This commit is fairly old, but this appears to cause a segfault for older versions of ldconfig: ``` $ docker run --rm --platform linux/arm64/v8 -ti docker.io/debian:bullseye-slim ldconfig qemu: uncaught target signal 11 (Segmentation fault) - core dumped Segmentation fault (core dumped) ```
The segfault happens inside ldconfig code (code_gen_buffer in qemu's backtrace), so I'm not sure how to debug that further, but it doesn't reproduce in bookworm's ldconfig so that is something that was "fixed" in glibc at some point. If someone needs to run older debian releases with a newer qemu that might be a problem in the future? [we might need to run old containers once every few years to rebuild old projects in a similar environment they were built on, so would eventually need to work around this problem somehow] The failure can be reproduced just running `qemu-aarch64 ./path/to/ldconfig` on an extracted container so it was easy to bisect and I've got down to this commit; hence replying here directly with involved people. ------ commit aec338d63bc28f1f13d5e64c561d7f1dd0e4b07e Author: Richard Henderson <[email protected]> Date: Wed Aug 16 10:32:18 2023 -0700 linux-user: Adjust brk for load_bias PIE executables are usually linked at offset 0 and are relocated somewhere during load. The hiaddr needs to be adjusted to keep the brk next to the executable. Cc: [email protected] Fixes: 1f356e8c013 ("linux-user: Adjust initial brk when interpreter is close to executable") Tested-by: Helge Deller <[email protected]> Reviewed-by: Ilya Leoshkevich <[email protected]> Reviewed-by: Philippe Mathieu-Daudé <[email protected]> Signed-off-by: Richard Henderson <[email protected]> ------ I've done my share of debugging linux-user last week[1] so I'll leave this as is for now, I've downgraded to (a non-static-pie build of) 7.1 for our build machine and am not in immediate trouble. [1] https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1053101 If this doesn't get much interest I might try to pick at it further in a couple of weeks, assuming it's something we can/want to fix on qemu side. Thanks, -- Dominique
