RE: [PATCH] mmc: card: Fixup request missing in mmc_blk_issue_rw_rq

[Justin Wang (çä)]

> -----Original Message-----
> From: Ulf Hansson [mailto:ulf.hansson@xxxxxxxxxx]
> Sent: Thursday, May 28, 2015 4:29 PM
> To: Justin Wang (çä)
> Cc: kuninori.morimoto.gx@xxxxxxxxxxx; jh80.chung@xxxxxxxxxxx;
> akpm@xxxxxxxxxxxxxxxxxxxx; JBottomley@xxxxxxxx; ben@xxxxxxxxxxxxxxx;
> chuanxiao.dong@xxxxxxxxx; linux-mmc@xxxxxxxxxxxxxxx;
> linux-kernel@xxxxxxxxxxxxxxx
> Subject: Re: [PATCH] mmc: card: Fixup request missing in mmc_blk_issue_rw_rq
> Importance: High
> 
> On 19 May 2015 at 08:32, Justin Wang (çä) <Justin.Wang@xxxxxxxxxxxxxx>
> wrote:
> > From 05849da563c80c20597ab6275d5881a8ed426f96 Mon Sep 17 00:00:00
> 2001
> > From: justin.wang <justin.wang@xxxxxxxxxxxxxx>
> > Date: Mon, 18 May 2015 20:14:15 +0800
> > Subject: [PATCH] mmc: card: Fixup request missing in mmc_blk_issue_rw_rq
> >
> > The current handler of MMC_BLK_CMD_ERR in mmc_blk_issue_rw_rq
> function
> > may cause new coming request permanent missing when the ongoing
> > request (previoulsy started) complete end.
> 
> I think you need to elaborate a bit more here. In exactly what
> scenario does this problem occur?
> 

Thanks, I will give a detailed description next patch.

> >
> > This would cause the process related to the missing request stay at 'D'
> > state forever.
> 
> 'D' state?
> 
> I suppose it waits for a response for its IO request. Doesn't also the
> block layer complain or timeout somehow? Would be interesting to know
> a bit more about what really happens when things goes wrong.
> 

Yes, process will wait the missed request complete forever, no timeout in block layer.
The related process will be blocked.
> >
> 
> Seems like we should have a fixes tag here as well, could you possibly
> try to find what commit that introduced this bug?
> 
> > Signed-off-by: Ding Wang <justin.wang@xxxxxxxxxxxxxx>
> > ---
> >  drivers/mmc/card/block.c |    8 +++++---
> >  1 files changed, 5 insertions(+), 3 deletions(-)
> >
> > diff --git a/drivers/mmc/card/block.c b/drivers/mmc/card/block.c
> > index 60f7141..f05cd1f 100644
> > --- a/drivers/mmc/card/block.c
> > +++ b/drivers/mmc/card/block.c
> > @@ -1910,9 +1910,11 @@ static int mmc_blk_issue_rw_rq(struct
> mmc_queue *mq, struct request *rqc)
> >                         break;
> >                 case MMC_BLK_CMD_ERR:
> >                         ret = mmc_blk_cmd_err(md, card, brq, req,
> ret);
> > -                       if (!mmc_blk_reset(md, card->host, type))
> > -                               break;
> > -                       goto cmd_abort;
> > +                       if (mmc_blk_reset(md, card->host, type))
> > +                               goto cmd_abort;
> > +                       if (!ret)
> > +                               goto start_new_req;
> > +                       break;
> >                 case MMC_BLK_RETRY:
> >                         if (retry++ < 5)
> >                                 break;
> > --
> > 1.7.4.1
> >
> 
> Kind regards
> Uffe