[Bug 57933] New: SAVIHost can take a long time to close, with it and wineserver eating the CPU
https://bugs.winehq.org/show_bug.cgi?id=57933 Bug ID: 57933 Summary: SAVIHost can take a long time to close, with it and wineserver eating the CPU Product: Wine Version: 7.5 Hardware: x86-64 OS: Linux Status: UNCONFIRMED Severity: minor Priority: P2 Component: winealsa.drv Assignee: wine-bugs(a)winehq.org Reporter: dark(a)darkok.xyz Distribution: --- Starting from 1fa0d491a87f5c7d9f600ea95ee2bfc37ca6f940 in Wine 7.5, SAVIhost can take a long time to close, with it and wineserver then using the CPU a lot, causing me to have to terminate it manually. It seems the amount of time it can take depends on system, on my main machine I've had it take a minute sometimes, whereas my laptop (which is much slower) can somehow do so in up to about 10 seconds. This still happens on Wine 10.3, which just released. Downloads: https://www.hermannseib.com/programs/savihostx64.zip https://cdn.roland.com/assets/media/zip/scva_win_trial.zip Steps to reproduce: 1. Execute "winetricks mfc90" to get a required DLL 2. Extract scva_win_trial.zip and install the Roland Sound Canvas VA VST 3. Extract savihostx64.zip, and move savihost.exe to "C:\Program Files\Roland\Sound Canvas VA" 4. Rename savihost.exe to "Sound Canvas VA.exe" 5. Run the renamed executable 6. Go to Devices>MIDI, and set Input Port 1 to "Midi Through Port-0" 7. Close the program (I mentioned this in passing a few years ago in 52828 but never made it its own report, sorry) -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.
https://bugs.winehq.org/show_bug.cgi?id=57933 DarkOK <dark(a)darkok.xyz> changed: What |Removed |Added ---------------------------------------------------------------------------- Regression SHA1| |1fa0d491a87f5c7d9f600ea95ee | |2bfc37ca6f940 Keywords| |download, regression -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.
http://bugs.winehq.org/show_bug.cgi?id=57933 --- Comment #1 from Sven Baars <sven.wine@gmail.com> --- Using ntsync actually fixes this issue for me. But why? This seems to be caused by a race condition in the application, where it calls midiInAddBuffer() from the MIDM_RESET callback. If we call midi_in_reset() (from the while loop in the commit that caused the regression) before MIDM_ADDBUFFER gets processed, everything works fine. If we call if after, we're stuck in an infinite loop. So it may just be that on Windows, and with ntsync, things happen fast enough that we never get stuck in the while loop. So I'm not sure if there's an actual bug to fix here. I'm also not sure how to verify that this is an application bug though, so I'll leave this for now. -- Do not reply to this email, post in Bugzilla using the above URL to reply. You are receiving this mail because: You are watching all bug changes.
participants (2)
-
WineHQ Bugzilla -
WineHQ Bugzilla