linux-kernel - Re: [PATCH v2] usb: gadget: ncm: Handle decoding of multiple NTB's in unwrap call

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]

Message-ID: <a2b9e73a-e4eb-4c21-9d85-bf643cb353ca@quicinc.com>
Date:   Tue, 26 Sep 2023 12:00:35 +0530
From:   Krishna Kurapati PSSNV <quic_kriskura@...cinc.com>
To:     Greg Kroah-Hartman <gregkh@...uxfoundation.org>
CC:     Linyu Yuan <quic_linyyuan@...cinc.com>,
        Maciej Żenczykowski <maze@...gle.com>,
        <linux-usb@...r.kernel.org>, <linux-kernel@...r.kernel.org>,
        <quic_ppratap@...cinc.com>, <quic_wcheng@...cinc.com>,
        <quic_jackp@...cinc.com>, <stable@...r.kernel.org>
Subject: Re: [PATCH v2] usb: gadget: ncm: Handle decoding of multiple NTB's in
 unwrap call



On 9/18/2023 1:07 PM, Krishna Kurapati PSSNV wrote:
> 
> 
> On 9/17/2023 1:34 PM, Greg Kroah-Hartman wrote:
>>> Cc: stable@...r.kernel.org
>>
>> What commit id does this fix?
>>
> 
> Hi Greg,
> 
> This fixes the initial patch that added the driver:
> 9f6ce4240a2bf456402c15c06768059e5973f28c
> 
>>> Reviewed-by: Maciej Żenczykowski <maze@...gle.com>
>>> Signed-off-by: Krishna Kurapati <quic_kriskura@...cinc.com>
>>> ---
>>>   drivers/usb/gadget/function/f_ncm.c | 26 +++++++++++++++++++-------
>>>   1 file changed, 19 insertions(+), 7 deletions(-)
>>>
>>> diff --git a/drivers/usb/gadget/function/f_ncm.c 
>>> b/drivers/usb/gadget/function/f_ncm.c
>>> index feccf4c8cc4f..f00f051438ec 100644
>>> --- a/drivers/usb/gadget/function/f_ncm.c
>>> +++ b/drivers/usb/gadget/function/f_ncm.c
>>> @@ -1156,7 +1156,8 @@ static int ncm_unwrap_ntb(struct gether *port,
>>>                 struct sk_buff_head *list)
>>>   {
>>>       struct f_ncm    *ncm = func_to_ncm(&port->func);
>>> -    __le16        *tmp = (void *) skb->data;
>>> +    unsigned char    *ntb_ptr = (void *) skb->data;
>>
>> Why persist with the extra ' ', didn't checkpatch complain about this?
>>
>> And why the cast at all?
>>
> My bad. I ran the checkpatch and got the following result:
> 
> kriskura@...kriskura-hyd:/local/mnt/workspace/krishna/510/testncm/kernel$ ./scripts/checkpatch.pl --strict 0001-usb-gadget-ncm-Handle-decoding-of-multiple-NTB-s-in-.patch
> WARNING: Prefer a maximum 75 chars per line (possible unwrapped commit 
> description?)
> #12:
> unwraps the obtained request data assuming only one NTB is present, we 
> loose
> 
> CHECK: No space is necessary after a cast
> #34: FILE: drivers/usb/gadget/function/f_ncm.c:1159:
> +       unsigned char   *ntb_ptr = (void *) skb->data;
> 
> CHECK: No space is necessary after a cast
> #46: FILE: drivers/usb/gadget/function/f_ncm.c:1176:
> +       tmp = (void *) ntb_ptr;
> 
> CHECK: No space is necessary after a cast
> #93: FILE: drivers/usb/gadget/function/f_ncm.c:1329:
> +               ntb_ptr = (unsigned char *) (ntb_ptr + block_len);
> 
> total: 0 errors, 1 warnings, 3 checks, 67 lines checked
> 
> 
> I ignored the checks and saw only that errors are 0. Seems like I missed 
> fixing the commit text wrapping to 75 chars (On line 12 it has 76 
> chars). Will fix it up in v3.
> 
> As per the cast, I initially didn't add any cast and saw that the code 
> was not able to parse the dwSignature of the NTH and decoding of all 
> packets was failing. Only when I added the cast, was the function able 
> to decode all packets properly.
> 
>>> +    __le16        *tmp;
>>>       unsigned    index, index2;
>>>       int        ndp_index;
>>>       unsigned    dg_len, dg_len2;
>>> @@ -1169,6 +1170,10 @@ static int ncm_unwrap_ntb(struct gether *port,
>>>       const struct ndp_parser_opts *opts = ncm->parser_opts;
>>>       unsigned    crc_len = ncm->is_crc ? sizeof(uint32_t) : 0;
>>>       int        dgram_counter;
>>> +    int        to_process = skb->len;
>>> +
>>> +parse_ntb:
>>> +    tmp = (void *) ntb_ptr;
>>
>> Again, no blank space please.
>>
>> And why the cast?
>>
> the second cast here was just to be in sync with the original code;
> __le16        *tmp = (void *) skb->data;
> 
> I didn't try removing this and running the test. Will check if the 
> second one is required or if decoding is proper without it or not.
> 
> Regards,
> Krishna,

Hi Greg,

  I rechecked the code and I don't see any error if I remove the 
typecast for ntb_ptr. If I remove the typecast for tmp, I see build 
errors as ntb_ptr is unsigned char and tmp is __le16. I have pushed v3 
removing the typecast for ntb_ptr and fixing the checkpatch errors. 
Thanks for pointing out that the typecast is not needed.

Regards,
Krishna,