Re: [f2fs-dev] [PATCH 3/5] f2fs: keep wait_ms if EAGAIN happens

[Chao Yu]

On 2022/5/7 7:20, Jaegeuk Kim wrote:
> In f2fs_gc thread, let's keep wait_ms when sec_freed was zero.

sec_freed won't increase for background GC due to below statement:

		if (gc_type == FG_GC &&
				get_valid_blocks(sbi, segno, false) == 0)
			seg_freed++;

It may cause gc thread migrates lots of segments in each round?

Thanks,

> Signed-off-by: Jaegeuk Kim <jaegeuk@xxxxxxxxxx>
> ---
>   fs/f2fs/gc.c | 4 ++--
>   1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/fs/f2fs/gc.c b/fs/f2fs/gc.c
> index aeffcc1d5c02..ec3f6f876e76 100644
> --- a/fs/f2fs/gc.c
> +++ b/fs/f2fs/gc.c
> @@ -37,7 +37,8 @@ static int gc_thread_func(void *data)
>   	unsigned int wait_ms;
>   	struct f2fs_gc_control gc_control = {
>   		.victim_segno = NULL_SEGNO,
> -		.should_migrate_blocks = false };
> +		.should_migrate_blocks = false,
> +		.err_gc_skipped = false };
>   
>   	wait_ms = gc_th->min_sleep_time;
>   
> @@ -146,7 +147,6 @@ static int gc_thread_func(void *data)
>   
>   		gc_control.init_gc_type = sync_mode ? FG_GC : BG_GC;
>   		gc_control.no_bg_gc = foreground;
> -		gc_control.err_gc_skipped = sync_mode;
>   
>   		/* if return value is not zero, no victim was selected */
>   		if (f2fs_gc(sbi, &gc_control))