On Tue, Oct 01, 2013 at 03:55:24PM -0400, Nicolas Pitre wrote:
> On Tue, 1 Oct 2013, Vyacheslav Tyrtov wrote:
>
> > From: Tarek Dakhran <t.dakhran@xxxxxxxxxxx>
[...]
> > + kfs_use_count[cpu][cluster]++;
> > + if (kfs_use_count[cpu][cluster] == 1) {
> > + ++core_count[cluster];
> > + if (core_count[cluster] == 1) {
> > + ret = exynos_cluster_power_up(cluster);
> > + if (ret) {
> > + pr_err("%s: cluster %u power up error\n",
> > + __func__, cluster);
> > + return ret;
> > + }
> > + __cci_control_port_by_index(MAX_NR_CLUSTERS
> > + - cluster, true);
>
> This is wrong and very racy. The state machine implemented in
> mcpm-head.S is there already to handle proper synchronization for you.
Maybe this issue didn't make itself obvious yet due to the lack of
suspend support.
Moving the CCI maintenance to power_up_setup() is essential for suspend/
resume to work, because then CPUs can power up randomly in response to
interrupts -- exynos_lock is not sufficient protection in that case.
The TC2 code should provide a good example of what to do.
[...]
Cheers
---Dave
--
To unsubscribe from this list: send the line "unsubscribe linux-doc" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
