Re: [PATCH v3 bpf-next 2/2] bpf: add tests for direct packet access from CGROUP_SKB

[Song Liu <songliubraving@fb.com>]

> On Oct 17, 2018, at 11:25 PM, Alexei Starovoitov <alexei.starovoitov@gmail.com> wrote:
> 
> On Wed, Oct 17, 2018 at 10:39:49PM -0700, Song Liu wrote:
>> Tests are added to make sure CGROUP_SKB cannot access:
>>  tc_classid, data_meta, flow_keys
>> 
>> and can read and write:
>>  mark, prority, and cb[0-4]
>> 
>> and can read other fields.
>> 
>> To make selftest with skb->sk work, a dummy sk is added in
>> bpf_prog_test_run_skb().
>> 
>> Signed-off-by: Song Liu <songliubraving@fb.com>
>> ---
>> net/bpf/test_run.c                          |   4 +
>> tools/testing/selftests/bpf/test_verifier.c | 170 ++++++++++++++++++++
>> 2 files changed, 174 insertions(+)
>> 
>> diff --git a/net/bpf/test_run.c b/net/bpf/test_run.c
>> index 0c423b8cd75c..c7210e2f1ae9 100644
>> --- a/net/bpf/test_run.c
>> +++ b/net/bpf/test_run.c
>> @@ -10,6 +10,7 @@
>> #include <linux/etherdevice.h>
>> #include <linux/filter.h>
>> #include <linux/sched/signal.h>
>> +#include <net/sock.h>
>> 
>> static __always_inline u32 bpf_test_run_one(struct bpf_prog *prog, void *ctx,
>> 		struct bpf_cgroup_storage *storage[MAX_BPF_CGROUP_STORAGE_TYPE])
>> @@ -115,6 +116,7 @@ int bpf_prog_test_run_skb(struct bpf_prog *prog, const union bpf_attr *kattr,
>> 	u32 retval, duration;
>> 	int hh_len = ETH_HLEN;
>> 	struct sk_buff *skb;
>> +	struct sock sk;
>> 	void *data;
>> 	int ret;
>> 
>> @@ -142,6 +144,8 @@ int bpf_prog_test_run_skb(struct bpf_prog *prog, const union bpf_attr *kattr,
>> 		kfree(data);
>> 		return -ENOMEM;
>> 	}
>> +	sock_init_data(NULL, &sk);
>> +	skb->sk = &sk;
> 
> I was about to apply it, but it crashes as:
> [   16.830822] BUG: unable to handle kernel paging request at 000000014427b974
> [   16.831363] PGD 8000000135ecf067 P4D 8000000135ecf067 PUD 0
> [   16.831792] Oops: 0000 [#1] SMP PTI
> [   16.832061] CPU: 1 PID: 1965 Comm: test_verifier Not tainted 4.19.0-rc7-02550-ga76dee97ff12 #1153
> [   16.832712] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.11.0-2.el7 04/01/2014
> [   16.833358] RIP: 0010:cmp_map_id+0x10/0x50
> [   16.835036] RSP: 0018:ffffc9000080faf8 EFLAGS: 00010246
> [   16.835429] RAX: 00000000ffffffff RBX: 0000000036069ee8 RCX: 0000000000000000
> [   16.835958] RDX: 000000014427b970 RSI: 000000014427b970 RDI: ffffc9000080fb44
> [   16.836496] RBP: 000000000000000c R08: ffffffff810f7330 R09: 0000000036069ee8
> [   16.837026] R10: 0000000000000000 R11: 0000000000000001 R12: 0000000000000000
> [   16.837554] R13: ffffffff810f7330 R14: 000000014427b970 R15: 000000001b034f74
> [   16.838083] FS:  00007fae50663700(0000) GS:ffff88013ba80000(0000) knlGS:0000000000000000
> [   16.838677] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> [   16.839105] CR2: 000000014427b974 CR3: 0000000135934005 CR4: 00000000003606e0
> [   16.839632] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
> [   16.840157] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
> [   16.840682] Call Trace:
> [   16.840897]  bsearch+0x50/0x90
> [   16.841144]  map_id_range_down+0x81/0xa0
> [   16.841438]  make_kuid+0xf/0x10
> [   16.841677]  sock_init_data+0x24f/0x260
> [   16.841979]  bpf_prog_test_run_skb+0x9e/0x270
> 
> I suspect sock_net_set(sk, &init_net) is necessary before sock_init_data() call.

I am not able to repro this, even with CONFIG_KASAN and CONFIG_PAGE_POISONING. 

Let me try a better approach on this.

Thanks,
Song