Created attachment 238981 [details] `git format-patch` patch file On arm64/aarch64, calling `longjmp(x, 0);` makes `setjmp(x)` return 0, which normally causes an infinite loop, and is against the ISO C standard for setjmp/longjmp. Instead, using a value of 0 should make `setjmp` return 1: > The `longjmp` function cannot cause the `setjmp` macro to return the > value 0; if `val` is 0, the `setjmp` macro returns the value 1. > > _Taken from §7.13.2.1.4 of the C99 spec_ This has already been reported in https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=255320, but the contributing docs weren't clear on what to do if I had a patch for an already existing problem report, so I thought I'd make another problem report so it has `[PATCH]` in the name. My commit is also available on GitHub https://github.com/aloisklink/freebsd-src/tree/fix-longjmp-with-0-val I also have a patch that adds tests for `longjmp(x, 0)` at https://github.com/aloisklink/freebsd-src/commit/007af6a46677b143f9544fd30e30a1b9f1048ae6 However, since there might be a few architectures that suffer from this bug, I'm not 100% sure if this okay to merge. I'll make a new PR for it.
Created attachment 238982 [details] Test case (warning, tests all architectures, not just aarch64) Oh, it looks like you can upload test cases to Bugzilla :) I'll upload the test case here then, and somebody can let me know if I should make another PR for it too. I've got a feeling that a bunch of other architectures have this same bug, so I'm not sure if it should yet be merged. Cherry-pick https://anonhg.netbsd.org/src/rev/1f14ece9dc46 from upstream (NetBSD) source. Original commit title was: > PR/56066: Jessica Clarke: > Add tests for calling {_,}longjmp with a zero value. See [NetBSD PR/56066](https://gnats.netbsd.org/cgi-bin/query-pr-single.pl?number=56066) --- The author of this patch is taken from the NetBSD code repo. However, the original author of this patch in the NetBSD PR is `Jessica Clarke <jrtc27@FreeBSD.org>`. Should we credit them as the commit author instead?
Triage: summary line tags such as [patch] are no longer used.
(In reply to Graham Perrin from comment #2) It may be worth updating https://docs.freebsd.org/en/articles/contributing/. It still recommends putting `[PATCH]` in the PR synopsis. --- I've also found out that RISC-V is also affected by this issue, and since I had a QEMU VM set up for it, I've made a fix for it in https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=268684 Also, is it worth CC-ing in `Jessica Clarke <jrtc27@FreeBSD.org>`? They submitted a patch for a test-case for the same issue in NetBSD, but I'm guessing from the `FreeBSD.org` domain in their email, they might also be interested in seeing this fixed in FreeBSD.
Didn't notice this bug when committing https://cgit.FreeBSD.org/src/commit/?id=9fb118bebced1452a46756a13be0161021b10905, so here's my best commit-hook@FreeBSD.org impression: A commit in branch main references this bug: URL: https://cgit.FreeBSD.org/src/commit/?id=9fb118bebced1452a46756a13be0161021b10905 commit 9fb118bebced1452a46756a13be0161021b10905 Author: Jessica Clarke <jrtc27@FreeBSD.org> AuthorDate: 2023-01-09 18:34:43 +0000 Commit: Jessica Clarke <jrtc27@FreeBSD.org> CommitDate: 2023-01-09 18:34:43 +0000 libc: Fix longjmp/_longjmp(buf, 0) for AArch64 and RISC-V These architectures fail to handle this special case, and will cause the corresponding setjmp/_setjmp to return 0 rather than 1. Fix this and add regression tests (also committed upstream). PR: 268684 Reviewed by: arichardson, jhb MFC after: 1 week Differential Revision: https://reviews.freebsd.org/D29363 contrib/netbsd-tests/lib/libc/setjmp/t_setjmp.c | 50 ++++++++++++++++++++++--- lib/libc/aarch64/gen/_setjmp.S | 3 +- lib/libc/aarch64/gen/setjmp.S | 3 +- lib/libc/riscv/gen/_setjmp.S | 3 ++ lib/libc/riscv/gen/setjmp.S | 3 ++ 5 files changed, 55 insertions(+), 7 deletions(-)