linux-kernel - Re: nct6775: UBSAN: shift exponent 32 is too large for 32-bit type 'long unsigned int'

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: <20180612221218.GA15406@roeck-us.net>
Date:   Tue, 12 Jun 2018 15:12:18 -0700
From:   Guenter Roeck <linux@...ck-us.net>
To:     Paul Menzel <pmenzel+linux-hwmon@...gen.mpg.de>
Cc:     Jean Delvare <jdelvare@...e.com>, linux-hwmon@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: nct6775: UBSAN: shift exponent 32 is too large for 32-bit type
 'long unsigned int'

Hi Paul,

On Tue, Jun 12, 2018 at 11:50:45PM +0200, Paul Menzel wrote:
> Dear Guenther, dear Jean,
> 
> 
> Building Linux with the undefined behavior sanitizer (UBSAN), loading the
> module *nct6775* on the ASRock E350M1 shows the error below.
> 
> ```
> [   28.322775] UBSAN: Undefined behaviour in drivers/hwmon/nct6775.c:4179:27
> [   28.322833] shift exponent 32 is too large for 32-bit type 'long unsigned
> int'
> ```
> 
> I believe this was introduced by commit cc66b3038254 (hwmon: (nct6775)
> Rework temperature source and label handling), but the same issue was
> already present before.
> 

I don't think so. The code used to be
	for (i = 0; i < data->temp_label_num - 1; i++) {
Notice the "- 1". So it should be
	for (i = 0; i < 31; i++) {

In practice it doesn't matter because "data->temp_mask & BIT(31 + 1)"
will always be false, causing the last loop iteration to abort, but
it is nevertheless wrong.

Thanks a lot for the report. I'll prepare a patch.

Guenter

> ```
> -       for (i = 0; i < data->temp_label_num - 1; i++) {
> +       for (i = 0; i < 32; i++) {
> +               if (!(data->temp_mask & BIT(i + 1)))
> +                       continue;
> ```
> 
> `include/linux/bitops.h` has the macros below.
> 
> ```
> #define BIT(nr)                 (1UL << (nr))
> #define BIT_ULL(nr)             (1ULL << (nr))
> ```
> 
> Maybe `BIT_ULL` should be used, but I do not know, if that’d would work with
> `data->temp_mask`, which has type `u32`.
> 
> 
> Kind regards,
> 
> Paul