Hi, an update of llama.cpp is in the works, it was just stalled by two bugs in the ggml build.
It should happen by the weekend, latest. Best, Christian On 2026-03-26 18:46, Philipp Klaus Krause wrote: > Package: llama.cpp > Version: 8064+dfsg-2 > Severity: normal > X-Debbugs-Cc: [email protected] > > Dear Maintainer, > > I am trying to use a Qwen3-coder model with llama.cpp and opencode. However, I > see frequent crashes of llama-server. One example can be found below. I > suspect > that this is upstream llama.cpp bug #19304 (https://github.com/ggml- > org/llama.cpp/issues/19304), which was fixed about a week after the release of > the currently packaged version, though the bug reports there used different > Qwen-Coder variants. > > […] > slot update_slots: id 3 | task 32108 | prompt done, n_tokens = 38850, > batch.n_tokens = 17 > slot init_sampler: id 3 | task 32108 | init sampler, took 5.85 ms, tokens: > text = 38850, total = 38850 > slot update_slots: id 3 | task 32108 | created context checkpoint 3 of 8 > (pos_min = 38832, pos_max = 38832, size = 75.376 MiB) > [New LWP 3895376] > [New LWP 3895375] > [New LWP 3895374] > [New LWP 3895373] > [New LWP 3895372] > [New LWP 3895371] > [New LWP 3895370] > [New LWP 3895369] > [New LWP 3895368] > [New LWP 3895367] > [New LWP 3895366] > [New LWP 3895365] > [New LWP 3895364] > [New LWP 3895363] > [New LWP 3895362] > [New LWP 3895128] > [New LWP 3895127] > [New LWP 3895126] > [New LWP 3895125] > [New LWP 3895124] > [New LWP 3895123] > [New LWP 3895122] > [New LWP 3895121] > [New LWP 3895120] > [New LWP 3895119] > [New LWP 3895118] > [New LWP 3895117] > [New LWP 3895116] > [New LWP 3895115] > [New LWP 3895114] > [New LWP 3895113] > [New LWP 3895112] > [New LWP 3895111] > [New LWP 3895110] > [New LWP 3895109] > [New LWP 3895108] > [New LWP 3895107] > [New LWP 3895106] > [New LWP 3895105] > [New LWP 3895104] > [New LWP 3895103] > [New LWP 3895102] > [New LWP 3895101] > [New LWP 3895100] > [New LWP 3895099] > [New LWP 3895098] > [New LWP 3895097] > [New LWP 3895096] > [New LWP 3895094] > [New LWP 3895093] > > This GDB supports auto-downloading debuginfo from the following URLs: > <https://debuginfod.debian.net> > Enable debuginfod for this session? (y or [n]) [answered N; input not from > terminal] > Debuginfod has been disabled. > To make this setting permanent, add 'set debuginfod enabled off' to .gdbinit. > [Thread debugging using libthread_db enabled] > Using host libthread_db library "/usr/lib/x86_64-linux-gnu/libthread_db.so.1". > __syscall_cancel_arch () at > ../sysdeps/unix/sysv/linux/x86_64/syscall_cancel.S:56 > ⚠️ warning: 56 ../sysdeps/unix/sysv/linux/x86_64/syscall_cancel.S: Datei oder > Verzeichnis nicht gefunden > #0 __syscall_cancel_arch () at > ../sysdeps/unix/sysv/linux/x86_64/syscall_cancel.S:56 > 56 in ../sysdeps/unix/sysv/linux/x86_64/syscall_cancel.S > #1 0x00007f9b9dc9be64 in __internal_syscall_cancel (a1=<optimized out>, > a2=<optimized out>, a3=<optimized out>, a4=<optimized out>, a5=a5@entry=0, > a6=a6@entry=0, nr=61) at ./nptl/cancellation.c:49 > ⚠️ warning: 49 ./nptl/cancellation.c: Datei oder Verzeichnis nicht gefunden > #2 0x00007f9b9dc9bead in __syscall_cancel (a1=<optimized out>, a2=<optimized > out>, a3=<optimized out>, a4=<optimized out>, a5=a5@entry=0, a6=a6@entry=0, > nr=61) at ./nptl/cancellation.c:75 > 75 in ./nptl/cancellation.c > #3 0x00007f9b9dd07c07 in __GI___wait4 (pid=<optimized out>, > stat_loc=<optimized out>, options=<optimized out>, usage=<optimized out>) at > ../sysdeps/unix/sysv/linux/wait4.c:30 > ⚠️ warning: 30 ../sysdeps/unix/sysv/linux/wait4.c: Datei oder Verzeichnis > nicht gefunden > #4 0x00007f9b9ec6aa53 in ggml_print_backtrace () from /usr/lib/x86_64-linux- > gnu/libggml-base.so.0 > #5 0x00007f9b9ec79c3f in ?? () from /usr/lib/x86_64-linux-gnu/libggml- > base.so.0 > #6 0x00007f9b9debb5fa in ?? () from /usr/lib/x86_64-linux-gnu/libstdc++.so.6 > #7 0x00007f9b9dea7749 in std::terminate() () from /usr/lib/x86_64-linux- > gnu/libstdc++.so.6 > #8 0x00007f9b9debb898 in __cxa_throw () from /usr/lib/x86_64-linux- > gnu/libstdc++.so.6 > #9 0x00007f9b9e2796ea in ?? () from /usr/lib/x86_64-linux- > gnu/llama/libllama.so.0 > #10 0x00007f9b9e2cd05c in llama_grammar_accept_impl(llama_grammar&, int) () > from /usr/lib/x86_64-linux-gnu/llama/libllama.so.0 > #11 0x000055c375409ba8 in ?? () > #12 0x000055c375277c94 in ?? () > #13 0x000055c3752c125e in ?? () > #14 0x000055c3751d5209 in ?? () > #15 0x00007f9b9dc33f75 in __libc_start_call_main > (main=main@entry=0x55c3751d1390, argc=argc@entry=19, > argv=argv@entry=0x7ffc39c3f588) at ../sysdeps/nptl/libc_start_call_main.h:58 > ⚠️ warning: 58 ../sysdeps/nptl/libc_start_call_main.h: Datei oder Verzeichnis > nicht gefunden > #16 0x00007f9b9dc34027 in __libc_start_main_impl (main=0x55c3751d1390, > argc=19, > argv=0x7ffc39c3f588, init=<optimized out>, fini=<optimized out>, > rtld_fini=<optimized out>, stack_end=0x7ffc39c3f578) at > ../csu/libc-start.c:360 > ⚠️ warning: 360 ../csu/libc-start.c: Datei oder Verzeichnis nicht gefunden > #17 0x000055c3751da7f1 in ?? () > [Inferior 1 (process 3895092) detached] > terminate called after throwing an instance of 'std::runtime_error' > what(): Unexpected empty grammar stack after accepting piece: =list (40972) > ./server-example: Zeile 11: 3895092 Abgebrochen llama-server > --model models/Qwen3-Coder-Next-Q4_K_S.gguf --ctx-size 131072 --alias > "Qwen3-Coder-Next" --seed 3407 --temp 1.0 --top-p 0.95 --min-p 0.01 --top-k 40 > --port 8081 > > > -- System Information: > Debian Release: forky/sid > APT prefers unstable > APT policy: (500, 'unstable'), (500, 'testing') > Architecture: amd64 (x86_64) > Foreign Architectures: i386 > > Kernel: Linux 6.19.8+deb14-amd64 (SMP w/32 CPU threads; PREEMPT) > Locale: LANG=de_DE.UTF-8, LC_CTYPE=de_DE.UTF-8 (charmap=UTF-8), LANGUAGE not > set > Shell: /bin/sh linked to /usr/bin/dash > Init: systemd (via /run/systemd/system) > LSM: AppArmor: enabled > > Versions of packages llama.cpp depends on: > ii llama.cpp-tools 8064+dfsg-2 > > Versions of packages llama.cpp recommends: > ii llama.cpp-tools-extra 8064+dfsg-2 > ii python3-gguf 8064+dfsg-2 > > Versions of packages llama.cpp suggests: > pn llama.cpp-examples <none> > > -- no debconf information
