[Qemu-devel] [PATCH] qemu-iotests: Honor ${CACHEMODE} in 091

[Qemu-devel] [PATCH] qemu-iotests: Honor ${CACHEMODE} in 091

[Fam Zheng]

We should allow testing this on tmpfs. Any cache setting in iotests
should try to obey $CACHEMODE.

Signed-off-by: Fam Zheng <famz@redhat.com>
---
 tests/qemu-iotests/091 | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/tests/qemu-iotests/091 b/tests/qemu-iotests/091
index 384b3ac..a7deea8 100755
--- a/tests/qemu-iotests/091
+++ b/tests/qemu-iotests/091
@@ -59,13 +59,13 @@ echo === Starting QEMU VM1 ===
 echo
 
 qemu_comm_method="monitor"
-_launch_qemu -drive file="${TEST_IMG}",cache=none,id=disk
+_launch_qemu -drive file="${TEST_IMG}",cache=${CACHEMODE},id=disk
 h1=$QEMU_HANDLE
 
 echo
 echo === Starting QEMU VM2 ===
 echo
-_launch_qemu -drive file="${TEST_IMG}",cache=none,id=disk \
+_launch_qemu -drive file="${TEST_IMG}",cache=${CACHEMODE},id=disk \
              -incoming "exec: cat '${MIG_FIFO}'"
 h2=$QEMU_HANDLE
 
-- 
1.9.2

Re: [Qemu-devel] [PATCH] qemu-iotests: Honor ${CACHEMODE} in 091

[Kevin Wolf]

Am 19.05.2014 um 14:24 hat Fam Zheng geschrieben:
> We should allow testing this on tmpfs. Any cache setting in iotests
> should try to obey $CACHEMODE.
> 
> Signed-off-by: Fam Zheng <famz@redhat.com>

Should we set _default_cache_mode=none then at least? We should probably
also refuse to run with unsafe.

Kevin

>  tests/qemu-iotests/091 | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/tests/qemu-iotests/091 b/tests/qemu-iotests/091
> index 384b3ac..a7deea8 100755
> --- a/tests/qemu-iotests/091
> +++ b/tests/qemu-iotests/091
> @@ -59,13 +59,13 @@ echo === Starting QEMU VM1 ===
>  echo
>  
>  qemu_comm_method="monitor"
> -_launch_qemu -drive file="${TEST_IMG}",cache=none,id=disk
> +_launch_qemu -drive file="${TEST_IMG}",cache=${CACHEMODE},id=disk
>  h1=$QEMU_HANDLE
>  
>  echo
>  echo === Starting QEMU VM2 ===
>  echo
> -_launch_qemu -drive file="${TEST_IMG}",cache=none,id=disk \
> +_launch_qemu -drive file="${TEST_IMG}",cache=${CACHEMODE},id=disk \
>               -incoming "exec: cat '${MIG_FIFO}'"
>  h2=$QEMU_HANDLE
>  
> -- 
> 1.9.2
> 

Re: [Qemu-devel] [PATCH] qemu-iotests: Honor ${CACHEMODE} in 091

[Fam Zheng]

On Mon, 05/19 16:40, Kevin Wolf wrote:
> Am 19.05.2014 um 14:24 hat Fam Zheng geschrieben:
> > We should allow testing this on tmpfs. Any cache setting in iotests
> > should try to obey $CACHEMODE.
> > 
> > Signed-off-by: Fam Zheng <famz@redhat.com>
> 
> Should we set _default_cache_mode=none then at least?

Yes we should set a default. We use _default_cache_mode=writethrough elsewhere,
why use none over writethrough here?

>                                                       We should probably
> also refuse to run with unsafe.

OK, I'll add _supported_cache_modes.

Fam

> >  tests/qemu-iotests/091 | 4 ++--
> >  1 file changed, 2 insertions(+), 2 deletions(-)
> > 
> > diff --git a/tests/qemu-iotests/091 b/tests/qemu-iotests/091
> > index 384b3ac..a7deea8 100755
> > --- a/tests/qemu-iotests/091
> > +++ b/tests/qemu-iotests/091
> > @@ -59,13 +59,13 @@ echo === Starting QEMU VM1 ===
> >  echo
> >  
> >  qemu_comm_method="monitor"
> > -_launch_qemu -drive file="${TEST_IMG}",cache=none,id=disk
> > +_launch_qemu -drive file="${TEST_IMG}",cache=${CACHEMODE},id=disk
> >  h1=$QEMU_HANDLE
> >  
> >  echo
> >  echo === Starting QEMU VM2 ===
> >  echo
> > -_launch_qemu -drive file="${TEST_IMG}",cache=none,id=disk \
> > +_launch_qemu -drive file="${TEST_IMG}",cache=${CACHEMODE},id=disk \
> >               -incoming "exec: cat '${MIG_FIFO}'"
> >  h2=$QEMU_HANDLE
> >  
> > -- 
> > 1.9.2
> > 

Re: [Qemu-devel] [PATCH] qemu-iotests: Honor ${CACHEMODE} in 091

[Kevin Wolf]

Am 20.05.2014 um 03:03 hat Fam Zheng geschrieben:
> On Mon, 05/19 16:40, Kevin Wolf wrote:
> > Am 19.05.2014 um 14:24 hat Fam Zheng geschrieben:
> > > We should allow testing this on tmpfs. Any cache setting in iotests
> > > should try to obey $CACHEMODE.
> > > 
> > > Signed-off-by: Fam Zheng <famz@redhat.com>
> > 
> > Should we set _default_cache_mode=none then at least?
> 
> Yes we should set a default. We use _default_cache_mode=writethrough elsewhere,
> why use none over writethrough here?

I think the test case set cache=none originally because it's a migration
test and this is the recommended setting for migration. As long as we're
on localhost, writethrough should be fine, though, so it's probably not
strictly necessary.

> >                                                       We should probably
> > also refuse to run with unsafe.
> 
> OK, I'll add _supported_cache_modes.

Thanks.

Kevin