diff options
Diffstat (limited to 'results/scraper/launchpad-without-comments/1499908')
| -rw-r--r-- | results/scraper/launchpad-without-comments/1499908 | 55 |
1 files changed, 55 insertions, 0 deletions
diff --git a/results/scraper/launchpad-without-comments/1499908 b/results/scraper/launchpad-without-comments/1499908 new file mode 100644 index 00000000..5ddee92e --- /dev/null +++ b/results/scraper/launchpad-without-comments/1499908 @@ -0,0 +1,55 @@ +hda sound capture broken with VNC + +QEmu is being used to run the Wine conformance tests in Windows virtual machines. Wine's conformance tests check the behavior of various Windows APIs and verify that they behave as expected. One of the tests checks the behavior of the mmdevapi sound capture APIs. This test works fine on real hardware and also works fine in various QEmu VMs but fails in some others. Further investigation showed that: + + * The mmdevapi:capture tests work on the two Vista VMs. Those use the ac97 sound card and are configured for VNC access to the VM. + + * The mmdevapi:capture tests fail in the Windows 7+ VMs. Those use an hda sound card and are configured for VNC access to the VM (so '-device intel-hda,id=sound0,bus=pci.0,addr=0x4 -device hda-duplex,id=sound0-codec0,bus=sound0.0,cad=0' and '-vnc 127.0.0.1:0'). + + * Furthermore configuring the VM for Spice access fixes the mmdevapi:capture test (so replacing -vnc with '-spice port=5900,addr=127.0.0.1,disable-ticketing,seamless-migration=on'), this even if no client connects to the VM. + +So in effect the -spice and -vnc options change the behavior of the sound device. + +To reproduce this bug: +1. Set up a Windows 7+ VM with an hda sound card ('ich6' in libvirt). +2. Set it up for access using VNC. +3. Copy the attached mmdevapi_test.exe file to it. (*) +4. Run the tests as follows: + mmdevapi_test.exe capture + +If you see these 'Test Failed' lines then the bug is still present: + +capture.c:586: Returned latency: 5.8050 ms +capture.c:178: Test failed: Position 1015 expected 0 +capture.c:186: Wait'ed position 1015 pad 0 flags 1, amount of frames locked: 448 +capture.c:228: Test failed: Position 2167 expected 1463 +capture.c:248: Sleep.1 position 2167 pad 4032 flags 1, amount of frames locked: 448 +capture.c:256: Test failed: Position 2167 expected 1463 +capture.c:292: GetBufferSize 21996 period size 448 +capture.c:302: Overrun position 4215 pad 8960 flags 1, amount of frames locked: 448 +capture.c:308: Test failed: GCP 8960 vs. BufferSize 21996 +capture.c:313: Test failed: Position 4215 gap 2304 +capture.c:329: Cont'ed position 5303 pad 8512 flags 1, amount of frames locked: 448 +capture.c:333: Test failed: Position 5303 expected 4663 +capture.c:334: Test failed: flags 1 +capture.c:353: Restart position 7351 pad 8064 flags 1, amount of frames locked: 448 +capture.c:358: Test failed: Position 7351 expected 5111 +capture.c:359: Test failed: flags 1 + +In case it helps, here is the source of mmdevapi_test.exe: +https://source.winehq.org/git/wine.git/?a=blob;f=dlls/mmdevapi/tests/capture.c;hb=60d1d6f5952e8b5d6fb0327a28c047058851fa70#l178 + + +So far I have confirmed that this bug is present in QEmu as shipped in the following Debian packages: + * qemu-kvm + qemu-system-x86 1:2.1+dfsg-12+deb8u2 + linux-image-3.16.0-4-amd64 3.16.7-ckt11-1+deb8u3 + * qemu-system-x86 1:2.3+dfsg-6a + linux-image-4.1.0-1-amd64 4.1.3-1 + + +(*) As alternatives to using the attached binary you can: +- Compile it from Wine's source. See: + https://source.winehq.org/git/wine.git/ + +- Or download the latest WineTest binary from https://test.winehq.org/builds/winetest-latest.exe + And extract the mmdevapi_test.exe from there: + winetest.exe -x tests + tests\mmdevapi_test.exe capture \ No newline at end of file |