Re: [PATCH] bpf: fix null ptr deref in dev_map_enqueue

["Toke Høiland-Jørgensen" <toke@kernel.org>]

Jesper Dangaard Brouer <hawk@kernel.org> writes:

> On 31/03/2024 11.08, Edward Adam Davis wrote:
>> [Syzbot reported]
>> general protection fault, probably for non-canonical address 0xdffffc0000000000: 0000 [#1] PREEMPT SMP KASAN PTI
>> KASAN: null-ptr-deref in range [0x0000000000000000-0x0000000000000007]
>> CPU: 0 PID: 5179 Comm: syz-executor120 Not tainted 6.8.0-syzkaller-05271-gf99c5f563c17 #0
>> Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 02/29/2024
>> RIP: 0010:dev_map_enqueue+0x31/0x3e0 kernel/bpf/devmap.c:539
>> Code: 41 56 41 55 41 54 53 48 83 ec 18 49 89 d4 49 89 f5 48 89 fd 49 be 00 00 00 00 00 fc ff df e8 e6 45 d8 ff 48 89 e8 48 c1 e8 03 <42> 80 3c 30 00 74 08 48 89 ef e8 d0 8b 3b 00 4c 8b 7d 00 48 83 c5
>> RSP: 0018:ffffc90003b0f688 EFLAGS: 00010246
>> RAX: 0000000000000000 RBX: 0000000000000000 RCX: ffff888025258000
>> RDX: 0000000000000000 RSI: ffff888024035070 RDI: 0000000000000000
>> RBP: 0000000000000000 R08: 0000000000000005 R09: ffffffff894ff55e
>> R10: 0000000000000004 R11: ffff888025258000 R12: ffff8880157d8000
>> R13: ffff888024035070 R14: dffffc0000000000 R15: ffff8880b943c088
>> FS:  00007fd0098e46c0(0000) GS:ffff8880b9400000(0000) knlGS:0000000000000000
>> CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
>> CR2: 00000000200009c0 CR3: 0000000025314000 CR4: 00000000003506f0
>> DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
>> DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
>> Call Trace:
>>   <TASK>
>>   __xdp_do_redirect_frame net/core/filter.c:4384 [inline]
>>   xdp_do_redirect_frame+0x20d/0x4d0 net/core/filter.c:4438
>>   xdp_test_run_batch net/bpf/test_run.c:336 [inline]
>>   bpf_test_run_xdp_live+0xe8a/0x1e90 net/bpf/test_run.c:384
>>   bpf_prog_test_run_xdp+0x813/0x11b0 net/bpf/test_run.c:1267
>>   bpf_prog_test_run+0x33a/0x3b0 kernel/bpf/syscall.c:4240
>>   __sys_bpf+0x48d/0x810 kernel/bpf/syscall.c:5649
>>   __do_sys_bpf kernel/bpf/syscall.c:5738 [inline]
>>   __se_sys_bpf kernel/bpf/syscall.c:5736 [inline]
>>   __x64_sys_bpf+0x7c/0x90 kernel/bpf/syscall.c:5736
>>   do_syscall_64+0xfb/0x240
>>   entry_SYSCALL_64_after_hwframe+0x6d/0x75
>> RIP: 0033:0x7fd00992a0d9
>> Code: 28 00 00 00 75 05 48 83 c4 28 c3 e8 81 18 00 00 90 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 b0 ff ff ff f7 d8 64 89 01 48
>> RSP: 002b:00007fd0098e4238 EFLAGS: 00000246 ORIG_RAX: 0000000000000141
>> RAX: ffffffffffffffda RBX: 00007fd0099b43e8 RCX: 00007fd00992a0d9
>> RDX: 0000000000000050 RSI: 0000000020000240 RDI: 000000000000000a
>> RBP: 00007fd0099b43e0 R08: 00007fd0098e46c0 R09: 00007fd0098e46c0
>> R10: 00007fd0098e46c0 R11: 0000000000000246 R12: 00007fd009981060
>> R13: 0000000000000016 R14: 00007fffcb70c160 R15: 00007fffcb70c248
>>   </TASK>
>> [Fix]
>> On the execution path of bpf_prog_test_run(), due to ri->map being NULL,
>> ri->tgtvalue was not set correctly.
>> 
>> Reported-and-tested-by: syzbot+af9492708df9797198d6@syzkaller.appspotmail.com
>> Signed-off-by: Edward Adam Davis <eadavis@qq.com>
>> ---
>>   kernel/bpf/devmap.c | 6 +++++-
>>   1 file changed, 5 insertions(+), 1 deletion(-)
>> 
>> diff --git a/kernel/bpf/devmap.c b/kernel/bpf/devmap.c
>> index 4e2cdbb5629f..ef20de14154a 100644
>> --- a/kernel/bpf/devmap.c
>> +++ b/kernel/bpf/devmap.c
>> @@ -86,6 +86,7 @@ struct bpf_dtab {
>>   static DEFINE_PER_CPU(struct list_head, dev_flush_list);
>>   static DEFINE_SPINLOCK(dev_map_lock);
>>   static LIST_HEAD(dev_map_list);
>> +static bool is_valid_dst(struct bpf_dtab_netdev *obj, struct xdp_frame *xdpf);
>>   
>>   static struct hlist_head *dev_map_create_hash(unsigned int entries,
>>   					      int numa_node)
>> @@ -536,7 +537,10 @@ int dev_xdp_enqueue(struct net_device *dev, struct xdp_frame *xdpf,
>>   int dev_map_enqueue(struct bpf_dtab_netdev *dst, struct xdp_frame *xdpf,
>>   		    struct net_device *dev_rx)
>>   {
>> -	struct net_device *dev = dst->dev;
>> +	struct net_device *dev;
>> +	if (!is_valid_dst(dst, xdpf))
>
> This is overkill, because __xdp_enqueue() already contains most of the
> checks in is_valid_dst().
>
> Why not:
>
>   if (!dst)
> 	return -EINVAL;
>
>
>> +		return -EINVAL;
>> +	dev = dst->dev;
>>   
>>   	return __xdp_enqueue(dev, xdpf, dev_rx, dst->xdp_prog);
>>   }
>
>
> Is this fix pampering over another issue?
>
> To repeat myself:
> I think something is wrong in xdp_test_run_batch().

I did spot a bug in the test_run code related to XDP_TX, but I don't
really see how that particular issue could trigger this bug, since
AFAICT the reproducer doesn't return XDP_TX. It looks like the program
is setting the BROADCAST flag, but I don't see how that can lead to
dev_map_enqueue() being run.

> The `ri->tgt_value` is being set in __bpf_xdp_redirect_map(), but I
> cannot see __bpf_xdp_redirect_map() being used in xdp_test_run_batch().

__bpf_xdp_redirect_map() is always being called from inside the BPF
programs (through the bpf_redirect_map()). I don't really think we have
a way of ensuring that we always follow such a call with an XDP_REDIRECT
return, so we're kinda relying on the XDP program to do the right thing,
or we could risk stale data being left in bpf_redirect_info, no? I am
not sure if that is what's going on here, though.

> Is this a case of XDP program returning XDP_REDIRECT without having
> called the BPF helper for redirect?

From trying to run the reproducer, I don't *think* that is the case, but
cf the above I don't think that is the case in this instance.

Anyway, let's see if I can get syzbot to test the fix to
xdp_test_run_batch():

#syz test

diff --git a/net/bpf/test_run.c b/net/bpf/test_run.c
index dfd919374017..a3f24486829e 100644
--- a/net/bpf/test_run.c
+++ b/net/bpf/test_run.c
@@ -281,9 +281,9 @@ static int xdp_recv_frames(struct xdp_frame **frames, int nframes,
 static int xdp_test_run_batch(struct xdp_test_data *xdp, struct bpf_prog *prog,
                              u32 repeat)
 {
-       struct bpf_redirect_info *ri = this_cpu_ptr(&bpf_redirect_info);
        int err = 0, act, ret, i, nframes = 0, batch_sz;
        struct xdp_frame **frames = xdp->frames;
+       struct bpf_redirect_info *ri;
        struct xdp_page_head *head;
        struct xdp_frame *frm;
        bool redirect = false;
@@ -294,6 +294,7 @@ static int xdp_test_run_batch(struct xdp_test_data *xdp, struct bpf_prog *prog,
 
        local_bh_disable();
        xdp_set_return_frame_no_direct();
+       ri = this_cpu_ptr(&bpf_redirect_info);
 
        for (i = 0; i < batch_sz; i++) {
                page = page_pool_dev_alloc_pages(xdp->pp);