2019-12-28 02:58:04

by chenqiwu

Subject: [PATCH] cpufreq: powernow-k8: avoid use after free issue in cpufreq_notify_transition()

From: chenqiwu <[email protected]>

There is a potential UAF issue in cpufreq_notify_transition() that the
cpufreq of current cpu has been released before using it. So we should
make a judgement and avoid it.

Signed-off-by: chenqiwu <[email protected]>
---
drivers/cpufreq/powernow-k8.c | 5 +++++
1 file changed, 5 insertions(+)

diff --git a/drivers/cpufreq/powernow-k8.c b/drivers/cpufreq/powernow-k8.c
index 2db2f17..7391eb0 100644
--- a/drivers/cpufreq/powernow-k8.c
+++ b/drivers/cpufreq/powernow-k8.c
@@ -913,6 +913,11 @@ static int transition_frequency_fidvid(struct powernow_k8_data *data,
freqs.new = find_khz_freq_from_fid(fid);

policy = cpufreq_cpu_get(smp_processor_id());
+ if (!policy) {
+ pr_debug("cpu %d: CPUFreq policy not found\n",
+ smp_processor_id());
+ return 1;
+ }
cpufreq_cpu_put(policy);

cpufreq_freq_transition_begin(policy, &freqs);
--
1.9.1

2020-01-06 05:45:56

by Viresh Kumar

Subject: Re: [PATCH] cpufreq: powernow-k8: avoid use after free issue in cpufreq_notify_transition()

On 28-12-19, 10:57, [email protected] wrote:
> From: chenqiwu <[email protected]>
>
> There is a potential UAF issue in cpufreq_notify_transition() that the
> cpufreq of current cpu has been released before using it. So we should
> make a judgement and avoid it.
>
> Signed-off-by: chenqiwu <[email protected]>
> ---
> drivers/cpufreq/powernow-k8.c | 5 +++++
> 1 file changed, 5 insertions(+)
>
> diff --git a/drivers/cpufreq/powernow-k8.c b/drivers/cpufreq/powernow-k8.c
> index 2db2f17..7391eb0 100644
> --- a/drivers/cpufreq/powernow-k8.c
> +++ b/drivers/cpufreq/powernow-k8.c
> @@ -913,6 +913,11 @@ static int transition_frequency_fidvid(struct powernow_k8_data *data,
> freqs.new = find_khz_freq_from_fid(fid);
>
> policy = cpufreq_cpu_get(smp_processor_id());
> + if (!policy) {
> + pr_debug("cpu %d: CPUFreq policy not found\n",
> + smp_processor_id());
> + return 1;
> + }

You $subject and this change doesn't look related to me. The cpufreq
policy shall never be NULL here as we are in the middle of changing
frequency, initiated by cpufreq core itself.

> cpufreq_cpu_put(policy);
>
> cpufreq_freq_transition_begin(policy, &freqs);
> --
> 1.9.1

--
viresh