Skip to contentSkip to conversation

SDHCI_QUIRK_NO_MULTIBLOCK for i.MX35/25

BUG/RFC Steffen Trumtrar 2 participants 5 messages Nov 15, 2013 – Jan 20, 2014 MMC, SD and SDIO

lore
Share & tools
lore
#0 lore raw

Hi!

1 Actual Use-Case
=================

I have a SDIO card with a Marvell 8787, that works fine on an i.MX53 with the drivers/net/wireless/mwifiex/sdio.c driver. On an i.MX35 however the loading of the firmware errors out on the very first single-block transfer with 1024 bytes payload.

2 Reason for failure
====================

sdhci defines SDHCI_QUIRK_NO_MULTIBLOCK to completely disable multi-block transfers for a host. The max_blk_count is simply set to 1 to achieve this. Currently only i.MX35 and i.MX25 seem to be the only users of that quirk. The quirk is used to fix an errata on these SoCs, that says, that "CMD12 abort operation does not abort data transfer on AHB". (ENGcm07207)

So, this is a problem regarding only SD and not SDIO. On this special SDIO card, it looks like multi-block transfer has to be used for anything >16 bytes. Removing the quirk fixes the firmware loading error. The device is detected successfully after loading and can connect to an AP.

3 Question
==========

And here comes the question: What would be the correct approach to fix that and differentiate those two cases, so that we can use multi-block transfers with CMD53 but not with CMD25(?!) ?

1a. As I see no other users of that quirk, it might be okay to rename
it, to better reflect the actual problem it fixes (broken CMD12 or something).

1b. When the host knows with what it deals with (SDIO or SD) set the
max_blk_count according to the quirk. And only in the SD case, that is.

or

2. Introduce a max_blk_count_sd and max_blk_count_sdio and use them everywhere. Seems rather invasive and over the top.

or

3. something better.

Regards,
Steffen

Signature · 4 lines
Pengutronix e.K.                           |                             |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |
#1 lore raw
Quoted 43 lines · Steffen Trumtrar

Hi!

1 Actual Use-Case
=================

I have a SDIO card with a Marvell 8787, that works fine on an i.MX53 with the drivers/net/wireless/mwifiex/sdio.c driver. On an i.MX35 however the loading of the firmware errors out on the very first single-block transfer with 1024 bytes payload.

2 Reason for failure
====================

sdhci defines SDHCI_QUIRK_NO_MULTIBLOCK to completely disable multi-block transfers for a host. The max_blk_count is simply set to 1 to achieve this. Currently only i.MX35 and i.MX25 seem to be the only users of that quirk. The quirk is used to fix an errata on these SoCs, that says, that "CMD12 abort operation does not abort data transfer on AHB". (ENGcm07207)

So, this is a problem regarding only SD and not SDIO. On this special SDIO card, it looks like multi-block transfer has to be used for anything >16 bytes. Removing the quirk fixes the firmware loading error. The device is detected successfully after loading and can connect to an AP.

3 Question
==========

And here comes the question: What would be the correct approach to fix that and differentiate those two cases, so that we can use multi-block transfers with CMD53 but not with CMD25(?!) ?

1a. As I see no other users of that quirk, it might be okay to rename
it, to better reflect the actual problem it fixes (broken CMD12 or something).

1b. When the host knows with what it deals with (SDIO or SD) set the
max_blk_count according to the quirk. And only in the SD case, that is.

Quoted 10 lines

or

2. Introduce a max_blk_count_sd and max_blk_count_sdio and use them everywhere. Seems rather invasive and over the top.

or

3. something better.

Ping...

Regards,
Steffen

Signature · 4 lines
Pengutronix e.K.                           |                             |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |
#2 lore raw
Quoted 49 lines · Steffen Trumtrar
Quoted 48 lines · Steffen Trumtrar

Hi!

1 Actual Use-Case
=================

I have a SDIO card with a Marvell 8787, that works fine on an i.MX53 with the drivers/net/wireless/mwifiex/sdio.c driver. On an i.MX35 however the loading of the firmware errors out on the very first single-block transfer with 1024 bytes payload.

2 Reason for failure
====================

sdhci defines SDHCI_QUIRK_NO_MULTIBLOCK to completely disable multi-block transfers for a host. The max_blk_count is simply set to 1 to achieve this. Currently only i.MX35 and i.MX25 seem to be the only users of that quirk. The quirk is used to fix an errata on these SoCs, that says, that "CMD12 abort operation does not abort data transfer on AHB". (ENGcm07207)

So, this is a problem regarding only SD and not SDIO. On this special SDIO card, it looks like multi-block transfer has to be used for anything >16 bytes. Removing the quirk fixes the firmware loading error. The device is detected successfully after loading and can connect to an AP.

3 Question
==========

And here comes the question: What would be the correct approach to fix that and differentiate those two cases, so that we can use multi-block transfers with CMD53 but not with CMD25(?!) ?

1a. As I see no other users of that quirk, it might be okay to rename
it, to better reflect the actual problem it fixes (broken CMD12 or something).

1b. When the host knows with what it deals with (SDIO or SD) set the
max_blk_count according to the quirk. And only in the SD case, that is.

or

2. Introduce a max_blk_count_sd and max_blk_count_sdio and use them everywhere. Seems rather invasive and over the top.

Quoted 8 lines
Quoted 5 lines

or

3. something better.

Ping...

Ping ping ...

Regards,
Steffen

Signature · 4 lines
Pengutronix e.K.                           |                             |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |
#3 lore raw
In reply to Steffen Trumtrar

Hi Steffen,

Quoted 43 lines · Steffen Trumtrar

Hi!

1 Actual Use-Case
=================

I have a SDIO card with a Marvell 8787, that works fine on an i.MX53 with the drivers/net/wireless/mwifiex/sdio.c driver. On an i.MX35 however the loading of the firmware errors out on the very first single-block transfer with 1024 bytes payload.

2 Reason for failure
====================

sdhci defines SDHCI_QUIRK_NO_MULTIBLOCK to completely disable multi-block transfers for a host. The max_blk_count is simply set to 1 to achieve this. Currently only i.MX35 and i.MX25 seem to be the only users of that quirk. The quirk is used to fix an errata on these SoCs, that says, that "CMD12 abort operation does not abort data transfer on AHB". (ENGcm07207)

So, this is a problem regarding only SD and not SDIO. On this special SDIO card, it looks like multi-block transfer has to be used for anything >16 bytes. Removing the quirk fixes the firmware loading error. The device is detected successfully after loading and can connect to an AP.

3 Question
==========

And here comes the question: What would be the correct approach to fix that and differentiate those two cases, so that we can use multi-block transfers with CMD53 but not with CMD25(?!) ?

1a. As I see no other users of that quirk, it might be okay to rename
it, to better reflect the actual problem it fixes (broken CMD12 or something).

1b. When the host knows with what it deals with (SDIO or SD) set the
max_blk_count according to the quirk. And only in the SD case, that is.

Quoted 9 lines

or

2. Introduce a max_blk_count_sd and max_blk_count_sdio and use them everywhere. Seems rather invasive and over the top.

or

3. something better.

There's an .init_card callback which is called in mmc_sdio_init_card. That means we can know current card is SDIO, then you can try to update the max_blk_count there.

Regards
Dong Aisheng

Quoted 13 lines

Regards,
Steffen

--

Pengutronix e.K.                           |                             |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

--
To unsubscribe from this list: send the line "unsubscribe linux-mmc" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

#4 lore raw

Hi Dong,

Quoted 57 lines · Dong Aisheng

Hi Steffen,

Quoted 43 lines · Steffen Trumtrar

Hi!

1 Actual Use-Case
=================

I have a SDIO card with a Marvell 8787, that works fine on an i.MX53 with the drivers/net/wireless/mwifiex/sdio.c driver. On an i.MX35 however the loading of the firmware errors out on the very first single-block transfer with 1024 bytes payload.

2 Reason for failure
====================

sdhci defines SDHCI_QUIRK_NO_MULTIBLOCK to completely disable multi-block transfers for a host. The max_blk_count is simply set to 1 to achieve this. Currently only i.MX35 and i.MX25 seem to be the only users of that quirk. The quirk is used to fix an errata on these SoCs, that says, that "CMD12 abort operation does not abort data transfer on AHB". (ENGcm07207)

So, this is a problem regarding only SD and not SDIO. On this special SDIO card, it looks like multi-block transfer has to be used for anything >16 bytes. Removing the quirk fixes the firmware loading error. The device is detected successfully after loading and can connect to an AP.

3 Question
==========

And here comes the question: What would be the correct approach to fix that and differentiate those two cases, so that we can use multi-block transfers with CMD53 but not with CMD25(?!) ?

1a. As I see no other users of that quirk, it might be okay to rename
it, to better reflect the actual problem it fixes (broken CMD12 or something).

1b. When the host knows with what it deals with (SDIO or SD) set the
max_blk_count according to the quirk. And only in the SD case, that is.

Quoted 9 lines

or

2. Introduce a max_blk_count_sd and max_blk_count_sdio and use them everywhere. Seems rather invasive and over the top.

or

3. something better.

Quoted 6 lines

There's an .init_card callback which is called in mmc_sdio_init_card. That means we can know current card is SDIO, then you can try to update the max_blk_count there.

From what I can see, this seems to be the correct approach.

So, leave the quirk and in the sdio-case write a different max_blk_count ?!

Any objections to that ?

Steffen

Signature · 4 lines
Pengutronix e.K.                           |                             |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |