| Message ID | 1353448867-15008-16-git-send-email-thomas.petazzoni@free-electrons.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Tue, Nov 20, 2012 at 11:01 PM, Thomas Petazzoni <thomas.petazzoni@free-electrons.com> wrote: > With all vic_of_init() users converted to the irqchip infrastructure, > there is no longer a reason for this function to be > exported. Therefore, mark it as static. > > Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com> It may be that I have missed some patches, but are really all VIC users converted in this patch series? In arch/arm/mach-u300/core.c I have this: vic_init((void __iomem *) U300_INTCON0_VBASE, IRQ_U300_INTCON0_START, mask[0], mask[0]); vic_init((void __iomem *) U300_INTCON1_VBASE, IRQ_U300_INTCON1_START, mask[1], mask[1]); arch/arm/mach-nomadik/cpu-8815.c: /* This modified VIC cell has two register blocks, at 0 and 0x20 */ vic_init(io_p2v(NOMADIK_IC_BASE + 0x00), IRQ_VIC_START + 0, ~0, 0); vic_init(io_p2v(NOMADIK_IC_BASE + 0x20), IRQ_VIC_START + 32, ~0, 0); And there are more still I think? I didn't see those converted so I guess I must have missed some patches or something? PS please make sure u300_defconfig and nhk8815_defconfig builds, or I will be a sad puppy ... Yours, Linus Walleij
Dear Linus Walleij, On Wed, 21 Nov 2012 13:40:57 +0100, Linus Walleij wrote: > On Tue, Nov 20, 2012 at 11:01 PM, Thomas Petazzoni > <thomas.petazzoni@free-electrons.com> wrote: > > > With all vic_of_init() users converted to the irqchip infrastructure, > > there is no longer a reason for this function to be > > exported. Therefore, mark it as static. > > > > Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com> > > It may be that I have missed some patches, but are really all VIC > users converted in this patch series? > > In arch/arm/mach-u300/core.c I have this: > vic_init((void __iomem *) U300_INTCON0_VBASE, IRQ_U300_INTCON0_START, > mask[0], mask[0]); > vic_init((void __iomem *) U300_INTCON1_VBASE, IRQ_U300_INTCON1_START, > mask[1], mask[1]); > > arch/arm/mach-nomadik/cpu-8815.c: > > /* This modified VIC cell has two register blocks, at 0 and 0x20 */ > vic_init(io_p2v(NOMADIK_IC_BASE + 0x00), IRQ_VIC_START + 0, ~0, 0); > vic_init(io_p2v(NOMADIK_IC_BASE + 0x20), IRQ_VIC_START + 32, ~0, 0); > > And there are more still I think? I didn't see those converted so > I guess I must have missed some patches or something? In this patch series, I'm making vic_of_init() static. vic_init() used by non-DT platforms is kept as is, with no change, so I'm not sure why you're worried about vic_init() being a problem. > PS please make sure u300_defconfig and nhk8815_defconfig builds, or > I will be a sad puppy ... Indeed. The whole thing needs more testing, for sure. I guess I should have labeled it as RFC, because it's really what it is. Best regards, Thomas
On Wed, Nov 21, 2012 at 1:53 PM, Thomas Petazzoni <thomas.petazzoni@free-electrons.com> wrote: > On Wed, 21 Nov 2012 13:40:57 +0100, Linus Walleij wrote: >> It may be that I have missed some patches, but are really all VIC >> users converted in this patch series? > > In this patch series, I'm making vic_of_init() static. vic_init() used > by non-DT platforms is kept as is, with no change, so I'm not sure why > you're worried about vic_init() being a problem. Probably because I didn't parse the _of_ part :-/ Sorry about the fuzz! But the idea is to use the same convenient mechanism for the non-DT platforms later, so we get rid of all the incldue headers I guess? Yours, Linus Walleij
Linus, On Wed, 21 Nov 2012 15:40:02 +0100, Linus Walleij wrote: > On Wed, Nov 21, 2012 at 1:53 PM, Thomas Petazzoni > <thomas.petazzoni@free-electrons.com> wrote: > > On Wed, 21 Nov 2012 13:40:57 +0100, Linus Walleij wrote: > > >> It may be that I have missed some patches, but are really all VIC > >> users converted in this patch series? > > > > In this patch series, I'm making vic_of_init() static. vic_init() used > > by non-DT platforms is kept as is, with no change, so I'm not sure why > > you're worried about vic_init() being a problem. > > Probably because I didn't parse the _of_ part :-/ > > Sorry about the fuzz! No problem :) > But the idea is to use the same convenient mechanism for > the non-DT platforms later, so we get rid of all the incldue > headers I guess? Yes. I personally haven't yet thought of this part of the problem, but I think Rob has a plan, and if not, we can certainly think of one without too much problems. Best regards, Thomas
On 11/21/2012 08:46 AM, Thomas Petazzoni wrote: > Linus, > > On Wed, 21 Nov 2012 15:40:02 +0100, Linus Walleij wrote: >> On Wed, Nov 21, 2012 at 1:53 PM, Thomas Petazzoni >> <thomas.petazzoni@free-electrons.com> wrote: >>> On Wed, 21 Nov 2012 13:40:57 +0100, Linus Walleij wrote: >> >>>> It may be that I have missed some patches, but are really all VIC >>>> users converted in this patch series? >>> >>> In this patch series, I'm making vic_of_init() static. vic_init() used >>> by non-DT platforms is kept as is, with no change, so I'm not sure why >>> you're worried about vic_init() being a problem. >> >> Probably because I didn't parse the _of_ part :-/ >> >> Sorry about the fuzz! > > No problem :) > >> But the idea is to use the same convenient mechanism for >> the non-DT platforms later, so we get rid of all the incldue >> headers I guess? > > Yes. I personally haven't yet thought of this part of the problem, but > I think Rob has a plan, and if not, we can certainly think of one > without too much problems. While we can trim down the headers, I don't think we will get rid of them in all cases. My only plan was everything gets converted to DT. Then we can remove gic_init, gic_cascade_irq, and vic_init. Rob
Rob, On Wed, 21 Nov 2012 09:07:59 -0600, Rob Herring wrote: > > Yes. I personally haven't yet thought of this part of the problem, but > > I think Rob has a plan, and if not, we can certainly think of one > > without too much problems. > > While we can trim down the headers, I don't think we will get rid of > them in all cases. My only plan was everything gets converted to DT. > Then we can remove gic_init, gic_cascade_irq, and vic_init. Sounds like a good plan :-) Thomas
diff --git a/drivers/irqchip/irq-vic.c b/drivers/irqchip/irq-vic.c index 98a0708..c8c8cab 100644 --- a/drivers/irqchip/irq-vic.c +++ b/drivers/irqchip/irq-vic.c @@ -440,7 +440,8 @@ void __init vic_init(void __iomem *base, unsigned int irq_start, } #ifdef CONFIG_OF -int __init vic_of_init(struct device_node *node, struct device_node *parent) +static int __init vic_of_init(struct device_node *node, + struct device_node *parent) { void __iomem *regs; int irq_base;
With all vic_of_init() users converted to the irqchip infrastructure, there is no longer a reason for this function to be exported. Therefore, mark it as static. Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com> --- drivers/irqchip/irq-vic.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)