Syscall trace, in case it helps: SYSCALL[7333,1](unix:470) fstatat64(FIXME)(fd:4294967294, path:0x10539ca50(/var/folders/_k/v97x1css2x93xldm8cg7cw5w0000gn/T), ub:0x10539b7d8, flag:0x0)[sync] --> Success(0x0:0x0) SYSCALL[7333,1](unix:464) --7333-- WARNING: unhandled amd64-darwin syscall: unix:464 --7333-- You may be able to write your own handler. --7333-- Read the file README_MISSING_SYSCALL_OR_IOCTL. --7333-- Nevertheless we consider this a bug. Please report --7333-- it at http://valgrind.org/support/bug_reports.html. --> [pre-fail] Failure(0x4e) SYSCALL[7333,1](unix:397) sys_write ( 1, 0x103800800, 266 ) --> [async] ... I think this is caused by calling mkstemp(), but haven't attached a debugger to make sure.
Thanks for the report Chris -- that unix syscall is openat_nocancel() on recent macOS kernels. Do you have a simple program that reproduces this bug report for you?
Yes, it was mkstemp. This example: int main(int argc, char** argv) { char buffer[] = "boxbackup.recombinetemp.1.XXXXXX"; if(mkstemp(buffer) < 0) { perror("mkstemp"); } return 0; } Outputs: --7593-- WARNING: unhandled amd64-darwin syscall: unix:464 --7593-- You may be able to write your own handler. --7593-- Read the file README_MISSING_SYSCALL_OR_IOCTL. --7593-- Nevertheless we consider this a bug. Please report --7593-- it at http://valgrind.org/support/bug_reports.html. mkstemp: Function not implemented
Plus a few commits following the main one below commit d79897e0aff3f5f3cf3e77760761ace3bb847d0f Author: Paul Floyd <pjfloyd@wanadoo.fr> Date: Mon Nov 20 12:59:12 2023 +0100 Bug 390269 - unhandled amd64-darwin syscall: unix:464 (openat_nocancel) Copied from https://github.com/LouisBrunner/valgrind-macos.git Regtest to come shortly