[Bug] Initial firmware installation intermittently fails [Windows 11 x64]

[EdHerdman]

Quick summary

In issue #12621 I reported RPCS3 would reliably crash to desktop after installing the firmware and clicking through the "firmware and LLE modules installed" message, before compiling the modules. Since that build, I've tried installing the firmware in multiple versions and occasionally RPCS3 will crash in the same way on the first attempt, but succeed on the second attempt by simply re-launching the emulator and overwriting the firmware. The behavior is improved, but still unreliable.

rpcs3-v0.0.24-14173-3581c5b0_win64 based on #12686 still has this behavior - I unzipped the files into a fresh directory, immediately tried to install the firmware through drag-and-drop, and it crashed immediately after the initial installation and showing the "firmware and LLE modules installed" message on the first run.

UPDATE: rpcs3-v0.0.24-14183-a604394b_win64 from #12700 succeeded on installing firmware on the first attempt. Hopefully this sticks...

Attach a log file

RPCS3.log.gz

System configuration

Windows 11 Pro 22H2 (build 22621.521).
Ryzen 7 3700X

[Augusto7743]

Please test 0.0.24-14123.
Since 0.0.24-14127 happening wrong huge memory allocation crashing Linux.

[EdHerdman]

In #12621 I found the commit which originally seemed to introduce the bug, and the last build without the bug; both are newer than 0.0.24-14123. I'm also only testing Windows builds for now.

I think maybe you've commented on the wrong issue? I will check the latest build too, though.

Edit: Latest build #12700 worked on the first attempt. Since this is an intermittent issue I can't confirm it's really 'fixed,' but so far it's looking good.

[EdHerdman]

I've tried this on some further builds, all succeeded first try unless otherwise noted:
rpcs3-v0.0.24-14226-87411da9_win64
rpcs3-v0.0.24-14232-3c884772_win64
rpcs3-v0.0.24-14247-36b1004e_win64 (I might have aborted firmware install on this one time. It crashed in one attempt to install firmware, but another, fresh copy succeeded)
rpcs3-v0.0.24-14263-0737c788_win64
rpcs3-v0.0.24-14266-cf86b6c1_win64

I'm going to watch it a bit further and see if it happens with another new build. Perhaps something unrelated to the firmware cache is causing a crash? But I also feel like perhaps this should be closed soon, because it's very easy to work around this by just trying again if it happens.

[Asinin3]

I might have aborted firmware install on this

Sounds like you are making entirely new RPCS3 folders each time. You don't need to. Just put the executable inside RPCS3 folder. I keep around 300 exe files in my folder of different builds for fast regression tracking.

[EdHerdman]

Yes, you're right, but tracking whether RPCS3 firmware install succeeds was the entire purpose of reporting this issue.

[EdHerdman]

I can confirm this is still happening occasionally: rpcs3-v0.0.24-14337-5210df68_win64

Not a deal-breaker, but something odd is still happening behind the scenes causing an unwelcome CTD.