https://bugs.winehq.org/show_bug.cgi?id=54833
Bug ID: 54833 Summary: Heroes of Might and Magic 5 crash on starting some builtin maps, since wine 7.13 Product: Wine Version: 8.5 Hardware: x86-64 OS: Linux Status: UNCONFIRMED Severity: normal Priority: P2 Component: -unknown Assignee: wine-bugs@winehq.org Reporter: galdralag@bk.ru Distribution: ---
While answering on https://bugs.winehq.org/show_bug.cgi?id=39742#c21 I found that I can't start many single player maps (Single Player -> Custom Game) - game crashes. These maps was playable earlier. And playing on maps that can be started game often occasionally crashes.
After investigation different versions I found this begins from wine 7.13 on 7.12 all ok.
https://bugs.winehq.org/show_bug.cgi?id=54833
Austin English austinenglish@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Keywords| |regression
https://bugs.winehq.org/show_bug.cgi?id=54833
--- Comment #1 from Rafał Mużyło galtgendo@o2.pl --- ...in short: logs, please.
https://bugs.winehq.org/show_bug.cgi?id=54833
--- Comment #2 from Vjacheslav galdralag@bk.ru --- (In reply to Rafał Mużyło from comment #1)
...in short: logs, please.
What logs I need give? Console output of crashed game?
https://bugs.winehq.org/show_bug.cgi?id=54833
--- Comment #3 from Rafał Mużyło galtgendo@o2.pl --- While the console spew is not always helpful in determining the cause of the problem, most of the time it's at very least a good starting point to narrow things down.
...and I mean all of it, not just the crash backtrace.
https://bugs.winehq.org/show_bug.cgi?id=54833
Zeb Figura z.figura12@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |z.figura12@gmail.com
--- Comment #4 from Zeb Figura z.figura12@gmail.com --- (In reply to Rafał Mużyło from comment #1)
...in short: logs, please.
This kind of comment isn't very helpful (or polite, for that matter), not least because "logs" is not nearly specific enough. Best is to provide a link to instructions:
https://wiki.winehq.org/FAQ#How_can_I_get_a_debugging_log_.28a.k.a._terminal...
https://bugs.winehq.org/show_bug.cgi?id=54833
--- Comment #5 from Rafał Mużyło galtgendo@o2.pl --- (In reply to comment #4)
Interestingly enough, on top of bug creation page (https://bugs.winehq.org/enter_bug.cgi?product=Wine) there's a link to bugs wiki page which - surprisingly enough - provides the very same link.
But of course nobody reads those.
https://bugs.winehq.org/show_bug.cgi?id=54833
--- Comment #6 from Vjacheslav galdralag@bk.ru --- Created attachment 74367 --> https://bugs.winehq.org/attachment.cgi?id=74367 Game crash output, wine 8.6
https://bugs.winehq.org/show_bug.cgi?id=54833
--- Comment #7 from Rafał Mużyło galtgendo@o2.pl ---
0120:err:d3d:wined3d_debug_callback 0124E308: "GL_OUT_OF_MEMORY error generated. Failed to allocate CPU address space mapping for buffer object (consider building 64-bit app).".
That's one of the nasty ones - some apps/games map excessively in wine. Some even far above what you'd expect from game size (IIRC I've seen below 1G in size games dying from this). It's relatively rare and would be nice if it got fixed, but chances are it won't be.
Perhaps once the new 32-on-64 mode becomes reliable things get changed, but that's not a given.
If you checked the range correctly, Zeb is most likely the author of the change that triggered this.
https://bugs.winehq.org/show_bug.cgi?id=54833
--- Comment #8 from Rafał Mużyło galtgendo@o2.pl --- ...PS: perhaps there's another gotcha about 831ff102008e2ba93a403344646b5ed67258eaeb...
https://bugs.winehq.org/show_bug.cgi?id=54833
--- Comment #9 from Zeb Figura z.figura12@gmail.com --- (In reply to Rafał Mużyło from comment #7)
0120:err:d3d:wined3d_debug_callback 0124E308: "GL_OUT_OF_MEMORY error generated. Failed to allocate CPU address space mapping for buffer object (consider building 64-bit app).".
That's one of the nasty ones - some apps/games map excessively in wine. Some even far above what you'd expect from game size (IIRC I've seen below 1G in size games dying from this). It's relatively rare and would be nice if it got fixed, but chances are it won't be.
It doesn't necessarily have to do with applications performing 3D mappings, or even if it does, that's an odd thing to ascribe blame to, seeing as those mappings happen on Windows as well and the latter usually does fine. More likely culprits include excessively sized system libraries, PulseAudio shared memory slices, or system 3D drivers not unmapping buffers even when we ask them to.
Perhaps once the new 32-on-64 mode becomes reliable things get changed, but that's not a given.
New-style wow64 won't help most of this, at least not by itself. Mappings accessible to the PE side still need to go in the bottom 2 (or 4) GiB.
If you checked the range correctly, Zeb is most likely the author of the change that triggered this.
What change?
(In reply to Rafał Mużyło from comment #8)
...PS: perhaps there's another gotcha about 831ff102008e2ba93a403344646b5ed67258eaeb...
In what way?
https://bugs.winehq.org/show_bug.cgi?id=54833
--- Comment #10 from Rafał Mużyło galtgendo@o2.pl ---
What change? In what way?
Well...if the version range is correct, the most likely culprit (going by symptoms) lies within your 2022-07-14 series for wined3d (at least going by commit titles). Looking specifically at "Respect the client map count even for suballocated BOs", which seems to made things more correct, but looks as potentially likely to result in increased memory consumption (by delaying memory releases - ...perhaps in some corner cases indefinitely ?).
https://bugs.winehq.org/show_bug.cgi?id=54833
--- Comment #11 from Rafał Mużyło galtgendo@o2.pl --- ...and yes, I'm doing my wild guesstimate shtick again, but with the given data, that set/commit seems most probable... (unless the reporter is willing/able to do full bisect or someone else does it for him, that's likely closest we'll get)
https://bugs.winehq.org/show_bug.cgi?id=54833
--- Comment #12 from Rafał Mużyło galtgendo@o2.pl --- As for the first part of your post, generally yes, but in this case we're dealing with a regression and that changes the equation quite a bit.
https://bugs.winehq.org/show_bug.cgi?id=54833
--- Comment #13 from Vjacheslav galdralag@bk.ru --- Some updates to my first report. Problem with this game is more complex. Every version beginning from 7.13 changes behavior.
7.12 - all ok. 7.13 - I have laptop with nvidia optimus. Using nvidia card game shows black screen. On intel card game starts and very slow(as expected) and problem map is ok. This behavior is from 7.12 to 8.0-rc1.
8.0-rc1 - On nvidia card game still shows black screen. On intel card game starts and problem map hangs. 8.0-rc4 - On nvidia card game still shows black screen. On intel card game starts without sound(!) and problem map crashes(!). 8.0 - Game starts with black screen for both cards. 8.4 - Game starts on nvidia card(!) with no sound, problem map crashes. On intel card game crashes on start. 8.5 - Game starts on nvidia card and sound appeared(!) again, problem map crashes. On intel card game crashes on start.
https://bugs.winehq.org/show_bug.cgi?id=54833
--- Comment #14 from Vjacheslav galdralag@bk.ru --- Link to download game https://disk.yandex.ee/d/jzJtWicpSI_EZQ
https://bugs.winehq.org/show_bug.cgi?id=54833
--- Comment #15 from Vjacheslav galdralag@bk.ru --- wine 9.3: still not fixed