Le 24/04/2020 à 22:48, Helge Deller a écrit :
> Drop the open-coded fcntl flags conversion in the eventfd2 syscall and
> replace it with the built-in conversion with fcntl_flags_tbl.
>
> Signed-off-by: Helge Deller <del...@gmx.de>
>
> diff --git a/linux-user/syscall.c b/linux-user/syscall.c
> index 05f03919ff..ebf0d38321 100644
> --- a/linux-user/syscall.c
> +++ b/linux-user/syscall.c
> @@ -11938,13 +11942,7 @@ static abi_long do_syscall1(void *cpu_env, int num,
> abi_long arg1,
> #if defined(TARGET_NR_eventfd2)
> case TARGET_NR_eventfd2:
> {
> - int host_flags = arg2 & (~(TARGET_O_NONBLOCK | TARGET_O_CLOEXEC));
> - if (arg2 & TARGET_O_NONBLOCK) {
> - host_flags |= O_NONBLOCK;
> - }
> - if (arg2 & TARGET_O_CLOEXEC) {
> - host_flags |= O_CLOEXEC;
> - }
> + int host_flags = target_to_host_bitmask(arg2, fcntl_flags_tbl);
> ret = get_errno(eventfd(arg1, host_flags));
> if (ret >= 0) {
> fd_trans_register(ret, &target_eventfd_trans);
>
The problem here is eventfd2 doesn't take O_ flags but EFD_ flags.
Most EFD_ flags are mapped to O_ flags, but one is not:
include/linux/eventfd.h:
/*
* CAREFUL: Check include/uapi/asm-generic/fcntl.h when defining
* new flags, since they might collide with O_* ones. We want
* to re-use O_* flags that couldn't possibly have a meaning
* from eventfd, in order to leave a free define-space for
* shared O_* flags.
*/
#define EFD_SEMAPHORE (1 << 0)
#define EFD_CLOEXEC O_CLOEXEC
#define EFD_NONBLOCK O_NONBLOCK
So I think it's better to convert them manually
Perhaps we can defined TARGET_EFD_ flags to make this clearer, I don't know.
Thanks,
Laurent
