| Message ID | 20171123020832.8165-10-mreitz@redhat.com |
|---|---|
| State | New |
| Headers | show |
| Series | iotests: Fix iotests for weird formats/options | expand |
On 11/22/2017 09:08 PM, Max Reitz wrote: > Tests 080, 130, 137, and 176 simply do not work with compat=0.10 for the > reasons stated there. > > 177 is a bit more interesting: Originally, it was actually very much > intended to work with compat=0.10 (it even had a special case for that). > However, it now prints the test image's map twice, and short of just not > doing that, there is no solution I can imagine that is both simple and > would leave compat=0.10 support intact. > So we lost that support in f0a9c18f9e7 and 81c219ac6ce Eric, any input before we downscope your test? > Signed-off-by: Max Reitz <mreitz@redhat.com> Without agonizing over it, I don't see an easy win either, so: Reviewed-by: John Snow <jsnow@redhat.com> (but it is a shame to lose the ability to test it.)
On 12/08/2017 07:46 PM, John Snow wrote: > > > On 11/22/2017 09:08 PM, Max Reitz wrote: >> Tests 080, 130, 137, and 176 simply do not work with compat=0.10 for the >> reasons stated there. >> >> 177 is a bit more interesting: Originally, it was actually very much >> intended to work with compat=0.10 (it even had a special case for that). >> However, it now prints the test image's map twice, and short of just not >> doing that, there is no solution I can imagine that is both simple and >> would leave compat=0.10 support intact. >> > > So we lost that support in > f0a9c18f9e7 > and > 81c219ac6ce > > Eric, any input before we downscope your test? Ouch, I broke my own test. Maybe the best thing would be to split 177 into two tests: the original test (as it was before f0a9c18) that works on both compats, and a new test that works on just compat=1.1 images for the things added in later commits. Since I'm the author for all commits (so far) to that file, I guess I can sign up for that work...
On 2017-12-09 17:53, Eric Blake wrote: > On 12/08/2017 07:46 PM, John Snow wrote: >> >> >> On 11/22/2017 09:08 PM, Max Reitz wrote: >>> Tests 080, 130, 137, and 176 simply do not work with compat=0.10 for the >>> reasons stated there. >>> >>> 177 is a bit more interesting: Originally, it was actually very much >>> intended to work with compat=0.10 (it even had a special case for that). >>> However, it now prints the test image's map twice, and short of just not >>> doing that, there is no solution I can imagine that is both simple and >>> would leave compat=0.10 support intact. >>> >> >> So we lost that support in >> f0a9c18f9e7 >> and >> 81c219ac6ce >> >> Eric, any input before we downscope your test? > > Ouch, I broke my own test. > > Maybe the best thing would be to split 177 into two tests: the original > test (as it was before f0a9c18) that works on both compats, and a new > test that works on just compat=1.1 images for the things added in later > commits. > > Since I'm the author for all commits (so far) to that file, I guess I > can sign up for that work... Did you make any progress? :-) Max
On 01/17/2018 07:41 AM, Max Reitz wrote: >>> So we lost that support in >>> f0a9c18f9e7 >>> and >>> 81c219ac6ce >>> >>> Eric, any input before we downscope your test? >> >> Ouch, I broke my own test. >> >> Maybe the best thing would be to split 177 into two tests: the original >> test (as it was before f0a9c18) that works on both compats, and a new >> test that works on just compat=1.1 images for the things added in later >> commits. >> >> Since I'm the author for all commits (so far) to that file, I guess I >> can sign up for that work... > > Did you make any progress? :-) Ah, thanks for the reminder. Posted as <20180117165420.15946-1-eblake@redhat.com>
diff --git a/tests/qemu-iotests/080 b/tests/qemu-iotests/080 index 55044c700b..1c2bd85742 100755 --- a/tests/qemu-iotests/080 +++ b/tests/qemu-iotests/080 @@ -41,8 +41,9 @@ trap "_cleanup; exit \$status" 0 1 2 3 15 _supported_fmt qcow2 _supported_proto file _supported_os Linux -# Internal snapshots are (currently) impossible with refcount_bits=1 -_unsupported_imgopts 'refcount_bits=1[^0-9]' +# - Internal snapshots are (currently) impossible with refcount_bits=1 +# - This is generally a test for compat=1.1 images +_unsupported_imgopts 'refcount_bits=1[^0-9]' 'compat=0.10' header_size=104 diff --git a/tests/qemu-iotests/130 b/tests/qemu-iotests/130 index e7e43de6d6..2c4b94da1b 100755 --- a/tests/qemu-iotests/130 +++ b/tests/qemu-iotests/130 @@ -45,6 +45,8 @@ _supported_fmt qcow2 _supported_proto generic _unsupported_proto vxhs _supported_os Linux +# We are going to use lazy-refcounts +_unsupported_imgopts 'compat=0.10' qemu_comm_method="monitor" diff --git a/tests/qemu-iotests/137 b/tests/qemu-iotests/137 index eb91e517d7..5a01250005 100755 --- a/tests/qemu-iotests/137 +++ b/tests/qemu-iotests/137 @@ -41,6 +41,8 @@ trap "_cleanup; exit \$status" 0 1 2 3 15 _supported_fmt qcow2 _supported_proto file _supported_os Linux +# We are going to use lazy-refcounts +_unsupported_imgopts 'compat=0.10' _make_test_img 64M diff --git a/tests/qemu-iotests/176 b/tests/qemu-iotests/176 index b8dc17c592..d38b3aeb91 100755 --- a/tests/qemu-iotests/176 +++ b/tests/qemu-iotests/176 @@ -48,6 +48,8 @@ trap "_cleanup; exit \$status" 0 1 2 3 15 _supported_fmt qcow2 _supported_proto file _supported_os Linux +# Persistent dirty bitmaps require compat=1.1 +_unsupported_imgopts 'compat=0.10' function run_qemu() { diff --git a/tests/qemu-iotests/177 b/tests/qemu-iotests/177 index 28990977f1..86cf25f855 100755 --- a/tests/qemu-iotests/177 +++ b/tests/qemu-iotests/177 @@ -39,6 +39,8 @@ trap "_cleanup; exit \$status" 0 1 2 3 15 _supported_fmt qcow2 _supported_proto file +# This test assumes that discard leaves zero clusters +_unsupported_imgopts 'compat=0.10' CLUSTER_SIZE=1M size=128M @@ -93,15 +95,6 @@ echo "== verify image content ==" function verify_io() { - if ($QEMU_IMG info -f "$IMGFMT" "$TEST_IMG" | - grep "compat: 0.10" > /dev/null); then - # For v2 images, discarded clusters are read from the backing file - discarded=11 - else - # Discarded clusters are zeroed for v3 or later - discarded=0 - fi - echo read -P 22 0 1000 echo read -P 33 1000 128k echo read -P 22 132072 7871512 @@ -109,7 +102,7 @@ function verify_io() echo read -P 22 10096640 23457792 echo read -P 0 32M 32M echo read -P 22 64M 13M - echo read -P $discarded 77M 29M + echo read -P 0 77M 29M echo read -P 22 106M 4M echo read -P 11 110M 18M }
Tests 080, 130, 137, and 176 simply do not work with compat=0.10 for the reasons stated there. 177 is a bit more interesting: Originally, it was actually very much intended to work with compat=0.10 (it even had a special case for that). However, it now prints the test image's map twice, and short of just not doing that, there is no solution I can imagine that is both simple and would leave compat=0.10 support intact. Signed-off-by: Max Reitz <mreitz@redhat.com> --- tests/qemu-iotests/080 | 5 +++-- tests/qemu-iotests/130 | 2 ++ tests/qemu-iotests/137 | 2 ++ tests/qemu-iotests/176 | 2 ++ tests/qemu-iotests/177 | 13 +++---------- 5 files changed, 12 insertions(+), 12 deletions(-)