Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > linux.kernel > #1637708 > unrolled thread

Re: [PATCH] android: binder: check result of binder_get_thread() in binder_poll()

Started byJohn Stultz <john.stultz@linaro.org>
First post2017-05-08 22:50 +0200
Last post2017-05-08 23:00 +0200
Articles 2 — 2 participants

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.


Contents

  Re: [PATCH] android: binder: check result of binder_get_thread() in binder_poll() John Stultz <john.stultz@linaro.org> - 2017-05-08 22:50 +0200
    Re: [PATCH] android: binder: check result of binder_get_thread() in binder_poll() Doug Anderson <dianders@chromium.org> - 2017-05-08 23:00 +0200

#1637708 — Re: [PATCH] android: binder: check result of binder_get_thread() in binder_poll()

FromJohn Stultz <john.stultz@linaro.org>
Date2017-05-08 22:50 +0200
SubjectRe: [PATCH] android: binder: check result of binder_get_thread() in binder_poll()
Message-ID<tEXLs-ls-9@gated-at.bofh.it>
On Mon, May 8, 2017 at 1:43 PM, Dmitry Torokhov
<dmitry.torokhov@gmail.com> wrote:
> If binder_get_thread() fails to give us a thread data, we should avoid
> dereferencing a NULL pointer and return POLLERR instead.
>
> Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

Pulling Todd Kjos in on this too.
-john

> ---
>  drivers/android/binder.c | 12 ++++++++----
>  1 file changed, 8 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/android/binder.c b/drivers/android/binder.c
> index aae4d8d4be36..66ed714fedd5 100644
> --- a/drivers/android/binder.c
> +++ b/drivers/android/binder.c
> @@ -3103,18 +3103,22 @@ static unsigned int binder_poll(struct file *filp,
>                                 struct poll_table_struct *wait)
>  {
>         struct binder_proc *proc = filp->private_data;
> -       struct binder_thread *thread = NULL;
> +       struct binder_thread *thread;
>         int wait_for_proc_work;
>
>         binder_lock(__func__);
>
>         thread = binder_get_thread(proc);
> -
> -       wait_for_proc_work = thread->transaction_stack == NULL &&
> -               list_empty(&thread->todo) && thread->return_error == BR_OK;
> +       if (thread)
> +               wait_for_proc_work = thread->transaction_stack == NULL &&
> +                                    list_empty(&thread->todo) &&
> +                                    thread->return_error == BR_OK;
>
>         binder_unlock(__func__);
>
> +       if (!thread)
> +               return POLLERR;
> +
>         if (wait_for_proc_work) {
>                 if (binder_has_proc_work(proc, thread))
>                         return POLLIN;
> --
> 2.13.0.rc1.294.g07d810a77f-goog
>
>
> --
> Dmitry

[toc] | [next] | [standalone]


#1637714

FromDoug Anderson <dianders@chromium.org>
Date2017-05-08 23:00 +0200
Message-ID<tEXV7-oX-7@gated-at.bofh.it>
In reply to#1637708
Dmitry,

On Mon, May 8, 2017 at 1:46 PM, John Stultz <john.stultz@linaro.org> wrote:
> On Mon, May 8, 2017 at 1:43 PM, Dmitry Torokhov
> <dmitry.torokhov@gmail.com> wrote:
>> If binder_get_thread() fails to give us a thread data, we should avoid
>> dereferencing a NULL pointer and return POLLERR instead.
>>
>> Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
>
> Pulling Todd Kjos in on this too.
> -john
>
>> ---
>>  drivers/android/binder.c | 12 ++++++++----
>>  1 file changed, 8 insertions(+), 4 deletions(-)
>>
>> diff --git a/drivers/android/binder.c b/drivers/android/binder.c
>> index aae4d8d4be36..66ed714fedd5 100644
>> --- a/drivers/android/binder.c
>> +++ b/drivers/android/binder.c
>> @@ -3103,18 +3103,22 @@ static unsigned int binder_poll(struct file *filp,
>>                                 struct poll_table_struct *wait)
>>  {
>>         struct binder_proc *proc = filp->private_data;
>> -       struct binder_thread *thread = NULL;
>> +       struct binder_thread *thread;
>>         int wait_for_proc_work;
>>
>>         binder_lock(__func__);
>>
>>         thread = binder_get_thread(proc);
>> -
>> -       wait_for_proc_work = thread->transaction_stack == NULL &&
>> -               list_empty(&thread->todo) && thread->return_error == BR_OK;
>> +       if (thread)
>> +               wait_for_proc_work = thread->transaction_stack == NULL &&
>> +                                    list_empty(&thread->todo) &&
>> +                                    thread->return_error == BR_OK;
>>
>>         binder_unlock(__func__);
>>
>> +       if (!thread)
>> +               return POLLERR;
>> +
>>         if (wait_for_proc_work) {
>>                 if (binder_has_proc_work(proc, thread))
>>                         return POLLIN;
>> --

I'm no expert on the poll function, but I agree that it's wise to
check the result of binder_get_thread() since it can return NULL.
FWIW:

Reviewed-by: Douglas Anderson <dianders@chromium.org>

[toc] | [prev] | [standalone]


Back to top | Article view | linux.kernel


csiph-web