Re: [GIT PULL] Update dw-mmc controller - regression!?

[Jaehoon Chung <jh80.chung@samsung.com>]

On 07/12/2016 07:13 PM, Ulf Hansson wrote:
> On 12 July 2016 at 11:39, Jaehoon Chung <jh80.chung@samsung.com> wrote:
>> On 07/12/2016 05:55 PM, Shawn Lin wrote:
>>> 在 2016/7/12 16:28, Ulf Hansson 写道:
>>>> On 6 July 2016 at 18:19, Ulf Hansson <ulf.hansson@linaro.org> wrote:
>>>>> On 29 June 2016 at 03:42, Jaehoon Chung <jh80.chung@samsung.com> wrote:
>>>>>> Dear Ulf,
>>>>>>
>>>>>> Could you pull these patches into your repository (next branch)?
>>>>>>
>>>>>> Best Regards,
>>>>>> Jaehoon Chung
>>>>>>
>>>>>> The following changes since commit 466780648db4aed7722de91e9d9749c6c5969327:
>>>>>>
>>>>>>   mmc: sh_mmcif: Use a 10s timeout in the error recovery path (2016-06-27 11:19:29 +0200)
>>>>>>
>>>>>> are available in the git repository at:
>>>>>>
>>>>>>   https://github.com/jh80chung/dw-mmc.git for-ulf
>>>>>>
>>>>>> for you to fetch changes up to 545585efee538cac5202b415a3d0c12b4556e96f:
>>>>>>
>>>>>>   mmc: dw_mmc: fix unmap sg twice when finding data err (2016-06-29 10:39:05 +0900)
>>>>>>
>>>>>> ----------------------------------------------------------------
>>>>>> Ben Dooks (1):
>>>>>>       mmc: dw_mmc: fix 32bit little-endian access of des1 field
>>>>>>
>>>>>> Doug Anderson (2):
>>>>>>       mmc: dw_mmc: Consider HLE errors to be data and command errors
>>>>>>       mmc: dw_mmc: Wait for data transfer after response errors.
>>>>>>
>>>>>> Jaehoon Chung (4):
>>>>>>       mmc: dw_mmc: exynos: fix the NULL pointer dereference error
>>>>>>       mmc: dw_mmc: remove the quirks flags
>>>>>>       mmc: dw_mmc: add the card write threshold for HS400 mode
>>>>>>       mmc: dw_mmc: prevent to set the wrong value
>>>>>>
>>>>>> Jérôme Forissier (1):
>>>>>>       mmc: dw_mmc: k3: add MMC_CAP_CMD23
>>>>>>
>>>>>> Seung-Woo Kim (1):
>>>>>>       mmc: dw_mmc: remove UBSAN warning in dw_mci_setup_bus()
>>>>>>
>>>>>> Shawn Lin (3):
>>>>>>       mmc: dw_mmc: remove redundant of set_bit and clear_bit
>>>>>>       mmc: dw_mmc: check card present before starting request
>>>>>>       mmc: dw_mmc: fix unmap sg twice when finding data err
>>>>>>
>>>>>>  drivers/mmc/host/dw_mmc-exynos.c   |   2 +-
>>>>>>  drivers/mmc/host/dw_mmc-k3.c       |   7 ++
>>>>>>  drivers/mmc/host/dw_mmc-rockchip.c |   3 -
>>>>>>  drivers/mmc/host/dw_mmc.c          | 157 ++++++++++++++++++++++---------------
>>>>>>  drivers/mmc/host/dw_mmc.h          |  10 +--
>>>>>>  include/linux/mmc/dw_mmc.h         |   9 ---
>>>>>>  6 files changed, 108 insertions(+), 80 deletions(-)
>>>>>
>>>>> Thanks Jaehoon, pulled into my next branch!
>>>>
>>>>
>>>> Reported by kernelci, It seems like some of these patches are causing
>>>> the exynos5250-arndale to fail to boot.
>>>> I looked briefly at the logs and the bisect report (which isn't
>>>> complete) and it turns out that the eMMC card at mmc0 isn't detected
>>>> properly.
>>>>
>>>> Here's the boot summary (it turns out there are more boards failing,
>>>> don't know if all are related to dw_mmc, yet):
>>>> https://kernelci.org/boot/all/job/ulfh/kernel/v4.7-rc5-92-g4e01b578ade9/
>>>
>>> I quickly look at the failed bootlogs of your reports.
>>> My three patches don't touch the behaviour for which I search dw_mmc
>>> code to find whether it will generate some possible to make mmc-core
>>> got -EINVAL. It's a little difficult for me to point out the root cause
>>> without more debug log or geting the real failed board.
>>>
>>>
>>> It will be great if Jaehoon have a exynos5250-* board to debug?
>>> Or Ulf, could you revert the commit 97f659a2e (mmc: dw_mmc: prevent to set the wrong value) to see if we could boot up exynos5250-arndale
>>> successfully?
>>
>> Right, when i have debugged, this commit 97f659a2e (mmc: dw_mmc: prevent to set the wrong value) is problem.
>> If Ulf can revert it, otherwise i needs to modify the dts file of arndale baord.
>>
>> In future, I will fix more exactly. Sorry for this.
> 
> No worries, these things happens and is the reason to why kernelci and
> other automated build/boot tools exists! :-)
> 
>>
>> When i have checked,
>>
>> Before reverting,
>> [    2.757696] mmc_host mmc0: vqmmc not available.(Skip the switching voltage)
>> [    2.757713] mmc0: error -22 whilst initialising MMC card
>>
>> After reverted,
>> [    2.480755] mmc0: new DDR MMC card at address 0001
>> [    2.481723] mmcblk0: mmc0:0001 SDW16G 14.7 GiB
>> [    2.482218] mmcblk0boot0: mmc0:0001 SDW16G partition 1 4.00 MiB
>> [    2.482681] mmcblk0boot1: mmc0:0001 SDW16G partition 2 4.00 MiB
>> [    2.483345] mmcblk0rpmb: mmc0:0001 SDW16G partition 3 4.00 MiB
>> [    2.486493]  mmcblk0: p1 p2 p3 p4 < p5 p6 >
>>
>> How about? revert? otherwise I will send to fix the patch.
>> I will accept Ulf's opinion. Sorry for this, again.
> 
> No revert needed, I have just dropped (through a re-base) from my next branch.

Thanks!

> 
> The changes are published now, so let's give kernelci some time to run
> a new build/boot test!

If there will occur other problem, let me know, plz.
Also thank Shawn and Ben!

Best Regards,
Jaehoon Chung

> 
> To all involved, I appreciate your quick responsiveness in solving this issue!
> 
> Kind regards
> Uffe
> 
>>
>> Best Regards,
>> Jaehoon Chung
>>
>>>
>>>>
>>>> Here's the bootlog:
>>>> https://storage.kernelci.org/ulfh/v4.7-rc5-92-g4e01b578ade9/arm-multi_v7_defconfig+CONFIG_EFI=y/lab-baylibre-seattle/boot-exynos5250-arndale.txt
>>>>
>>>> Please, could any of you dw_mmc experts have a look at this as soon as
>>>> possible!?
>>>>
>>>> Kind regards
>>>> Uffe
>>>> --
>>>> To unsubscribe from this list: send the line "unsubscribe linux-mmc" in
>>>> the body of a message to majordomo@vger.kernel.org
>>>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>>>>
>>>
>>>
>>
> 
> 
>