linux-kernel - Re: [PATCH] tty: vt: keyboard: do not copy an extra-byte in copy_to

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

Date:   Sun, 7 Nov 2021 01:16:18 +0530
From:   Ajay Garg <ajaygargnsit@...il.com>
To:     Andy Shevchenko <andy.shevchenko@...il.com>,
        Pavel Skripkin <paskripkin@...il.com>,
        Greg KH <gregkh@...uxfoundation.org>, jirislaby@...nel.org,
        kernel@...il.dk, David Laight <David.Laight@...lab.com>
Cc:     "linux-serial@...r.kernel.org" <linux-serial@...r.kernel.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] tty: vt: keyboard: do not copy an extra-byte in copy_to_user

Actually, on further thoughts, even David's solution will require an
extra check, if -E2BIG is returned.

So, I guess the solution suggested by me looks the best
(https://lore.kernel.org/linux-serial/868025b485b94480ad17d0ec971b3ee9@AcuMS.aculab.com/T/#m1c4aaa4347b02fd4c11ce611ff5029fcb71c37a1)
:

1.
== Do not use the return value from strlcpy. ==

                len = strlcpy(kbs, func_table[kb_func] ? : "", len);
=>
                strlcpy(kbs, func_table[kb_func] ? : "", len);


2.
== Calculate the actual length of kbs, add 1, and then copy those many
bytes to user-buffer ==

ret = copy_to_user(user_kdgkb->kb_string, kbs, len + 1) ?
                        -EFAULT : 0;
=>
ret = copy_to_user(user_kdgkb->kb_string, kbs, strlen(kbs) + 1) ?
                        -EFAULT : 0;

On Sun, Nov 7, 2021 at 12:50 AM Ajay Garg <ajaygargnsit@...il.com> wrote:
>
> Thanks Pavel, Andy, David for the help.
>
> Andy,
>
> There is no compilation/runtime blocker.
> There were warnings reported by smatch.
>
> My intention is to make the method "vt_do_kdgkb_ioctl" bullet-proof in
> itself, without depending upon external clients.
>
> Pavel has explained that currently things are fine, as per :
> https://lore.kernel.org/linux-serial/868025b485b94480ad17d0ec971b3ee9@AcuMS.aculab.com/T/#m740fffb7c6ee52fdc98b9ef0b4e32a060b6a3be3
>
> but it seems that there is a big flaw - we are dependent on the length
> of "func_table[kb_func]" being ok. If func_table[kb_func] goes awry,
> the method will cause overflow.
>
> Since func_table[kb_func]" is not managed by the method, so the method
> must not depend on func_table[kb_func]" length-correctness. Instead,
> "vt_do_kdgkb_ioctl" must ensure no overflow, without depending how
> external entities (func_table[kb_func] behave.
>
>
>
> The issue with strlcpy, along with a potential "fix", has been explained in :
> https://lore.kernel.org/linux-serial/868025b485b94480ad17d0ec971b3ee9@AcuMS.aculab.com/T/#m1c4aaa4347b02fd4c11ce611ff5029fcb71c37a1
>
> David has provided a simpler fix (usage of strscpy), as in :
> https://lore.kernel.org/linux-serial/868025b485b94480ad17d0ec971b3ee9@AcuMS.aculab.com/T/#m63dab1137e593f2030920a53272f71866b442f40
>
>
> So, we could go with one of the above changes (mine/David's), or
> nothing at all (since there is no blocker).
>
> I vote for David's strscpy "fix", as it is simple, and does away with
> the dependency on the length of "func_table[kb_func]".
>
>
> Would like to know what the maintainers think.
> If there is a consensus that the method "vt_do_kdgkb_ioctl" be made
> bullet-proof in itself, please let me know, I will float the next
> version of patch.
>
>
> Thanks again Pavel, David, Andy.
>
>
> Thanks and Regards,
> Ajay