Re: [PATCH] f2fs: fix long latency due to discard during umount

[Jaegeuk Kim]

On 03/13, Sahitya Tummala wrote:
> On Thu, Mar 12, 2020 at 06:45:35PM -0700, Jaegeuk Kim wrote:
> > On 03/13, Sahitya Tummala wrote:
> > > On Thu, Mar 12, 2020 at 10:02:42AM -0700, Jaegeuk Kim wrote:
> > > > On 03/12, Sahitya Tummala wrote:
> > > > > F2FS already has a default timeout of 5 secs for discards that
> > > > > can be issued during umount, but it can take more than the 5 sec
> > > > > timeout if the underlying UFS device queue is already full and there
> > > > > are no more available free tags to be used. In that case, submit_bio()
> > > > > will wait for the already queued discard requests to complete to get
> > > > > a free tag, which can potentially take way more than 5 sec.
> > > > > 
> > > > > Fix this by submitting the discard requests with REQ_NOWAIT
> > > > > flags during umount. This will return -EAGAIN for UFS queue/tag full
> > > > > scenario without waiting in the context of submit_bio(). The FS can
> > > > > then handle these requests by retrying again within the stipulated
> > > > > discard timeout period to avoid long latencies.
> > > > > 
> > > > > Signed-off-by: Sahitya Tummala <stummala@xxxxxxxxxxxxxx>
> > > > > ---
> > > > >  fs/f2fs/segment.c | 14 +++++++++++++-
> > > > >  1 file changed, 13 insertions(+), 1 deletion(-)
> > > > > 
> > > > > diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
> > > > > index fb3e531..a06bbac 100644
> > > > > --- a/fs/f2fs/segment.c
> > > > > +++ b/fs/f2fs/segment.c
> > > > > @@ -1124,10 +1124,13 @@ static int __submit_discard_cmd(struct f2fs_sb_info *sbi,
> > > > >  	struct discard_cmd_control *dcc = SM_I(sbi)->dcc_info;
> > > > >  	struct list_head *wait_list = (dpolicy->type == DPOLICY_FSTRIM) ?
> > > > >  					&(dcc->fstrim_list) : &(dcc->wait_list);
> > > > > -	int flag = dpolicy->sync ? REQ_SYNC : 0;
> > > > > +	int flag;
> > > > >  	block_t lstart, start, len, total_len;
> > > > >  	int err = 0;
> > > > >  
> > > > > +	flag = dpolicy->sync ? REQ_SYNC : 0;
> > > > > +	flag |= dpolicy->type == DPOLICY_UMOUNT ? REQ_NOWAIT : 0;
> > > > > +
> > > > >  	if (dc->state != D_PREP)
> > > > >  		return 0;
> > > > >  
> > > > > @@ -1203,6 +1206,11 @@ static int __submit_discard_cmd(struct f2fs_sb_info *sbi,
> > > > >  		bio->bi_end_io = f2fs_submit_discard_endio;
> > > > >  		bio->bi_opf |= flag;
> > > > >  		submit_bio(bio);
> > > > > +		if ((flag & REQ_NOWAIT) && (dc->error == -EAGAIN)) {
> > > > > +			dc->state = D_PREP;
> > > > > +			err = dc->error;
> > > > > +			break;
> > > > > +		}
> > > > >  
> > > > >  		atomic_inc(&dcc->issued_discard);
> > > > >  
> > > > > @@ -1510,6 +1518,10 @@ static int __issue_discard_cmd(struct f2fs_sb_info *sbi,
> > > > >  			}
> > > > >  
> > > > >  			__submit_discard_cmd(sbi, dpolicy, dc, &issued);
> > > > > +			if (dc->error == -EAGAIN) {
> > > > > +				congestion_wait(BLK_RW_ASYNC, HZ/50);
> > > > 
> > > > 						--> need to be DEFAULT_IO_TIMEOUT
> > > 
> > > Yes, i will update it.
> > > 
> > > > 
> > > > > +				__relocate_discard_cmd(dcc, dc);
> > > > 
> > > > It seems we need to submit bio first, and then move dc to wait_list, if there's
> > > > no error, in __submit_discard_cmd().
> > > 
> > > Yes, that is not changed and it still happens for the failed request
> > > that is re-queued here too when it gets submitted again later.
> > > 
> > > I am requeuing the discard request failed with -EAGAIN error back to 
> > > dcc->pend_list[] from wait_list. It will call submit_bio() for this request
> > > and also move to wait_list when it calls __submit_discard_cmd() again next
> > > time. Please let me know if I am missing anything?
> > 
> > This patch has no problem, but I'm thinking that __submit_discard_cmd() needs
> > to return with any values by assumption where the waiting list should have
> > submitted commands.
> 
> I think dc->queued will indicated that dc is moved to wait_list. This can be
> used along with return value to take right action. Can you check if this
> works?

I mean why can't do this *in* __submit_discard_cmd()? Otherwise, existing and
future callers should consider to handle the errors everytime.

> 
> diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
> index a06bbac..91df060 100644
> --- a/fs/f2fs/segment.c
> +++ b/fs/f2fs/segment.c
> @@ -1478,7 +1478,7 @@ static int __issue_discard_cmd(struct f2fs_sb_info *sbi,
>         struct list_head *pend_list;
>         struct discard_cmd *dc, *tmp;
>         struct blk_plug plug;
> -       int i, issued = 0;
> +       int i, err, issued = 0;
>         bool io_interrupted = false;
> 
>         if (dpolicy->timeout != 0)
> @@ -1517,8 +1517,10 @@ static int __issue_discard_cmd(struct f2fs_sb_info *sbi,
>                                 break;
>                         }
> 
> -                       __submit_discard_cmd(sbi, dpolicy, dc, &issued);
> -                       if (dc->error == -EAGAIN) {
> +                       err = __submit_discard_cmd(sbi, dpolicy, dc, &issued);
> +                       if (err && err != -EAGAIN) {
> +                               __remove_discard_cmd(sbi, dc);
> +                       } else if (err == -EAGAIN && dc->queued) {
>                                 congestion_wait(BLK_RW_ASYNC, HZ/50);
>                                 __relocate_discard_cmd(dcc, dc);
>                         }
> 
> thanks,
> > 
> > > 
> > > Thanks,
> > > 
> > > > 
> > > > > +			}
> > > > >  
> > > > >  			if (issued >= dpolicy->max_requests)
> > > > >  				break;
> > > > > -- 
> > > > > Qualcomm India Private Limited, on behalf of Qualcomm Innovation Center, Inc.
> > > > > Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a Linux Foundation Collaborative Project.
> > > 
> > > -- 
> > > --
> > > Sent by a consultant of the Qualcomm Innovation Center, Inc.
> > > The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.
> 
> -- 
> --
> Sent by a consultant of the Qualcomm Innovation Center, Inc.
> The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.