linux-kernel - Re: [PATCH RFC] ARM: dts: mxs: leave card detect out of common mmc pins config

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date:	Tue, 9 Apr 2013 09:18:12 +0200
From:	Hector Palacios <hector.palacios@...i.com>
To:	Marek Vasut <marex@...x.de>
CC:	Shawn Guo <shawn.guo@...aro.org>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
	"maxime.ripard@...e-electrons.com" <maxime.ripard@...e-electrons.com>,
	"fabio.estevam@...escale.com" <fabio.estevam@...escale.com>,
	<linux-mmc@...r.kernel.org>
Subject: Re: [PATCH RFC] ARM: dts: mxs: leave card detect out of common mmc
 pins config

Dear Marek Vasut,

On 04/08/2013 06:28 PM, Marek Vasut wrote:
> Dear Shawn Guo,
>
>> On Mon, Apr 08, 2013 at 03:58:05PM +0200, Hector Palacios wrote:
>>> On 04/08/2013 02:48 PM, Shawn Guo wrote:
>>>> On Mon, Apr 08, 2013 at 12:12:20PM +0200, Hector Palacios wrote:
>>>>> MicroSD card sockets don't usually have card detect line. This pin
>>>>> is actually not needed for the MMC to work and it is more of a
>>>>> platform design decission to have it.
>>>>> The card detect pin already has a configuration entry of its own:
>>>>> 'mmc0_cd_cfg' so we complete the iomux configuration here and let
>>>>> platforms to include it or not depending on whether the card detect
>>>>> line is routed to the SD socket.
>>>>
>>>> Sounds sensible.
>>>>
>>>>> Signed-off-by: Hector Palacios <hector.palacios@...i.com>
>>>>> ---
>>>>>
>>>>> Hello,
>>>>>
>>>>> All imx28 based platforms except 'bluegiga,apx4devkit' and
>>>>> 'schulercontrol,imx28-sps1', use 'mmc0_cd_cfg' in their mmc
>>>>> configuration so please check whether this patch would break these
>>>>> platforms.
>>>>
>>>> I just tested the patch on imx28-evk and card-detection still works.  So
>>>> patches applied, thanks.
>>>
>>> The EVK and most platforms will work because they are using
>>> 'mmc0_cd_cfg' so actually this patch does not change anything on
>>> them.
>>> Platforms 'bluegiga,apx4devkit' and 'schulercontrol,imx28-sps1'
>>> however are not referencing 'mmc0_cd_cfg' so after applying this
>>> patch they will have unconfigured CD line and they may break.
>>
>> Ah, yes.  I thought that any board that has CD support has to reference
>> 'mmc0_cd_cfg'.  That's not necessarily true.
>>
>>> The driver will call get_cd() upon probing, which returns the status of
>>> the CD line. Please check these two platforms before applying.
>>
>> Ok, let's wait for people owning the boards to confirm.
>
> Maybe you want to use MMC_CAP_NEEDS_POLL as was noted by someone before on the
> olinuxino -- the slot is there, it's just the CD line that's missing.

I'm not sure of what you mean. The mxs-mmc.c driver already sets the 
MMC_CAP_NEEDS_POLL flag by default in the probe() function. My platform does not even 
route the CD line because the microSD socket does not have it.
So what I have done is modify the driver to parse the property 'non-removable' from 
the device tree in order to set the MMC_CAP_NONREMOVABLE flag:

diff --git a/drivers/mmc/host/mxs-mmc.c b/drivers/mmc/host/mxs-mmc.c
index 206fe49..2a3b9c9 100644
--- a/drivers/mmc/host/mxs-mmc.c
+++ b/drivers/mmc/host/mxs-mmc.c
@@ -679,15 +682,20 @@ static int mxs_mmc_probe(struct platform_device *pdev)
         /* set mmc core parameters */
         mmc->ops = &mxs_mmc_ops;
         mmc->caps = MMC_CAP_SD_HIGHSPEED | MMC_CAP_MMC_HIGHSPEED |
-                   MMC_CAP_SDIO_IRQ | MMC_CAP_NEEDS_POLL;
+                   MMC_CAP_SDIO_IRQ;

         of_property_read_u32(np, "bus-width", &bus_width);
         if (bus_width == 4)
                 mmc->caps |= MMC_CAP_4_BIT_DATA;
         else if (bus_width == 8)
                 mmc->caps |= MMC_CAP_4_BIT_DATA | MMC_CAP_8_BIT_DATA;
-       host->wp_gpio = of_get_named_gpio_flags(np, "wp-gpios", 0, &flags);

+       if (of_find_property(np, "non-removable", NULL))
+               mmc->caps |= MMC_CAP_NONREMOVABLE;
+       else
+               mmc->caps |= MMC_CAP_NEEDS_POLL;
+
+       host->wp_gpio = of_get_named_gpio_flags(np, "wp-gpios", 0, &flags);
         if (flags & OF_GPIO_ACTIVE_LOW)
                 host->wp_inverted = 1;


In theory, and according to the mmc bindings doc, this flag must assume the card is 
always present. However the MMC core function mmc_rescan() will still call at least 
once the get_cd() hook, which in the mxs-mmc.c returns the value of the CD line. In my 
platform (where the CD line is not connected) this is returning 0, so I needed to do 
the following:

diff --git a/drivers/mmc/host/mxs-mmc.c b/drivers/mmc/host/mxs-mmc.c
index 206fe49..2a3b9c9 100644
--- a/drivers/mmc/host/mxs-mmc.c
+++ b/drivers/mmc/host/mxs-mmc.c
@@ -95,6 +95,9 @@ static int mxs_mmc_get_cd(struct mmc_host *mmc)
         struct mxs_mmc_host *host = mmc_priv(mmc);
         struct mxs_ssp *ssp = &host->ssp;

+       if (mmc->caps & MMC_CAP_NONREMOVABLE)
+               return 1;
+
         return !(readl(ssp->base + HW_SSP_STATUS(ssp)) &
                  BM_SSP_STATUS_CARD_DETECT);
  }

to return a 1 when the driver should assume that the card is there. I wonder however 
if this should be a global mmc patch instead, and have mmc_rescan() not call the host 
get_cd() hook at all if MMC_CAP_NONREMOVABLE is set.

Regards,
-- 
Héctor Palacios
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/