mbox series

[v2,0/8] Port am335 and am437 devices to sdhci-omap

Message ID 20190215192033.24203-1-faiz_abbas@ti.com (mailing list archive)
Headers show
Series Port am335 and am437 devices to sdhci-omap | expand

Message

Faiz Abbas Feb. 15, 2019, 7:20 p.m. UTC 
The following add driver patches for porting TI's am335x and am437x
devices to the sdhci-omap driver.

This involves adding external DMA support to sdhci (first 3 patches from
Chunyan) plus some miscellaneous patches to take care of deviations of
the controllers from the sdhci model.

DT changes will be posted in a separate series.

Untested versions of Chunyan's patches were posted before[1].

Tested on: am335x-evm, am335x-boneblack, am335x-sk, am437x-gpevm,
am43xx-gpevm, am437x-idk, dra7xx-evm, dra72x-evm

v2:
1. sdhci is using two bottom halves. One threaded_rq for card detect and a
   tasklet for finishing mmc requests. Patch 1 removes the tasklet and
   moves its function to the threaded_irq. This enables me to
   terminate_sync() in sdhci_request_done()

2. Factored out common code for between the normal adn external dma case

3. Using existing API sdhci_data_timeout_irq for disabling DTO during
   erase commands.

4. Fixed subject line for dt-bindings patch.

[1] https://patchwork.kernel.org/project/linux-mmc/list/?series=54897

Chunyan Zhang (3):
  mmc: sdhci: add support for using external DMA devices
  dt-bindings: sdhci-omap: Add properties for using external dma
  mmc: sdhci-omap: Add using external dma

Faiz Abbas (5):
  mmc: sdhci: Get rid of finish_tasklet
  mmc: sdhci: Add quirk for disabling DTO during erase command
  mmc: sdhci-omap: Add DISABLE_DTO_FOR_ERASE Quirk
  dt-bindings: sdhci-omap: Add am335x and am437x specific bindings
  mmc: sdhci-omap: Add am335x and am437x specific compatibles

 .../devicetree/bindings/mmc/sdhci-omap.txt    |   9 +
 drivers/mmc/host/Kconfig                      |   4 +
 drivers/mmc/host/sdhci-omap.c                 |  27 +-
 drivers/mmc/host/sdhci.c                      | 333 +++++++++++++++---
 drivers/mmc/host/sdhci.h                      |  12 +-
 5 files changed, 328 insertions(+), 57 deletions(-)

Comments

Tony Lindgren Feb. 15, 2019, 8:02 p.m. UTC | #1 
* Faiz Abbas <faiz_abbas@ti.com> [190215 19:17]:
> The following add driver patches for porting TI's am335x and am437x
> devices to the sdhci-omap driver.
> 
> This involves adding external DMA support to sdhci (first 3 patches from
> Chunyan) plus some miscellaneous patches to take care of deviations of
> the controllers from the sdhci model.

Good to see this happening :)

I think am437x should have also the ADMA as it's mostly
omap4 based? See the old omap4 ADMA series at [0] below.

Are you seeing any improvment in SDIO card read/write speeds
between external DMA and ADM BTW?

Regards,

Tony

[0] https://linux-omap.vger.kernel.narkive.com/H4EbLW96/patch-0-4-omap4-hsmmc-adding-adma-support
Faiz Abbas Feb. 18, 2019, 1:49 p.m. UTC | #2 
Hi Tony,

On 16/02/19 1:32 AM, Tony Lindgren wrote:
> * Faiz Abbas <faiz_abbas@ti.com> [190215 19:17]:
>> The following add driver patches for porting TI's am335x and am437x
>> devices to the sdhci-omap driver.
>>
>> This involves adding external DMA support to sdhci (first 3 patches from
>> Chunyan) plus some miscellaneous patches to take care of deviations of
>> the controllers from the sdhci model.
> 
> Good to see this happening :)
> 
> I think am437x should have also the ADMA as it's mostly
> omap4 based? See the old omap4 ADMA series at [0] below.
> 
> Are you seeing any improvment in SDIO card read/write speeds
> between external DMA and ADM BTW

am437 is using dmaengine even in the old omap_hsmmc driver. I haven't
tried to enable ADMA in am437x.

Thanks,
Faiz
Tony Lindgren Feb. 18, 2019, 4:25 p.m. UTC | #3 
* Faiz Abbas <faiz_abbas@ti.com> [190218 13:46]:
> Hi Tony,
> 
> On 16/02/19 1:32 AM, Tony Lindgren wrote:
> > * Faiz Abbas <faiz_abbas@ti.com> [190215 19:17]:
> >> The following add driver patches for porting TI's am335x and am437x
> >> devices to the sdhci-omap driver.
> >>
> >> This involves adding external DMA support to sdhci (first 3 patches from
> >> Chunyan) plus some miscellaneous patches to take care of deviations of
> >> the controllers from the sdhci model.
> > 
> > Good to see this happening :)
> > 
> > I think am437x should have also the ADMA as it's mostly
> > omap4 based? See the old omap4 ADMA series at [0] below.
> > 
> > Are you seeing any improvment in SDIO card read/write speeds
> > between external DMA and ADM BTW
> 
> am437 is using dmaengine even in the old omap_hsmmc driver. I haven't
> tried to enable ADMA in am437x.

Right, it's probably safest to keep it that way at least initially
to avoid changing multiple things at once. Then ADMA can be
configured later on if needed.

Regards,

Tony