https://bugs.kde.org/show_bug.cgi?id=523186
Fabian Rensen <[email protected]> changed: What |Removed |Added ---------------------------------------------------------------------------- Resolution|FIXED |--- Status|RESOLVED |REOPENED Ever confirmed|0 |1 --- Comment #5 from Fabian Rensen <[email protected]> --- (In reply to Albert Astals Cid from comment #4) > 6.28.1 should fix it. > > Reopen if it still happens after you update. I updated to 6.28.1 today on my Fedora 44 (kf6-kimageformats-6.28.1-1.fc44.src.rpm). I still get the same crashes when Dolphin builds thumbnails or when I open an image in Gwenview. I dug a bit through the crashdump, and here is the interesting part: > #0 __pthread_kill_implementation (threadid=<optimized out>, > signo=signo@entry=11, no_tid=no_tid@entry=0) at pthread_kill.c:44 > 44 return INTERNAL_SYSCALL_ERROR_P (ret) ? INTERNAL_SYSCALL_ERRNO > (ret) : 0; > [Current thread is 1 (Thread 0x7fc6916016c0 (LWP 38005))] > (gdb) frame 6 > #6 0x00007fc6cba7180b in StreamReader_CApi::request_range > (this=0x7fc6c0008610, start=<optimized out>, end_pos=2612) at > /usr/src/debug/libheif-1.21.2-1.fc44.x86_64/libheif/bitstream.h:168 > 168 heif_reader_range_request_result result = > m_func_table->request_range(start, end_pos, m_userdata); > (gdb) p m_func_table->reader_api_version > $1 = 2 So, in my live image, the reader api version is set to 2, while the code clearly sets it to 1 as added in 6.28.0 (heif.cpp): > static heif_reader create_heif_reader_for_qiodevice() > { > heif_reader reader = {}; > > reader.reader_api_version = 1; But I think that is the reason why libheif tries to go through the request_range route. -- You are receiving this mail because: You are watching all bug changes.
