[PATCH 1/2] dmatest: implement two helpers to unmap dma memory

[PATCH 1/2] dmatest: implement two helpers to unmap dma memory

[Andy Shevchenko]

The unmap_src() and unmap_dst() will be used later as well.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
---
 drivers/dma/dmatest.c |   27 +++++++++++++++++----------
 1 file changed, 17 insertions(+), 10 deletions(-)

diff --git a/drivers/dma/dmatest.c b/drivers/dma/dmatest.c
index 3e777d2..b515343 100644
--- a/drivers/dma/dmatest.c
+++ b/drivers/dma/dmatest.c
@@ -236,6 +236,20 @@ static unsigned int min_odd(unsigned int x, unsigned int y)
 	return val % 2 ? val : val - 1;
 }
 
+static inline void unmap_src(struct device *dev, dma_addr_t *addr, size_t len,
+			     unsigned int count)
+{
+	while (count--)
+		dma_unmap_single(dev, addr[count], len, DMA_TO_DEVICE);
+}
+
+static inline void unmap_dst(struct device *dev, dma_addr_t *addr, size_t len,
+			     unsigned int count)
+{
+	while (count--)
+		dma_unmap_single(dev, addr[count], len, DMA_BIDIRECTIONAL);
+}
+
 /*
  * This function repeatedly tests DMA transfers of various lengths and
  * offsets for a given operation type until it is told to exit by
@@ -394,13 +408,8 @@ static int dmatest_func(void *data)
 		}
 
 		if (!tx) {
-			for (i = 0; i < src_cnt; i++)
-				dma_unmap_single(dev->dev, dma_srcs[i], len,
-						 DMA_TO_DEVICE);
-			for (i = 0; i < dst_cnt; i++)
-				dma_unmap_single(dev->dev, dma_dsts[i],
-						 test_buf_size,
-						 DMA_BIDIRECTIONAL);
+			unmap_src(dev->dev, dma_srcs, len, src_cnt);
+			unmap_dst(dev->dev, dma_dsts, test_buf_size, dst_cnt);
 			pr_warning("%s: #%u: prep error with src_off=0x%x "
 					"dst_off=0x%x len=0x%x\n",
 					thread_name, total_tests - 1,
@@ -454,9 +463,7 @@ static int dmatest_func(void *data)
 		}
 
 		/* Unmap by myself (see DMA_COMPL_SKIP_DEST_UNMAP above) */
-		for (i = 0; i < dst_cnt; i++)
-			dma_unmap_single(dev->dev, dma_dsts[i], test_buf_size,
-					 DMA_BIDIRECTIONAL);
+		unmap_dst(dev->dev, dma_dsts, test_buf_size, dst_cnt);
 
 		error_count = 0;
 
-- 
1.7.10.4

[PATCH 2/2] dmatest: check for dma mapping error

[Andy Shevchenko]

We get a warning if CONFIG_DMA_API_DEBUG=y

[   28.150631] WARNING: at lib/dma-debug.c:933 check_unmap+0x5d6/0x6ac()
[   28.157058] dw_dmac dw_dmac.0: DMA-API: device driver failed to check map error[device address=0x0000000035698305] [size=14365 bytes] [mapped as single]

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
---
 drivers/dma/dmatest.c |   22 +++++++++++++++++++++-
 1 file changed, 21 insertions(+), 1 deletion(-)

diff --git a/drivers/dma/dmatest.c b/drivers/dma/dmatest.c
index b515343..7be60f1 100644
--- a/drivers/dma/dmatest.c
+++ b/drivers/dma/dmatest.c
@@ -378,15 +378,35 @@ static int dmatest_func(void *data)
 
 			dma_srcs[i] = dma_map_single(dev->dev, buf, len,
 						     DMA_TO_DEVICE);
+			ret = dma_mapping_error(dev->dev, dma_srcs[i]);
+			if (ret) {
+				unmap_src(dev->dev, dma_srcs, len, i);
+				pr_warn("%s: #%u: mapping error %d with "
+					"src_off=0x%x len=0x%x\n",
+					thread_name, total_tests - 1, ret,
+					src_off, len);
+				failed_tests++;
+				continue;
+			}
 		}
 		/* map with DMA_BIDIRECTIONAL to force writeback/invalidate */
 		for (i = 0; i < dst_cnt; i++) {
 			dma_dsts[i] = dma_map_single(dev->dev, thread->dsts[i],
 						     test_buf_size,
 						     DMA_BIDIRECTIONAL);
+			ret = dma_mapping_error(dev->dev, dma_dsts[i]);
+			if (ret) {
+				unmap_src(dev->dev, dma_srcs, len, src_cnt);
+				unmap_dst(dev->dev, dma_dsts, test_buf_size, i);
+				pr_warn("%s: #%u: mapping error %d with "
+					"dst_off=0x%x len=0x%x\n",
+					thread_name, total_tests - 1, ret,
+					dst_off, test_buf_size);
+				failed_tests++;
+				continue;
+			}
 		}
 
-
 		if (thread->type == DMA_MEMCPY)
 			tx = dev->device_prep_dma_memcpy(chan,
 							 dma_dsts[0] + dst_off,
-- 
1.7.10.4

Re: [PATCH 1/2] dmatest: implement two helpers to unmap dma memory

[Viresh Kumar]

On 10 December 2012 17:07, Andy Shevchenko
<andriy.shevchenko@linux.intel.com> wrote:
> The unmap_src() and unmap_dst() will be used later as well.
>
> Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
> ---
>  drivers/dma/dmatest.c |   27 +++++++++++++++++----------
>  1 file changed, 17 insertions(+), 10 deletions(-)

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>

Re: [PATCH 2/2] dmatest: check for dma mapping error

[Viresh Kumar]

On 10 December 2012 17:07, Andy Shevchenko
<andriy.shevchenko@linux.intel.com> wrote:
> We get a warning if CONFIG_DMA_API_DEBUG=y
>
> [   28.150631] WARNING: at lib/dma-debug.c:933 check_unmap+0x5d6/0x6ac()
> [   28.157058] dw_dmac dw_dmac.0: DMA-API: device driver failed to check map error[device address=0x0000000035698305] [size=14365 bytes] [mapped as single]
>
> Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>

Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>

Re: [PATCH 2/2] dmatest: check for dma mapping error

[Andrew Morton]

On Mon, 10 Dec 2012 13:37:44 +0200
Andy Shevchenko <andriy.shevchenko@linux.intel.com> wrote:

> We get a warning if CONFIG_DMA_API_DEBUG=y
> 
> [   28.150631] WARNING: at lib/dma-debug.c:933 check_unmap+0x5d6/0x6ac()
> [   28.157058] dw_dmac dw_dmac.0: DMA-API: device driver failed to check map error[device address=0x0000000035698305] [size=14365 bytes] [mapped as single]
> 
> ...
>
> --- a/drivers/dma/dmatest.c
> +++ b/drivers/dma/dmatest.c
> @@ -378,15 +378,35 @@ static int dmatest_func(void *data)
>  
>  			dma_srcs[i] = dma_map_single(dev->dev, buf, len,
>  						     DMA_TO_DEVICE);
> +			ret = dma_mapping_error(dev->dev, dma_srcs[i]);
> +			if (ret) {
> +				unmap_src(dev->dev, dma_srcs, len, i);
> +				pr_warn("%s: #%u: mapping error %d with "
> +					"src_off=0x%x len=0x%x\n",
> +					thread_name, total_tests - 1, ret,
> +					src_off, len);
> +				failed_tests++;
> +				continue;
> +			}

The changelog and the code don't match.  Which one is out of date?

Re: [PATCH 2/2] dmatest: check for dma mapping error

[Andy Shevchenko]

On Fri, Dec 14, 2012 at 1:34 AM, Andrew Morton
<akpm@linux-foundation.org> wrote:
> On Mon, 10 Dec 2012 13:37:44 +0200
> Andy Shevchenko <andriy.shevchenko@linux.intel.com> wrote:
>
>> We get a warning if CONFIG_DMA_API_DEBUG=y
>>
>> [   28.150631] WARNING: at lib/dma-debug.c:933 check_unmap+0x5d6/0x6ac()
>> [   28.157058] dw_dmac dw_dmac.0: DMA-API: device driver failed to check map error[device address=0x0000000035698305] [size=14365 bytes] [mapped as single]

>> --- a/drivers/dma/dmatest.c
>> +++ b/drivers/dma/dmatest.c
>> @@ -378,15 +378,35 @@ static int dmatest_func(void *data)
>>
>>                       dma_srcs[i] = dma_map_single(dev->dev, buf, len,
>>                                                    DMA_TO_DEVICE);
>> +                     ret = dma_mapping_error(dev->dev, dma_srcs[i]);
>> +                     if (ret) {
>> +                             unmap_src(dev->dev, dma_srcs, len, i);
>> +                             pr_warn("%s: #%u: mapping error %d with "
>> +                                     "src_off=0x%x len=0x%x\n",
>> +                                     thread_name, total_tests - 1, ret,
>> +                                     src_off, len);
>> +                             failed_tests++;
>> +                             continue;
>> +                     }
>
> The changelog and the code don't match.  Which one is out of date?

I'm afraid I didn't get what is wrong.

dmatest maps memory via dma_map_single() call. And it asks dmaengine
(at the end the DMA controller driver) to unmap them after usage.
Exactly at that time user gets a warning. And this warning is
independent of DMA controller in use.

-- 
With Best Regards,
Andy Shevchenko

Re: [PATCH 2/2] dmatest: check for dma mapping error

[Andrew Morton]

On Fri, 14 Dec 2012 09:06:03 +0200 Andy Shevchenko <andy.shevchenko@gmail.com> wrote:

> On Fri, Dec 14, 2012 at 1:34 AM, Andrew Morton
> <akpm@linux-foundation.org> wrote:
> > On Mon, 10 Dec 2012 13:37:44 +0200
> > Andy Shevchenko <andriy.shevchenko@linux.intel.com> wrote:
> >
> >> We get a warning if CONFIG_DMA_API_DEBUG=y
> >>
> >> [   28.150631] WARNING: at lib/dma-debug.c:933 check_unmap+0x5d6/0x6ac()
> >> [   28.157058] dw_dmac dw_dmac.0: DMA-API: device driver failed to check map error[device address=0x0000000035698305] [size=14365 bytes] [mapped as single]
> 
> >> --- a/drivers/dma/dmatest.c
> >> +++ b/drivers/dma/dmatest.c
> >> @@ -378,15 +378,35 @@ static int dmatest_func(void *data)
> >>
> >>                       dma_srcs[i] = dma_map_single(dev->dev, buf, len,
> >>                                                    DMA_TO_DEVICE);
> >> +                     ret = dma_mapping_error(dev->dev, dma_srcs[i]);
> >> +                     if (ret) {
> >> +                             unmap_src(dev->dev, dma_srcs, len, i);
> >> +                             pr_warn("%s: #%u: mapping error %d with "
> >> +                                     "src_off=0x%x len=0x%x\n",
> >> +                                     thread_name, total_tests - 1, ret,
> >> +                                     src_off, len);
> >> +                             failed_tests++;
> >> +                             continue;
> >> +                     }
> >
> > The changelog and the code don't match.  Which one is out of date?
> 
> I'm afraid I didn't get what is wrong.
> 

The above code will not emit the string "dw_dmac dw_dmac.0: DMA-API:
device driver failed to check map error[device
address=0x0000000035698305] [size=14365 bytes] [mapped as single]".

Oh I see what you did.  That string is emitted by the kernel at
present, and the patch prevents this.  And it also emits a new warning
if the mapping attempt failed, and that new warning wasn't described.

I updated the changelog:


: The kernel emits a warning if CONFIG_DMA_API_DEBUG=y:
: 
: [   28.150631] WARNING: at lib/dma-debug.c:933 check_unmap+0x5d6/0x6ac()
: [   28.157058] dw_dmac dw_dmac.0: DMA-API: device driver failed to check map error[device address=0x0000000035698305] [size=14365 bytes] [mapped as single]
: 
: Fix this by adding the required checking of the dma_map_single() return
: value.

Re: [PATCH 2/2] dmatest: check for dma mapping error

[Andy Shevchenko]

On Fri, 2012-12-14 at 00:33 -0800, Andrew Morton wrote: 
> On Fri, 14 Dec 2012 09:06:03 +0200 Andy Shevchenko <andy.shevchenko@gmail.com> wrote:
> 
> > On Fri, Dec 14, 2012 at 1:34 AM, Andrew Morton
> > <akpm@linux-foundation.org> wrote:
> > > On Mon, 10 Dec 2012 13:37:44 +0200
> > > Andy Shevchenko <andriy.shevchenko@linux.intel.com> wrote:
> > >
> > >> We get a warning if CONFIG_DMA_API_DEBUG=y
> > >>
> > >> [   28.150631] WARNING: at lib/dma-debug.c:933 check_unmap+0x5d6/0x6ac()
> > >> [   28.157058] dw_dmac dw_dmac.0: DMA-API: device driver failed to check map error[device address=0x0000000035698305] [size=14365 bytes] [mapped as single]
> > 
> > >> --- a/drivers/dma/dmatest.c
> > >> +++ b/drivers/dma/dmatest.c
> > >> @@ -378,15 +378,35 @@ static int dmatest_func(void *data)
> > >>
> > >>                       dma_srcs[i] = dma_map_single(dev->dev, buf, len,
> > >>                                                    DMA_TO_DEVICE);
> > >> +                     ret = dma_mapping_error(dev->dev, dma_srcs[i]);
> > >> +                     if (ret) {
> > >> +                             unmap_src(dev->dev, dma_srcs, len, i);
> > >> +                             pr_warn("%s: #%u: mapping error %d with "
> > >> +                                     "src_off=0x%x len=0x%x\n",
> > >> +                                     thread_name, total_tests - 1, ret,
> > >> +                                     src_off, len);
> > >> +                             failed_tests++;
> > >> +                             continue;
> > >> +                     }
> > >
> > > The changelog and the code don't match.  Which one is out of date?
> > 
> > I'm afraid I didn't get what is wrong.
> > 
> 
> The above code will not emit the string "dw_dmac dw_dmac.0: DMA-API:
> device driver failed to check map error[device
> address=0x0000000035698305] [size=14365 bytes] [mapped as single]".
> 
> Oh I see what you did.  That string is emitted by the kernel at
> present, and the patch prevents this.  And it also emits a new warning
> if the mapping attempt failed, and that new warning wasn't described.

Ah, okay, thank you.

> I updated the changelog:

Yes, you right, the changelog in my case describes  only part of the
problem.

> : The kernel emits a warning if CONFIG_DMA_API_DEBUG=y:
> : 
> : [   28.150631] WARNING: at lib/dma-debug.c:933 check_unmap+0x5d6/0x6ac()
> : [   28.157058] dw_dmac dw_dmac.0: DMA-API: device driver failed to check map error[device address=0x0000000035698305] [size=14365 bytes] [mapped as single]
> : 
> : Fix this by adding the required checking of the dma_map_single() return
> : value.

-- 
Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Intel Finland Oy