SDIO3.0 + 4.1.15 kernel

[gary_bisson]

Hi,

As SDIO3.0 been tested on the 4.1.15 kernel?

Asking the question because we are using a new WiFi chip which works in SDIO3.0 mode on our i.MX6Q Nitrogen6x boards. This worked flawlessly on the 3.14.52 kernel. It is working on 4.1.15 kernel but it is now throwing the following errors every few seconds:

mmc2: Got data interrupt 0x00000002 even though no data operation was in progress.

Doing a git bisect showed this commit is the one introducing the issue:

MLK-12391-4 mmc: sdhci-esdhc-imx: enable hw auto retuning for MAN_TUNING

linux-2.6-imx.git - Freescale i.MX Linux Tree

After reverting this patch all the warnings are gone. Can the NXP Linux team confirm if SDIO3.0 has been tested and if this patch is mandatory?

Regards,

Gary

[Yuri]

Hello,

  According to "i.MX_Linux_Release_Notes.pdf", SDIO 3.0 was not checked.

As for the patch, please use the recent BSP :

http://www.nxp.com/webapp/Download?colCode=L4.1.15_1.2.0_LINUX_DOCS&Parent_nodeId=133769948107170617...

http://www.nxp.com/webapp/Download?colCode=L4.1.15_1.1.0_LINUX_DOCS&Parent_nodeId=133769948107170617...

Summary page :

i.MX 6 Series Software and Development Tool|NXP

Have a great day,
Yuri

-----------------------------------------------------------------------------------------------------------------------
Note: If this post answers your question, please click the Correct Answer button. Thank you!
-----------------------------------------------------------------------------------------------------------------------

[gary_bisson]

Hi Yuri.

What do you mean by "please use the recent BSP"? The patch I am pointing to is from the latest 4.1.15_1.0.0_ga branch  which is the same as the 4.1.15_1.2.0 tag. Please clarify but as of today it cannot be more recent than that.

Anyway, can you pass the message along to the Linux team so that they know the patch causes issue in SDIO3.0 mode.

Regards,

Gary

[gary_bisson]

Hi YuriMuhin_ng​,

Did you get any feedback from the Linux team?

Regards,

Gary

[Yuri · Accepted solution]

Hello,

  From the team :

Patch 7fd37e87d was tested on 4.1.15. Testing was performed on the MCIMX7x-SDB,
which has an SDIO 3.0 card integrated on the board. IMX6 boards do not have an SDIO 3.0
device on them and, therefore, were not tested against this patch.

   In most cases, hardware auto-tuning is unnecessary. It is very rare but possible that, e.g.,

a large temperature change might cause a timing shift which can be detected and corrected by

the hardware auto-tuning feature. This potential should be evaluated by customers.

It is confirmed: it is possible remove the patch and the MMC core driver will retune timing in software
after each system suspend/resume. This may slightly affect performance until the next suspend/resume

in the case where timing is reduced.

Regards,

Yuri.

[gary_bisson]

Hi Yuri,

Thanks for the follow-up.

It'd be nice for the team to try i.MX6 with SDIO3.0 someday, so maybe the patch can be reverted in the tree directly, especially if hw auto-tuning is unnecessary in most cases.

Here is the device with which the issue has been seen on i.MX6Q:

SX-SDCAC | Radio Modules | Silex Technology America

Regards,

Gary