[Qemu-devel] [PATCH] qemu-iotests: fix pattern for write test

[Qemu-devel] [PATCH] qemu-iotests: fix pattern for write test

[Stefan Weil]

The write pattern value 0axb is invalid and evaluates to 0,
so the read check (which uses a correct value of 0xab) will fail.

This failure will only be detected with a separate patch for
qemu-io. Without it, qemu-io cannot interpret hex values
and always uses a pattern value of 0.

Signed-off-by: Stefan Weil <weil@mail.berlios.de>
---
 002 |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/002 b/002
index adbd85f..14025e2 100755
--- a/002
+++ b/002
@@ -61,7 +61,7 @@ $QEMU_IO -c "read -pP 0xa 0 $size" $TEST_IMG | _filter_qemu_io
 
 echo
 echo "unaligned pwrite"
-$QEMU_IO -c 'write -pP 0axb 66 42' $TEST_IMG | _filter_qemu_io
+$QEMU_IO -c 'write -pP 0xab 66 42' $TEST_IMG | _filter_qemu_io
 
 echo
 echo "verify pattern"
-- 
1.5.6.5

[Qemu-devel] [PATCH] qemu-io: Fix parsing of pattern arguments

[Stefan Weil]

atoi won't parse sedecimal pattern values (it always returns 0).
The qemu-iotests use such pattern values, to strtol is needed
to get correct results.

Signed-off-by: Stefan Weil <weil@mail.berlios.de>
---
 qemu-io.c |   12 ++++++------
 1 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/qemu-io.c b/qemu-io.c
index d0927ba..36fb77a 100644
--- a/qemu-io.c
+++ b/qemu-io.c
@@ -286,7 +286,7 @@ read_f(int argc, char **argv)
 			break;
 		case 'P':
 			Pflag = 1;
-			pattern = atoi(optarg);
+			pattern = strtol(optarg, NULL, 0);
 			break;
 		case 'q':
 			qflag = 1;
@@ -444,7 +444,7 @@ readv_f(int argc, char **argv)
 			break;
 		case 'P':
 			Pflag = 1;
-			pattern = atoi(optarg);
+			pattern = strtol(optarg, NULL, 0);
 			break;
 		case 'q':
 			qflag = 1;
@@ -565,7 +565,7 @@ write_f(int argc, char **argv)
 			pflag = 1;
 			break;
 		case 'P':
-			pattern = atoi(optarg);
+			pattern = strtol(optarg, NULL, 0);
 			break;
 		case 'q':
 			qflag = 1;
@@ -685,7 +685,7 @@ writev_f(int argc, char **argv)
 			qflag = 1;
 			break;
 		case 'P':
-			pattern = atoi(optarg);
+			pattern = strtol(optarg, NULL, 0);
 			break;
 		default:
 			return command_usage(&writev_cmd);
@@ -859,7 +859,7 @@ aio_read_f(int argc, char **argv)
 			break;
 		case 'P':
 			ctx->Pflag = 1;
-			ctx->pattern = atoi(optarg);
+			ctx->pattern = strtol(optarg, NULL, 0);
 			break;
 		case 'q':
 			ctx->qflag = 1;
@@ -959,7 +959,7 @@ aio_write_f(int argc, char **argv)
 			ctx->qflag = 1;
 			break;
 		case 'P':
-			pattern = atoi(optarg);
+			pattern = strtol(optarg, NULL, 0);
 			break;
 		default:
 			free(ctx);
-- 
1.5.6.5

Re: [Qemu-devel] [PATCH] qemu-io: Fix parsing of pattern arguments

[Christoph Hellwig]

On Sat, Jul 11, 2009 at 02:35:44PM +0200, Stefan Weil wrote:
> atoi won't parse sedecimal pattern values (it always returns 0).
> The qemu-iotests use such pattern values, to strtol is needed
> to get correct results.

Indeed.  But now that we're using strtol which can return useful
errors I think we should also handle those errors.  Can you respin
the patch with that error handling added?

Re: [Qemu-devel] [PATCH] qemu-io: Fix parsing of pattern arguments

[Stefan Weil]

Christoph Hellwig schrieb:
> On Sat, Jul 11, 2009 at 02:35:44PM +0200, Stefan Weil wrote:
>   
>> atoi won't parse sedecimal pattern values (it always returns 0).
>> The qemu-iotests use such pattern values, to strtol is needed
>> to get correct results.
>>     
>
> Indeed.  But now that we're using strtol which can return useful
> errors I think we should also handle those errors.  Can you respin
> the patch with that error handling added?

Yes, but I'll finish the vdi driver first, so an improved patch will
have to wait a little.

I'd prefer to see the current version of the patch applied as
first step.

Regards,
Stefan

Re: [Qemu-devel] [PATCH] qemu-iotests: fix pattern for write test

[Christoph Hellwig]

On Sat, Jul 11, 2009 at 02:34:56PM +0200, Stefan Weil wrote:
> The write pattern value 0axb is invalid and evaluates to 0,
> so the read check (which uses a correct value of 0xab) will fail.
> 
> This failure will only be detected with a separate patch for
> qemu-io. Without it, qemu-io cannot interpret hex values
> and always uses a pattern value of 0.

Thanks a lot, I've put this patch in.