Re: [RFCv2 13/14] irq_domain: Remove 'new' irq_domain in favour of the ppc one

On Tue, Jan 24, 2012 at 3:10 PM, Rob Herring [off-list ref] wrote:

On 01/23/2012 03:07 PM, Grant Likely wrote:

quoted

This patch removes the simplistic implementation of irq_domains and enab=

les

quoted

the powerpc infrastructure for all irq_domain users. =A0The powerpc
infrastructure includes support for complex mappings between Linux and
hardware irq numbers, and can manage allocation of irq_descs.

This patch also converts the few users of irq_domain_add()/irq_domain_de=

l()

quoted

to call irq_domain_add_legacy() instead.

v2: Fix removal of irq_alloc_descs() call in gic driver

Signed-off-by: Grant Likely <redacted>
---
=A0arch/arm/common/gic.c =A0 =A0 =A0 =A0 =A0 =A0 =A0 | =A0 85 ++++++++--=

---------

quoted

=A0arch/arm/common/vic.c =A0 =A0 =A0 =A0 =A0 =A0 =A0 | =A0 16 +---
=A0arch/arm/include/asm/hardware/gic.h | =A0 =A04 +-
=A0arch/arm/include/asm/hardware/vic.h | =A0 =A02 +
=A0arch/arm/mach-exynos/common.c =A0 =A0 =A0 | =A0 =A02 +-
=A0arch/arm/mach-versatile/core.c =A0 =A0 =A0| =A0 =A05 +-
=A0drivers/mfd/twl-core.c =A0 =A0 =A0 =A0 =A0 =A0 =A0| =A0 12 +--
=A0include/linux/irqdomain.h =A0 =A0 =A0 =A0 =A0 | =A0 45 +---------
=A0kernel/irq/irqdomain.c =A0 =A0 =A0 =A0 =A0 =A0 =A0| =A0159 +++-------=

-------------------------

quoted

=A09 files changed, 69 insertions(+), 261 deletions(-)

diff --git a/arch/arm/common/gic.c b/arch/arm/common/gic.c

snip

quoted

@@ -716,17 +708,17 @@ void __init gic_init_bases(unsigned int gic_nr, in=

t irq_start,

quoted

=A0 =A0 =A0 =A0 =A0 =A0 =A0 gic_irqs =3D 1020;
=A0 =A0 =A0 gic->gic_irqs =3D gic_irqs;

- =A0 =A0 domain->nr_irq =3D gic_irqs - domain->hwirq_base;
- =A0 =A0 domain->irq_base =3D irq_alloc_descs(irq_start, 16, domain->nr=

_irq,

quoted

- =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =

=A0 =A0 =A0numa_node_id());

quoted

- =A0 =A0 if (IS_ERR_VALUE(domain->irq_base)) {
+ =A0 =A0 irq_base =3D irq_alloc_descs(irq_start, 16, gic_irqs - hwirq_b=

ase,

quoted

+ =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0numa_no=

de_id());

quoted

+ =A0 =A0 if (IS_ERR_VALUE(irq_base)) {
=A0 =A0 =A0 =A0 =A0 =A0 =A0 WARN(1, "Cannot allocate irq_descs @ IRQ%d, =

assuming pre-allocated\n",

quoted

=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0irq_start);
- =A0 =A0 =A0 =A0 =A0 =A0 domain->irq_base =3D irq_start;
+ =A0 =A0 =A0 =A0 =A0 =A0 irq_base =3D irq_start;
=A0 =A0 =A0 }
- =A0 =A0 domain->host_data =3D gic;
- =A0 =A0 domain->ops =3D &gic_irq_domain_ops;
- =A0 =A0 irq_domain_add(domain);
+ =A0 =A0 gic->domain =3D irq_domain_add_legacy(node, gic_irqs, irq_base=

,

gic_irqs is wrong here. It needs 16 or 32 subtracted off. This patch
will fix things:

Good catch, I've integrated this into my series.

g.