Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1346360 > unrolled thread

[PATCH v2 1/2] mmc: sdhci: Set DMA mask

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  [PATCH v2 1/2] mmc: sdhci: Set DMA mask Alexandre Courbot <acourbot@nvidia.com> - 2016-03-01 05:50 +0100
    Re: [PATCH v2 1/2] mmc: sdhci: Set DMA mask Arnd Bergmann <arnd@arndb.de> - 2016-03-01 22:40 +0100
      Re: [PATCH v2 1/2] mmc: sdhci: Set DMA mask Alexandre Courbot <gnurou@gmail.com> - 2016-03-04 07:20 +0100

#1346360 — [PATCH v2 1/2] mmc: sdhci: Set DMA mask

Set the DMA mask if specified by the host's dma_mask member, and
not only if the DMA flags are set.

Use dma_set_mask_and_coherent() to set it instead of messing directly
with the device's dma_mask.

Signed-off-by: Alexandre Courbot <acourbot@nvidia.com>
---
 drivers/mmc/host/sdhci.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/drivers/mmc/host/sdhci.c b/drivers/mmc/host/sdhci.c
index fd9139947fa3..0693f52d238a 100644
--- a/drivers/mmc/host/sdhci.c
+++ b/drivers/mmc/host/sdhci.c
@@ -2997,10 +2997,12 @@ int sdhci_add_host(struct sdhci_host *host)
 	 * mask, but PIO does not need the hw shim so we set a new
 	 * mask here in that case.
 	 */
-	if (!(host->flags & (SDHCI_USE_SDMA | SDHCI_USE_ADMA))) {
+	if (!(host->flags & (SDHCI_USE_SDMA | SDHCI_USE_ADMA)))
 		host->dma_mask = DMA_BIT_MASK(64);
-		mmc_dev(mmc)->dma_mask = &host->dma_mask;
-	}
+
+	if (host->dma_mask &&
+	    dma_set_mask_and_coherent(mmc_dev(mmc), host->dma_mask))
+		pr_warn("%s: cannot set DMA mask\n", mmc_hostname(mmc));
 
 	if (host->version >= SDHCI_SPEC_300)
 		host->max_clk = (caps[0] & SDHCI_CLOCK_V3_BASE_MASK)
-- 
2.7.2

[toc] | [next] | [standalone]

#1347044

On Tuesday 01 March 2016 13:32:43 Alexandre Courbot wrote:
>          */
> -       if (!(host->flags & (SDHCI_USE_SDMA | SDHCI_USE_ADMA))) {
> +       if (!(host->flags & (SDHCI_USE_SDMA | SDHCI_USE_ADMA)))
>                 host->dma_mask = DMA_BIT_MASK(64);
> -               mmc_dev(mmc)->dma_mask = &host->dma_mask;
> -       }
> +
> +       if (host->dma_mask &&
> +           dma_set_mask_and_coherent(mmc_dev(mmc), host->dma_mask))
> +               pr_warn("%s: cannot set DMA mask\n", mmc_hostname(mmc));
> 

Looks good, 

Acked-by: Arnd Bergmann <arnd@arndb.de>

[toc] | [prev] | [next] | [standalone]

#1349874

On Wed, Mar 2, 2016 at 6:30 AM, Arnd Bergmann <arnd@arndb.de> wrote:
> On Tuesday 01 March 2016 13:32:43 Alexandre Courbot wrote:
>>          */
>> -       if (!(host->flags & (SDHCI_USE_SDMA | SDHCI_USE_ADMA))) {
>> +       if (!(host->flags & (SDHCI_USE_SDMA | SDHCI_USE_ADMA)))
>>                 host->dma_mask = DMA_BIT_MASK(64);
>> -               mmc_dev(mmc)->dma_mask = &host->dma_mask;
>> -       }
>> +
>> +       if (host->dma_mask &&
>> +           dma_set_mask_and_coherent(mmc_dev(mmc), host->dma_mask))
>> +               pr_warn("%s: cannot set DMA mask\n", mmc_hostname(mmc));
>>
>
> Looks good,
>
> Acked-by: Arnd Bergmann <arnd@arndb.de>

Withdrawing this patch after discussing with Arnd. The test detects
devices that are not capable of DMA, so it makes no sense to call a
DMA function right after that.

A better solution to the problem this patch tried to solve is on the way.

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web