Hello,

While at it, we'd better also fix sysdeps/mach/hurd/dup3.c the exact
same way (which will also fix dup2).

Thanks,
Samuel

Zhaoming Luo, le jeu. 06 mars 2025 06:39:04 +0800, a ecrit:
> Ignoring the return value of mach_port_mod_ref() causes the situation

> +         if (err)
> +           {
> +             /* When an error occurs during giving a user ref to the
> +                io server port */
> +             result = -1;
> +
> +             if (err == KERN_UREFS_OVERFLOW)
> +               errno = EMFILE;
> +             else
> +               errno = EINVAL;

The idiomatic way is to use __hurd_fail:
                if (err == KERN_UREFS_OVERFLOW)
                  result = __hurd_failed (EMFILE);
                else
                  result = __hurd_failed (EINVAL);

> +           }
> +         else
> +           {
> +             if (ctty != MACH_PORT_NULL)
> +               __mach_port_mod_refs (__mach_task_self (), ctty,
> +                       MACH_PORT_RIGHT_SEND, 1);

This one could fail too, we want to fail the whole thing in that case
(and __mach_port_mod_refs(port, -1) to clean the ref successfully
acquired above).

> +
> +             /* Install the ports and flags in the new descriptor.  */
> +             if (ctty != MACH_PORT_NULL)
> +               _hurd_port_set (&new->ctty, ctty);
> +             new->flags = flags;
> +             _hurd_port_locked_set (&new->port, port); /* Unlocks NEW.  */
> +           }
>         }
>  
>       HURD_CRITICAL_END;

Reply via email to