Closed Joermungand closed 6 years ago
I will first try to reproduce the issue with CR2 files I have, and eventually will ask for an example of failing RAW file if I do not see the crash...
Thanks!
Would you be able to compile photoflow with -DCMAKE_BUILD_TYPE=RelWithDebInfo
cmake flag, run the program through gdb
and provide me the full backtrace of the crash?
Also, could you provide me an example of failing CR2 file? I tried with few that I have, and I could never observe any crash...
Thanks!
Sure, right away. It’ll take a while to compile, though. Meanwhile, an example it crashes with is right here.
Here’s the backtrace. Hope it’s useful. bt-full.log
Could you tell me the version of VIPS (or libvips) that is installed on your system? It might be some bug in VIPS itself...
VIPS is currently at 8.6.1. As a matter of fact the crash occurred the same day as VIPS was updated from 8.5.9 – however, I doubt this should be the case, since I also tried downgrading VIPS to 8.5.9 and the result was exactly the same: no crash with Photoflow prior to that commit, crash from then on.
Interesting... in any case, the culprit seems to be the vips_image_remove()
function, so I will dig into it.
I have just committed some additional diagnostics and, hopefully, a fix. Could you please update your sources and try again?
If it still fails, I would need the full terminal output after the crash...
Thanks!
Yeah, looks like that to me. Please let me know if I can help in any way.
On it.
Crashed again. Here’s the complete log: crash.log. Please let me know if you need the gdb backtrace.
Please let me know if you need the gdb backtrace.
Yes, please!
OK. Here’s the gdb log: bt-full.log. And here’s the terminal output of running gdb: gdb-output.log.
Thanks! I have hopefully finally found the mistake: the use of a deleted image (out
) here.
Please update again, and cross fingers...
That would explain it. Updating.
Yap. It looks like that was the culprit. Working fine now. Thanks a lot.
Good job, thanks a lot!
I'm closing the issue.
Photoflow crashes (segfaults) when opening CR2 files. I’ve managed to trace the regression down to commit 30bc9c5b2b4a567a307b92816864f831084e9e26 on January 9th. Please let me know what logs you need added.