Some more review:
net/if_types.h: OK
os/policy.c: OK
os/priv_defs: OK
os/zone.c: There is a comment
6034 * Add an data link name for the zone. Does not check for duplicates.
but the code now checks for duplicates for the same zone.
Makes sense to update the comment.
I assume that there is some user-space logic so that bge3 can be
assigned to the zone for both IPv4 and IPv6 (which is what the old
comments say is the need for allowing duplicates for the same zone.) Or
was that logic just unnecessary even before your changes?
You've made zone_add_datalink and zone_remove_datalink no longer be
static. But they are a bit hard to use from the kernel in general since
they assume a syscall called them (the use set_errno). Do you use them
from elsewhere in the kernel? If so it probably make sense to have them
return an errno just like zone_check_datalink. Otherwise keep them as
static functions.
KM_SLEEP allocation while holding zone_lock isn't a good idea at
6225 idarray = kmem_alloc(sizeof (datalink_id_t) * idcount,
KM_SLEEP);
sys/dlpi.h: OK
sys/netstack.h: OK
sys/policy.h: OK
sys/sockio.h: OK
sys/zone.h: OK (and it doesn't show zone_add/remove_datalink, hence they
can be static?)
*/modstubs.s: OK
usr/src/lib/libc/port/sys/zone.c: OK
usr/src/lib/brand/native/zone/platform.xml: OK
usr/src/lib/brand/native/zone/config.xml: OK
Post integration you need to make sure those two file changes also
appear in the ipkg brand for opensolaris.
usr/src/head/zone.h: OK
usr/src/cmd/zoneadmd/zoneadmd.h: OK
zoneadmd.c: OK
vplat.c: Can we remove the dladm_setzid() from the system as a result of
vplat.c no longer using it?
zoneadm.c: OK
usr/src/cmd/zoneadm/Makefile: OK
Erik
_______________________________________________
networking-discuss mailing list
[email protected]
