mail archive of the barebox mailing list
 help / color / mirror / Atom feed
* [PATCH] mci: dw_mmc: set dma mask to work correctly on 64bit platforms
@ 2021-06-10 13:10 Sascha Hauer
  2021-06-11  8:23 ` Ahmad Fatoum
  2021-06-11  8:26 ` Ahmad Fatoum
  0 siblings, 2 replies; 5+ messages in thread
From: Sascha Hauer @ 2021-06-10 13:10 UTC (permalink / raw)
  To: Barebox List

The variant we support currently can only do 32bit DMA. Adjust dma mask
accordingly. Also use dma_map_single() rather than dma_sync_single() to
actually get errors when the mapping fails.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
---
 drivers/mci/dw_mmc.c | 53 ++++++++++++++++++++++++++------------------
 1 file changed, 31 insertions(+), 22 deletions(-)

diff --git a/drivers/mci/dw_mmc.c b/drivers/mci/dw_mmc.c
index 7979568841..139193ccf5 100644
--- a/drivers/mci/dw_mmc.c
+++ b/drivers/mci/dw_mmc.c
@@ -25,6 +25,7 @@
 
 struct dwmci_host {
 	struct mci_host mci;
+	struct device_d *dev;
 	struct clk *clk_biu, *clk_ciu;
 	void *ioaddr;
 	unsigned int fifo_size_bytes;
@@ -34,6 +35,7 @@ struct dwmci_host {
 	int ciu_div;
 	u32 fifoth_val;
 	u32 pwren_value;
+	dma_addr_t idmac_dma;
 };
 
 struct dwmci_idmac {
@@ -110,12 +112,12 @@ static int dwmci_prepare_data_pio(struct dwmci_host *host,
 }
 
 static int dwmci_prepare_data_dma(struct dwmci_host *host,
-		struct mci_data *data)
+		struct mci_data *data, dma_addr_t dma)
 {
 	unsigned long ctrl;
 	unsigned int i = 0, flags, cnt, blk_cnt;
-	unsigned start_addr;
 	struct dwmci_idmac *desc = host->idmac;
+	dma_addr_t desc_dma = host->idmac_dma;
 
 	blk_cnt = data->blocks;
 
@@ -124,12 +126,7 @@ static int dwmci_prepare_data_dma(struct dwmci_host *host,
 
 	dwmci_wait_reset(host, DWMCI_CTRL_FIFO_RESET);
 
-	dwmci_writel(host, DWMCI_DBADDR, (uint32_t)desc);
-
-	if (data->flags & MMC_DATA_READ)
-		start_addr = (uint32_t)data->dest;
-	else
-		start_addr = (uint32_t)data->src;
+	dwmci_writel(host, DWMCI_DBADDR, desc_dma);
 
 	do {
 		flags = DWMCI_IDMAC_OWN | DWMCI_IDMAC_CH;
@@ -142,10 +139,12 @@ static int dwmci_prepare_data_dma(struct dwmci_host *host,
 			cnt = data->blocksize * 8;
 		}
 
+		desc_dma += sizeof(*desc);
+
 		desc->flags = flags;
 		desc->cnt = cnt;
-		desc->addr = start_addr + (i * PAGE_SIZE);
-		desc->next_addr = (uint32_t)(desc + 1);
+		desc->addr = dma + (i * PAGE_SIZE);
+		desc->next_addr = desc_dma;
 
 		dev_dbg(host->mci.hw_dev, "desc@ 0x%p 0x%08x 0x%08x 0x%08x 0x%08x\n",
 				desc, flags, cnt, desc->addr, desc->next_addr);
@@ -172,12 +171,12 @@ static int dwmci_prepare_data_dma(struct dwmci_host *host,
 }
 
 static int dwmci_prepare_data(struct dwmci_host *host,
-		struct mci_data *data)
+		struct mci_data *data, dma_addr_t dma)
 {
 	if (dwmci_use_pio(host))
 		return dwmci_prepare_data_pio(host, data);
 	else
-		return dwmci_prepare_data_dma(host, data);
+		return dwmci_prepare_data_dma(host, data, dma);
 }
 
 static int dwmci_set_transfer_mode(struct dwmci_host *host,
@@ -272,6 +271,7 @@ dwmci_cmd(struct mci_host *mci, struct mci_cmd *cmd, struct mci_data *data)
 	uint64_t start;
 	int ret;
 	unsigned int num_bytes = 0;
+	dma_addr_t dma = 0;
 
 	start = get_time_ns();
 	while (1) {
@@ -287,16 +287,20 @@ dwmci_cmd(struct mci_host *mci, struct mci_cmd *cmd, struct mci_data *data)
 	dwmci_writel(host, DWMCI_RINTSTS, DWMCI_INTMSK_ALL);
 
 	if (data) {
+
 		num_bytes = data->blocks * data->blocksize;
 
 		if (data->flags & MMC_DATA_WRITE)
-			dma_sync_single_for_device((unsigned long)data->src,
-						   num_bytes, DMA_TO_DEVICE);
+			dma = dma_map_single(host->dev, (void *)data->src, num_bytes,
+					     DMA_TO_DEVICE);
 		else
-			dma_sync_single_for_device((unsigned long)data->dest,
-						   num_bytes, DMA_FROM_DEVICE);
+			dma = dma_map_single(host->dev, data->dest, num_bytes,
+					     DMA_FROM_DEVICE);
 
-		ret = dwmci_prepare_data(host, data);
+		if (dma_mapping_error(host->dev, dma))
+			return -EFAULT;
+
+		ret = dwmci_prepare_data(host, data, dma);
 		if (ret)
 			return ret;
 	}
@@ -400,11 +404,11 @@ dwmci_cmd(struct mci_host *mci, struct mci_cmd *cmd, struct mci_data *data)
 			dwmci_writel(host, DWMCI_CTRL, ctrl);
 
 			if (data->flags & MMC_DATA_WRITE)
-				dma_sync_single_for_cpu((unsigned long)data->src,
-							num_bytes, DMA_TO_DEVICE);
+				dma_unmap_single(host->dev, dma, num_bytes,
+						 DMA_TO_DEVICE);
 			else
-				dma_sync_single_for_cpu((unsigned long)data->dest,
-							num_bytes, DMA_FROM_DEVICE);
+				dma_unmap_single(host->dev, dma, num_bytes,
+						 DMA_FROM_DEVICE);
 		}
 	}
 
@@ -550,6 +554,9 @@ static int dw_mmc_probe(struct device_d *dev)
 
 	host = xzalloc(sizeof(*host));
 
+	dma_set_mask(dev, DMA_BIT_MASK(32));
+	host->dev = dev;
+
 	host->clk_biu = clk_get(dev, "biu");
 	if (IS_ERR(host->clk_biu))
 		return PTR_ERR(host->clk_biu);
@@ -567,7 +574,9 @@ static int dw_mmc_probe(struct device_d *dev)
 	host->ioaddr = IOMEM(iores->start);
 
 	host->idmac = dma_alloc_coherent(sizeof(*host->idmac) * DW_MMC_NUM_IDMACS,
-					 DMA_ADDRESS_BROKEN);
+					 &host->idmac_dma);
+	if (!host->idmac)
+		return PTR_ERR(-ENOMEM);
 
 	host->mci.send_cmd = dwmci_cmd;
 	host->mci.set_ios = dwmci_set_ios;
-- 
2.29.2


_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox


^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [PATCH] mci: dw_mmc: set dma mask to work correctly on 64bit platforms
  2021-06-10 13:10 [PATCH] mci: dw_mmc: set dma mask to work correctly on 64bit platforms Sascha Hauer
@ 2021-06-11  8:23 ` Ahmad Fatoum
  2021-06-11 10:02   ` Sascha Hauer
  2021-06-11  8:26 ` Ahmad Fatoum
  1 sibling, 1 reply; 5+ messages in thread
From: Ahmad Fatoum @ 2021-06-11  8:23 UTC (permalink / raw)
  To: Sascha Hauer, Barebox List

Hello Sascha,

On 10.06.21 15:10, Sascha Hauer wrote:
> The variant we support currently can only do 32bit DMA. Adjust dma mask
> accordingly. Also use dma_map_single() rather than dma_sync_single() to
> actually get errors when the mapping fails.
> 
> Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

Tested-by: Ahmad Fatoum <a.fatoum@pengutronix.de>

On a StarFive JH7100 (64-bit CPU, 32-bit MMC controller, non-1:1 mapping for
coherent DMA).

>  	host->idmac = dma_alloc_coherent(sizeof(*host->idmac) * DW_MMC_NUM_IDMACS,
> -					 DMA_ADDRESS_BROKEN);
> +					 &host->idmac_dma);
> +	if (!host->idmac)

That works for Linux, but not for barebox: barebox dma_alloc_coherent doesn't have

a dev parameter, so it can't check for dma_mapping_error() internally.

dma_alloc_coherent also never returns NULL in barebox, all implementations, except

for kvx, abort if no memory could be allocated.

> +		return PTR_ERR(-ENOMEM);

-ENOMEM is no pointer.

>  
>  	host->mci.send_cmd = dwmci_cmd;
>  	host->mci.set_ios = dwmci_set_ios;
> 

-- 
Pengutronix e.K.                           |                             |
Steuerwalder Str. 21                       | http://www.pengutronix.de/  |
31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox


^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [PATCH] mci: dw_mmc: set dma mask to work correctly on 64bit platforms
  2021-06-10 13:10 [PATCH] mci: dw_mmc: set dma mask to work correctly on 64bit platforms Sascha Hauer
  2021-06-11  8:23 ` Ahmad Fatoum
@ 2021-06-11  8:26 ` Ahmad Fatoum
  1 sibling, 0 replies; 5+ messages in thread
From: Ahmad Fatoum @ 2021-06-11  8:26 UTC (permalink / raw)
  To: Sascha Hauer, Barebox List

On 10.06.21 15:10, Sascha Hauer wrote:
> The variant we support currently can only do 32bit DMA. Adjust dma mask
> accordingly. Also use dma_map_single() rather than dma_sync_single() to
> actually get errors when the mapping fails.
> 
> Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
> ---
>  drivers/mci/dw_mmc.c | 53 ++++++++++++++++++++++++++------------------
>  1 file changed, 31 insertions(+), 22 deletions(-)

>  static int dwmci_set_transfer_mode(struct dwmci_host *host,
> @@ -272,6 +271,7 @@ dwmci_cmd(struct mci_host *mci, struct mci_cmd *cmd, struct mci_data *data)
>  	uint64_t start;
>  	int ret;
>  	unsigned int num_bytes = 0;
> +	dma_addr_t dma = 0;
>  
>  	start = get_time_ns();
>  	while (1) {
> @@ -287,16 +287,20 @@ dwmci_cmd(struct mci_host *mci, struct mci_cmd *cmd, struct mci_data *data)
>  	dwmci_writel(host, DWMCI_RINTSTS, DWMCI_INTMSK_ALL);
>  
>  	if (data) {
> +

stray new line



-- 
Pengutronix e.K.                           |                             |
Steuerwalder Str. 21                       | http://www.pengutronix.de/  |
31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox


^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [PATCH] mci: dw_mmc: set dma mask to work correctly on 64bit platforms
  2021-06-11  8:23 ` Ahmad Fatoum
@ 2021-06-11 10:02   ` Sascha Hauer
  2021-06-11 10:40     ` Ahmad Fatoum
  0 siblings, 1 reply; 5+ messages in thread
From: Sascha Hauer @ 2021-06-11 10:02 UTC (permalink / raw)
  To: Ahmad Fatoum; +Cc: Barebox List

On Fri, Jun 11, 2021 at 10:23:25AM +0200, Ahmad Fatoum wrote:
> Hello Sascha,
> 
> On 10.06.21 15:10, Sascha Hauer wrote:
> > The variant we support currently can only do 32bit DMA. Adjust dma mask
> > accordingly. Also use dma_map_single() rather than dma_sync_single() to
> > actually get errors when the mapping fails.
> > 
> > Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
> 
> Tested-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
> 
> On a StarFive JH7100 (64-bit CPU, 32-bit MMC controller, non-1:1 mapping for
> coherent DMA).
> 
> >  	host->idmac = dma_alloc_coherent(sizeof(*host->idmac) * DW_MMC_NUM_IDMACS,
> > -					 DMA_ADDRESS_BROKEN);
> > +					 &host->idmac_dma);
> > +	if (!host->idmac)
> 
> That works for Linux, but not for barebox: barebox dma_alloc_coherent doesn't have
> 
> a dev parameter, so it can't check for dma_mapping_error() internally.

Yeah, I know. I have that on my mental todo list and hoped nobody would
realize.
Adding a dev parameter to dma_alloc_coherent() is one thing. With that
we can check for errors. The next step of course would be to allocate
memory in the allowed area, not only to complain.

> 
> dma_alloc_coherent also never returns NULL in barebox, all implementations, except
> 
> for kvx, abort if no memory could be allocated.
> 
> > +		return PTR_ERR(-ENOMEM);
> 
> -ENOMEM is no pointer.

Yes, fixed.

Sascha

-- 
Pengutronix e.K.                           |                             |
Steuerwalder Str. 21                       | http://www.pengutronix.de/  |
31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox


^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [PATCH] mci: dw_mmc: set dma mask to work correctly on 64bit platforms
  2021-06-11 10:02   ` Sascha Hauer
@ 2021-06-11 10:40     ` Ahmad Fatoum
  0 siblings, 0 replies; 5+ messages in thread
From: Ahmad Fatoum @ 2021-06-11 10:40 UTC (permalink / raw)
  To: Sascha Hauer; +Cc: Barebox List

Hi,

On 11.06.21 12:02, Sascha Hauer wrote:
> On Fri, Jun 11, 2021 at 10:23:25AM +0200, Ahmad Fatoum wrote:
>> Hello Sascha,
>>
>> On 10.06.21 15:10, Sascha Hauer wrote:
>>> The variant we support currently can only do 32bit DMA. Adjust dma mask
>>> accordingly. Also use dma_map_single() rather than dma_sync_single() to
>>> actually get errors when the mapping fails.
>>>
>>> Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
>>
>> Tested-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
>>
>> On a StarFive JH7100 (64-bit CPU, 32-bit MMC controller, non-1:1 mapping for
>> coherent DMA).
>>
>>>  	host->idmac = dma_alloc_coherent(sizeof(*host->idmac) * DW_MMC_NUM_IDMACS,
>>> -					 DMA_ADDRESS_BROKEN);
>>> +					 &host->idmac_dma);
>>> +	if (!host->idmac)
>>
>> That works for Linux, but not for barebox: barebox dma_alloc_coherent doesn't have
>>
>> a dev parameter, so it can't check for dma_mapping_error() internally.
> 
> Yeah, I know. I have that on my mental todo list and hoped nobody would
> realize.

Tss. tss. ;)

> Adding a dev parameter to dma_alloc_coherent() is one thing. With that
> we can check for errors. The next step of course would be to allocate
> memory in the allowed area, not only to complain.

What you can do for now is checking for dma_mapping_error here and abort
the probe if you exceed the mask.

> 
>>
>> dma_alloc_coherent also never returns NULL in barebox, all implementations, except
>>
>> for kvx, abort if no memory could be allocated.
>>
>>> +		return PTR_ERR(-ENOMEM);
>>
>> -ENOMEM is no pointer.
> 
> Yes, fixed.
> 
> Sascha
> 

-- 
Pengutronix e.K.                           |                             |
Steuerwalder Str. 21                       | http://www.pengutronix.de/  |
31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox


^ permalink raw reply	[flat|nested] 5+ messages in thread

end of thread, other threads:[~2021-06-11 10:42 UTC | newest]

Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-06-10 13:10 [PATCH] mci: dw_mmc: set dma mask to work correctly on 64bit platforms Sascha Hauer
2021-06-11  8:23 ` Ahmad Fatoum
2021-06-11 10:02   ` Sascha Hauer
2021-06-11 10:40     ` Ahmad Fatoum
2021-06-11  8:26 ` Ahmad Fatoum

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox