linux-kernel - Re: [PATCH] futex: Fix a faulty comment.

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: <8a7cc760-b1e2-0da6-f3cc-4e7baa3d956c@collabora.com>
Date:   Mon, 6 Dec 2021 17:34:56 -0300
From:   André Almeida <andrealmeid@...labora.com>
To:     Jangwoong Kim <6812skiii@...il.com>
Cc:     peterz@...radead.org, mingo@...hat.com, tglx@...utronix.de,
        dvhart@...radead.org, dave@...olabs.net,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH] futex: Fix a faulty comment.

Hi Jangwoong,

Please don't top post when replying to emails in this list (read more
here: https://www.mediawiki.org/wiki/Mailing_list_etiquette)

Às 14:44 de 06/12/21, Jangwoong Kim escreveu:
> Hi André.
> 
> That patch was definitely wrong, I apologize.
> 
> However, since futex_wait_multiple_setup() sets the last index of
> futex that was woken up,
> shouldn't the comment be modified as below?
> 
> If so, I will resend a patch.
> 
>                 /*
> -                * Even if something went wrong, if we find out that a futex
> -                * was woken, we don't return error and return this index to
> -                * userspace
> +                * Even if something went wrong, if we find out that any futex
> +                * was woken, we don't return error and return the last index
> +                * awoken to userspace

Indeed, this is more correct. You can send a patch to clarify this comment.

>                  */
>                 *woken = unqueue_multiple(vs, i);
>                 if (*woken >= 0)
> 
> I sent the patch because I thought this was important enough to be corrected.
> 
> Let me know If this is not crucial enough to be patched, so I won't
> keep sending comment-fixing patches.
> 
> Thank you.
> Jangwoong Kim.
> 
> 2021년 12월 6일 (월) 오후 9:12, André Almeida <andrealmeid@...labora.com>님이 작성:
>>
>> Hi Jangwoong,
>>
>> Thanks for your patch! However...
>>
>> Às 15:14 de 04/12/21, 6812skiii@...il.com escreveu:
>>> From: Jangwoong Kim <6812skiii@...il.com>
>>>
>>> We return 1, not the index of futex woken up.
>>>
>>> Signed-off-by: Jangwoong Kim <6812skiii@...il.com>
>>> ---
>>>  kernel/futex/waitwake.c | 4 ++--
>>>  1 file changed, 2 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/kernel/futex/waitwake.c b/kernel/futex/waitwake.c
>>> index 4ce0923f1ce3..d148e5d4956b 100644
>>> --- a/kernel/futex/waitwake.c
>>> +++ b/kernel/futex/waitwake.c
>>> @@ -455,8 +455,8 @@ static int futex_wait_multiple_setup(struct futex_vector *vs, int count, int *wo
>>>
>>>               /*
>>>                * Even if something went wrong, if we find out that a futex
>>> -              * was woken, we don't return error and return this index to
>>> -              * userspace
>>> +              * was woken, we don't return error and make userspace aware
>>> +              * of this by returning 1.
>>
>> We return to userspace the value at *woken, so your fix is wrong. Have a
>> look at futex_wait_multiple():
>>
>>                 ret = futex_wait_multiple_setup(vs, count, &hint);
>>                 if (ret) {
>>                         if (ret > 0) {
>>                                 /* A futex was woken during setup */
>>                                 ret = hint;
>>                         }
>>                         return ret;
>>                 }
>>
>> When we return 1 at futex_wait_multiple_setup(), we end up returning the
>> hint/woken value to userspace.
>>
>> Let me know if you have questions.
>>
>>         André