Skip to contentSkip to conversation

OMAP HSMMC: fix MMC3 dma

PATCH In review Grazvydas Ignotas 4 participants 6 messages Jan 27, 2009 – Feb 8, 2009

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 OMAP HSMMC: fix MMC3 dma

Data transfers on third OMAP3 MMC controller don't work because DMA line numbers are only defined for MMC1 and MMC2. Fix that and store line numbers in mmc_omap_host structure to reduce code size.
Tested on OMAP3 pandora board.

---
 drivers/mmc/host/omap_hsmmc.c |   31 +++++++++++++++++++++++--------
 1 files changed, 23 insertions(+), 8 deletions(-)
drivers/mmc/host/omap_hsmmc.c +23 -8
index 70d8d78..a2c4aee 100644
@@ -97,6 +97,7 @@
  */
 #define OMAP_MMC1_DEVID		0
 #define OMAP_MMC2_DEVID		1
+#define OMAP_MMC3_DEVID		2
 
 #define OMAP_MMC_DATADIR_NONE	0
 #define OMAP_MMC_DATADIR_READ	1
@@ -148,6 +149,7 @@ struct mmc_omap_host {
 	int			irq;
 	int			carddetect;
 	int			use_dma, dma_ch;
+	int			dma_line_tx, dma_line_rx;
 	int			initstr;
 	int			slot_id;
 	int			dbclk_enabled;
@@ -672,16 +674,10 @@ mmc_omap_start_dma_transfer(struct mmc_omap_host *host, struct mmc_request *req)
 
 	if (!(data->flags & MMC_DATA_WRITE)) {
 		host->dma_dir = DMA_FROM_DEVICE;
-		if (host->id == OMAP_MMC1_DEVID)
-			sync_dev = OMAP24XX_DMA_MMC1_RX;
-		else
-			sync_dev = OMAP24XX_DMA_MMC2_RX;
+		sync_dev = host->dma_line_rx;
 	} else {
 		host->dma_dir = DMA_TO_DEVICE;
-		if (host->id == OMAP_MMC1_DEVID)
-			sync_dev = OMAP24XX_DMA_MMC1_TX;
-		else
-			sync_dev = OMAP24XX_DMA_MMC2_TX;
+		sync_dev = host->dma_line_tx;
 	}
 
 	ret = omap_request_dma(sync_dev, "MMC/SD", mmc_omap_dma_cb,
@@ -1058,6 +1054,25 @@ static int __init omap_mmc_probe(struct platform_device *pdev)
 	OMAP_HSMMC_WRITE(host->base, HCTL,
 			OMAP_HSMMC_READ(host->base, HCTL) | SDBP);
 
+	/* Select DMA lines */
+	switch (host->id) {
+	case OMAP_MMC1_DEVID:
+		host->dma_line_tx = OMAP24XX_DMA_MMC1_TX;
+		host->dma_line_rx = OMAP24XX_DMA_MMC1_RX;
+		break;
+	case OMAP_MMC2_DEVID:
+		host->dma_line_tx = OMAP24XX_DMA_MMC2_TX;
+		host->dma_line_rx = OMAP24XX_DMA_MMC2_RX;
+		break;
+	case OMAP_MMC3_DEVID:
+		host->dma_line_tx = OMAP34XX_DMA_MMC3_TX;
+		host->dma_line_rx = OMAP34XX_DMA_MMC3_RX;
+		break;
+	default:
+		dev_err(mmc_dev(host->mmc), "Invalid MMC id\n");
+		goto err_irq;
+	}
+
 	/* Request IRQ for MMC operations */
 	ret = request_irq(host->irq, mmc_omap_irq, IRQF_DISABLED,
 			mmc_hostname(mmc), host);
Signature · 1 line
1.5.4.3
#1 lore raw
Quoted 11 lines · Grazvydas Ignotas

Data transfers on third OMAP3 MMC controller don't work because DMA line numbers are only defined for MMC1 and MMC2. Fix that and store line numbers in mmc_omap_host structure to reduce code size.
Tested on OMAP3 pandora board.

---
 drivers/mmc/host/omap_hsmmc.c |   31 +++++++++++++++++++++++--------
 1 files changed, 23 insertions(+), 8 deletions(-)

Have you considered using the scatter-gather emulation patch:

	http://marc.info/?l=linux-omap&m=122726069922195&w=2

It was reverted:

	http://marc.info/?l=linux-omap&m=122904725322787&w=2

because of a hardware problem with OMAP3430 before ES3.0, but applying:

	http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=6a79e391df295bd7c2aa1309ea5031f361c197fd

seems to overcome even that.

#2 lore raw
Quoted 20 lines · Adrian Hunter
Quoted 7 lines · Grazvydas Ignotas

Data transfers on third OMAP3 MMC controller don't work because DMA line numbers are only defined for MMC1 and MMC2. Fix that and store line numbers in mmc_omap_host structure to reduce code size.
Tested on OMAP3 pandora board.

Quoted 5 lines
---
 drivers/mmc/host/omap_hsmmc.c |   31 +++++++++++++++++++++++--------
 1 files changed, 23 insertions(+), 8 deletions(-)

Have you considered using the scatter-gather emulation patch:

       http://marc.info/?l=linux-omap&m=122726069922195&w=2

It was reverted:

Quoted 8 lines
       http://marc.info/?l=linux-omap&m=122904725322787&w=2

because of a hardware problem with OMAP3430 before ES3.0, but applying:

http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=6a79e391df295bd7c2aa1309ea5031f361c197fd

seems to overcome even that.

Hmm, but it still doesn't know about MMC3. But in case that SG patch goes first, I can resend.

#3 lore raw
Quoted 27 lines · Grazvydas Ignotas
Quoted 25 lines · Adrian Hunter
Quoted 7 lines · Grazvydas Ignotas

Data transfers on third OMAP3 MMC controller don't work because DMA line numbers are only defined for MMC1 and MMC2. Fix that and store line numbers in mmc_omap_host structure to reduce code size.
Tested on OMAP3 pandora board.

Quoted 5 lines
---
 drivers/mmc/host/omap_hsmmc.c |   31 +++++++++++++++++++++++--------
 1 files changed, 23 insertions(+), 8 deletions(-)

Have you considered using the scatter-gather emulation patch:

       http://marc.info/?l=linux-omap&m=122726069922195&w=2

It was reverted:

       http://marc.info/?l=linux-omap&m=122904725322787&w=2

because of a hardware problem with OMAP3430 before ES3.0, but applying:

Quoted 6 lines
Quoted 3 lines

http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=6a79e391df295bd7c2aa1309ea5031f361c197fd

seems to overcome even that.

Hmm, but it still doesn't know about MMC3. But in case that SG patch goes first, I can resend.

On related news, looks like the hsmmc.c has finally hit the mainline tree!

Tony

#4 lore raw
In reply to Grazvydas Ignotas
Quoted 25 lines · Grazvydas Ignotas
Quoted 23 lines · Adrian Hunter
Quoted 11 lines · Grazvydas Ignotas

Data transfers on third OMAP3 MMC controller don't work because DMA line numbers are only defined for MMC1 and MMC2. Fix that and store line numbers in mmc_omap_host structure to reduce code size.
Tested on OMAP3 pandora board.

---
 drivers/mmc/host/omap_hsmmc.c |   31 +++++++++++++++++++++++--------
 1 files changed, 23 insertions(+), 8 deletions(-)

Have you considered using the scatter-gather emulation patch:

       http://marc.info/?l=linux-omap&m=122726069922195&w=2

It was reverted:

       http://marc.info/?l=linux-omap&m=122904725322787&w=2

because of a hardware problem with OMAP3430 before ES3.0, but applying:

Quoted 6 lines
Quoted 3 lines

http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=6a79e391df295bd7c2aa1309ea5031f361c197fd

seems to overcome even that.

Hmm, but it still doesn't know about MMC3. But in case that SG patch goes first, I can resend.

I just meant, you might be interested because it gives better performance.

No, it doesn't know about MMC3, but it does hit some of the same lines of code.

#5 lore raw
In reply to Grazvydas Ignotas
Quoted 10 lines · Grazvydas Ignotas
drivers/mmc/host/omap_hsmmc.c +0 -0
@@ -1058,6 +1054,25 @@ static int __init omap_mmc_probe(struct platform_device *pdev)
         OMAP_HSMMC_WRITE(host->base, HCTL,
                         OMAP_HSMMC_READ(host->base, HCTL) | SDBP);


+       /* Select DMA lines */
+       switch (host->id) {
+       case OMAP_MMC1_DEVID:
+               host->dma_line_tx = OMAP24XX_DMA_MMC1_TX; +               host->dma_line_rx = OMAP24XX_DMA_MMC1_RX; +               break;

Actually the cleanest way to do that is through IORESOURCE_DMA type resources associated with the platform devices ... but doing it that way requires reworking the MMC platform device setup logic. Which I can understand wanting to avoid here.

(I think that rework is worth doing for multiple reasons; this is just one more.)

Quoted 8 lines

+       case OMAP_MMC2_DEVID:
+               host->dma_line_tx = OMAP24XX_DMA_MMC2_TX; +               host->dma_line_rx = OMAP24XX_DMA_MMC2_RX; +               break;
+       case OMAP_MMC3_DEVID:
+               host->dma_line_tx = OMAP34XX_DMA_MMC3_TX; +               host->dma_line_rx = OMAP34XX_DMA_MMC3_RX; +               break;

Quoted 8 lines

+       default:
+               dev_err(mmc_dev(host->mmc), "Invalid MMC id\n"); +               goto err_irq;
+       }
+
        /* Request IRQ for MMC operations */
        ret = request_irq(host->irq, mmc_omap_irq, IRQF_DISABLED, mmc_hostname(mmc), host);

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