN criteria for non-regular files request.
>
> Fixes: 33122aa930f1 ("ext4: Add allocation criteria 1.5 (CR1_5)")
> Reported-by: Eric Whitney <[email protected]>
> Signed-off-by: Ritesh Harjani (IBM) <[email protected]>
> ---
Hi Ritesh,

patch looks good to me. Feel free to add:

Reviewed-by: Ojaswin Mujoo <ojaswin.linux.ibm.com>

Regards,
ojaswin
> fs/ext4/mballoc.c | 13 ++++++++++++-
> 1 file changed, 12 insertions(+), 1 deletion(-)
>
> diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c
> index 3ab37533349f..bc004f5d3f3c 100644
> --- a/fs/ext4/mballoc.c
> +++ b/fs/ext4/mballoc.c
> @@ -975,7 +975,18 @@ static void ext4_mb_choose_next_group_goal_fast(struct ext4_allocation_context *
> *group = grp->bb_group;
> ac->ac_flags |= EXT4_MB_CR_GOAL_LEN_FAST_OPTIMIZED;
> } else {
> - *new_cr = CR_BEST_AVAIL_LEN;
> + /*
> + * CR_BEST_AVAIL_LEN works based on the concept that we have
> + * a larger normalized goal len request which can be trimmed to
> + * a smaller goal len such that it can still satisfy original
> + * request len. However, allocation request for non-regular
> + * files never gets normalized.
> + * See function ext4_mb_normalize_request() (EXT4_MB_HINT_DATA).
> + */
> + if (ac->ac_flags & EXT4_MB_HINT_DATA)
> + *new_cr = CR_BEST_AVAIL_LEN;
> + else
> + *new_cr = CR_GOAL_LEN_SLOW;
> }
> }
>
> --
> 2.40.1
>

2023-07-17 21:40:59

by Eric Whitney

[permalink] [raw]
Subject: Re: [PATCH] ext4: Don't use CR_BEST_AVAIL_LEN for non-regular files

* Ritesh Harjani (IBM) <[email protected]>:
> Using CR_BEST_AVAIL_LEN only make sense for regular files, as for
> non-regular files we never normalize the allocation request length i.e.
> goal len is same as original length (ac_g_ex.fe_len == ac_o_ex.fe_len).
>
> Hence there is no scope of trimming the goal length to make it
> satisfy original request len. Thus this patch avoids using
> CR_BEST_AVAIL_LEN criteria for non-regular files request.
>
> Fixes: 33122aa930f1 ("ext4: Add allocation criteria 1.5 (CR1_5)")
> Reported-by: Eric Whitney <[email protected]>
> Signed-off-by: Ritesh Harjani (IBM) <[email protected]>
> ---
> fs/ext4/mballoc.c | 13 ++++++++++++-
> 1 file changed, 12 insertions(+), 1 deletion(-)
>
> diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c
> index 3ab37533349f..bc004f5d3f3c 100644
> --- a/fs/ext4/mballoc.c
> +++ b/fs/ext4/mballoc.c
> @@ -975,7 +975,18 @@ static void ext4_mb_choose_next_group_goal_fast(struct ext4_allocation_context *
> *group = grp->bb_group;
> ac->ac_flags |= EXT4_MB_CR_GOAL_LEN_FAST_OPTIMIZED;
> } else {
> - *new_cr = CR_BEST_AVAIL_LEN;
> + /*
> + * CR_BEST_AVAIL_LEN works based on the concept that we have
> + * a larger normalized goal len request which can be trimmed to
> + * a smaller goal len such that it can still satisfy original
> + * request len. However, allocation request for non-regular
> + * files never gets normalized.
> + * See function ext4_mb_normalize_request() (EXT4_MB_HINT_DATA).
> + */
> + if (ac->ac_flags & EXT4_MB_HINT_DATA)
> + *new_cr = CR_BEST_AVAIL_LEN;
> + else
> + *new_cr = CR_GOAL_LEN_SLOW;
> }
> }
>
> --
> 2.40.1
>

Works for me on 6.5-rc2 with this patch applied - 500/500 generic/269 trials
passed on bigalloc_1k.

Tested-by: Eric Whitney <[email protected]>

Thanks!
Eric

2023-08-05 12:26:35

by Theodore Ts'o

[permalink] [raw]
Subject: Re: [PATCH] ext4: Don't use CR_BEST_AVAIL_LEN for non-regular files


On Sun, 16 Jul 2023 19:33:34 +0530, Ritesh Harjani (IBM) wrote:
> Using CR_BEST_AVAIL_LEN only make sense for regular files, as for
> non-regular files we never normalize the allocation request length i.e.
> goal len is same as original length (ac_g_ex.fe_len == ac_o_ex.fe_len).
>
> Hence there is no scope of trimming the goal length to make it
> satisfy original request len. Thus this patch avoids using
> CR_BEST_AVAIL_LEN criteria for non-regular files request.
>
> [...]

Applied, thanks!

[1/1] ext4: Don't use CR_BEST_AVAIL_LEN for non-regular files
commit: 772c9f691dcf3a487f29ddb90a5a15c78d7328e1

Best regards,
--
Theodore Ts'o <[email protected]>