[PATCH] D124212: [sanitizer] Use canonical syscalls everywhere

[Ulrich Weigand via Phabricator via llvm-commits]

uweigand added a comment.

In D124212#3480335 <https://reviews.llvm.org/D124212#3480335>, @eugenis wrote:

> Thanks. Ulrich, could you help me with the s390 failures? I can reland with a `#define SANITIZER_USES_CANONICAL_LINUX_SYSCALLS !SANITIZER_S390` for now, but it would be great to remove the non-canonical code path completely.

Looks like the problem is in `internal_fork`.   This code path:

  return internal_syscall(SYSCALL(clone), SIGCHLD, 0);

fails on s390x because the low-level `clone` syscall interface is different between architectures.  To quote the Linux manpage:

         The raw system call interface on x86-64 and some other architectures (including sh, tile, and alpha) is roughly:
  
             long clone(unsigned long flags, void *child_stack,
                        int *ptid, int *ctid,
                        unsigned long newtls);
  [... ]
         On the cris and s390 architectures, the order of the first two arguments is reversed:
  
             long clone(void *child_stack, unsigned long flags,
                        int *ptid, int *ctid,
                        unsigned long newtls);

This means that instead of passing NULL as child stack pointer, we're passing SIGCHLD (0x11), and so the child crashes when accessing the stack.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D124212/new/

https://reviews.llvm.org/D124212