netdev - Re: [PATCH bpf-next] bpf: remove useless variable

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date:   Sat, 15 Dec 2018 02:13:18 +0100
From:   Daniel Borkmann <daniel@...earbox.net>
To:     Martin Lau <kafai@...com>, Andrea Claudi <aclaudi@...hat.com>
Cc:     "netdev@...r.kernel.org" <netdev@...r.kernel.org>,
        "ast@...nel.org" <ast@...nel.org>,
        "sbrivio@...hat.com" <sbrivio@...hat.com>,
        John Fastabend <john.fastabend@...il.com>
Subject: Re: [PATCH bpf-next] bpf: remove useless variable

On 12/14/2018 06:03 PM, Martin Lau wrote:
> On Thu, Dec 13, 2018 at 10:44:57PM +0100, Andrea Claudi wrote:
>> bytes is initialized to end - start at the beginning of this function,
>> and is never changed. Remove it making the code a bit more readable.
>>
>> Suggested-by: Stefano Brivio <sbrivio@...hat.com>
>> Signed-off-by: Andrea Claudi <aclaudi@...hat.com>
> The change looks correct.
> I found the original code more intuitive to read though.
> Daniel/John, thoughts?

Change is correct, but I kind of agree with Martin that it seems to
make it slightly less intuitive to follow the logic. Probably just
personal preference. I think the change is mainly about the first
part since 'msg->data + bytes' is perhaps slightly more obvious than
'msg->data + end - start', so that second chunk was kind of a necessity
due to getting rid of bytes var. The first part is on finding the
right scatterlist ring entry based on the user provided data /start/.
And the /offset/ is the accumulation of the sg entries in the skmsg
we needed to walk to get to this point. So the /start - offset/ is
the headroom of the last skmsg entry. This and /bytes/ is needed for
checking whether it's single or multi skmsg element. 'end - offset'
feels a bit less intuitive to follow here given the previous logic.

Thanks,
Daniel

>> ---
>>  net/core/filter.c | 6 +++---
>>  1 file changed, 3 insertions(+), 3 deletions(-)
>>
>> diff --git a/net/core/filter.c b/net/core/filter.c
>> index aa274679965d..f31a0de14216 100644
>> --- a/net/core/filter.c
>> +++ b/net/core/filter.c
>> @@ -2170,8 +2170,8 @@ static const struct bpf_func_proto bpf_msg_cork_bytes_proto = {
>>  BPF_CALL_4(bpf_msg_pull_data, struct sk_msg *, msg, u32, start,
>>  	   u32, end, u64, flags)
>>  {
>> -	u32 len = 0, offset = 0, copy = 0, poffset = 0, bytes = end - start;
>>  	u32 first_sge, last_sge, i, shift, bytes_sg_total;
>> +	u32 len = 0, offset = 0, copy = 0, poffset = 0;
>>  	struct scatterlist *sge;
>>  	u8 *raw, *to, *from;
>>  	struct page *page;
>> @@ -2196,7 +2196,7 @@ BPF_CALL_4(bpf_msg_pull_data, struct sk_msg *, msg, u32, start,
>>  	/* The start may point into the sg element so we need to also
>>  	 * account for the headroom.
>>  	 */
>> -	bytes_sg_total = start - offset + bytes;
>> +	bytes_sg_total = end - offset;
>>  	if (!msg->sg.copy[i] && bytes_sg_total <= len)
>>  		goto out;
>>  
>> @@ -2279,7 +2279,7 @@ BPF_CALL_4(bpf_msg_pull_data, struct sk_msg *, msg, u32, start,
>>  		      msg->sg.end - shift;
>>  out:
>>  	msg->data = sg_virt(&msg->sg.data[first_sge]) + start - offset;
>> -	msg->data_end = msg->data + bytes;
>> +	msg->data_end = msg->data + end - start;
>>  	return 0;
>>  }
>>  
>> -- 
>> 2.19.2