Re: [Qemu-devel] [PATCH 07/12] linux-user: Handle NULL argument to sched_{get, set}param

[Peter Maydell]

On 4 August 2014 17:45, Tom Musta <address@hidden> wrote:
> The sched_getparam and sched_setparam system calls take a pointer
> argument to a sched_param structure.  When this pointer is null,
> errno should be set to EINVAL.

Don't we also need this for the other syscall which takes a
sched_param pointer, sched_setscheduler?

(Strictly speaking I think we could get away without the check
in sched_getparam, because the host syscall will return EINVAL
for us, but I'm happy for us to put the explicit check in for
consistency.)

>
> Signed-off-by: Tom Musta <address@hidden>
>
> diff --git a/linux-user/syscall.c b/linux-user/syscall.c
> index f524a39..5f193cd 100644
> --- a/linux-user/syscall.c
> +++ b/linux-user/syscall.c
> @@ -7699,6 +7699,9 @@ abi_long do_syscall(void *cpu_env, int num, abi_long 
> arg1,
>              struct sched_param *target_schp;
>              struct sched_param schp;
>
> +            if (arg2 == 0) {
> +                return -TARGET_EINVAL;
> +            }
>              if (!lock_user_struct(VERIFY_READ, target_schp, arg2, 1))
>                  goto efault;
>              schp.sched_priority = tswap32(target_schp->sched_priority);
> @@ -7710,6 +7713,10 @@ abi_long do_syscall(void *cpu_env, int num, abi_long 
> arg1,
>          {
>              struct sched_param *target_schp;
>              struct sched_param schp;
> +
> +            if (arg2 == 0) {
> +                return -TARGET_EINVAL;
> +            }
>              ret = get_errno(sched_getparam(arg1, &schp))
>              if (!is_error(ret)) {
>                  if (!lock_user_struct(VERIFY_WRITE, target_schp, arg2, 0))
> --
> 1.7.1

thanks
-- PMM