Misc. bug: llama-parallel segmentation fault

[gnusupport]

Name and Version

llama-cli --version
ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
Device 0: NVIDIA GeForce RTX 3090, compute capability 8.6, VMM: yes
version: 854 (eaea325)
built with cc (Debian 12.2.0-14) 12.2.0 for x86_64-linux-gnu

Operating systems

Linux

Which llama.cpp modules do you know to be affected?

Other (Please specify in the next section)

Command line

llama-parallel 
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:    no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
  Device 0: NVIDIA GeForce RTX 3090, compute capability 8.6, VMM: yes
build: 854 (eaea325) with cc (Debian 12.2.0-14) 12.2.0 for x86_64-linux-gnu
gguf_init_from_file: failed to open GGUF file 'models/7B/ggml-model-f16.gguf'
llama_model_load_from_file_impl: using device CUDA0 (NVIDIA GeForce RTX 3090) - 8580 MiB free
Segmentation fault

Problem description & steps to reproduce

just run it

First Bad Commit

No response

Relevant log output

[152970.213392] llama-parallel[398992]: segfault at 8 ip 00007f8fb6e5bde0 sp 00007ffd3eda32e8 error 4 in libllama.so[7f8fb6e2c000+103000] likely on CPU 10 (core 20, socket 0)
[152970.213399] Code: 78 03 00 00 48 8b 83 88 03 00 00 eb c5 90 66 66 2e 0f 1f 84 00 00 00 00 00 90 48 8b 07 c3 66 66 2e 0f 1f 84 00 00 00 00 00 90 <8b> 47 08 c3 66 66 2e 0f 1f 84 00 00 00 00 00 90 8b 47 08 31 d2 f7
[153216.665971] audit: type=1400 audit(1745908345.234:549): apparmor="STATUS" operation="profile_replace" info="same as current profile, skipping" profile="unconfined" name="libreoffice-oosplash" pid=433604 comm="apparmor_parser"
[153216.682449] audit: type=1400 audit(1745908345.250:550): apparmor="STATUS" operation="profile_replace" info="same as current profile, skipping" profile="unconfined" name="libreoffice-senddoc" pid=433607 comm="apparmor_parser"
[153219.057131] audit: type=1400 audit(1745908347.626:551): apparmor="STATUS" operation="profile_replace" profile="unconfined" name="libreoffice-soffice" pid=433610 comm="apparmor_parser"
[153219.072592] audit: type=1400 audit(1745908347.642:552): apparmor="STATUS" operation="profile_replace" profile="unconfined" name="libreoffice-soffice//gpg" pid=433610 comm="apparmor_parser"
[153219.092082] audit: type=1400 audit(1745908347.662:553): apparmor="STATUS" operation="profile_replace" info="same as current profile, skipping" profile="unconfined" name="libreoffice-xpdfimport" pid=433613 comm="apparmor_parser"
[153715.905905] llama-parallel[435794]: segfault at 8 ip 00007f52356bede0 sp 00007ffd70522758 error 4 in libllama.so[7f523568f000+103000] likely on CPU 10 (core 20, socket 0)
[153715.905912] Code: 78 03 00 00 48 8b 83 88 03 00 00 eb c5 90 66 66 2e 0f 1f 84 00 00 00 00 00 90 48 8b 07 c3 66 66 2e 0f 1f 84 00 00 00 00 00 90 <8b> 47 08 c3 66 66 2e 0f 1f 84 00 00 00 00 00 90 8b 47 08 31 d2 f7