Hello,
Mikhail Karpov, le mer. 21 oct. 2026 18:02:23 +0700, a ecrit:
> Also, I’d like to know whether the return value of
> mach_port_mod_refs should be checked as well.
Yes, we'd want to check that rather than letting refs miscounted.
> diff --git a/lwip/port/netif/hurdethif.c b/lwip/port/netif/hurdethif.c
> index 0370281..6444662 100644
> --- a/lwip/port/netif/hurdethif.c
> +++ b/lwip/port/netif/hurdethif.c
> @@ -174,8 +174,11 @@ hurdethif_device_open (struct netif *netif)
> else
> {
> ethif->readptname = ports_get_right (ethif->readpt);
> - mach_port_insert_right (mach_task_self (), ethif->readptname,
> - ethif->readptname, MACH_MSG_TYPE_MAKE_SEND);
> + err = mach_port_insert_right (mach_task_self (), ethif->readptname,
> + ethif->readptname,
> + MACH_MSG_TYPE_MAKE_SEND);
> + if (err)
> + error (0, err, "mach_port_insert_right");
We'd also want to avoid doing the rest of the function. Actually the
function could be simplified first in a preparatory patch, with some
return err; rather than "else" statements.
> diff --git a/trans/streamio.c b/trans/streamio.c
> index 694da06..57bbf04 100644
> --- a/trans/streamio.c
> +++ b/trans/streamio.c
> @@ -834,14 +834,13 @@ dev_open (const char *name, dev_mode_t mode)
> err = ports_create_port (phys_reply_class, streamdev_bucket,
> sizeof (struct port_info), &phys_reply_pi);
> if (err)
> - {
> - mach_port_deallocate (mach_task_self (), device_master);
> - return err;
> - }
> + goto device_master_deallocate;
Yes, that makes things simpler.
Samuel
