linux-kernel - Re: [PATCH] PM: QoS: Add check to make sure CPU freq is non-negative

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:   Tue, 12 Jul 2022 20:37:03 +0200
From:   "Rafael J. Wysocki" <rafael@...nel.org>
To:     Shivnandan Kumar <quic_kshivnan@...cinc.com>
Cc:     "Rafael J. Wysocki" <rafael@...nel.org>,
        Len Brown <len.brown@...el.com>, Pavel Machek <pavel@....cz>,
        Linux PM <linux-pm@...r.kernel.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] PM: QoS: Add check to make sure CPU freq is non-negative

On Thu, Jun 23, 2022 at 8:47 AM Shivnandan Kumar
<quic_kshivnan@...cinc.com> wrote:
>
>         CPU frequency should never be non-negative.

Do you mean "always be non-negative"?

>         If some client driver calls freq_qos_update_request with some
>         value greater than INT_MAX, then it will set max CPU freq at
>         fmax but it will add plist node with some negative priority.
>         plist node has priority from INT_MIN (highest) to INT_MAX
>         (lowest). Once priority is set as negative, another client
>         will not be able to reduce max CPU frequency. Adding check
>         to make sure CPU freq is non-negative will fix this problem.
> Signed-off-by: Shivnandan Kumar <quic_kshivnan@...cinc.com>
>
> ---
>  kernel/power/qos.c | 6 ++++--
>  1 file changed, 4 insertions(+), 2 deletions(-)
>
> diff --git a/kernel/power/qos.c b/kernel/power/qos.c
> index ec7e1e85923e..41e96fe34bfd 100644
> --- a/kernel/power/qos.c
> +++ b/kernel/power/qos.c
> @@ -531,7 +531,8 @@ int freq_qos_add_request(struct freq_constraints *qos,
>  {
>         int ret;
>
> -       if (IS_ERR_OR_NULL(qos) || !req)
> +       if (IS_ERR_OR_NULL(qos) || !req || value < FREQ_QOS_MIN_DEFAULT_VALUE
> +               || value > FREQ_QOS_MAX_DEFAULT_VALUE)

Why do you check against the defaults?

>                 return -EINVAL;
>
>         if (WARN(freq_qos_request_active(req),
> @@ -563,7 +564,8 @@ EXPORT_SYMBOL_GPL(freq_qos_add_request);
>   */
>  int freq_qos_update_request(struct freq_qos_request *req, s32 new_value)
>  {
> -       if (!req)
> +       if (!req || new_value < FREQ_QOS_MIN_DEFAULT_VALUE ||
> +               new_value > FREQ_QOS_MAX_DEFAULT_VALUE)
>                 return -EINVAL;
>
>         if (WARN(!freq_qos_request_active(req),
> --

I agree that it should guard against adding negative values, but I
don't see why s32 can be greater than INT_MAX.

Also why don't you put the guard into freq_qos_apply() instead of
duplicating it in the callers of that function?