Re: [PATCH v2] cpufreq: brcmstb-avs-cpufreq: Fix -Warray-bounds bug

From: "Gustavo A. R. Silva" <gustavo@embeddedor.com>
To: Kees Cook <keescook@chromium.org>,
	"Gustavo A. R. Silva" <gustavoars@kernel.org>
Cc: Markus Mayer <mmayer@broadcom.com>,
	Broadcom internal kernel review list
	<bcm-kernel-feedback-list@broadcom.com>,
	"Rafael J. Wysocki" <rafael@kernel.org>,
	Viresh Kumar <viresh.kumar@linaro.org>,
	Florian Fainelli <florian.fainelli@broadcom.com>,
	linux-pm@vger.kernel.org, linux-arm-kernel@lists.infradead.org,
	linux-kernel@vger.kernel.org, linux-hardening@vger.kernel.org
Subject: Re: [PATCH v2] cpufreq: brcmstb-avs-cpufreq: Fix -Warray-bounds bug
Date: Mon, 31 Jul 2023 21:20:17 -0600	[thread overview]
Message-ID: <8d83bd2e-9e66-c5fc-e351-1d57f710296c@embeddedor.com> (raw)
In-Reply-To: <202307311610.B1EB796684@keescook>

On 7/31/23 17:16, Kees Cook wrote:
> On Mon, Jul 31, 2023 at 03:07:20PM -0600, Gustavo A. R. Silva wrote:
>> Update the iteration conditions in the for() loop to avoid writing in
>> array `table` beyond its allocated size at:
>>
>> drivers/cpufreq/brcmstb-avs-cpufreq.c:
>> 449         table[i].frequency = CPUFREQ_TABLE_END;
>>
>> This fixes the following -Warray-bounds warning seen after building
>> ARM with multi_v7_defconfig (GCC 13):
>> In function 'brcm_avs_get_freq_table',
>>      inlined from 'brcm_avs_cpufreq_init' at drivers/cpufreq/brcmstb-avs-cpufreq.c:623:15:
>> drivers/cpufreq/brcmstb-avs-cpufreq.c:449:28: warning: array subscript 5 is outside array bounds of 'void[60]' [-Warray-bounds=]
>>    449 |         table[i].frequency = CPUFREQ_TABLE_END;
> 
> #define AVS_PSTATE_P0           0x0
> #define AVS_PSTATE_P1           0x1
> #define AVS_PSTATE_P2           0x2
> #define AVS_PSTATE_P3           0x3
> #define AVS_PSTATE_P4           0x4
> #define AVS_PSTATE_MAX          AVS_PSTATE_P4
> 
>          table = devm_kcalloc(dev, AVS_PSTATE_MAX + 1, sizeof(*table),
>                               GFP_KERNEL);
> 	...
>          for (i = AVS_PSTATE_P0; i <= AVS_PSTATE_MAX; i++) {
> 		...
> 	}
>          table[i].frequency = CPUFREQ_TABLE_END;
> 
> I see "AVS_PSTATE_MAX + 1" being used for the allocation, and so the
> loop is likely correctly doing P0 through P4. If there is supposed to be
> a terminating element in the table, I think the correct fix would be to
> allocate an additional element, not stop the loop from processing P4.

Yeah; I think you're right. And it seems that this function header makes it
clear, too:

drivers/cpufreq/brcmstb-avs-cpufreq.c:
421 /*
422  * We determine which frequencies are supported by cycling through all P-states
423  * and reading back what frequency we are running at for each P-state.
424  */
425 static struct cpufreq_frequency_table *
426 brcm_avs_get_freq_table(struct device *dev, struct private_data *priv)

I just sent v3:
https://lore.kernel.org/linux-hardening/ZMh45KH2iPIpNktr@work/

Let's see what the maintainers say.

Thanks for the feedback!
--
Gustavo

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel