[PATCH 0/5] x86/pci: patches for tip/x86/unify-pci

[PATCH 0/5] x86/pci: patches for tip/x86/unify-pci

[Robert Richter]

This patch series contains patches for tip/x86/unify-pci. It fixes
NUMA initialization code reflecting latest changes in tip/master. Also
there are build fixes and minor cleanups. Patch #5 is a cherry pick
that fixes build on UP systems. Code reboots for smp/nosmp,
acpi/noacpi, numaq, 32/64bit. It compiles for visws.

[PATCH 1/5] x86/pci: Fix build error for X86_VISWS

[Robert Richter]

Signed-off-by: Robert Richter <robert.richter@amd.com>
---
 arch/x86/Kconfig |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
index bb0c0d0..51329ee 100644
--- a/arch/x86/Kconfig
+++ b/arch/x86/Kconfig
@@ -283,7 +283,7 @@ config X86_VOYAGER
 
 config X86_VISWS
 	bool "SGI 320/540 (Visual Workstation)"
-	depends on X86_32 && !PCI
+	depends on X86_32 && PCI
 	help
 	  The SGI Visual Workstation series is an IA32-based workstation
 	  based on SGI systems chips with some legacy PC hardware attached.
-- 
1.5.5.4

[PATCH 2/5] x86/pci: Makefile merge: Changing 32bit ordering

[Robert Richter]

This should be save since VISWS and NUMAQ exclud each other.

Signed-off-by: Robert Richter <robert.richter@amd.com>
---
 arch/x86/pci/Makefile |    3 +--
 1 files changed, 1 insertions(+), 2 deletions(-)

diff --git a/arch/x86/pci/Makefile b/arch/x86/pci/Makefile
index 99d9f09..2a08d2d 100644
--- a/arch/x86/pci/Makefile
+++ b/arch/x86/pci/Makefile
@@ -8,12 +8,11 @@ obj-$(CONFIG_PCI_OLPC)		+= olpc.o
 pci-y				:= fixup.o
 pci-$(CONFIG_ACPI)		+= acpi.o
 pci-y				+= legacy.o irq.o
+pci-$(CONFIG_X86_NUMAQ)		+= numa.o
 
 # Careful: VISWS overrule the pci-y above. The colons are
 # therefor correct. This needs a proper fix by distangling the code.
 pci-$(CONFIG_X86_VISWS)		:= visws.o fixup.o
 
-pci-$(CONFIG_X86_NUMAQ)		+= numa.o
-
 obj-y				+= $(pci-y) common.o early.o
 obj-y				+= amd_bus.o
-- 
1.5.5.4

[PATCH 3/5] x86/pci merge: Fixing numaq initialization

[Robert Richter]

Patch d49c4288 (tip/x86/mpparse) introduced some changes in calling
subsys_init calls if CONFIG_X86_NUMAQ option is set. This patch
updates subsystem initalization according to this changes.

Signed-off-by: Robert Richter <robert.richter@amd.com>
---
 arch/x86/pci/legacy.c |    3 +++
 arch/x86/pci/numa.c   |   10 +---------
 arch/x86/pci/pci.h    |    1 +
 3 files changed, 5 insertions(+), 9 deletions(-)

diff --git a/arch/x86/pci/legacy.c b/arch/x86/pci/legacy.c
index 3c1d795..d8b6e92 100644
--- a/arch/x86/pci/legacy.c
+++ b/arch/x86/pci/legacy.c
@@ -62,6 +62,9 @@ int __init pci_subsys_init(void)
 #endif
 	pci_legacy_init();
 	pcibios_irq_init();
+#ifdef CONFIG_X86_NUMAQ
+	pci_numa_init();
+#endif
 	pcibios_init();
 }
 subsys_initcall(pci_subsys_init);
diff --git a/arch/x86/pci/numa.c b/arch/x86/pci/numa.c
index e1620dc..8b5ca19 100644
--- a/arch/x86/pci/numa.c
+++ b/arch/x86/pci/numa.c
@@ -151,7 +151,7 @@ static void __devinit pci_fixup_i450nx(struct pci_dev *d)
 }
 DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_82451NX, pci_fixup_i450nx);
 
-static int __init pci_numa_init(void)
+int __init pci_numa_init(void)
 {
 	int quad;
 
@@ -176,11 +176,3 @@ static int __init pci_numa_init(void)
 		}
 	return 0;
 }
-
-static __init int pci_subsys_init(void)
-{
-	pci_numa_init();
-	pcibios_irq_init();
-	pcibios_init();
-}
-subsys_initcall(pci_subsys_init);
diff --git a/arch/x86/pci/pci.h b/arch/x86/pci/pci.h
index 5824174..b2270a5 100644
--- a/arch/x86/pci/pci.h
+++ b/arch/x86/pci/pci.h
@@ -107,6 +107,7 @@ extern void __init dmi_check_skip_isa_align(void);
 /* some common used subsys_initcalls */
 extern int __init pci_acpi_init(void);
 extern int __init pcibios_irq_init(void);
+extern int __init pci_numa_init(void);
 extern int __init pcibios_init(void);
 
 /* pci-mmconfig.c */
-- 
1.5.5.4

Re: [PATCH 3/5] x86/pci merge: Fixing numaq initialization

[Yinghai Lu]

On Thu, Jul 10, 2008 at 9:58 AM, Robert Richter <robert.richter@amd.com> wrote:
> Patch d49c4288 (tip/x86/mpparse) introduced some changes in calling
> subsys_init calls if CONFIG_X86_NUMAQ option is set. This patch
> updates subsystem initalization according to this changes.
>
> Signed-off-by: Robert Richter <robert.richter@amd.com>
> ---
>  arch/x86/pci/legacy.c |    3 +++
>  arch/x86/pci/numa.c   |   10 +---------
>  arch/x86/pci/pci.h    |    1 +
>  3 files changed, 5 insertions(+), 9 deletions(-)
>
> diff --git a/arch/x86/pci/legacy.c b/arch/x86/pci/legacy.c
> index 3c1d795..d8b6e92 100644
> --- a/arch/x86/pci/legacy.c
> +++ b/arch/x86/pci/legacy.c
> @@ -62,6 +62,9 @@ int __init pci_subsys_init(void)
>  #endif
>        pci_legacy_init();
>        pcibios_irq_init();
> +#ifdef CONFIG_X86_NUMAQ
> +       pci_numa_init();
> +#endif
>        pcibios_init();
>  }

please change sequence to

+#ifdef CONFIG_X86_NUMAQ
+       pci_numa_init();
+#endif
#ifdef CONFIG_ACPI
        pci_acpi_init();
#endif
        pci_legacy_init();

YH

[PATCH 4/5] x86/pci: Fix warnings in subsys_initcall functions

[Robert Richter]

Signed-off-by: Robert Richter <robert.richter@amd.com>
---
 arch/x86/pci/legacy.c |    2 ++
 arch/x86/pci/visws.c  |    2 ++
 2 files changed, 4 insertions(+), 0 deletions(-)

diff --git a/arch/x86/pci/legacy.c b/arch/x86/pci/legacy.c
index d8b6e92..132876c 100644
--- a/arch/x86/pci/legacy.c
+++ b/arch/x86/pci/legacy.c
@@ -66,5 +66,7 @@ int __init pci_subsys_init(void)
 	pci_numa_init();
 #endif
 	pcibios_init();
+
+	return 0;
 }
 subsys_initcall(pci_subsys_init);
diff --git a/arch/x86/pci/visws.c b/arch/x86/pci/visws.c
index 16e5206..852998b 100644
--- a/arch/x86/pci/visws.c
+++ b/arch/x86/pci/visws.c
@@ -109,5 +109,7 @@ static __init int pci_subsys_init(void)
 {
 	pci_visws_init();
 	pcibios_init();
+
+	return 0;
 }
 subsys_initcall(pci_subsys_init);
-- 
1.5.5.4

[PATCH 5/5] x86: build fix for "x86: fix C1E && nx6325 stability problem"

[Robert Richter]

From: Ingo Molnar <mingo@elte.hu>

fix:

 arch/x86/kernel/built-in.o: In function `dmi_ignore_irq0_timer_override':
 boot.c:(.init.text+0x3ea4): undefined reference to `force_mask_ioapic_irq_2'

Signed-off-by: Ingo Molnar <mingo@elte.hu>
---
 include/asm-x86/genapic_32.h |    4 ++++
 include/asm-x86/genapic_64.h |    4 ++++
 2 files changed, 8 insertions(+), 0 deletions(-)

diff --git a/include/asm-x86/genapic_32.h b/include/asm-x86/genapic_32.h
index 8d4c8bd..33a73f5 100644
--- a/include/asm-x86/genapic_32.h
+++ b/include/asm-x86/genapic_32.h
@@ -119,6 +119,10 @@ enum uv_system_type {UV_NONE, UV_LEGACY_APIC, UV_X2APIC, UV_NON_UNIQUE_APIC};
 #define is_uv_system()			0
 #define uv_wakeup_secondary(a, b)	1
 
+#ifdef CONFIG_X86_IO_APIC
 extern void force_mask_ioapic_irq_2(void);
+#else
+static inline void force_mask_ioapic_irq_2(void) { }
+#endif
 
 #endif
diff --git a/include/asm-x86/genapic_64.h b/include/asm-x86/genapic_64.h
index 082ad02..647e4e5 100644
--- a/include/asm-x86/genapic_64.h
+++ b/include/asm-x86/genapic_64.h
@@ -46,6 +46,10 @@ extern int uv_wakeup_secondary(int phys_apicid, unsigned int start_rip);
 
 extern void setup_apic_routing(void);
 
+#ifdef CONFIG_X86_IO_APIC
 extern void force_mask_ioapic_irq_2(void);
+#else
+static inline void force_mask_ioapic_irq_2(void) { }
+#endif
 
 #endif
-- 
1.5.5.4

Re: [PATCH 0/5] x86/pci: patches for tip/x86/unify-pci

[Ingo Molnar]

* Robert Richter <robert.richter@amd.com> wrote:

> This patch series contains patches for tip/x86/unify-pci. It fixes 
> NUMA initialization code reflecting latest changes in tip/master. Also 
> there are build fixes and minor cleanups. Patch #5 is a cherry pick 
> that fixes build on UP systems. Code reboots for smp/nosmp, 
> acpi/noacpi, numaq, 32/64bit. It compiles for visws.

thanks Robert. Today i've eliminated the VISWS subarch which greatly 
simplified your series. I've picked up these two fixes:

 Robert Richter (2):
       x86/pci merge: fixing numaq initialization
       x86/pci: fix warnings in subsys_initcall functions

the others were moot i think.

Please check the tip/x86/generalize-visws, i just pushed them all out.

	Ingo

Re: [PATCH 0/5] x86/pci: patches for tip/x86/unify-pci

[Ingo Molnar]

* Ingo Molnar <mingo@elte.hu> wrote:

> * Robert Richter <robert.richter@amd.com> wrote:
> 
> > This patch series contains patches for tip/x86/unify-pci. It fixes 
> > NUMA initialization code reflecting latest changes in tip/master. 
> > Also there are build fixes and minor cleanups. Patch #5 is a cherry 
> > pick that fixes build on UP systems. Code reboots for smp/nosmp, 
> > acpi/noacpi, numaq, 32/64bit. It compiles for visws.
> 
> thanks Robert. Today i've eliminated the VISWS subarch which greatly 
> simplified your series. I've picked up these two fixes:
> 
>  Robert Richter (2):
>        x86/pci merge: fixing numaq initialization
>        x86/pci: fix warnings in subsys_initcall functions

btw., this is how arch/x86/pci/Makefile looks like now in tip/master:

----------------------------------------->
pci-y                           := fixup.o
pci-$(CONFIG_ACPI)              += acpi.o
pci-y                           += legacy.o irq.o

pci-$(CONFIG_X86_VISWS)         += visws.o

pci-$(CONFIG_X86_NUMAQ)         += numa.o

obj-y                           += $(pci-y) common.o early.o
obj-y                           += amd_bus.o
<-----------------------------------------

it's _really_ clean now, and we dont have those weird overrides anymore. 
Both VisWS and NUMAQ are now two straightforward runtime PCI overrides - 
not build-time complications. Much more maintainable i think.

	Ingo

Re: [PATCH 0/5] x86/pci: patches for tip/x86/unify-pci

[Sam Ravnborg]

On Thu, Jul 10, 2008 at 08:12:34PM +0200, Ingo Molnar wrote:
> 
> * Ingo Molnar <mingo@elte.hu> wrote:
> 
> > * Robert Richter <robert.richter@amd.com> wrote:
> > 
> > > This patch series contains patches for tip/x86/unify-pci. It fixes 
> > > NUMA initialization code reflecting latest changes in tip/master. 
> > > Also there are build fixes and minor cleanups. Patch #5 is a cherry 
> > > pick that fixes build on UP systems. Code reboots for smp/nosmp, 
> > > acpi/noacpi, numaq, 32/64bit. It compiles for visws.
> > 
> > thanks Robert. Today i've eliminated the VISWS subarch which greatly 
> > simplified your series. I've picked up these two fixes:
> > 
> >  Robert Richter (2):
> >        x86/pci merge: fixing numaq initialization
> >        x86/pci: fix warnings in subsys_initcall functions
> 
> btw., this is how arch/x86/pci/Makefile looks like now in tip/master:
> 
> ----------------------------------------->
> pci-y                           := fixup.o
> pci-$(CONFIG_ACPI)              += acpi.o
> pci-y                           += legacy.o irq.o
> 
> pci-$(CONFIG_X86_VISWS)         += visws.o
> 
> pci-$(CONFIG_X86_NUMAQ)         += numa.o
> 
> obj-y                           += $(pci-y) common.o early.o
> obj-y                           += amd_bus.o
> <-----------------------------------------

So if I am not mistaken you can kill pci-y entirely:

obj-y                           := fixup.o
obj-$(CONFIG_ACPI)              += acpi.o
obj-y                           += legacy.o irq.o

obj-$(CONFIG_X86_VISWS)         += visws.o

obj-$(CONFIG_X86_NUMAQ)         += numa.o

obj-y                           += common.o early.o
obj-y                           += amd_bus.o

	Sam

Re: [PATCH 0/5] x86/pci: patches for tip/x86/unify-pci

[Ingo Molnar]

* Sam Ravnborg <sam@ravnborg.org> wrote:

> On Thu, Jul 10, 2008 at 08:12:34PM +0200, Ingo Molnar wrote:
> > 
> > * Ingo Molnar <mingo@elte.hu> wrote:
> > 
> > > * Robert Richter <robert.richter@amd.com> wrote:
> > > 
> > > > This patch series contains patches for tip/x86/unify-pci. It fixes 
> > > > NUMA initialization code reflecting latest changes in tip/master. 
> > > > Also there are build fixes and minor cleanups. Patch #5 is a cherry 
> > > > pick that fixes build on UP systems. Code reboots for smp/nosmp, 
> > > > acpi/noacpi, numaq, 32/64bit. It compiles for visws.
> > > 
> > > thanks Robert. Today i've eliminated the VISWS subarch which greatly 
> > > simplified your series. I've picked up these two fixes:
> > > 
> > >  Robert Richter (2):
> > >        x86/pci merge: fixing numaq initialization
> > >        x86/pci: fix warnings in subsys_initcall functions
> > 
> > btw., this is how arch/x86/pci/Makefile looks like now in tip/master:
> > 
> > ----------------------------------------->
> > pci-y                           := fixup.o
> > pci-$(CONFIG_ACPI)              += acpi.o
> > pci-y                           += legacy.o irq.o
> > 
> > pci-$(CONFIG_X86_VISWS)         += visws.o
> > 
> > pci-$(CONFIG_X86_NUMAQ)         += numa.o
> > 
> > obj-y                           += $(pci-y) common.o early.o
> > obj-y                           += amd_bus.o
> > <-----------------------------------------
> 
> So if I am not mistaken you can kill pci-y entirely:
> 
> obj-y                           := fixup.o
> obj-$(CONFIG_ACPI)              += acpi.o
> obj-y                           += legacy.o irq.o
> 
> obj-$(CONFIG_X86_VISWS)         += visws.o
> 
> obj-$(CONFIG_X86_NUMAQ)         += numa.o
> 
> obj-y                           += common.o early.o
> obj-y                           += amd_bus.o

ah, indeed :-) Very nice.

	Ingo

Re: [PATCH 0/5] x86/pci: patches for tip/x86/unify-pci

[Robert Richter]

On 10.07.08 21:54:25, Ingo Molnar wrote:
> 
> * Sam Ravnborg <sam@ravnborg.org> wrote:
> 
> > On Thu, Jul 10, 2008 at 08:12:34PM +0200, Ingo Molnar wrote:
> > > 
> > > * Ingo Molnar <mingo@elte.hu> wrote:
> > > 
> > > > * Robert Richter <robert.richter@amd.com> wrote:
> > > > 
> > > > > This patch series contains patches for tip/x86/unify-pci. It fixes 
> > > > > NUMA initialization code reflecting latest changes in tip/master. 
> > > > > Also there are build fixes and minor cleanups. Patch #5 is a cherry 
> > > > > pick that fixes build on UP systems. Code reboots for smp/nosmp, 
> > > > > acpi/noacpi, numaq, 32/64bit. It compiles for visws.
> > > > 
> > > > thanks Robert. Today i've eliminated the VISWS subarch which greatly 
> > > > simplified your series. I've picked up these two fixes:
> > > > 
> > > >  Robert Richter (2):
> > > >        x86/pci merge: fixing numaq initialization
> > > >        x86/pci: fix warnings in subsys_initcall functions
> > > 
> > > btw., this is how arch/x86/pci/Makefile looks like now in tip/master:
> > > 
> > > ----------------------------------------->
> > > pci-y                           := fixup.o
> > > pci-$(CONFIG_ACPI)              += acpi.o
> > > pci-y                           += legacy.o irq.o
> > > 
> > > pci-$(CONFIG_X86_VISWS)         += visws.o
> > > 
> > > pci-$(CONFIG_X86_NUMAQ)         += numa.o
> > > 
> > > obj-y                           += $(pci-y) common.o early.o
> > > obj-y                           += amd_bus.o
> > > <-----------------------------------------
> > 
> > So if I am not mistaken you can kill pci-y entirely:

Yes, but...

> > 
> > obj-y                           := fixup.o

... this is with a plus:

obj-y                           += fixup.o

> > obj-$(CONFIG_ACPI)              += acpi.o
> > obj-y                           += legacy.o irq.o
> > 
> > obj-$(CONFIG_X86_VISWS)         += visws.o
> > 
> > obj-$(CONFIG_X86_NUMAQ)         += numa.o
> > 
> > obj-y                           += common.o early.o
> > obj-y                           += amd_bus.o
> 
> ah, indeed :-) Very nice.
> 
> 	Ingo
> 

-- 
Advanced Micro Devices, Inc.
Operating System Research Center
email: robert.richter@amd.com

Re: [PATCH 0/5] x86/pci: patches for tip/x86/unify-pci

[Sam Ravnborg]

> > > 
> > > So if I am not mistaken you can kill pci-y entirely:
> 
> Yes, but...
> 
> > > 
> > > obj-y                           := fixup.o
> 
> ... this is with a plus:
> 
> obj-y                           += fixup.o

obj-y is empty when you enter the Makefile.
So if you start out with ':=' or '+=' is no difference.

	Sam

Re: [PATCH 0/5] x86/pci: patches for tip/x86/unify-pci

[Ingo Molnar]

* Sam Ravnborg <sam@ravnborg.org> wrote:

> > > > 
> > > > So if I am not mistaken you can kill pci-y entirely:
> > 
> > Yes, but...
> > 
> > > > 
> > > > obj-y                           := fixup.o
> > 
> > ... this is with a plus:
> > 
> > obj-y                           += fixup.o
> 
> obj-y is empty when you enter the Makefile.
> So if you start out with ':=' or '+=' is no difference.

i guess it's safer to use +=, because it's ordering invariant.

We should use := only if we absolutely want to override the current 
rules.

	Ingo

Re: [PATCH 0/5] x86/pci: patches for tip/x86/unify-pci

[Sam Ravnborg]

> > 
> > obj-y is empty when you enter the Makefile.
> > So if you start out with ':=' or '+=' is no difference.
> 
> i guess it's safer to use +=, because it's ordering invariant.
> 
> We should use := only if we absolutely want to override the current 
> rules.

In the past I was carefully to use ':=' for the first assignmnet so the
variable was a " simply expanded variables" variable.
But as I have added a 'obj-y := ' in kbuild we should
be safe.

" simply expanded variables" is some special make stuff.

A typical:

FOO = bar

is a "recursively expanded variable".
Here the assingment is evaluated on each usage.

	Sam

Re: [PATCH 0/5] x86/pci: patches for tip/x86/unify-pci

[Robert Richter]

On 10.07.08 20:12:34, Ingo Molnar wrote:
> 
> * Ingo Molnar <mingo@elte.hu> wrote:
> 
> > * Robert Richter <robert.richter@amd.com> wrote:
> > 
> > > This patch series contains patches for tip/x86/unify-pci. It fixes 
> > > NUMA initialization code reflecting latest changes in tip/master. 
> > > Also there are build fixes and minor cleanups. Patch #5 is a cherry 
> > > pick that fixes build on UP systems. Code reboots for smp/nosmp, 
> > > acpi/noacpi, numaq, 32/64bit. It compiles for visws.
> > 
> > thanks Robert. Today i've eliminated the VISWS subarch which greatly 
> > simplified your series. I've picked up these two fixes:
> > 
> >  Robert Richter (2):
> >        x86/pci merge: fixing numaq initialization
> >        x86/pci: fix warnings in subsys_initcall functions
> 
> btw., this is how arch/x86/pci/Makefile looks like now in tip/master:
> 
> ----------------------------------------->
> pci-y                           := fixup.o
> pci-$(CONFIG_ACPI)              += acpi.o
> pci-y                           += legacy.o irq.o
> 
> pci-$(CONFIG_X86_VISWS)         += visws.o
> 
> pci-$(CONFIG_X86_NUMAQ)         += numa.o
> 
> obj-y                           += $(pci-y) common.o early.o
> obj-y                           += amd_bus.o
> <-----------------------------------------
> 
> it's _really_ clean now, and we dont have those weird overrides anymore. 
> Both VisWS and NUMAQ are now two straightforward runtime PCI overrides - 
> not build-time complications. Much more maintainable i think.

Yes, it really looks nice now.

-Robert

-- 
Advanced Micro Devices, Inc.
Operating System Research Center
email: robert.richter@amd.com