[PATCH] x86/mrst: move sfi timer irqs after gsi_top

[PATCH] x86/mrst: move sfi timer irqs after gsi_top

[Jacob Pan]

Moorestown borrows mp_irqs for its early SFI device IRQs, it used
to be 1:1 mapped between irq # and ioapic pin#. After gsi_top is
introduced, we have to allocate irqs on top of gsi_top since below
that irqs are not set to ioapic chip whereas Moorestown SFI device
irqs are all routed via ioapic chip.

Without this fix, Moorestown fails to boot at secondary clock setup
since it cannot do request_irq on cpu 1 timer which does not have a
valid irq_chip.

Also use correct MP configuration table entry type INTSRC, which is
used for IO interrupt assignment.

Signed-off-by: Jacob Pan <jacob.jun.pan@linux.intel.com>
---
 arch/x86/kernel/apb_timer.c   |    2 +-
 arch/x86/platform/mrst/mrst.c |    4 ++--
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/arch/x86/kernel/apb_timer.c b/arch/x86/kernel/apb_timer.c
index cd1ffed..f9f82cb 100644
--- a/arch/x86/kernel/apb_timer.c
+++ b/arch/x86/kernel/apb_timer.c
@@ -630,7 +630,7 @@ void __init apbt_time_init(void)
 		p_mtmr = sfi_get_mtmr(i);
 		if (p_mtmr) {
 			adev->tick = p_mtmr->freq_hz;
-			adev->irq = p_mtmr->irq;
+			adev->irq = p_mtmr->irq + gsi_top;
 		} else
 			printk(KERN_ERR "Failed to get timer for cpu %d\n", i);
 		adev->count = 0;
diff --git a/arch/x86/platform/mrst/mrst.c b/arch/x86/platform/mrst/mrst.c
index 5c0207b..448c20d 100644
--- a/arch/x86/platform/mrst/mrst.c
+++ b/arch/x86/platform/mrst/mrst.c
@@ -97,12 +97,12 @@ static int __init sfi_parse_mtmr(struct sfi_table_header *table)
 			pentry->freq_hz, pentry->irq);
 			if (!pentry->irq)
 				continue;
-			mp_irq.type = MP_IOAPIC;
+			mp_irq.type = MP_INTSRC;
 			mp_irq.irqtype = mp_INT;
 /* triggering mode edge bit 2-3, active high polarity bit 0-1 */
 			mp_irq.irqflag = 5;
 			mp_irq.srcbus = 0;
-			mp_irq.srcbusirq = pentry->irq;	/* IRQ */
+			mp_irq.srcbusirq = pentry->irq + gsi_top; /* IRQ */
 			mp_irq.dstapic = MP_APIC_ALL;
 			mp_irq.dstirq = pentry->irq;
 			mp_save_irq(&mp_irq);
-- 
1.7.1

Re: [PATCH] x86/mrst: move sfi timer irqs after gsi_top

[Feng Tang]

Hi Jacob,

The mrst vrtc driver has the similar irq number problem in 2.6.39-rc2,
how about the following patch to solve it in ioapic code.

Also I saw NR_IRQS_LEGACY are used in several other places of io_apic code,
it may has some potential issue for mrst which I'm not sure.

Thanks,
Feng
---------------------

>From 9bfe95ca6b1737a0418c6403f9684da9fb844cbf Mon Sep 17 00:00:00 2001
From: Feng Tang <feng.tang@intel.com>
Date: Thu, 7 Apr 2011 09:08:00 +0800
Subject: [PATCH] x86: ioapic: use legacy_pic->nr_legacy_irqs instead of NR_IRQS_LEGACY for pin_2_irq()

legacy_pic->nr_legacy_irqs has the correct number of legacy irqs for
a x86 platform (legacy PC, Moorestown etc).

Signed-off-by: Feng Tang <feng.tang@intel.com>
---
 arch/x86/kernel/apic/io_apic.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
index 68df09b..2dcad66 100644
--- a/arch/x86/kernel/apic/io_apic.c
+++ b/arch/x86/kernel/apic/io_apic.c
@@ -954,7 +954,7 @@ static int pin_2_irq(int idx, int apic, int pin)
 	} else {
 		u32 gsi = mp_gsi_routing[apic].gsi_base + pin;
 
-		if (gsi >= NR_IRQS_LEGACY)
+		if (gsi >= legacy_pic->nr_legacy_irqs)
 			irq = gsi;
 		else
 			irq = gsi_top + gsi;
-- 
1.7.0.4

> From: Jacob Pan <jacob.jun.pan@linux.intel.com>
> Date: 2011/4/5
> Subject: [PATCH] x86/mrst: move sfi timer irqs after gsi_top
> To: LKML <linux-kernel@vger.kernel.org>, "H. Peter Anvin"
> <hpa@zytor.com>, Ingo Molnar <mingo@elte.hu>, Thomas Gleixner
> <tglx@linutronix.de>, Arjan van de Ven <arjan@linux.intel.com>, Alan
> Cox <alan@linux.intel.com>
> 抄送： Jacob Pan <jacob.jun.pan@linux.intel.com>
> 
> 
> Moorestown borrows mp_irqs for its early SFI device IRQs, it used
> to be 1:1 mapped between irq # and ioapic pin#. After gsi_top is
> introduced, we have to allocate irqs on top of gsi_top since below
> that irqs are not set to ioapic chip whereas Moorestown SFI device
> irqs are all routed via ioapic chip.
> 
> Without this fix, Moorestown fails to boot at secondary clock setup
> since it cannot do request_irq on cpu 1 timer which does not have a
> valid irq_chip.
> 
> Also use correct MP configuration table entry type INTSRC, which is
> used for IO interrupt assignment.
> 
> Signed-off-by: Jacob Pan <jacob.jun.pan@linux.intel.com>
> ---
>  arch/x86/kernel/apb_timer.c   |    2 +-
>  arch/x86/platform/mrst/mrst.c |    4 ++--
>  2 files changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/arch/x86/kernel/apb_timer.c b/arch/x86/kernel/apb_timer.c
> index cd1ffed..f9f82cb 100644
> --- a/arch/x86/kernel/apb_timer.c
> +++ b/arch/x86/kernel/apb_timer.c
> @@ -630,7 +630,7 @@ void __init apbt_time_init(void)
>                p_mtmr = sfi_get_mtmr(i);
>                if (p_mtmr) {
>                        adev->tick = p_mtmr->freq_hz;
> -                       adev->irq = p_mtmr->irq;
> +                       adev->irq = p_mtmr->irq + gsi_top;
>                } else
>                        printk(KERN_ERR "Failed to get timer for cpu
> %d\n", i); adev->count = 0;
> diff --git a/arch/x86/platform/mrst/mrst.c
> b/arch/x86/platform/mrst/mrst.c index 5c0207b..448c20d 100644
> --- a/arch/x86/platform/mrst/mrst.c
> +++ b/arch/x86/platform/mrst/mrst.c
> @@ -97,12 +97,12 @@ static int __init sfi_parse_mtmr(struct
> sfi_table_header *table)
>                        pentry->freq_hz, pentry->irq);
>                        if (!pentry->irq)
>                                continue;
> -                       mp_irq.type = MP_IOAPIC;
> +                       mp_irq.type = MP_INTSRC;
>                        mp_irq.irqtype = mp_INT;
>  /* triggering mode edge bit 2-3, active high polarity bit 0-1 */
>                        mp_irq.irqflag = 5;
>                        mp_irq.srcbus = 0;
> -                       mp_irq.srcbusirq = pentry->irq; /* IRQ */
> +                       mp_irq.srcbusirq = pentry->irq + gsi_top; /*
> IRQ */ mp_irq.dstapic = MP_APIC_ALL;
>                        mp_irq.dstirq = pentry->irq;
>                        mp_save_irq(&mp_irq);
> --
> 1.7.1
> 
> --
> To unsubscribe from this list: send the line "unsubscribe
> linux-kernel" in the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/

Re: [PATCH] x86/mrst: move sfi timer irqs after gsi_top

[jacob pan]

On Thu, 7 Apr 2011 09:30:21 +0800
Feng Tang <feng.tang@intel.com> wrote:

> Hi Jacob,
> 
> The mrst vrtc driver has the similar irq number problem in 2.6.39-rc2,
> how about the following patch to solve it in ioapic code.
Just dug up the old thread in https://lkml.org/lkml/2010/6/10/70. I
actually have been through this before, I just sent out another patch
for this as suggested by Eric.

This patch is withdrawn now.

Thanks,

Jacob