[meta-intel] Valleyisland SDIO support

Wed May 14 02:58:17 PDT 2014

Hi Chiau Ee,

we use a preliminary hardware of a COM Express Module. It is the Conga TCA3 from Congatec (http://www.congatec.com/de/produkte/com-express-typ6/conga-tca3.html) with the E3825 Atom CPU.
The installed BIOS is AMI Aptio UEFI firmware based. The version is TA31R010 from 2014- 04-10.
Currently there is no documented BIOS Setup Description available. The BIOS setup itselfs got a configuration option named "SDIO Configuration". With this the "SDIO Access Mode" can be set to {Auto|PIO|DMA}. This is the only option I have discovered yet which got to do something with the SDIO.

Falk

Von: Chew, Chiau Ee [mailto:chiau.ee.chew at intel.com]
Gesendet: Mittwoch, 14. Mai 2014 07:58
An: Chang, Rebecca Swee Fun; IGI/Falk Bauer
Cc: meta-intel at yoctoproject.org
Betreff: RE: Valleyisland SDIO support

Hi Falk,

The reason that we dropped this workaround patch is because this DDR50 feature shall be disabled in BIOS instead.
Which BIOS you are using currently?

Thanks,
Chiau Ee

From: meta-intel-bounces at yoctoproject.org [mailto:meta-intel-bounces at yoctoproject.org] On Behalf Of Chang, Rebecca Swee Fun
Sent: Monday, May 12, 2014 7:32 PM
To: 'IGI/Falk Bauer'
Cc: meta-intel at yoctoproject.org
Subject: Re: [meta-intel] Valleyisland SDIO support

Hi,

I'm aware that the DDR50 mode is not fully supported. The clock appears to be unstable when SDCARD host running with DDR50 mode. We have a workaround that introduce a new quirk to force host with broken DDR50 mode to run with SDR25 mode. As we think that this workaround might not be suitable to be upstream to mainline kernel, this workaround will be implemented in our valleyisland-io feature branch in future. We are now working on backporting our features from mainline kernel to LTSI/LTS kernel. And eventually all those workaround that do not fit upstream criteria will be implemented temporary in feature branch.

I'm not sure whether I can provide you the workaround patch here. I will work with our driver developers in order to help you in this. You may file a bug in bugzilla so that we got track on your enquiry. Thanks a lot.

Rebecca

From: IGI/Falk Bauer [mailto:f.bauer at igi-systems.com]
Sent: 12 May, 2014 5:37 PM
To: Chang, Rebecca Swee Fun
Cc: meta-intel at yoctoproject.org
Subject: Valleyisland SDIO support

Hello Rebecca,
thanks for your recently work with the Baytrail board. I can confirm that the image in dora branch is working with full OpenGL support.
Here is another question relating to the SDIO controller embedded into the atom processor.
In the intel specifiction update from january 2014 (http://www.intel.com/content/dam/www/public/us/en/documents/specification-updates/atom-Z36xxx-Z37xxx-spec-update.pdf) there is an entry named "VLT55 SD Card UHS - I Mode is Not Fully Supported".
We have issues here with sd cards from sandisk which are running into error messages like
**********
May 7 16:27:53 valleyisland-32 user.debug kernel: [ 125.254962] sdhci-pci 0000:00:12.0: setting latency timer to 64
May 7 16:27:55 valleyisland-32 user.info kernel: [ 127.510449] mmc1: new ultra high speed DDR50 SDHC card at address e624
May 7 16:27:55 valleyisland-32 user.info kernel: [ 127.511010] mmcblk0: mmc1:e624 SD08G 7.40 GiB
May 7 16:27:55 valleyisland-32 user.info kernel: [ 127.520193] mmcblk0: p1
May 7 16:27:55 valleyisland-32 user.err kernel: [ 127.526524] mmcblk0: error -84 transferring data, sector 15523712, nr 8, cmd response 0x900, card status 0xb00
May 7 16:27:55 valleyisland-32 user.warn kernel: [ 127.526534] mmcblk0: retrying using single block read
May 7 16:27:55 valleyisland-32 user.err kernel: [ 127.529319] mmcblk0: error -84 transferring data, sector 15523712, nr 8, cmd response 0x900, card status 0x0
May 7 16:27:55 valleyisland-32 user.warn kernel: [ 127.529329] blk_update_request: 773 callbacks suppressed
May 7 16:27:55 valleyisland-32 user.err kernel: [ 127.529335] end_request: I/O error, dev mmcblk0, sector 15523712
May 7 16:27:55 valleyisland-32 user.err kernel: [ 127.531980] mmcblk0: error -84 transferring data, sector 15523713, nr 7, cmd response 0x900, card status 0x0
May 7 16:27:55 valleyisland-32 user.err kernel: [ 127.531990] end_request: I/O error, dev mmcblk0, sector 15523713
**********

These messages appears in the kernel log buffer after inserting an sd card SanDisk Extreme 30 MB/s, class 10 UHS 1 8GB (vendor name in sysfs is SU08G).
Further investigations reveals that the error reproduce is related to the kind of sd card, filesystem on the partition aso.
I think that the errata vlt55 leads to the messages in the log. Have you experienced some similar issues with the Z3000 atom series? Or is there some plans or possibilities to disable the UHS feature for the SDIO device embedded in the atom cpu? I know that in the specification update there is currently the "no fix" entry. But maybe there is some "quick-and-dirty" fix like a "SDHCI_QUIRK2_BROKEN_UHS" define or so in the sdhci module in the kernel.
If you wish i can also fill a bug in bugzilla.

Falk
________________________________________
IGI - Ingenieur-Gesellschaft fuer Interfaces mbH
Dipl.-Ing. Falk Bauer
Applications Engineer
Langenauer Str. 46
57223 Kreuztal, Germany
Tel.:
+49 2732 5525 0
Fax:
+49 2732 5525 25
E-mail:
f.bauer at igi-systems.com
Internet:
www.igi-systems.com
Follow us:

________________________________________
Meet us:
Geospatial World Forum in Geneve (5-9 May) * Geo Business in London (28+29 May)
________________________________________

------------------------------------------------------
IGI - Ingenieur-Gesellschaft fuer Interfaces mbH
Dipl.-Ing. Falk Bauer Applications Engineer
Langenauer Str. 46
57223 Kreuztal, Germany
Tel.: +49 2732 5525 0
Fax: +49 2732 5525 25
E-mail: f.bauer at igi-systems.com
Internet: www.igi-systems.com
------------------------------------------------------