[PATCH bpf v2 6/6] bpf: fix JITed dump for multi-function programs via syscall

[Daniel Borkmann]

On 05/21/2018 09:42 PM, Sandipan Das wrote:
> On 05/18/2018 09:21 PM, Daniel Borkmann wrote:
>> On 05/18/2018 02:50 PM, Sandipan Das wrote:
>>> Currently, for multi-function programs, we cannot get the JITed
>>> instructions using the bpf system call's BPF_OBJ_GET_INFO_BY_FD
>>> command. Because of this, userspace tools such as bpftool fail
>>> to identify a multi-function program as being JITed or not.
>>>
>>> With the JIT enabled and the test program running, this can be
>>> verified as follows:
>>>
>>>   # cat /proc/sys/net/core/bpf_jit_enable
>>>   1
>>>
>>> Before applying this patch:
>>>
>>>   # bpftool prog list
>>>   1: kprobe  name foo  tag b811aab41a39ad3d  gpl
>>>           loaded_at 2018-05-16T11:43:38+0530  uid 0
>>>           xlated 216B  not jited  memlock 65536B
>>>   ...
>>>
>>>   # bpftool prog dump jited id 1
>>>   no instructions returned
>>>
>>> After applying this patch:
>>>
>>>   # bpftool prog list
>>>   1: kprobe  name foo  tag b811aab41a39ad3d  gpl
>>>           loaded_at 2018-05-16T12:13:01+0530  uid 0
>>>           xlated 216B  jited 308B  memlock 65536B
>>>   ...
>>
>> That's really nice! One comment inline below:
>>
>>>   # bpftool prog dump jited id 1
>>>      0:   nop
>>>      4:   nop
>>>      8:   mflr    r0
>>>      c:   std     r0,16(r1)
>>>     10:   stdu    r1,-112(r1)
>>>     14:   std     r31,104(r1)
>>>     18:   addi    r31,r1,48
>>>     1c:   li      r3,10
>>>   ...
>>>
>>> Signed-off-by: Sandipan Das <sandipan at linux.vnet.ibm.com>
>>> ---
>>>  kernel/bpf/syscall.c | 38 ++++++++++++++++++++++++++++++++------
>>>  1 file changed, 32 insertions(+), 6 deletions(-)
>>>
>>> diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
>>> index 54a72fafe57c..2430d159078c 100644
>>> --- a/kernel/bpf/syscall.c
>>> +++ b/kernel/bpf/syscall.c
>>> @@ -1896,7 +1896,7 @@ static int bpf_prog_get_info_by_fd(struct bpf_prog *prog,
>>>  	struct bpf_prog_info info = {};
>>>  	u32 info_len = attr->info.info_len;
>>>  	char __user *uinsns;
>>> -	u32 ulen;
>>> +	u32 ulen, i;
>>>  	int err;
>>>  
>>>  	err = check_uarg_tail_zero(uinfo, sizeof(info), info_len);
>>> @@ -1922,7 +1922,6 @@ static int bpf_prog_get_info_by_fd(struct bpf_prog *prog,
>>>  	ulen = min_t(u32, info.nr_map_ids, ulen);
>>>  	if (ulen) {
>>>  		u32 __user *user_map_ids = u64_to_user_ptr(info.map_ids);
>>> -		u32 i;
>>>  
>>>  		for (i = 0; i < ulen; i++)
>>>  			if (put_user(prog->aux->used_maps[i]->id,
>>> @@ -1970,13 +1969,41 @@ static int bpf_prog_get_info_by_fd(struct bpf_prog *prog,
>>>  	 * for offload.
>>>  	 */
>>>  	ulen = info.jited_prog_len;
>>> -	info.jited_prog_len = prog->jited_len;
>>> +	if (prog->aux->func_cnt) {
>>> +		info.jited_prog_len = 0;
>>> +		for (i = 0; i < prog->aux->func_cnt; i++)
>>> +			info.jited_prog_len += prog->aux->func[i]->jited_len;
>>> +	} else {
>>> +		info.jited_prog_len = prog->jited_len;
>>> +	}
>>> +
>>>  	if (info.jited_prog_len && ulen) {
>>>  		if (bpf_dump_raw_ok()) {
>>>  			uinsns = u64_to_user_ptr(info.jited_prog_insns);
>>>  			ulen = min_t(u32, info.jited_prog_len, ulen);
>>> -			if (copy_to_user(uinsns, prog->bpf_func, ulen))
>>> -				return -EFAULT;
>>> +
>>> +			/* for multi-function programs, copy the JITed
>>> +			 * instructions for all the functions
>>> +			 */
>>> +			if (prog->aux->func_cnt) {
>>> +				u32 len, free;
>>> +				u8 *img;
>>> +
>>> +				free = ulen;
>>> +				for (i = 0; i < prog->aux->func_cnt; i++) {
>>> +					len = prog->aux->func[i]->jited_len;
>>> +					img = (u8 *) prog->aux->func[i]->bpf_func;
>>> +					if (len > free)
>>> +						break;
>>> +					if (copy_to_user(uinsns, img, len))
>>> +						return -EFAULT;
>>> +					uinsns += len;
>>> +					free -= len;
>>
>> Is there any way we can introduce a delimiter between the different
>> images such that they could be more easily correlated with the call
>> from the main (or other sub-)program instead of having one contiguous
>> dump blob?
> 
> Can we have another member in bpf_prog_info that points to a list of the lengths of the
> JITed images for each subprogram? We can use this information to split up the dump.

Seems okay to me.

Thanks,
Daniel