bpf: fix map not being uncharged during map creation failure
authorDaniel Borkmann <daniel@iogearbox.net>
Thu, 3 Nov 2016 23:56:31 +0000 (00:56 +0100)
committerDavid S. Miller <davem@davemloft.net>
Mon, 7 Nov 2016 18:22:26 +0000 (13:22 -0500)
commit20b2b24f91f70e7d3f0918c077546cb21bd73a87
treeecdffd693fa3fd2d368def47c1ea510330119265
parent483bed2b0ddd12ec33fc9407e0c6e1088e77a97c
bpf: fix map not being uncharged during map creation failure

In map_create(), we first find and create the map, then once that
suceeded, we charge it to the user's RLIMIT_MEMLOCK, and then fetch
a new anon fd through anon_inode_getfd(). The problem is, once the
latter fails f.e. due to RLIMIT_NOFILE limit, then we only destruct
the map via map->ops->map_free(), but without uncharging the previously
locked memory first. That means that the user_struct allocation is
leaked as well as the accounted RLIMIT_MEMLOCK memory not released.
Make the label names in the fix consistent with bpf_prog_load().

Fixes: aaac3ba95e4c ("bpf: charge user for creation of BPF maps and programs")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
kernel/bpf/syscall.c