On 2020-10-22 19:04, Peter Zijlstra wrote:
On Thu, Oct 22, 2020 at 06:19:37PM +0530, Sai Prakash Ranjan wrote:
On 2020-10-22 17:02, Peter Zijlstra wrote:
> On Thu, Oct 22, 2020 at 04:27:52PM +0530, Sai Prakash Ranjan wrote:
>
> > Looking at the ETR and other places in the kernel, ETF and the
> > ETB are the only places trying to dereference the task(owner)
> > in tmc_enable_etf_sink_perf() which is also called from the
> > sched_in path as in the call trace.
>
> > @@ -391,6 +392,10 @@ static void *tmc_alloc_etf_buffer(struct
> > coresight_device *csdev,
> > {
> > int node;
> > struct cs_buffers *buf;
> > + struct task_struct *task = READ_ONCE(event->owner);
> > +
> > + if (!task || is_kernel_event(event))
> > + return NULL;
>
>
> This is *wrong*... why do you care about who owns the events?
>
The original issue was the owner being NULL and causing
a NULL pointer dereference. I did ask some time back
if it is valid for the owner to be NULL [1] and should
probably be handled in events core?
No, what I asked is why do you care about ->owner to begin with? That
seems wrong. A driver should not touch ->owner _at_all_.
Ah ok, so Suzuki explained that in other reply and if there is
some other better way?
Thanks,
Sai
--
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a
member
of Code Aurora Forum, hosted by The Linux Foundation
