Thunderbird hang

[Neil Waldhauer]

Someone suggested this experiment, and I tried it. I configured Thunderbird to not check for new mail. Then I opened Thunderbird and left it open, but doing nothing. With the build of Thunderbird in ArcaOS 5.0.7, dated March 23, 2020, the computer will hang after three or four days. By hang I mean no response to mouse or keyboard input, no response to network.

With the thunderbird-45.8.0.en-US-i686-06-06-22.os2.zip  build, dated June 6, 2022, no hang occurs even after eight days.

[Dave Yeo]

I've lost track. Could you post the buildconfig from the hanging Thunderbird?
Help-->Troubleshooting Information, then click the about:buildconfig URL under Application Basics.

[Dave Yeo]

I've lost track. Could you post the buildconfig from the hanging Thunderbird?
Help-->Troubleshooting Information, then click the about:buildconfig URL under Application Basics.

Never mind. I take it that the screensaver was installed?
Anyways there were at least two problems with that build. It used the system NSPR4 and NSS, which was fine until Bitwise updated them, as about:troubleshooting shows,

Code: [Select]

Library Versions
Expected minimum version Version in use
NSPR       4.12                        4.23
NSS               3.23 Basic ECC        3.47
NSS Util       3.23                        3.47
NSS SSL       3.23 Basic ECC        3.47
NSS S/MIME  3.23 Basic ECC        3.47
This mostly caused the browsers (SM and FF) to vanish now and again, never noticed TB vanishing. Fixed by reverting to using the intree versions of NSPR4 and NSS.
The other issue is to do with optimizing. That build (Pentium M) used -mfpmath=sse, which caused a weird problem in the NSIdleServices() class, without the screensaver installed, caused FF and SM to start using one core at a hundred percent, it was consistent with different values for FF and SM, something like 7 or 11 minutes. Never noticed it with TB but didn't experiment that much and it was only an issue without the screensaver installed. NSIdleServices() does various cleanup stuff and tries to do it when screensaving.
This was fixed by not using -mfpmath=sse, which was a shame as it was magnitudes faster then using the co-processor for floating point math.
5.1 will also ship with i686 builds to avoid some optimizations.
Anyways seems you found another issue, perhaps caused by optimization or the updated NSS and NSPR. I'm using a Pentium 4 build without issue, though I haven't left it idle for days and use the screensaver.
Thanks for the interesting report.

[Neil Waldhauer]

Thanks for your comments. I didn't know about help->troubleshooting information.

You are correct about the screensaver installed. The screen saver was not enabled, however.

I'm trying to chase down some other system hangs to do with Lotus SmartSuite, but when the system hangs, it's sometimes difficult to see what caused it. Then I have to leave a system up for a long time to see if it hangs. This is a testing scheme that takes a long time to resolve.