On Tue, Oct 31 2017 at 9:41:43 am GMT, Paul Burton <paul.bur...@mips.com>
wrote:
> This series continues cleaning & fixing up the MIPS GIC irqchip driver
> whilst laying groundwork to support multi-cluster systems.
>
> Patch 1 refactors in order to reduce some duplication and prepare us for
> the following patches.
>
> Patches 2-4 move per-CPU GIC configuration away from being performed all
> at once when the driver is probed or when interrupts are masked &
> unmasked, instead performing configuration as CPUs are brought online.
> This allows us to support reconfiguring after clusters are powered down
> & back up, generally cleans up and fixes bugs in the process.
>
> Patch 5 makes use of num_possible_cpus() to reserve IPIs, rather than
> the gic_vpes variable. This prepares us for multi-cluster in which
> gic_vpes is mostly meaningless since it only reflects the local cluster,
> and it generally makes more sense to use the more standard
> num_possible_cpus().
>
> Patch 6 removes the now unused gic_vpes variable.
>
> Patch 7 is a general clean up but also prepares us for later patches as
> described in its commit message.
>
> Patch 8 is a general clean up marking some variables static.
>
> This series by itself continues along the path towards supporting
> multi-cluster systems such as the MIPS I6500, but does not yet get us
> the whole way there. If you wish to see my current work in progress
> which builds out multi-cluster support atop these patches then that can
> be found in the multicluster branch of:
>
> git://git.linux-mips.org/pub/scm/paul/linux.git
>
> Or browsed at:
>
> https://git.linux-mips.org/cgit/paul/linux.git/log/?h=multicluster
>
> This series applies cleanly atop v4.14-rc7.
Are those targeting 4.14 or 4.15? It is getting quite late for the
former, and it doesn't seem to cleanly apply on tip/irq/core (or my
irqchip-4.15 branch) if that's for the latter (patch 6 shouts at me).
Thanks,
M.
--
Jazz is not dead. It just smells funny.
