socfpga: mmc problems

[ml@communistcode.co.uk (Jack Mitchell)]

Morning Dinh,

On 03/07/13 16:09, Dinh Nguyen wrote:
> Hi Jack,
>
> On Wed, 2013-07-03 at 16:01 +0100, Jack Mitchell wrote:
>> Having received one of the new Arrow SoCkit boards at a recent training
>> day I set about seeing how far mainline was in successfully booting this
>> board.
>>
>> As I wanted to boot from mmc, I first checked out linux-next [1] in
>> order to get the newly merged mmc driver, I then applied the device-tree
>> enablement series [2] on top before finally attempting to boot. I then
>> found that the standard socfpga_defconfig didn't enable the MMC drivers,
>> which I also did (should this be part of the socfpga_defconfig...?).
>
> I was going to update this when I enable ethernet as well to avoid less
> churn on the defconfig.

That's great!

>
>>
>> However, the boot hangs at:
>>
>> Synopsys Designware Multimedia Card Interface Driver
>> dwmmc_socfpga ff704000.dwmmc0: Using internal DMA controller.
>> dwmmc_socfpga ff704000.dwmmc0: DW MMC controller at irq 171, 32 bit host
>> data width, 1024 deep fifo
>>
>> To fix this I checked out the diff between the Altera vendor mmc driver
>> and the mainline driver and found that if I added this small change:
>>
>>
>>     if (of_property_read_u32(dev->of_node, "pwr-en", &pwr_en)) {
>>       dev_info(dev, "couldn't determine pwr-en, assuming pwr-en = 0\n");
>>       pwr_en = 0;
>>     }
>>
>>     /* Set PWREN bit */
>>     mci_writel(host, PWREN, pwr_en);
>
> I tested this on my devkit, which was not made by Arrow, but is _almost_
> the same. I'll try to see if I can get my hands on an Arrow board to
> reproduce.

Now, I think I made a _slight_ mistake, in that I accidentally copied 
the Altera vendor kernel onto the sdcard instead of the linux-next 
kernel. Re-testing this morning with a fresh mind, I checked the config 
again and now I find the Arrow board still won't boot from MMC, even 
with the above change, with the code hanging at the same place.

I know you don't have an Arrow board, but if you have a tree which I 
could pull from with your amalgamated changes; that would rule out me 
missing any patches and it actually being a board problem. I can't seem 
to see if the V8 device tree patches were pulled into anyones tree, 
otherwise I would just wait till they appeared in 3.11-rc1...?

Sorry for the confusion!

>
>>
>> In the dw_mci_socfpga_priv_init function, the board would boot.
>>
>> Synopsys Designware Multimedia Card Interface Driver
>> dwmmc_socfpga ff704000.dwmmc0: couldn't determine pwr-en, assuming
>> pwr-en = 0
>> dwmmc_socfpga ff704000.dwmmc0: Using internal DMA controller.
>> dwmmc_socfpga ff704000.dwmmc0: DW MMC controller at irq 171, 32 bit host
>> data width, 1024 deep fifo
>> mmc_host mmc0: Bus speed (slot 0) = 100000000Hz (slot req 400000Hz,
>> actual 400000HZ div = 125)
>> dwmmc_socfpga ff704000.dwmmc0: 1 slots initialized
>> dwmmc_socfpga ff704000.dwmmc0: Version ID is 240a
>> .....
>> .....
>> .....
>>
>>
>> Has this been overlooked, or are there some other changes elsewhere that
>> I haven't yet come across which addresses this?
>
> I'm not aware of any other changes.
>
> Dinh
>>
>> Cheers,
>>
>> [1] fa1383200d289afdff2f5678eb89483daa537465
>> [2] http://comments.gmane.org/gmane.linux.ports.arm.kernel/243529
>>
>
>
>


-- 

   Jack Mitchell (jack at embed.me.uk)
   Embedded Systems Engineer
   http://www.embed.me.uk

--