[U-Boot] [PATCH 1/7] mmc: atmel_sdhci: fix uninitialized pointer deref on probe

[Peng Fan]

> 
> >> Subject: Re: [U-Boot] [PATCH 1/7] mmc: atmel_sdhci: fix uninitialized
> >> pointer deref on probe
> >>
> >>
> >>
> >> On 07.08.2019 13:04, Peng Fan wrote:
> >>
> >>>
> >>>> Subject: Re: [U-Boot] [PATCH 1/7] mmc: atmel_sdhci: fix
> >>>> uninitialized pointer deref on probe
> >>>>
> >>>> Tested-by: Eugen Hristev <eugen.hristev at microchip.com>
> >>>>
> >>>> This has to go as soon as possible into the tree.
> >>>> The at91 boards do not boot without this. The commit checking for
> >>>> sdhci-caps-mask introduces a crash at this point.
> >>>>
> >>>> I can take it through atmel tree if needed.
> >>>
> >>>
> >>> I need wait R-b and A-b for other patches in the patchset, Then PR
> >>> to Tom. So please take this patch if hurry, and mark patchwork as
> >>> accepted.
> >>>
> >>
> >> As it looks to me all the other platforms have the same issue. So
> >> they are just as urgent. In this case I assume it won't take long for
> >> all of them to be taken, so, it's fine.
> >>
> >> Just one question, why did you not move all the 4 initialization
> >> lines before the sdhci_setup_cfg ? Other drivers initialize all 4 before the
> call.
> >
> > Only dev is required to address the issue.
> >
> 
> True. But other drivers perform all 4 initializations. In the future the sdhci core
> might again make wrong assumptions about them, and we might get another
> crash?
> Why not move all the 4 lines now ? It looks like they are just backpointers. So
> I would say the initialization of the pointers is incomplete before the
> sdhci_setup_cfg call, would be better to do it completely

After checking the code, sdhci.c is common code, priv pointer should be not
used there. So no need to move priv here.
and upriv is used by mmc uclass.

The remaining 2 has no chance to be used by sdhci.c.

Regards,
Peng.

> 
> Eugen
> 
> > Regards,
> > Peng.
> >
> >>
> >> Eugen
> >>
> >>
> >>> Thanks,
> >>> Peng.
> >>>
> >>>>
> >>>> Thanks,
> >>>> Eugen
> >>>>
> >>>> On 06.08.2019 05:47, Peng Fan wrote:
> >>>>
> >>>>>
> >>>>> Commit 3d296365e4e8 ("mmc: sdhci: Add support for
> >>>>> sdhci-caps-mask") sdhci_setup_cfg() expects a valid sdhci_host mmc
> >>>>> field.
> >>>>>
> >>>>> Fixes: 3d296365e4e8 ("mmc: sdhci: Add support for
> >>>>> sdhci-caps-mask")
> >>>>> Cc: Faiz Abbas <faiz_abbas at ti.com>
> >>>>> Cc: Wenyou Yang <wenyou.yang at atmel.com>
> >>>>> Signed-off-by: Peng Fan <peng.fan at nxp.com>
> >>>>> ---
> >>>>>     drivers/mmc/atmel_sdhci.c | 4 ++--
> >>>>>     1 file changed, 2 insertions(+), 2 deletions(-)
> >>>>>
> >>>>> diff --git a/drivers/mmc/atmel_sdhci.c b/drivers/mmc/atmel_sdhci.c
> >>>>> index 4be47ba75e..d930ed8da0 100644
> >>>>> --- a/drivers/mmc/atmel_sdhci.c
> >>>>> +++ b/drivers/mmc/atmel_sdhci.c
> >>>>> @@ -88,13 +88,13 @@ static int atmel_sdhci_probe(struct udevice
> *dev)
> >>>>>     		return -EINVAL;
> >>>>>
> >>>>>     	host->max_clk = max_clk;
> >>>>> +	host->mmc = &plat->mmc;
> >>>>> +	host->mmc->dev = dev;
> >>>>>
> >>>>>     	ret = sdhci_setup_cfg(&plat->cfg, host, 0,
> >>>> ATMEL_SDHC_MIN_FREQ);
> >>>>>     	if (ret)
> >>>>>     		return ret;
> >>>>>
> >>>>> -	host->mmc = &plat->mmc;
> >>>>> -	host->mmc->dev = dev;
> >>>>>     	host->mmc->priv = host;
> >>>>>     	upriv->mmc = host->mmc;
> >>>>>
> >>>>>