You are not logged in.
By activating OpenCL in GIMP, I get
Thread 1 "gimp" received signal SIGSEGV, Segmentation fault.
0x00007fffd3416227 in ?? () from /usr/lib/libigc.so.2All ok if OpenCL is disabled.
Maybe Gimp should be rebuilded? Or GEGL? Or is it a GIMP bug?
Last edited by guiodic (2025-11-07 17:25:24)
Offline
https://gitlab.archlinux.org/archlinux/ … -/issues/2 ?
https://github.com/bdwgc/bdwgc/issues/783
Patch seems on the way but also https://gitlab.archlinux.org/archlinux/ … 5059f32a4e should™ have fixed that?
pacman -Qikk gcOffline
https://gitlab.archlinux.org/archlinux/ … -/issues/2 ?
https://github.com/bdwgc/bdwgc/issues/783
Patch seems on the way but also https://gitlab.archlinux.org/archlinux/ … 5059f32a4e should™ have fixed that?pacman -Qikk gc
Nome : gc
Versione : 8.2.10-2
Descrizione : A garbage collector for C and C++
Architettura : x86_64
Offline
Do you have a better backtrace that isn't just the last stack position and ideally using https://wiki.archlinux.org/title/Debuginfod ?
Offline
seth wrote:https://gitlab.archlinux.org/archlinux/ … -/issues/2 ?
https://github.com/bdwgc/bdwgc/issues/783
Patch seems on the way but also https://gitlab.archlinux.org/archlinux/ … 5059f32a4e should™ have fixed that?pacman -Qikk gcNome : gc
Versione : 8.2.10-2
Descrizione : A garbage collector for C and C++
Architettura : x86_64
https://bbs.archlinux.org/viewtopic.php?id=57855'
Picking 4 lines to post doesn't help.
Offline
guiodic wrote:seth wrote:https://gitlab.archlinux.org/archlinux/ … -/issues/2 ?
https://github.com/bdwgc/bdwgc/issues/783
Patch seems on the way but also https://gitlab.archlinux.org/archlinux/ … 5059f32a4e should™ have fixed that?pacman -Qikk gcNome : gc
Versione : 8.2.10-2
Descrizione : A garbage collector for C and C++
Architettura : x86_64https://bbs.archlinux.org/viewtopic.php?id=57855'
Picking 4 lines to post doesn't help.
Nome : gc
Versione : 8.2.10-2
Descrizione : A garbage collector for C and C++
Architettura : x86_64
URL : https://www.hboehm.info/gc/
Licenze : LicenseRef-GC MIT
Gruppi : Nessuno
Fornisce : Nessuno
Dipenda da : gcc-libs
Dipendenze opzionali : Nessuno
Richiesto da : guile inkscape
Opzionale per : Nessuno
Conflitti con : Nessuno
Sostituisce : Nessuno
Spazio richiesto : 781,23 KiB
Pacchettizzatore : Frederik Schwan <freswa@archlinux.org>
Data di creazione : mer 22 ott 2025, 19:28:23
Data di installazione : gio 23 ott 2025, 10:27:22
Motivo dell'installazione : Installato come dipendenza di un altro pacchetto
Script di install : No
Convalidato da : Firma
gc: 81 file totali, 0 file alteratiNow that you know the licence for the package, you'll surely understand how to fix the crash, right?
Offline
The line he was looking (and I also somewhat asking) for was
gc: 81 file totali, 0 file alteratiPlease avoid bloating the thread with full quotes of previous posts and then
Do you have a better backtrace that isn't just the last stack position and ideally using https://wiki.archlinux.org/title/Debuginfod ?
Offline
Do you have a better backtrace that isn't just the last stack position and ideally using https://wiki.archlinux.org/title/Debuginfod ?
Starting program: /usr/bin/gimp
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/usr/lib/libthread_db.so.1".
[New Thread 0x7ffff1a446c0 (LWP 167930)]
[New Thread 0x7ffff09016c0 (LWP 167931)]
[New Thread 0x7fffebfff6c0 (LWP 167932)]
[New Thread 0x7fffeb7fe6c0 (LWP 167933)]
[New Thread 0x7fffeaffd6c0 (LWP 167934)]
[New Thread 0x7fffea7fc6c0 (LWP 167935)]
[New Thread 0x7fffe9ffb6c0 (LWP 167936)]
[New Thread 0x7fffe97fa6c0 (LWP 167937)]
[New Thread 0x7fffe6f006c0 (LWP 167938)]
Thread 1 "gimp" received signal SIGSEGV, Segmentation fault.
0x00007fffd32d4227 in IGC::SOALayoutChecker::visitBitCastInst (this=this@entry=0x7fffffff7dd0,
BI=...)
at /usr/src/debug/intel-graphics-compiler/intel-graphics-compiler-2.20.3/IGC/Compiler/CISACodeGen/LowerGEPForPrivMem.cpp:524
524 Type *baseT = GetBaseType(IGCLLVM::getNonOpaquePtrEltTy(BI.getType()), true);Offline
If I unset GEGL_USE_OPENCL=gpu, Gimp starts.
Then, it seems a bug in GEGL with the new intel opencl drivers.
Offline
Yup.
The backtrace is gonna be much longer than that, can you post the rest of it?
Offline
Yup.
The backtrace is gonna be much longer than that, can you post the rest of it?
gdb only gives this after downloading the symbols.
Offline
It uses a pager, have you tried to move up/down w/ the arrow or page keys?
Offline
It uses a pager, have you tried to move up/down w/ the arrow or page keys?
I'm stupid, but not that stupid.
Offline
You're looking at the stored coredump there?
And that's the *entire* output when you enter
thread apply all backtrace fullinto gdb???
Offline
You're looking at the stored coredump there?
And that's the *entire* output when you enterthread apply all backtrace fullinto gdb???
No, that's what comes out after `run` in gdb. But I'm not at my computer right now, so I'll post the entire coredump tomorrow when I get back to my PC.
I doubt Arch can do anything about it, though, as it's probably an upstream issue.
Offline
The longer trace should make it easier (also for the upstream devs) to spot the cause or an existing bug.
Offline