linux-kernel - Re: [PATCH] amdgpu: fix gcc -Wrestrict warning

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]

Message-ID: <4e63dbbc-0aa3-2950-dda1-1e6aa19d7d5d@rasmusvillemoes.dk>
Date:   Wed, 24 Mar 2021 15:14:36 +0100
From:   Rasmus Villemoes <linux@...musvillemoes.dk>
To:     Arnd Bergmann <arnd@...nel.org>
Cc:     Alex Deucher <alexander.deucher@....com>,
        Christian König <christian.koenig@....com>,
        David Airlie <airlied@...ux.ie>,
        Daniel Vetter <daniel@...ll.ch>, Huang Rui <ray.huang@....com>,
        Jinzhou Su <Jinzhou.Su@....com>,
        amd-gfx list <amd-gfx@...ts.freedesktop.org>,
        dri-devel <dri-devel@...ts.freedesktop.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] amdgpu: fix gcc -Wrestrict warning

On 24/03/2021 14.33, Arnd Bergmann wrote:
> On Tue, Mar 23, 2021 at 4:57 PM Rasmus Villemoes
> <linux@...musvillemoes.dk> wrote:
>> On 23/03/2021 14.04, Arnd Bergmann wrote:
>>>                       if (securedisplay_cmd->status == TA_SECUREDISPLAY_STATUS__SUCCESS) {
>>> +                             int pos = 0;
>>>                               memset(i2c_output,  0, sizeof(i2c_output));
>>>                               for (i = 0; i < TA_SECUREDISPLAY_I2C_BUFFER_SIZE; i++)
>>> -                                     sprintf(i2c_output, "%s 0x%X", i2c_output,
>>> +                                     pos += sprintf(i2c_output + pos, " 0x%X",
>>>                                               securedisplay_cmd->securedisplay_out_message.send_roi_crc.i2c_buf[i]);
>>>                               dev_info(adev->dev, "SECUREDISPLAY: I2C buffer out put is :%s\n", i2c_output);
>>
>> Eh, why not get rid of the 256 byte stack allocation and just replace
>> all of this by
>>
>>   dev_info(adev->dev, ""SECUREDISPLAY: I2C buffer out put is: %*ph\n",
>> TA_SECUREDISPLAY_I2C_BUFFER_SIZE,
>> securedisplay_cmd->securedisplay_out_message.send_roi_crc.i2c_buf);
>>
>> That's much less code (both in #LOC and .text), and avoids adding yet
>> another place that will be audited over and over for "hm, yeah, that
>> sprintf() is actually not gonna overflow".
>>
>> Yeah, it'll lose the 0x prefixes for each byte and use lowercase hex chars.
> 
> Ah, I didn't know the kernel's sprintf could do that, that's really nice.

If you're bored, you can "git grep -E -C4 '%[0.]2[xX]'" and find places
that are inside a small loop, many can trivially be converted to %ph,
though often with some small change in formatting. If you're lucky, you
even get to fix real bugs when people pass a "char" to %02x and "know"
that that will produce precisely two bytes of output, so they've sized
their stack buffer accordingly - boom when "char" happens to be signed
and one of the bytes have a value beyond ascii and %02x produces 0xffffffXX.

%ph has a hard-coded upper bound of 64 bytes, I think that's silly
because people instead do these inefficient and very verbose loops
instead, wasting stack, .text and runtime.

Rasmus