On Sun, Aug 14 2022, Vitaliy Makkoveev <[email protected]> wrote:
> I propose to avoid netlock relocking for SIOCGIFSFFPAGE too.
sigh, that seems to be needed indeed.
> Also, since the relock should be avoided for many commands, I prefer
> to have something like below:
>
> switch(cmd){
> case SIOCGIFMEDIA:
> case ...:
> relock = 1;
> break;
> }
>
> if (relock)
> NET_UNLOCK();
Fine with me. "netlock_held" made more sense to me as the variable
name, I hope you won't mind. ;)
Works fine on GENERIC.MP.
Index: dev/fdt/if_cad.c
===================================================================
RCS file: /cvs/src/sys/dev/fdt/if_cad.c,v
retrieving revision 1.11
diff -u -p -r1.11 if_cad.c
--- dev/fdt/if_cad.c 8 Mar 2022 16:13:08 -0000 1.11
+++ dev/fdt/if_cad.c 14 Aug 2022 20:20:00 -0000
@@ -550,12 +550,22 @@ cad_ioctl(struct ifnet *ifp, u_long cmd,
{
struct cad_softc *sc = ifp->if_softc;
struct ifreq *ifr = (struct ifreq *)data;
- int error = 0;
+ int error = 0, netlock_held = 1;
int s;
- NET_UNLOCK();
+ switch (cmd) {
+ case SIOCGIFMEDIA:
+ case SIOCSIFMEDIA:
+ case SIOCGIFSFFPAGE:
+ netlock_held = 0;
+ break;
+ }
+
+ if (netlock_held)
+ NET_UNLOCK();
rw_enter_write(&sc->sc_cfg_lock);
- NET_LOCK();
+ if (netlock_held)
+ NET_LOCK();
s = splnet();
switch (cmd) {
--
jca | PGP : 0x1524E7EE / 5135 92C1 AD36 5293 2BDF DDCC 0DFA 74AE 1524 E7EE
