[U-Boot] [PATCH v2] mmc: fsl_esdhc: Avoid infinite loop in esdhc_send_cmd_common()

[Ricardo Salveti]

On Mon, Nov 19, 2018 at 12:03 PM Ricardo Salveti <rsalveti at rsalveti.net> wrote:
>
> Hi Fabio,
>
> On Mon, Nov 19, 2018 at 10:31 AM Fabio Estevam <festevam at gmail.com> wrote:
> >
> > The following hang is observed on a Hummingboard 2 MicroSOM
> > i2eX iMX6D - rev 1.3 with no eMMC populated on board:
> >
> > U-Boot SPL 2018.11+gf6206f8587 (Nov 16 2018 - 00:56:34 +0000)
> > Trying to boot from MMC1
> >
> > U-Boot 2018.11+gf6206f8587 (Nov 16 2018 - 00:56:34 +0000)
> >
> > CPU:   Freescale i.MX6D rev1.5 996 MHz (running at 792 MHz)
> > CPU:   Extended Commercial temperature grade (-20C to 105C) at 33C
> > Reset cause: POR
> > Board: MX6 HummingBoard2
> > DRAM:  1 GiB
> > MMC:   FSL_SDHC: 0, FSL_SDHC: 1
> > Loading Environment from MMC... *** Warning - bad CRC, using default environment
> >
> > No panel detected: default to HDMI
> > Display: HDMI (1024x768)
> > In:    serial
> > Out:   serial
> > Err:   serial
> > ---> hangs
> >
> > which is caused by the following infinite loop inside esdhc_send_cmd_common()
> >
> >         while (!(esdhc_read32(&regs->irqstat) & flags))
> >                 ;
> >
> > Instead of looping forever, provide an exit path so that a timeout
> > error can be propagated in the case irqstat does not report
> > any interrupts, which may happen when no eMMC is populated on
> > board.
> >
> > Reported-by: Ricardo Salveti <rsalveti at rsalveti.net>
> > Signed-off-by: Fabio Estevam <festevam at gmail.com>
> > ---
> > Changes since v1:
> > - Jump to the label out on error path (Baruch).
> >
> >  drivers/mmc/fsl_esdhc.c | 10 ++++++++--
> >  1 file changed, 8 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/mmc/fsl_esdhc.c b/drivers/mmc/fsl_esdhc.c
> > index 3cdfa7f..b8d0b00 100644
> > --- a/drivers/mmc/fsl_esdhc.c
> > +++ b/drivers/mmc/fsl_esdhc.c
> > @@ -396,6 +396,7 @@ static int esdhc_send_cmd_common(struct fsl_esdhc_priv *priv, struct mmc *mmc,
> >         uint    irqstat;
> >         u32     flags = IRQSTAT_CC | IRQSTAT_CTOE;
> >         struct fsl_esdhc *regs = priv->esdhc_regs;
> > +       unsigned long start;
> >
> >  #ifdef CONFIG_SYS_FSL_ERRATUM_ESDHC111
> >         if (cmd->cmdidx == MMC_CMD_STOP_TRANSMISSION)
> > @@ -453,8 +454,13 @@ static int esdhc_send_cmd_common(struct fsl_esdhc_priv *priv, struct mmc *mmc,
> >                 flags = IRQSTAT_BRR;
> >
> >         /* Wait for the command to complete */
> > -       while (!(esdhc_read32(&regs->irqstat) & flags))
> > -               ;
> > +       start = get_timer(0);
> > +       while (!(esdhc_read32(&regs->irqstat) & flags)) {
> > +               if (get_timer(start) > 1000) {
> > +                       err = -ETIMEDOUT;
> > +                       goto out;
> > +               }
> > +       }
>
> This fixes the hang I was having, but adds a significant boot delay
> because it needs to wait until it gets the timeout (1000). Is there a
> better timeout value to be used here?

Looking a bit further, this ends up adding a significant boot delay
because the mmc driver still tries to send another two additional
commands, so it ends executing this timeout logic 3 times in total.
Guess this can be improved at
http://git.denx.de/?p=u-boot.git;a=blob;f=drivers/mmc/mmc.c;h=d6b9cdc99229d8bfe0c8be3adfd6e5209a11601c;hb=HEAD#l2633
by either checking for mmc first or by simply just checking and
handling the return code from mmc_send_if_cond, but this could be
another patch.

Would still be nice to confirm if 1000 is the minimal timeout value to
be used here.

Thanks,
-- 
Ricardo Salveti de Araujo