* [PATCH] powerpc: Update MPIC workarounds
@ 2005-12-13 7:04 Benjamin Herrenschmidt
2005-12-13 16:58 ` Olof Johansson
0 siblings, 1 reply; 3+ messages in thread
From: Benjamin Herrenschmidt @ 2005-12-13 7:04 UTC (permalink / raw)
To: Paul Mackerras; +Cc: linuxppc64-dev, linuxppc-dev list
From: Segher Boessenkool <segher@kernel.crashing.org>
Cleanup the MPIC IO-APIC workarounds, make them a bit more generic,
smaller and faster.
Signed-off-by: Segher Boessenkool <segher@kernel.crashing.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Index: linux-work/arch/powerpc/sysdev/mpic.c
===================================================================
--- linux-work.orig/arch/powerpc/sysdev/mpic.c 2005-12-06 16:17:43.000000000 +1100
+++ linux-work/arch/powerpc/sysdev/mpic.c 2005-12-07 13:30:45.000000000 +1100
@@ -175,57 +175,57 @@ static inline int mpic_is_ht_interrupt(s
return mpic->fixups[source_no].base != NULL;
}
+
static inline void mpic_apic_end_irq(struct mpic *mpic, unsigned int source_no)
{
struct mpic_irq_fixup *fixup = &mpic->fixups[source_no];
- u32 tmp;
spin_lock(&mpic->fixup_lock);
- writeb(0x11 + 2 * fixup->irq, fixup->base);
- tmp = readl(fixup->base + 2);
- writel(tmp | 0x80000000ul, fixup->base + 2);
- /* config writes shouldn't be posted but let's be safe ... */
- (void)readl(fixup->base + 2);
+ writeb(0x11 + 2 * fixup->irq, fixup->base + 2);
+ writel(fixup->data, fixup->base + 4);
spin_unlock(&mpic->fixup_lock);
}
-static void __init mpic_amd8111_read_irq(struct mpic *mpic, u8 __iomem *devbase)
+static void __init mpic_scan_ioapic(struct mpic *mpic, u8 __iomem *devbase)
{
- int i, irq;
+ int i, irq, n;
u32 tmp;
+ u8 pos;
- printk(KERN_INFO "mpic: - Workarounds on AMD 8111 @ %p\n", devbase);
+ for (pos = readb(devbase + 0x34); pos; pos = readb(devbase + pos + 1)) {
+ u8 id = readb(devbase + pos);
- for (i=0; i < 24; i++) {
- writeb(0x10 + 2*i, devbase + 0xf2);
- tmp = readl(devbase + 0xf4);
- if ((tmp & 0x1) || !(tmp & 0x20))
- continue;
- irq = (tmp >> 16) & 0xff;
- mpic->fixups[irq].irq = i;
- mpic->fixups[irq].base = devbase + 0xf2;
+ if (id == 0x08) {
+ id = readb(devbase + pos + 3);
+ if (id == 0x80)
+ break;
+ }
}
-}
-
-static void __init mpic_amd8131_read_irq(struct mpic *mpic, u8 __iomem *devbase)
-{
- int i, irq;
- u32 tmp;
+ if (pos == 0)
+ return;
+
+ printk(KERN_INFO "mpic: - Workarounds @ %p, pos = 0x%02x\n", devbase, pos);
- printk(KERN_INFO "mpic: - Workarounds on AMD 8131 @ %p\n", devbase);
+ devbase += pos;
- for (i=0; i < 4; i++) {
- writeb(0x10 + 2*i, devbase + 0xba);
- tmp = readl(devbase + 0xbc);
- if ((tmp & 0x1) || !(tmp & 0x20))
+ writeb(0x01, devbase + 2);
+ n = (readl(devbase + 4) >> 16) & 0xff;
+
+ for (i = 0; i <= n; i++) {
+ writeb(0x10 + 2 * i, devbase + 2);
+ tmp = readl(devbase + 4);
+ if ((tmp & 0x21) != 0x20)
continue;
irq = (tmp >> 16) & 0xff;
mpic->fixups[irq].irq = i;
- mpic->fixups[irq].base = devbase + 0xba;
+ mpic->fixups[irq].base = devbase;
+ writeb(0x11 + 2 * i, devbase + 2);
+ mpic->fixups[irq].data = readl(devbase + 4) | 0x80000000;
}
}
+
static void __init mpic_scan_ioapics(struct mpic *mpic)
{
unsigned int devfn;
@@ -241,21 +241,19 @@ static void __init mpic_scan_ioapics(str
/* Init spinlock */
spin_lock_init(&mpic->fixup_lock);
- /* Map u3 config space. We assume all IO-APICs are on the primary bus
- * and slot will never be above "0xf" so we only need to map 32k
+ /* Map U3 config space. We assume all IO-APICs are on the primary bus
+ * so we only need to map 64kB.
*/
- cfgspace = (unsigned char __iomem *)ioremap(0xf2000000, 0x8000);
+ cfgspace = ioremap(0xf2000000, 0x10000);
BUG_ON(cfgspace == NULL);
/* Now we scan all slots. We do a very quick scan, we read the header type,
* vendor ID and device ID only, that's plenty enough
*/
- for (devfn = 0; devfn < PCI_DEVFN(0x10,0); devfn ++) {
+ for (devfn = 0; devfn < 0x100; devfn++) {
u8 __iomem *devbase = cfgspace + (devfn << 8);
u8 hdr_type = readb(devbase + PCI_HEADER_TYPE);
u32 l = readl(devbase + PCI_VENDOR_ID);
- u16 vendor_id, device_id;
- int multifunc = 0;
DBG("devfn %x, l: %x\n", devfn, l);
@@ -264,21 +262,11 @@ static void __init mpic_scan_ioapics(str
l == 0x0000ffff || l == 0xffff0000)
goto next;
- /* Check if it's a multifunction device (only really used
- * to function 0 though
- */
- multifunc = !!(hdr_type & 0x80);
- vendor_id = l & 0xffff;
- device_id = (l >> 16) & 0xffff;
-
- /* If a known device, go to fixup setup code */
- if (vendor_id == PCI_VENDOR_ID_AMD && device_id == 0x7460)
- mpic_amd8111_read_irq(mpic, devbase);
- if (vendor_id == PCI_VENDOR_ID_AMD && device_id == 0x7450)
- mpic_amd8131_read_irq(mpic, devbase);
+ mpic_scan_ioapic(mpic, devbase);
+
next:
/* next device, if function 0 */
- if ((PCI_FUNC(devfn) == 0) && !multifunc)
+ if (PCI_FUNC(devfn) == 0 && (hdr_type & 0x80) == 0)
devfn += 7;
}
}
Index: linux-work/include/asm-powerpc/mpic.h
===================================================================
--- linux-work.orig/include/asm-powerpc/mpic.h 2005-11-24 17:18:48.000000000 +1100
+++ linux-work/include/asm-powerpc/mpic.h 2005-12-07 10:32:51.000000000 +1100
@@ -117,7 +117,8 @@ typedef int (*mpic_cascade_t)(struct pt_
struct mpic_irq_fixup
{
u8 __iomem *base;
- unsigned int irq;
+ u32 data;
+ unsigned int irq;
};
#endif /* CONFIG_MPIC_BROKEN_U3 */
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH] powerpc: Update MPIC workarounds
2005-12-13 7:04 [PATCH] powerpc: Update MPIC workarounds Benjamin Herrenschmidt
@ 2005-12-13 16:58 ` Olof Johansson
2005-12-13 20:23 ` Benjamin Herrenschmidt
0 siblings, 1 reply; 3+ messages in thread
From: Olof Johansson @ 2005-12-13 16:58 UTC (permalink / raw)
To: Benjamin Herrenschmidt; +Cc: linuxppc64-dev, linuxppc-dev list
On Tue, Dec 13, 2005 at 06:04:29PM +1100, Benjamin Herrenschmidt wrote:
> From: Segher Boessenkool <segher@kernel.crashing.org>
>
> Cleanup the MPIC IO-APIC workarounds, make them a bit more generic,
> smaller and faster.
I really don't like all the hand-coded constants in this code. They're
all over the place, and there's no descriptions of what they are there
for. Lots of hardcoded offsets, etc. Since this is a cleanup, wouldn't
it be a good time to use symbolic constands and/or comment them up a
bit?
> Index: linux-work/arch/powerpc/sysdev/mpic.c
> ===================================================================
> --- linux-work.orig/arch/powerpc/sysdev/mpic.c 2005-12-06 16:17:43.000000000 +1100
> +++ linux-work/arch/powerpc/sysdev/mpic.c 2005-12-07 13:30:45.000000000 +1100
> @@ -175,57 +175,57 @@ static inline int mpic_is_ht_interrupt(s
> return mpic->fixups[source_no].base != NULL;
> }
>
> +
> static inline void mpic_apic_end_irq(struct mpic *mpic, unsigned int source_no)
> {
> struct mpic_irq_fixup *fixup = &mpic->fixups[source_no];
> - u32 tmp;
>
> spin_lock(&mpic->fixup_lock);
> - writeb(0x11 + 2 * fixup->irq, fixup->base);
> - tmp = readl(fixup->base + 2);
> - writel(tmp | 0x80000000ul, fixup->base + 2);
> - /* config writes shouldn't be posted but let's be safe ... */
> - (void)readl(fixup->base + 2);
> + writeb(0x11 + 2 * fixup->irq, fixup->base + 2);
> + writel(fixup->data, fixup->base + 4);
This seems like a functional change: Previous code wrote at base, new at
base+2?
-Olof
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH] powerpc: Update MPIC workarounds
2005-12-13 16:58 ` Olof Johansson
@ 2005-12-13 20:23 ` Benjamin Herrenschmidt
0 siblings, 0 replies; 3+ messages in thread
From: Benjamin Herrenschmidt @ 2005-12-13 20:23 UTC (permalink / raw)
To: Olof Johansson; +Cc: linuxppc64-dev, linuxppc-dev list
On Tue, 2005-12-13 at 08:58 -0800, Olof Johansson wrote:
> On Tue, Dec 13, 2005 at 06:04:29PM +1100, Benjamin Herrenschmidt wrote:
> > From: Segher Boessenkool <segher@kernel.crashing.org>
> >
> > Cleanup the MPIC IO-APIC workarounds, make them a bit more generic,
> > smaller and faster.
>
> I really don't like all the hand-coded constants in this code. They're
> all over the place, and there's no descriptions of what they are there
> for. Lots of hardcoded offsets, etc. Since this is a cleanup, wouldn't
> it be a good time to use symbolic constands and/or comment them up a
> bit?
Heh, welcome to segher world :) I'll add more comments & symbolic
constants in the other patch that applies on top of this one.
> > Index: linux-work/arch/powerpc/sysdev/mpic.c
> > ===================================================================
> > --- linux-work.orig/arch/powerpc/sysdev/mpic.c 2005-12-06 16:17:43.000000000 +1100
> > +++ linux-work/arch/powerpc/sysdev/mpic.c 2005-12-07 13:30:45.000000000 +1100
> > @@ -175,57 +175,57 @@ static inline int mpic_is_ht_interrupt(s
> > return mpic->fixups[source_no].base != NULL;
> > }
> >
> > +
> > static inline void mpic_apic_end_irq(struct mpic *mpic, unsigned int source_no)
> > {
> > struct mpic_irq_fixup *fixup = &mpic->fixups[source_no];
> > - u32 tmp;
> >
> > spin_lock(&mpic->fixup_lock);
> > - writeb(0x11 + 2 * fixup->irq, fixup->base);
> > - tmp = readl(fixup->base + 2);
> > - writel(tmp | 0x80000000ul, fixup->base + 2);
> > - /* config writes shouldn't be posted but let's be safe ... */
> > - (void)readl(fixup->base + 2);
> > + writeb(0x11 + 2 * fixup->irq, fixup->base + 2);
> > + writel(fixup->data, fixup->base + 4);
>
> This seems like a functional change: Previous code wrote at base, new at
> base+2?
Hrm... I have to double check but I think the previous code hard coded
the bases for the 2 known APICs (incuding the +2) while the new code
properly scans the PCI capabilities and thus gets the capability base
instead, which is better (but needs +2 / +4). I suppose I could
pre-offset by 2 at discovery time to avoid an addition but I'm not sure
it's worth it.
Ben.
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2005-12-13 20:23 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2005-12-13 7:04 [PATCH] powerpc: Update MPIC workarounds Benjamin Herrenschmidt
2005-12-13 16:58 ` Olof Johansson
2005-12-13 20:23 ` Benjamin Herrenschmidt
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.