rtnetlink: fix data overflow in rtnl_calcit()
authorDi Zhu <zhudi21@huawei.com>
Wed, 21 Oct 2020 02:00:53 +0000 (10:00 +0800)
committerJakub Kicinski <kuba@kernel.org>
Thu, 22 Oct 2020 01:24:08 +0000 (18:24 -0700)
commitebfe3c5183733f784264450a41646a482f964e5e
treebd86f2a5823a70d137f0aec26d196d9f3fe0dfbf
parent5403caf21648d739bf2b1266c33e34384c313379
rtnetlink: fix data overflow in rtnl_calcit()

"ip addr show" command execute error when we have a physical
network card with a large number of VFs

The return value of if_nlmsg_size() in rtnl_calcit() will exceed
range of u16 data type when any network cards has a larger number of
VFs. rtnl_vfinfo_size() will significant increase needed dump size when
the value of num_vfs is larger.

Eventually we get a wrong value of min_ifinfo_dump_size because of overflow
which decides the memory size needed by netlink dump and netlink_dump()
will return -EMSGSIZE because of not enough memory was allocated.

So fix it by promoting  min_dump_alloc data type to u32 to
avoid whole netlink message size overflow and it's also align
with the data type of struct netlink_callback{}.min_dump_alloc
which is assigned by return value of rtnl_calcit()

Signed-off-by: Di Zhu <zhudi21@huawei.com>
Link: https://lore.kernel.org/r/20201021020053.1401-1-zhudi21@huawei.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
include/linux/netlink.h
net/core/rtnetlink.c