Since commit 5ea2fc8 ("linux-user: Sanity check clone flags"), trying to run fork() fails with old distro on some architectures.
This is the case with HP-PA and Debian 5 (Lenny). It fails on: if ((flags & CSIGNAL) != TARGET_SIGCHLD) { return -TARGET_EINVAL; } because flags is 17, whereas on HP-PA, SIGCHLD is 18. 17 is the SIGCHLD value of my host (x86_64). It appears that for TARGET_NR_fork and TARGET_NR_vfork, QEMU calls do_fork() with SIGCHLD instead of TARGET_SIGCHLD. Signed-off-by: Laurent Vivier <laur...@vivier.eu> --- linux-user/syscall.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/linux-user/syscall.c b/linux-user/syscall.c index f569f82..4d85355 100644 --- a/linux-user/syscall.c +++ b/linux-user/syscall.c @@ -7680,7 +7680,7 @@ abi_long do_syscall(void *cpu_env, int num, abi_long arg1, break; #ifdef TARGET_NR_fork case TARGET_NR_fork: - ret = get_errno(do_fork(cpu_env, SIGCHLD, 0, 0, 0, 0)); + ret = get_errno(do_fork(cpu_env, TARGET_SIGCHLD, 0, 0, 0, 0)); break; #endif #ifdef TARGET_NR_waitpid @@ -10490,7 +10490,8 @@ abi_long do_syscall(void *cpu_env, int num, abi_long arg1, #endif #ifdef TARGET_NR_vfork case TARGET_NR_vfork: - ret = get_errno(do_fork(cpu_env, CLONE_VFORK | CLONE_VM | SIGCHLD, + ret = get_errno(do_fork(cpu_env, + CLONE_VFORK | CLONE_VM | TARGET_SIGCHLD, 0, 0, 0, 0)); break; #endif -- 2.9.3