[tip:x86/irq] x86, acpi/irq: Teach acpi_get_override_irq to take a gsi not an isa_irq

view
thread

Previous thread: [PATCH 04/10] x86: kill smpboot_hooks.h by Yinghai Lu on Sunday, March 21, 2010 - 6:36 pm. (2 messages)

Next thread: linux-next: manual merge of the libata tree with Linus' tree by Stephen Rothwell on Sunday, March 21, 2010 - 6:44 pm. (3 messages)

[view original message]

From: Yinghai Lu

Subject: [PATCH 00/10] x86/irq

Date: Sunday, March 21, 2010 - 6:36 pm

only irq related...

Thanks

Yinghai
--

[ message continues ]

[view original message]

From: Yinghai Lu

Subject: [PATCH 08/10] genericirq: add set_irq_desc_chip/data ...

Date: Sunday, March 21, 2010 - 6:36 pm

with set_irq_chip/data...

to take desc instead of irq

Signed-off-by: Yinghai Lu &lt;yinghai@kernel.org&gt;
---
 arch/x86/kernel/apic/io_apic.c |    4 +-
 include/linux/irq.h            |   10 +++-
 include/linux/irqnr.h          |    1 +
 kernel/irq/chip.c              |  127 +++++++++++++++++++++++-----------------
 4 files changed, 85 insertions(+), 57 deletions(-)

diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
index f908af5..432bea1 100644
--- a/arch/x86/kernel/apic/io_apic.c
+++ b/arch/x86/kernel/apic/io_apic.c
@@ -3302,7 +3302,7 @@ unsigned int create_irq_nr(unsigned int irq_want, int node)
 	raw_spin_unlock_irqrestore(&amp;vector_lock, flags);
 
 	if (irq &gt; 0)
-		dynamic_irq_init_keep_chip_data(irq);
+		dynamic_irq_init_keep_chip_data(irq_to_desc(irq));
 
 	return irq;
 }
@@ -3328,7 +3328,7 @@ void destroy_irq(unsigned int irq)
 	struct irq_desc *desc;
 	struct irq_cfg *cfg;
 
-	dynamic_irq_cleanup_keep_chip_data(irq);
+	dynamic_irq_cleanup_keep_chip_data(irq_to_desc(irq));
 
 	free_irte(irq);
 	raw_spin_lock_irqsave(&amp;vector_lock, flags);
diff --git a/include/linux/irq.h b/include/linux/irq.h
index 36ea6ac..1f5d112 100644
--- a/include/linux/irq.h
+++ b/include/linux/irq.h
@@ -420,9 +420,9 @@ static inline int irq_has_action(unsigned int irq)
 
 /* Dynamic irq helper functions */
 extern void dynamic_irq_init(unsigned int irq);
-void dynamic_irq_init_keep_chip_data(unsigned int irq);
+void dynamic_irq_init_keep_chip_data(struct irq_desc *desc);
 extern void dynamic_irq_cleanup(unsigned int irq);
-void dynamic_irq_cleanup_keep_chip_data(unsigned int irq);
+void dynamic_irq_cleanup_keep_chip_data(struct irq_desc *desc);
 
 /* Set/get chip/data for an IRQ: */
 extern int set_irq_chip(unsigned int irq, struct irq_chip *chip);
@@ -431,6 +431,12 @@ extern int set_irq_chip_data(unsigned int irq, void *data);
 extern int set_irq_type(unsigned int irq, unsigned int type);
 extern int set_irq_msi(unsigned int irq, struct msi_desc ...[ message continues ]

[view original message]

From: Yinghai Lu

Subject: [PATCH 10/10] x86: remove arch_probe_nr_irqs

Date: Sunday, March 21, 2010 - 6:36 pm

so keep nr_irqs == NR_IRQS

Signed-off-by: Yinghai Lu &lt;yinghai@kernel.org&gt;
---
 arch/x86/kernel/apic/io_apic.c |   22 ----------------------
 1 files changed, 0 insertions(+), 22 deletions(-)

diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
index ce93428..b9c4749 100644
--- a/arch/x86/kernel/apic/io_apic.c
+++ b/arch/x86/kernel/apic/io_apic.c
@@ -3921,28 +3921,6 @@ void __init probe_nr_irqs_gsi(void)
 	printk(KERN_DEBUG &quot;nr_irqs_gsi: %d\n&quot;, nr_irqs_gsi);
 }
 
-#ifdef CONFIG_SPARSE_IRQ
-int __init arch_probe_nr_irqs(void)
-{
-	int nr;
-
-	if (nr_irqs &gt; (NR_VECTORS * nr_cpu_ids))
-		nr_irqs = NR_VECTORS * nr_cpu_ids;
-
-	nr = nr_irqs_gsi + 8 * nr_cpu_ids;
-#if defined(CONFIG_PCI_MSI) || defined(CONFIG_HT_IRQ)
-	/*
-	 * for MSI and HT dyn irq
-	 */
-	nr += nr_irqs_gsi * 16;
-#endif
-	if (nr &lt; nr_irqs)
-		nr_irqs = nr;
-
-	return 0;
-}
-#endif
-
 static int __io_apic_set_pci_routing(struct device *dev, int irq,
 				struct io_apic_irq_attr *irq_attr)
 {
-- 
1.6.4.2

--

[ message continues ]

[view original message]

From: Yinghai Lu

Subject: [PATCH 02/10] x86: fix out of order of gsi - full

Date: Sunday, March 21, 2010 - 6:36 pm

Iranna D Ankad reported that IBM x3950 systems have boot problems
after this commit:

 |
 | commit b9c61b70075c87a8612624736faf4a2de5b1ed30
 |
 |    x86/pci: update pirq_enable_irq() to setup io apic routing
 |
  ACPI: IOAPIC (id[0x10] address[0xfecff000] gsi_base[0])
  IOAPIC[0]: apic_id 16, version 0, address 0xfecff000, GSI 0-2
  ACPI: IOAPIC (id[0x0f] address[0xfec00000] gsi_base[3])
  IOAPIC[1]: apic_id 15, version 0, address 0xfec00000, GSI 3-38
  ACPI: IOAPIC (id[0x0e] address[0xfec01000] gsi_base[39])
  IOAPIC[2]: apic_id 14, version 0, address 0xfec01000, GSI 39-74

  As explained in the previous patch (&quot;x86: Fix out of order gsi)

  need to remap those gsis

  This patch adds boot_ioapic_idx and gsi_to_irq/irq_to_gsi
  So we could make sure for those kind of system will have
        irq: 0 - 15 for legacy irq
        irq:  16 after will be gsi + 16

 -v13: move gsi_to_irq/irq_to_gsi to acpi/boot.c

Reported-by: Iranna D Ankad &lt;iranna.ankad@in.ibm.com&gt;
Bisected-by: Iranna D Ankad &lt;iranna.ankad@in.ibm.com&gt;
Tested-by: Gary Hade &lt;garyhade@us.ibm.com&gt;
Signed-off-by: Yinghai Lu &lt;yinghai@kernel.org&gt;
Cc: Thomas Renninger &lt;trenn@suse.de&gt;
Cc: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
Cc: Suresh Siddha &lt;suresh.b.siddha@intel.com&gt;
Cc: len.brown@intel.com
---
 arch/x86/include/asm/io_apic.h |    2 +-
 arch/x86/include/asm/mpspec.h  |   14 ++++++++
 arch/x86/kernel/acpi/boot.c    |   64 +++++++++++++++++++++++++++++------
 arch/x86/kernel/apic/io_apic.c |   72 +++++++++++++++++++++++++++++----------
 drivers/pnp/pnpacpi/rsparser.c |    4 ++
 5 files changed, 125 insertions(+), 31 deletions(-)

diff --git a/arch/x86/include/asm/io_apic.h b/arch/x86/include/asm/io_apic.h
index 35832a0..c4683b9 100644
--- a/arch/x86/include/asm/io_apic.h
+++ b/arch/x86/include/asm/io_apic.h
@@ -158,7 +158,7 @@ extern int io_apic_get_redir_entries(int ioapic);
 struct io_apic_irq_attr;
 extern int io_apic_set_pci_routing(struct device *dev, int irq,
 		 struct io_apic_irq_attr ...[ message continues ]

[view original message]

From: Thomas Gleixner

Subject: Re: [PATCH 02/10] x86: fix out of order of gsi - full

Date: Monday, March 22, 2010 - 4:14 am

Please do not propagate that pointer.

    	 *irq = setup_IO_APIC_irq_extra(gsi);

  Also these changes have a total lack of comments. Why do we modify

  Please make this static and provide a function to modify it from
  probe_ioapic_i8259().

  Also the variable name is horrible. What's the delta here ? It's an


  Please make that:



  Why is mp_register_gsi global ? It's only used in



  Can we please do a cleanup of irq variables to use unsigned

  Can we simply return mp_irqs[idx].srcbusirq here and get rid of the


  boot_ioapic_idx is an apic id. Why is the variable name suggesting

  How is this change related to this patch ? It looks more like an
  independent fix.

Thanks,

	tglx
--

[ message continues ]

[view original message]

From: Yinghai Lu

Subject: Re: [PATCH 02/10] x86: fix out of order of gsi - full

Date: Monday, March 22, 2010 - 12:45 pm

yes.

should adjust that function position.

arch/x86/kernel/acpi/boot.c:            plat_gsi = mp_register_gsi(dev, gsi, trigger, polarity);



ok another patch.

YH
--

[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: Re: [PATCH 02/10] x86: fix out of order of gsi - full

Date: Monday, March 29, 2010 - 6:40 am

I have played with this a bit and I have a cleaner patchset that
addresses this issue.  I plan to sleep on it, and then go through it
again before I send it.

Thomas is the procedure for working on this stuff write clean patches
and send them, and they will land in an x86 topic branch?

Eric
--

[ message continues ]

[view original message]

From: H. Peter Anvin

Subject: Re: [PATCH 02/10] x86: fix out of order of gsi - full

Date: Monday, March 29, 2010 - 10:57 am

Pretty much, yes.

	-hpa
--

[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: [PATCH 0/14] Start coping gsis < 16 that are not isa irqs.

Date: Monday, March 29, 2010 - 4:19 pm

It appears there are a few systems in the wild that use acpi
interrupt source overrides to report a gsi &gt; 16 is an isa irq.

This breaks all kinds of assumptions I figure any BIOS doing that
probably should be shot as that is very much not a conservative position.

That said acpi appears to allow this insanity and on the basis of being
liberal in what we accept we should try to do something reasonable, and
keep our notions of gsi, isa irq, and linux irq as straight as we can.

To that end this patchset slightly modifies the gsi to linux irq
mapping.  Making it a 1 to 1 identity mapping except for the first 16
gsis.  If those gsis are isa irqs (the typical and default acpi
configuration) they will be mapped into the first 16 irqs.  If those
first 16 gsis are not isa irqs they will be given a linux irq number
just past the last gsi.  Allowing us the chance to use them.

This patchset is my attempt to straighten out our understanding of which
kind of irq name is used where and the cleanups used to get there.

Thanks to Yinghai Lu &lt;yinghai@kernel.org&gt; for taking the first stab at
this and finding that there is a real world problem here.  This patchset
is inspired by his work, but little of it remains the same.

This patchset lies in the weird world between the acpi and x86 and ia64.
Since most of the changes are x86 related I think it makes most sense
to go via the x86 tree.


Eric W. Biederman (14):
      x86 acpi/irq: Introduce apci_isa_irq_to_gsi
      x86 acpi/irq: Teach acpi_get_override_irq to take a gsi not an isa_irq
      x86 acpi/irq: pci device dev-&gt;irq is an isa irq not a gsi
      x86 acpi/irq: Fix acpi_sci_ioapic_setup so it has both bus_irq and gsi
      x86 acpi/irq: Generalize mp_config_acpi_legacy_irqs
      x86 ioapic: Only export mp_find_ioapic and mp_find_ioapic_pin in io_apic.h
      x86 ioapic: Fix the types of gsi values
      x86 ioapic: Teach mp_register_ioapic to compute a global gsi_end
      x86 ioapic: In mpparse use mp_register_ioapic
      ...[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: [PATCH 03/14] x86 acpi/irq: pci device dev->irq is an isa irq not a gsi

Date: Monday, March 29, 2010 - 4:20 pm

From: Eric W. Biederman &lt;ebiederm@xmission.com&gt;

Strictly speaking on x86 (where acpi is used) dev-&gt;irq must be
a dual i8259 irq input aka an isa irq.  Therefore we should translate
that isa irq into a gsi before passing it to a function that
takes a gsi.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
---
 drivers/acpi/pci_irq.c |    8 +++++---
 1 files changed, 5 insertions(+), 3 deletions(-)

diff --git a/drivers/acpi/pci_irq.c b/drivers/acpi/pci_irq.c
index 843699e..703541d 100644
--- a/drivers/acpi/pci_irq.c
+++ b/drivers/acpi/pci_irq.c
@@ -400,11 +400,13 @@ int acpi_pci_irq_enable(struct pci_dev *dev)
 	 * driver reported one, then use it. Exit in any case.
 	 */
 	if (gsi &lt; 0) {
+		u32 dev_gsi;
 		dev_warn(&amp;dev-&gt;dev, &quot;PCI INT %c: no GSI&quot;, pin_name(pin));
 		/* Interrupt Line values above 0xF are forbidden */
-		if (dev-&gt;irq &gt; 0 &amp;&amp; (dev-&gt;irq &lt;= 0xF)) {
-			printk(&quot; - using IRQ %d\n&quot;, dev-&gt;irq);
-			acpi_register_gsi(&amp;dev-&gt;dev, dev-&gt;irq,
+		if (dev-&gt;irq &gt; 0 &amp;&amp; (dev-&gt;irq &lt;= 0xF) &amp;&amp;
+		    (acpi_isa_irq_to_gsi(dev-&gt;irq, &amp;dev_gsi) == 0)) {
+			printk(&quot; - using ISA IRQ %d\n&quot;, dev-&gt;irq);
+			acpi_register_gsi(&amp;dev-&gt;dev, dev_gsi,
 					  ACPI_LEVEL_SENSITIVE,
 					  ACPI_ACTIVE_LOW);
 			return 0;
-- 
1.6.5.2.143.g8cc62

--

[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: [PATCH 07/14] x86 ioapic: Fix the types of gsi values

Date: Monday, March 29, 2010 - 4:20 pm

From: Eric W. Biederman &lt;ebiederm@xmission.com&gt;

This patches fixes the types of gsi_base and gsi_end values in
struct mp_ioapic_gsi, and the gsi parameter of mp_find_ioapic
and mp_find_ioapic_pin

A gsi is cannonically a u32, not an int.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
---
 arch/x86/include/asm/io_apic.h |   10 +++++-----
 arch/x86/kernel/apic/io_apic.c |    4 ++--
 2 files changed, 7 insertions(+), 7 deletions(-)

diff --git a/arch/x86/include/asm/io_apic.h b/arch/x86/include/asm/io_apic.h
index 35832a0..feeaf0d 100644
--- a/arch/x86/include/asm/io_apic.h
+++ b/arch/x86/include/asm/io_apic.h
@@ -180,12 +180,12 @@ extern void ioapic_write_entry(int apic, int pin,
 extern void setup_ioapic_ids_from_mpc(void);
 
 struct mp_ioapic_gsi{
-	int gsi_base;
-	int gsi_end;
+	u32 gsi_base;
+	u32 gsi_end;
 };
 extern struct mp_ioapic_gsi  mp_gsi_routing[];
-int mp_find_ioapic(int gsi);
-int mp_find_ioapic_pin(int ioapic, int gsi);
+int mp_find_ioapic(u32 gsi);
+int mp_find_ioapic_pin(int ioapic, u32 gsi);
 void __init mp_register_ioapic(int id, u32 address, u32 gsi_base);
 extern void __init pre_init_apic_IRQ0(void);
 
@@ -197,7 +197,7 @@ static const int timer_through_8259 = 0;
 static inline void ioapic_init_mappings(void)	{ }
 static inline void ioapic_insert_resources(void) { }
 static inline void probe_nr_irqs_gsi(void)	{ }
-static inline int mp_find_ioapic(int gsi) { return 0; }
+static inline int mp_find_ioapic(u32 gsi) { return 0; }
 
 struct io_apic_irq_attr;
 static inline int io_apic_set_pci_routing(struct device *dev, int irq,
diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
index 54ba1f7..32cea57 100644
--- a/arch/x86/kernel/apic/io_apic.c
+++ b/arch/x86/kernel/apic/io_apic.c
@@ -4234,7 +4234,7 @@ void __init ioapic_insert_resources(void)
 	}
 }
 
-int mp_find_ioapic(int gsi)
+int mp_find_ioapic(u32 gsi)
 {
 	int i = 0;
 
@@ -4249,7 +4249,7 @@ int mp_find_ioapic(int gsi)
 	return -1;
 }
 
-int ...[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: [PATCH 12/14] x86 ioapic: Simplify probe_nr_irqs_gsi.

Date: Monday, March 29, 2010 - 4:20 pm

From: Eric W. Biederman &lt;ebiederm@xmission.com&gt;

Use the global gsi_end value now that all ioapics have
valid gsi numbers instead of a combination of acpi_probe_gsi
and walking all of the ioapics and couting their number of
entries by hand if acpi_probe_gsi gave us an answer we did
not like.

This fixes a small bug in probe_nr_irqs_gsi.  Previously
acpi_probe_gsi unnecessarily added 1 to the maximum
gsi_end value.  gsi_end is already one past the end of
the number of gsi's so the additional increment was
superfluous.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
---
 arch/x86/include/asm/mpspec.h  |    6 ------
 arch/x86/kernel/acpi/boot.c    |   23 -----------------------
 arch/x86/kernel/apic/io_apic.c |   17 +++--------------
 3 files changed, 3 insertions(+), 43 deletions(-)

diff --git a/arch/x86/include/asm/mpspec.h b/arch/x86/include/asm/mpspec.h
index 29994f0..c82868e 100644
--- a/arch/x86/include/asm/mpspec.h
+++ b/arch/x86/include/asm/mpspec.h
@@ -105,12 +105,6 @@ extern void mp_config_acpi_legacy_irqs(void);
 struct device;
 extern int mp_register_gsi(struct device *dev, u32 gsi, int edge_level,
 				 int active_high_low);
-extern int acpi_probe_gsi(void);
-#else /* !CONFIG_ACPI: */
-static inline int acpi_probe_gsi(void)
-{
-	return 0;
-}
 #endif /* CONFIG_ACPI */
 
 #define PHYSID_ARRAY_SIZE	BITS_TO_LONGS(MAX_APICS)
diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
index 9c48e99..0e514a1 100644
--- a/arch/x86/kernel/acpi/boot.c
+++ b/arch/x86/kernel/acpi/boot.c
@@ -875,29 +875,6 @@ static int __init acpi_parse_madt_lapic_entries(void)
 extern int es7000_plat;
 #endif
 
-int __init acpi_probe_gsi(void)
-{
-	int idx;
-	int gsi;
-	int max_gsi = 0;
-
-	if (acpi_disabled)
-		return 0;
-
-	if (!acpi_ioapic)
-		return 0;
-
-	max_gsi = 0;
-	for (idx = 0; idx &lt; nr_ioapics; idx++) {
-		gsi = mp_gsi_routing[idx].gsi_end;
-
-		if (gsi &gt; max_gsi)
-			max_gsi = gsi;
-	}
-
-	return max_gsi + 1;
-}
-
 static ...[ message continues ]

[view original message]

From: Yinghai Lu

Subject: Re: [PATCH 12/14] x86 ioapic: Simplify probe_nr_irqs_gsi.

Date: Monday, March 29, 2010 - 7:16 pm

On Mon, Mar 29, 2010 at 4:20 PM, Eric W. Biederman

you may need +1 here

--

[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: Re: [PATCH 12/14] x86 ioapic: Simplify probe_nr_irqs_gsi.

Date: Monday, March 29, 2010 - 9:43 pm

As documented in my comment that extra +1 has every appearance of a
bug.  Nothing is at gsi_end.  gsi_end is already at 1 past the last in
use gsi.  Therefore an extra +1 puts us two past the end for no
apparent reason.

Eric
--

[ message continues ]

[view original message]

From: Yinghai Lu

Subject: Re: [PATCH 12/14] x86 ioapic: Simplify probe_nr_irqs_gsi.

Date: Monday, March 29, 2010 - 9:55 pm

io_apic_get_redir_entries(), and io apic register readingout will return 23 if the total entries is 24.

[    0.000000] ACPI: IOAPIC (id[0x08] address[0xfec00000] gsi_base[0])
[    0.000000] IOAPIC[0]: apic_id 8, version 32, address 0xfec00000, GSI 0-23
[    0.000000] ACPI: IOAPIC (id[0x09] address[0xfec02000] gsi_base[24])
[    0.000000] IOAPIC[1]: apic_id 9, version 32, address 0xfec02000, GSI 24-47
[    0.000000] ACPI: IOAPIC (id[0x0a] address[0xfec04000] gsi_base[48])
[    0.000000] IOAPIC[2]: apic_id 10, version 32, address 0xfec04000, GSI 48-71
[    0.000000] ACPI: IOAPIC (id[0x0b] address[0xfec08000] gsi_base[72])
[    0.000000] IOAPIC[3]: apic_id 11, version 32, address 0xfec08000, GSI 72-95
[    0.000000] ACPI: IOAPIC (id[0x0c] address[0xfec0c000] gsi_base[96])
[    0.000000] IOAPIC[4]: apic_id 12, version 32, address 0xfec0c000, GSI 96-119

        mp_gsi_routing[idx].gsi_end = gsi_base +
            io_apic_get_redir_entries(idx);

        printk(KERN_INFO &quot;IOAPIC[%d]: apic_id %d, version %d, address 0x%x, &quot;
               &quot;GSI %d-%d\n&quot;, idx, mp_ioapics[idx].apicid,
               mp_ioapics[idx].apicver, mp_ioapics[idx].apicaddr,
               mp_gsi_routing[idx].gsi_base, mp_gsi_routing[idx].gsi_end);

YH
--

[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: Re: [PATCH 12/14] x86 ioapic: Simplify probe_nr_irqs_gsi.

Date: Monday, March 29, 2010 - 10:41 pm

Good catch.  I don't know if I have ever seen a function with a name to function
correspondence.  Apparently sfi.c also got this wrong when it was written.
and I missed the +1 in other places.

I guess this calls for another patch way at the beginning of my series that fixes
this brain damage.

Eric
--

[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: [PATCH 01/14] x86 acpi/irq: Introduce apci_isa_irq_to_gsi

Date: Monday, March 29, 2010 - 4:20 pm

From: Eric W. Biederman &lt;ebiederm@xmission.com&gt;

There are a number of cases where the current code makes the assumption
that isa irqs identity map to the first 16 acpi global system intereupts.
In most instances that assumption is correct as that is the required
behaviour in dual i8259 mode and the default behavior in ioapic mode.

However there are some systems out there that take advantage of acpis
interrupt remapping  for the isa irqs to have a completely different
mapping of isa_irq to gsi.

Introduce acpi_isa_irq_to_gsi to perform this mapping explicitly in the
code that needs it.  Initially this will be just the current assumed
identity mapping to ensure it's introduction does not cause regressions.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
---
 arch/ia64/kernel/acpi.c     |    8 ++++++++
 arch/x86/kernel/acpi/boot.c |    8 ++++++++
 include/linux/acpi.h        |    1 +
 3 files changed, 17 insertions(+), 0 deletions(-)

diff --git a/arch/ia64/kernel/acpi.c b/arch/ia64/kernel/acpi.c
index f1c9f70..16921bd 100644
--- a/arch/ia64/kernel/acpi.c
+++ b/arch/ia64/kernel/acpi.c
@@ -784,6 +784,14 @@ int acpi_gsi_to_irq(u32 gsi, unsigned int *irq)
 	return 0;
 }
 
+int acpi_isa_irq_to_gsi(unsigned isa_irq, u32 *gsi)
+{
+	if (isa_irq &gt;= 16)
+		return -1;
+	*gsi = isa_irq;
+	return 0;
+}
+
 /*
  *  ACPI based hotplug CPU support
  */
diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
index 0061ea2..f96060d 100644
--- a/arch/x86/kernel/acpi/boot.c
+++ b/arch/x86/kernel/acpi/boot.c
@@ -457,6 +457,14 @@ int acpi_gsi_to_irq(u32 gsi, unsigned int *irq)
 	return 0;
 }
 
+int acpi_isa_irq_to_gsi(unsigned isa_irq, u32 *gsi)
+{
+	if (isa_irq &gt;= 16)
+		return -1;
+	*gsi = isa_irq;
+	return 0;
+}
+
 /*
  * success: return IRQ number (&gt;=0)
  * failure: return &lt; 0
diff --git a/include/linux/acpi.h b/include/linux/acpi.h
index b926afe..7a937da 100644
--- a/include/linux/acpi.h
+++ b/include/linux/acpi.h
@@ -116,6 +116,7 @@ ...[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: [PATCH 06/14] x86 ioapic: Only export mp_find_ioapic and mp_find_ioapic_pin in io_apic.h

Date: Monday, March 29, 2010 - 4:20 pm

From: Eric W. Biederman &lt;ebiederm@xmission.com&gt;

Multiple declarations of the same function in different headers
is a pain to maintain.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
---
 arch/x86/include/asm/mpspec.h |    4 ----
 1 files changed, 0 insertions(+), 4 deletions(-)

diff --git a/arch/x86/include/asm/mpspec.h b/arch/x86/include/asm/mpspec.h
index d8bf23a..29994f0 100644
--- a/arch/x86/include/asm/mpspec.h
+++ b/arch/x86/include/asm/mpspec.h
@@ -106,10 +106,6 @@ struct device;
 extern int mp_register_gsi(struct device *dev, u32 gsi, int edge_level,
 				 int active_high_low);
 extern int acpi_probe_gsi(void);
-#ifdef CONFIG_X86_IO_APIC
-extern int mp_find_ioapic(int gsi);
-extern int mp_find_ioapic_pin(int ioapic, int gsi);
-#endif
 #else /* !CONFIG_ACPI: */
 static inline int acpi_probe_gsi(void)
 {
-- 
1.6.5.2.143.g8cc62

--

[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: [PATCH 14/14] x86 irq: Kill io_apic_renumber_irq

Date: Monday, March 29, 2010 - 4:20 pm

From: Eric W. Biederman &lt;ebiederm@xmission.com&gt;

Now that the generic irq layer is performing the exact same remapping as
io_apic_renumber_irq we can kill this weird  es7000 specific function.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
---
 arch/x86/include/asm/io_apic.h   |    1 -
 arch/x86/kernel/acpi/boot.c      |    5 -----
 arch/x86/kernel/apic/es7000_32.c |   19 -------------------
 arch/x86/kernel/apic/io_apic.c   |    6 ------
 4 files changed, 0 insertions(+), 31 deletions(-)

diff --git a/arch/x86/include/asm/io_apic.h b/arch/x86/include/asm/io_apic.h
index 37b0f2b..9da192a 100644
--- a/arch/x86/include/asm/io_apic.h
+++ b/arch/x86/include/asm/io_apic.h
@@ -159,7 +159,6 @@ struct io_apic_irq_attr;
 extern int io_apic_set_pci_routing(struct device *dev, int irq,
 		 struct io_apic_irq_attr *irq_attr);
 void setup_IO_APIC_irq_extra(u32 gsi);
-extern int (*ioapic_renumber_irq)(int ioapic, int irq);
 extern void ioapic_init_mappings(void);
 extern void ioapic_insert_resources(void);
 
diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
index b372aba..a48dfa9 100644
--- a/arch/x86/kernel/acpi/boot.c
+++ b/arch/x86/kernel/acpi/boot.c
@@ -1116,11 +1116,6 @@ int mp_register_gsi(struct device *dev, u32 gsi, int trigger, int polarity)
 
 	ioapic_pin = mp_find_ioapic_pin(ioapic, gsi);
 
-#ifdef CONFIG_X86_32
-	if (ioapic_renumber_irq)
-		gsi = ioapic_renumber_irq(ioapic, gsi);
-#endif
-
 	if (ioapic_pin &gt; MP_MAX_IOAPIC_PIN) {
 		printk(KERN_ERR &quot;Invalid reference to IOAPIC pin &quot;
 		       &quot;%d-%d\n&quot;, mp_ioapics[ioapic].apicid,
diff --git a/arch/x86/kernel/apic/es7000_32.c b/arch/x86/kernel/apic/es7000_32.c
index dd2b5f2..7a63d47 100644
--- a/arch/x86/kernel/apic/es7000_32.c
+++ b/arch/x86/kernel/apic/es7000_32.c
@@ -130,24 +130,6 @@ int					es7000_plat;
 
 static unsigned int			base;
 
-static int
-es7000_rename_gsi(int ioapic, int gsi)
-{
-	if (es7000_plat == ES7000_ZORRO)
-		return gsi;
-
-	if (!base) {
-		int i;
-		for ...[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: [PATCH 13/14] x86 acpi/irq: Handle isa irqs that are not identity mapped to gsi's.

Date: Monday, March 29, 2010 - 4:20 pm

From: Eric W. Biederman &lt;ebiederm@xmission.com&gt;

ACPI irq source overrides are allowed for the 16 isa irqs and are
allowed to map any gsi to any isa irq.  A few motherboards have been
seen to take advantage of this and put the isa irqs on the 2nd or
3rd ioapic.  This causes some problems, most notably the fact
that we can not use any gsi &lt; 16.

To correct this move the gsis that are not isa irqs and have
a gsi number &lt; 16 into the linux irq space just past gsi_end.
This is what the es7000 platform is doing today.  Moving only the
low 16 gsis above the rest of the gsi's only penalizes weird
platforms, leaving sane acpi implementations with a 1-1 mapping
of gsis and irqs.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
---
 arch/x86/kernel/acpi/boot.c    |   57 +++++++++++++++++++++++++++++++++++++---
 arch/x86/kernel/apic/io_apic.c |    8 ++++-
 2 files changed, 59 insertions(+), 6 deletions(-)

diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
index 0e514a1..b372aba 100644
--- a/arch/x86/kernel/acpi/boot.c
+++ b/arch/x86/kernel/acpi/boot.c
@@ -92,6 +92,53 @@ static u64 acpi_lapic_addr __initdata = APIC_DEFAULT_PHYS_BASE;
 enum acpi_irq_model_id acpi_irq_model = ACPI_IRQ_MODEL_PIC;
 
 
+/* 
+ * ISA irqs by default are the first 16 gsis but can be
+ * any gsi as specified by an interrupt source override.
+ */
+static u32 isa_irq_to_gsi[NR_IRQS_LEGACY] __read_mostly = {
+	0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15
+};
+
+static unsigned int gsi_to_irq(unsigned int gsi)
+{
+	unsigned int irq = gsi + NR_IRQS_LEGACY;
+	unsigned int i;
+
+	for (i = 0; i &lt; NR_IRQS_LEGACY; i++) {
+		if (isa_irq_to_gsi[i] == gsi) {
+			return i;
+		}
+	}
+
+	/* Provide an identity mapping of gsi == irq
+	 * except on truly weird platforms that have
+	 * non isa irqs in the first 16 gsis.
+	 */
+	if (gsi &gt;= NR_IRQS_LEGACY)
+		irq = gsi;
+	else
+		irq = gsi_end + gsi;
+
+	return irq;
+}
+
+static u32 irq_to_gsi(int irq)
+{
+	unsigned ...[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: [PATCH 05/14] x86 acpi/irq: Generalize mp_config_acpi_legacy_irqs

Date: Monday, March 29, 2010 - 4:20 pm

From: Eric W. Biederman &lt;ebiederm@xmission.com&gt;

Remove the assumption that there is not an override for isa irq 0.
Instead lookup the gsi and from that lookup the ioapic and pin of each
isa irq indivdually.

In general this should not have any behavioural affect but in
perverse cases this gets all of the details correct, instead of
doing something weird.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
---
 arch/x86/kernel/acpi/boot.c |   30 ++++++++++++++++++------------
 1 files changed, 18 insertions(+), 12 deletions(-)

diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
index 1742888..9c48e99 100644
--- a/arch/x86/kernel/acpi/boot.c
+++ b/arch/x86/kernel/acpi/boot.c
@@ -960,8 +960,6 @@ void __init mp_override_legacy_irq(u8 bus_irq, u8 polarity, u8 trigger, u32 gsi)
 void __init mp_config_acpi_legacy_irqs(void)
 {
 	int i;
-	int ioapic;
-	unsigned int dstapic;
 	struct mpc_intsrc mp_irq;
 
 #if defined (CONFIG_MCA) || defined (CONFIG_EISA)
@@ -982,19 +980,27 @@ void __init mp_config_acpi_legacy_irqs(void)
 #endif
 
 	/*
-	 * Locate the IOAPIC that manages the ISA IRQs (0-15).
-	 */
-	ioapic = mp_find_ioapic(0);
-	if (ioapic &lt; 0)
-		return;
-	dstapic = mp_ioapics[ioapic].apicid;
-
-	/*
 	 * Use the default configuration for the IRQs 0-15.  Unless
 	 * overridden by (MADT) interrupt source override entries.
 	 */
 	for (i = 0; i &lt; 16; i++) {
+		int ioapic, pin;
+		unsigned int dstapic;
 		int idx;
+		u32 gsi;
+
+		/* Locate the gsi that irq i maps to. */
+		if (acpi_isa_irq_to_gsi(i, &amp;gsi))
+			continue;
+
+		/*
+		 * Locate the IOAPIC that manages the ISA IRQ.
+		 */
+		ioapic = mp_find_ioapic(gsi);
+		if (ioapic &lt; 0)
+			continue;
+		pin = mp_find_ioapic_pin(ioapic, gsi);
+		dstapic = mp_ioapics[ioapic].apicid;
 
 		for (idx = 0; idx &lt; mp_irq_entries; idx++) {
 			struct mpc_intsrc *irq = mp_irqs + idx;
@@ -1004,7 +1010,7 @@ void __init mp_config_acpi_legacy_irqs(void)
 				break;
 
 			/* Do we already have a ...[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: [PATCH 11/14] x86 ioapic: Optimize pin_2_irq

Date: Monday, March 29, 2010 - 4:20 pm

From: Eric W. Biederman &lt;ebiederm@xmission.com&gt;

Now that all ioapics have valid gsi_base values use this to
accellerate pin_2_irq.  In the case of acpi this also ensures
that pin_2_irq will compute the same irq value for an ioapic
pin as acpi will.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
---
 arch/x86/kernel/apic/io_apic.c |   13 ++++---------
 1 files changed, 4 insertions(+), 9 deletions(-)

diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
index 0515f26..996cf8f 100644
--- a/arch/x86/kernel/apic/io_apic.c
+++ b/arch/x86/kernel/apic/io_apic.c
@@ -1018,7 +1018,7 @@ static inline int irq_trigger(int idx)
 int (*ioapic_renumber_irq)(int ioapic, int irq);
 static int pin_2_irq(int idx, int apic, int pin)
 {
-	int irq, i;
+	int irq;
 	int bus = mp_irqs[idx].srcbus;
 
 	/*
@@ -1030,18 +1030,13 @@ static int pin_2_irq(int idx, int apic, int pin)
 	if (test_bit(bus, mp_bus_not_pci)) {
 		irq = mp_irqs[idx].srcbusirq;
 	} else {
-		/*
-		 * PCI IRQs are mapped in order
-		 */
-		i = irq = 0;
-		while (i &lt; apic)
-			irq += nr_ioapic_registers[i++];
-		irq += pin;
+		u32 gsi = mp_gsi_routing[apic].gsi_base + pin;
 		/*
                  * For MPS mode, so far only needed by ES7000 platform
                  */
 		if (ioapic_renumber_irq)
-			irq = ioapic_renumber_irq(apic, irq);
+			gsi = ioapic_renumber_irq(apic, gsi);
+		irq = gsi;
 	}
 
 #ifdef CONFIG_X86_32
-- 
1.6.5.2.143.g8cc62

--

[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: [PATCH 10/14] x86 ioapic: Move nr_ioapic_registers calculation to mp_register_ioapic.

Date: Monday, March 29, 2010 - 4:20 pm

From: Eric W. Biederman &lt;ebiederm@xmission.com&gt;

Now that all ioapic registration happens in mp_register_ioapic we can
move the calculation of nr_ioapic_registers there from enable_IO_APIC.
The number of ioapic registers is already calucated in mp_register_ioapic
so all that really needs to be done is to save the caluclated value
in nr_ioapic_registers.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
---
 arch/x86/kernel/apic/io_apic.c |   22 ++++++++--------------
 1 files changed, 8 insertions(+), 14 deletions(-)

diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
index a979a00..0515f26 100644
--- a/arch/x86/kernel/apic/io_apic.c
+++ b/arch/x86/kernel/apic/io_apic.c
@@ -1944,20 +1944,8 @@ static struct { int pin, apic; } ioapic_i8259 = { -1, -1 };
 
 void __init enable_IO_APIC(void)
 {
-	union IO_APIC_reg_01 reg_01;
 	int i8259_apic, i8259_pin;
 	int apic;
-	unsigned long flags;
-
-	/*
-	 * The number of IO-APIC IRQ registers (== #pins):
-	 */
-	for (apic = 0; apic &lt; nr_ioapics; apic++) {
-		raw_spin_lock_irqsave(&amp;ioapic_lock, flags);
-		reg_01.raw = io_apic_read(apic, 1);
-		raw_spin_unlock_irqrestore(&amp;ioapic_lock, flags);
-		nr_ioapic_registers[apic] = reg_01.bits.entries+1;
-	}
 
 	if (!legacy_pic-&gt;nr_legacy_irqs)
 		return;
@@ -4280,6 +4268,7 @@ static int bad_ioapic(unsigned long address)
 void __init mp_register_ioapic(int id, u32 address, u32 gsi_base)
 {
 	int idx = 0;
+	int entries;
 
 	if (bad_ioapic(address))
 		return;
@@ -4298,9 +4287,14 @@ void __init mp_register_ioapic(int id, u32 address, u32 gsi_base)
 	 * Build basic GSI lookup table to facilitate gsi-&gt;io_apic lookups
 	 * and to prevent reprogramming of IOAPIC pins (PCI GSIs).
 	 */
+	entries = io_apic_get_redir_entries(idx);
 	mp_gsi_routing[idx].gsi_base = gsi_base;
-	mp_gsi_routing[idx].gsi_end = gsi_base +
-	    io_apic_get_redir_entries(idx);
+	mp_gsi_routing[idx].gsi_end = gsi_base + entries;
+
+	/*
+	 * The number of IO-APIC IRQ registers ...[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: [PATCH 04/14] x86 acpi/irq: Fix acpi_sci_ioapic_setup so it has both bus_irq and gsi

Date: Monday, March 29, 2010 - 4:20 pm

From: Eric W. Biederman &lt;ebiederm@xmission.com&gt;

Currently acpi_sci_ioapic_setup calls mp_override_legacy_irq with
bus_irq == gsi, which is wrong if we are comming from an override
Instead pass the bus_irq into acpi_sci_ioapic_setup.

This fix was inspired by a similar fix from:
Yinghai Lu &lt;yinghai@kernel.org&gt;

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
---
 arch/x86/kernel/acpi/boot.c |   12 +++++++-----
 1 files changed, 7 insertions(+), 5 deletions(-)

diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
index f96060d..1742888 100644
--- a/arch/x86/kernel/acpi/boot.c
+++ b/arch/x86/kernel/acpi/boot.c
@@ -312,7 +312,7 @@ acpi_parse_ioapic(struct acpi_subtable_header * header, const unsigned long end)
 /*
  * Parse Interrupt Source Override for the ACPI SCI
  */
-static void __init acpi_sci_ioapic_setup(u32 gsi, u16 polarity, u16 trigger)
+static void __init acpi_sci_ioapic_setup(u8 bus_irq, u16 polarity, u16 trigger, u32 gsi)
 {
 	if (trigger == 0)	/* compatible SCI trigger is level */
 		trigger = 3;
@@ -332,7 +332,7 @@ static void __init acpi_sci_ioapic_setup(u32 gsi, u16 polarity, u16 trigger)
 	 * If GSI is &lt; 16, this will update its flags,
 	 * else it will create a new mp_irqs[] entry.
 	 */
-	mp_override_legacy_irq(gsi, polarity, trigger, gsi);
+	mp_override_legacy_irq(bus_irq, polarity, trigger, gsi);
 
 	/*
 	 * stash over-ride to indicate we've been here
@@ -356,9 +356,10 @@ acpi_parse_int_src_ovr(struct acpi_subtable_header * header,
 	acpi_table_print_madt_entry(header);
 
 	if (intsrc-&gt;source_irq == acpi_gbl_FADT.sci_interrupt) {
-		acpi_sci_ioapic_setup(intsrc-&gt;global_irq,
+		acpi_sci_ioapic_setup(intsrc-&gt;source_irq,
 				      intsrc-&gt;inti_flags &amp; ACPI_MADT_POLARITY_MASK,
-				      (intsrc-&gt;inti_flags &amp; ACPI_MADT_TRIGGER_MASK) &gt;&gt; 2);
+				      (intsrc-&gt;inti_flags &amp; ACPI_MADT_TRIGGER_MASK) &gt;&gt; 2,
+				      intsrc-&gt;global_irq);
 		return 0;
 	}
 
@@ -1161,7 +1162,8 @@ static int __init ...[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: [PATCH 09/14] x86 ioapic: In mpparse use mp_register_ioapic

Date: Monday, March 29, 2010 - 4:20 pm

From: Eric W. Biederman &lt;ebiederm@xmission.com&gt;

Long ago MP_ioapic_info was the primary way of setting up our
ioapic data structures and mp_register_ioapic was a compatibility
shim for acpi code.  Now the situation is reversed and
and mp_register_ioapic is the primary way of setting up our
ioapic data structures.

Keep the setting up of ioapic data structures uniform by
having mp_register_ioapic call mp_register_ioapic.

This changes a few fields:

- type: is now hardset to MP_IOAPIC but type had to
  bey MP_IOAPIC or MP_ioapic_info would not have been called.

- flags: is now hard coded to MPC_APIC_USABLE.
  We require flags to contain at least MPC_APIC_USEBLE in
  MP_ioapic_info and we don't ever examine flags so dropping
  a few flags that might possibly exist that we have never
  used is harmless.

- apicaddr: Unchanged

- apicver: Read from the ioapic instead of using the cached
  hardware value in the MP table.  The real hardware value
  will be more accurate.

- apicid: Now verified to be unique and changed if it is not.
  If the BIOS got this right this is a noop.  If the BIOS did
  not fixing things appears to be the better solution.

This adds gsi_base and gsi_end values to our ioapics defined with
the mpatable, which will make our lives simpler later since
we can always assume gsi_base and gsi_end are valid.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
---
 arch/x86/kernel/mpparse.c |   25 +------------------------
 1 files changed, 1 insertions(+), 24 deletions(-)

diff --git a/arch/x86/kernel/mpparse.c b/arch/x86/kernel/mpparse.c
index a2c1edd..c0de938 100644
--- a/arch/x86/kernel/mpparse.c
+++ b/arch/x86/kernel/mpparse.c
@@ -115,21 +115,6 @@ static void __init MP_bus_info(struct mpc_bus *m)
 		printk(KERN_WARNING &quot;Unknown bustype %s - ignoring\n&quot;, str);
 }
 
-static int bad_ioapic(unsigned long address)
-{
-	if (nr_ioapics &gt;= MAX_IO_APICS) {
-		printk(KERN_ERR &quot;ERROR: Max # of I/O APICs (%d) exceeded &quot;
-		       &quot;(found %d)\n&quot;, ...[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: [PATCH 08/14] x86 ioapic: Teach mp_register_ioapic to compute a global gsi_end

Date: Monday, March 29, 2010 - 4:20 pm

From: Eric W. Biederman &lt;ebiederm@xmission.com&gt;

Add the global variable gsi_end and teach mp_register_ioapic
to keep it uptodate as we add more ioapics into the system.

ioapics can only be added early in boot so the code that
runs later can treat gsi_end as a constant.

Remove the have hacks in sfi.c to second guess mp_register_ioapic
by keeping t's own running total of how many gsi's have been seen,
and instead use the gsi_end.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
---
 arch/x86/include/asm/io_apic.h |    1 +
 arch/x86/kernel/apic/io_apic.c |    6 ++++++
 arch/x86/kernel/sfi.c          |    4 +---
 3 files changed, 8 insertions(+), 3 deletions(-)

diff --git a/arch/x86/include/asm/io_apic.h b/arch/x86/include/asm/io_apic.h
index feeaf0d..37b0f2b 100644
--- a/arch/x86/include/asm/io_apic.h
+++ b/arch/x86/include/asm/io_apic.h
@@ -184,6 +184,7 @@ struct mp_ioapic_gsi{
 	u32 gsi_end;
 };
 extern struct mp_ioapic_gsi  mp_gsi_routing[];
+extern u32 gsi_end;
 int mp_find_ioapic(u32 gsi);
 int mp_find_ioapic_pin(int ioapic, u32 gsi);
 void __init mp_register_ioapic(int id, u32 address, u32 gsi_base);
diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
index 32cea57..a979a00 100644
--- a/arch/x86/kernel/apic/io_apic.c
+++ b/arch/x86/kernel/apic/io_apic.c
@@ -88,6 +88,9 @@ int nr_ioapics;
 /* IO APIC gsi routing info */
 struct mp_ioapic_gsi  mp_gsi_routing[MAX_IO_APICS];
 
+/* One past the last gsi number used */
+u32 gsi_end;
+
 /* MP IRQ source entries */
 struct mpc_intsrc mp_irqs[MAX_IRQ_SOURCES];
 
@@ -4299,6 +4302,9 @@ void __init mp_register_ioapic(int id, u32 address, u32 gsi_base)
 	mp_gsi_routing[idx].gsi_end = gsi_base +
 	    io_apic_get_redir_entries(idx);
 
+	if (mp_gsi_routing[idx].gsi_end &gt; gsi_end)
+		gsi_end = mp_gsi_routing[idx].gsi_end;
+
 	printk(KERN_INFO &quot;IOAPIC[%d]: apic_id %d, version %d, address 0x%x, &quot;
 	       &quot;GSI %d-%d\n&quot;, idx, mp_ioapics[idx].apicid,
 	       mp_ioapics[idx].apicver, ...[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: [PATCH 02/14] x86 acpi/irq: Teach acpi_get_override_irq to take a gsi not an isa_irq

Date: Monday, March 29, 2010 - 4:20 pm

From: Eric W. Biederman &lt;ebiederm@xmission.com&gt;

In perverse acpi implementations the isa irqs are not identity mapped
to the first 16 gsi.  Furthermore at least the extended interrupt
resource capability may return gsi's and not isa irqs.  So since
what we get from acpi is a gsi teach acpi_get_overrride_irq to
operate on a gsi instead of an isa_irq.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
---
 arch/x86/kernel/apic/io_apic.c |   23 ++++++++++++++---------
 include/linux/acpi.h           |    4 ++--
 2 files changed, 16 insertions(+), 11 deletions(-)

diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
index e4e0ddc..54ba1f7 100644
--- a/arch/x86/kernel/apic/io_apic.c
+++ b/arch/x86/kernel/apic/io_apic.c
@@ -4073,22 +4073,27 @@ int __init io_apic_get_version(int ioapic)
 	return reg_01.bits.version;
 }
 
-int acpi_get_override_irq(int bus_irq, int *trigger, int *polarity)
+int acpi_get_override_irq(u32 gsi, int *trigger, int *polarity)
 {
-	int i;
+	int ioapic, pin, idx;
 
 	if (skip_ioapic_setup)
 		return -1;
 
-	for (i = 0; i &lt; mp_irq_entries; i++)
-		if (mp_irqs[i].irqtype == mp_INT &amp;&amp;
-		    mp_irqs[i].srcbusirq == bus_irq)
-			break;
-	if (i &gt;= mp_irq_entries)
+	ioapic = mp_find_ioapic(gsi);
+	if (ioapic &lt; 0)
+		return -1;
+
+	pin = mp_find_ioapic_pin(ioapic, gsi);
+	if (pin &lt; 0)
+		return -1;
+
+	idx = find_irq_entry(ioapic, pin, mp_INT);
+	if (idx &lt; 0)
 		return -1;
 
-	*trigger = irq_trigger(i);
-	*polarity = irq_polarity(i);
+	*trigger = irq_trigger(idx);
+	*polarity = irq_polarity(idx);
 	return 0;
 }
 
diff --git a/include/linux/acpi.h b/include/linux/acpi.h
index 7a937da..3da73f5 100644
--- a/include/linux/acpi.h
+++ b/include/linux/acpi.h
@@ -119,9 +119,9 @@ int acpi_gsi_to_irq (u32 gsi, unsigned int *irq);
 int acpi_isa_irq_to_gsi (unsigned isa_irq, u32 *gsi);
 
 #ifdef CONFIG_X86_IO_APIC
-extern int acpi_get_override_irq(int bus_irq, int *trigger, int *polarity);
+extern int ...[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: [PATCH 0/15] Start coping gsis < 16 that are not isa irqs. v2

Date: Tuesday, March 30, 2010 - 1:06 am

It appears there are a few systems in the wild that use acpi
interrupt source overrides to report a gsi &gt; 16 is an isa irq.

This breaks all kinds of assumptions I figure any BIOS doing that
probably should be shot as that is very much not a conservative position.

That said acpi appears to allow this insanity and on the basis of being
liberal in what we accept we should try to do something reasonable, and
keep our notions of gsi, isa irq, and linux irq as straight as we can.

To that end this patchset slightly modifies the gsi to linux irq
mapping.  Making it a 1 to 1 identity mapping except for the first 16
gsis.  If those gsis are isa irqs (the typical and default acpi
configuration) they will be mapped into the first 16 irqs.  If those
first 16 gsis are not isa irqs they will be given a linux irq number
just past the last gsi.  Allowing us the chance to use them.

This patchset is my attempt to straighten out our understanding of which
kind of irq name is used where and the cleanups used to get there.

Thanks to Yinghai Lu &lt;yinghai@kernel.org&gt; for taking the first stab at
this and finding that there is a real world problem here.  This patchset
is inspired by his work, but little of it remains the same.

This patchset lies in the weird world between the acpi and x86 and ia64.
Since most of the changes are x86 related I think it makes most sense
to go via the x86 tree.

v2: Recognize that gsi_end is last gsi for an ioapic not one past the end.
    Since I use gsi_end widely in my patchset a respin of many of the patches
    made sense.

Eric W. Biederman (15):
      x86 acpi/irq: Introduce apci_isa_irq_to_gsi
      x86 acpi/irq: Teach acpi_get_override_irq to take a gsi not an isa_irq
      x86 acpi/irq: pci device dev-&gt;irq is an isa irq not a gsi
      x86 acpi/irq: Fix acpi_sci_ioapic_setup so it has both bus_irq and gsi
      x86 acpi/irq: Generalize mp_config_acpi_legacy_irqs
      x86 ioapic: Only export mp_find_ioapic and mp_find_ioapic_pin in io_apic.h
      ...[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: [PATCH 05/15] x86 acpi/irq: Generalize mp_config_acpi_legacy_irqs

Date: Tuesday, March 30, 2010 - 1:07 am

From: Eric W. Biederman &lt;ebiederm@xmission.com&gt;

Remove the assumption that there is not an override for isa irq 0.
Instead lookup the gsi and from that lookup the ioapic and pin of each
isa irq indivdually.

In general this should not have any behavioural affect but in
perverse cases this gets all of the details correct, instead of
doing something weird.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
---
 arch/x86/kernel/acpi/boot.c |   30 ++++++++++++++++++------------
 1 files changed, 18 insertions(+), 12 deletions(-)

diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
index 1742888..9c48e99 100644
--- a/arch/x86/kernel/acpi/boot.c
+++ b/arch/x86/kernel/acpi/boot.c
@@ -960,8 +960,6 @@ void __init mp_override_legacy_irq(u8 bus_irq, u8 polarity, u8 trigger, u32 gsi)
 void __init mp_config_acpi_legacy_irqs(void)
 {
 	int i;
-	int ioapic;
-	unsigned int dstapic;
 	struct mpc_intsrc mp_irq;
 
 #if defined (CONFIG_MCA) || defined (CONFIG_EISA)
@@ -982,19 +980,27 @@ void __init mp_config_acpi_legacy_irqs(void)
 #endif
 
 	/*
-	 * Locate the IOAPIC that manages the ISA IRQs (0-15).
-	 */
-	ioapic = mp_find_ioapic(0);
-	if (ioapic &lt; 0)
-		return;
-	dstapic = mp_ioapics[ioapic].apicid;
-
-	/*
 	 * Use the default configuration for the IRQs 0-15.  Unless
 	 * overridden by (MADT) interrupt source override entries.
 	 */
 	for (i = 0; i &lt; 16; i++) {
+		int ioapic, pin;
+		unsigned int dstapic;
 		int idx;
+		u32 gsi;
+
+		/* Locate the gsi that irq i maps to. */
+		if (acpi_isa_irq_to_gsi(i, &amp;gsi))
+			continue;
+
+		/*
+		 * Locate the IOAPIC that manages the ISA IRQ.
+		 */
+		ioapic = mp_find_ioapic(gsi);
+		if (ioapic &lt; 0)
+			continue;
+		pin = mp_find_ioapic_pin(ioapic, gsi);
+		dstapic = mp_ioapics[ioapic].apicid;
 
 		for (idx = 0; idx &lt; mp_irq_entries; idx++) {
 			struct mpc_intsrc *irq = mp_irqs + idx;
@@ -1004,7 +1010,7 @@ void __init mp_config_acpi_legacy_irqs(void)
 				break;
 
 			/* Do we already have a ...[ message continues ]

[view original message]

From: tip-bot for Eric W. Biederman

Subject: [tip:x86/irq] x86, acpi/irq: Generalize mp_config_acpi_legacy_irqs

Date: Tuesday, May 4, 2010 - 7:07 pm

Commit-ID:  0fd52670fb6400be0996ac492b5ed77f3d83d69a
Gitweb:     http://git.kernel.org/tip/0fd52670fb6400be0996ac492b5ed77f3d83d69a
Author:     Eric W. Biederman &lt;ebiederm@xmission.com&gt;
AuthorDate: Tue, 30 Mar 2010 01:07:06 -0700
Committer:  H. Peter Anvin &lt;hpa@zytor.com&gt;
CommitDate: Tue, 4 May 2010 13:34:38 -0700

x86, acpi/irq: Generalize mp_config_acpi_legacy_irqs

Remove the assumption that there is not an override for isa irq 0.
Instead lookup the gsi and from that lookup the ioapic and pin of each
isa irq indivdually.

In general this should not have any behavioural affect but in
perverse cases this gets all of the details correct, instead of
doing something weird.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
LKML-Reference: &lt;1269936436-7039-5-git-send-email-ebiederm@xmission.com&gt;
Signed-off-by: H. Peter Anvin &lt;hpa@zytor.com&gt;
---
 arch/x86/kernel/acpi/boot.c |   30 ++++++++++++++++++------------
 1 files changed, 18 insertions(+), 12 deletions(-)

diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
index 0a036dc..3ee92f2 100644
--- a/arch/x86/kernel/acpi/boot.c
+++ b/arch/x86/kernel/acpi/boot.c
@@ -961,8 +961,6 @@ void __init mp_override_legacy_irq(u8 bus_irq, u8 polarity, u8 trigger, u32 gsi)
 void __init mp_config_acpi_legacy_irqs(void)
 {
 	int i;
-	int ioapic;
-	unsigned int dstapic;
 	struct mpc_intsrc mp_irq;
 
 #if defined (CONFIG_MCA) || defined (CONFIG_EISA)
@@ -983,19 +981,27 @@ void __init mp_config_acpi_legacy_irqs(void)
 #endif
 
 	/*
-	 * Locate the IOAPIC that manages the ISA IRQs (0-15).
-	 */
-	ioapic = mp_find_ioapic(0);
-	if (ioapic &lt; 0)
-		return;
-	dstapic = mp_ioapics[ioapic].apicid;
-
-	/*
 	 * Use the default configuration for the IRQs 0-15.  Unless
 	 * overridden by (MADT) interrupt source override entries.
 	 */
 	for (i = 0; i &lt; 16; i++) {
+		int ioapic, pin;
+		unsigned int dstapic;
 		int idx;
+		u32 gsi;
+
+		/* Locate the gsi that irq i maps to. */
+		if (acpi_isa_irq_to_gsi(i, ...[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: [PATCH 13/15] x86 ioapic: Simplify probe_nr_irqs_gsi.

Date: Tuesday, March 30, 2010 - 1:07 am

From: Eric W. Biederman &lt;ebiederm@xmission.com&gt;

Use the global gsi_end value now that all ioapics have
valid gsi numbers instead of a combination of acpi_probe_gsi
and walking all of the ioapics and couting their number of
entries by hand if acpi_probe_gsi gave us an answer we did
not like.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
---
 arch/x86/include/asm/mpspec.h  |    6 ------
 arch/x86/kernel/acpi/boot.c    |   23 -----------------------
 arch/x86/kernel/apic/io_apic.c |   17 +++--------------
 3 files changed, 3 insertions(+), 43 deletions(-)

diff --git a/arch/x86/include/asm/mpspec.h b/arch/x86/include/asm/mpspec.h
index 29994f0..c82868e 100644
--- a/arch/x86/include/asm/mpspec.h
+++ b/arch/x86/include/asm/mpspec.h
@@ -105,12 +105,6 @@ extern void mp_config_acpi_legacy_irqs(void);
 struct device;
 extern int mp_register_gsi(struct device *dev, u32 gsi, int edge_level,
 				 int active_high_low);
-extern int acpi_probe_gsi(void);
-#else /* !CONFIG_ACPI: */
-static inline int acpi_probe_gsi(void)
-{
-	return 0;
-}
 #endif /* CONFIG_ACPI */
 
 #define PHYSID_ARRAY_SIZE	BITS_TO_LONGS(MAX_APICS)
diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
index 9c48e99..0e514a1 100644
--- a/arch/x86/kernel/acpi/boot.c
+++ b/arch/x86/kernel/acpi/boot.c
@@ -875,29 +875,6 @@ static int __init acpi_parse_madt_lapic_entries(void)
 extern int es7000_plat;
 #endif
 
-int __init acpi_probe_gsi(void)
-{
-	int idx;
-	int gsi;
-	int max_gsi = 0;
-
-	if (acpi_disabled)
-		return 0;
-
-	if (!acpi_ioapic)
-		return 0;
-
-	max_gsi = 0;
-	for (idx = 0; idx &lt; nr_ioapics; idx++) {
-		gsi = mp_gsi_routing[idx].gsi_end;
-
-		if (gsi &gt; max_gsi)
-			max_gsi = gsi;
-	}
-
-	return max_gsi + 1;
-}
-
 static void assign_to_mp_irq(struct mpc_intsrc *m,
 				    struct mpc_intsrc *mp_irq)
 {
diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
index 764c74c..6d6def8 100644
--- a/arch/x86/kernel/apic/io_apic.c
+++ ...[ message continues ]

[view original message]

From: tip-bot for Eric W. Biederman

Subject: [tip:x86/irq] x86, ioapic: Simplify probe_nr_irqs_gsi.

Date: Tuesday, May 4, 2010 - 7:10 pm

Commit-ID:  4afc51a835d3aeba11c35090f524e05c84586d27
Gitweb:     http://git.kernel.org/tip/4afc51a835d3aeba11c35090f524e05c84586d27
Author:     Eric W. Biederman &lt;ebiederm@xmission.com&gt;
AuthorDate: Tue, 30 Mar 2010 01:07:14 -0700
Committer:  H. Peter Anvin &lt;hpa@zytor.com&gt;
CommitDate: Tue, 4 May 2010 13:35:11 -0700

x86, ioapic: Simplify probe_nr_irqs_gsi.

Use the global gsi_end value now that all ioapics have
valid gsi numbers instead of a combination of acpi_probe_gsi
and walking all of the ioapics and couting their number of
entries by hand if acpi_probe_gsi gave us an answer we did
not like.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
LKML-Reference: &lt;1269936436-7039-13-git-send-email-ebiederm@xmission.com&gt;
Signed-off-by: H. Peter Anvin &lt;hpa@zytor.com&gt;
---
 arch/x86/include/asm/mpspec.h  |    6 ------
 arch/x86/kernel/acpi/boot.c    |   23 -----------------------
 arch/x86/kernel/apic/io_apic.c |   17 +++--------------
 3 files changed, 3 insertions(+), 43 deletions(-)

diff --git a/arch/x86/include/asm/mpspec.h b/arch/x86/include/asm/mpspec.h
index 29994f0..c82868e 100644
--- a/arch/x86/include/asm/mpspec.h
+++ b/arch/x86/include/asm/mpspec.h
@@ -105,12 +105,6 @@ extern void mp_config_acpi_legacy_irqs(void);
 struct device;
 extern int mp_register_gsi(struct device *dev, u32 gsi, int edge_level,
 				 int active_high_low);
-extern int acpi_probe_gsi(void);
-#else /* !CONFIG_ACPI: */
-static inline int acpi_probe_gsi(void)
-{
-	return 0;
-}
 #endif /* CONFIG_ACPI */
 
 #define PHYSID_ARRAY_SIZE	BITS_TO_LONGS(MAX_APICS)
diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
index 3ee92f2..07a63ce 100644
--- a/arch/x86/kernel/acpi/boot.c
+++ b/arch/x86/kernel/acpi/boot.c
@@ -876,29 +876,6 @@ static int __init acpi_parse_madt_lapic_entries(void)
 extern int es7000_plat;
 #endif
 
-int __init acpi_probe_gsi(void)
-{
-	int idx;
-	int gsi;
-	int max_gsi = 0;
-
-	if (acpi_disabled)
-		return 0;
-
-	if ...[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: [PATCH 15/15] x86 irq: Kill io_apic_renumber_irq

Date: Tuesday, March 30, 2010 - 1:07 am

From: Eric W. Biederman &lt;ebiederm@xmission.com&gt;

Now that the generic irq layer is performing the exact same remapping as
io_apic_renumber_irq we can kill this weird  es7000 specific function.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
---
 arch/x86/include/asm/io_apic.h   |    1 -
 arch/x86/kernel/acpi/boot.c      |    5 -----
 arch/x86/kernel/apic/es7000_32.c |   19 -------------------
 arch/x86/kernel/apic/io_apic.c   |    6 ------
 4 files changed, 0 insertions(+), 31 deletions(-)

diff --git a/arch/x86/include/asm/io_apic.h b/arch/x86/include/asm/io_apic.h
index 37b0f2b..9da192a 100644
--- a/arch/x86/include/asm/io_apic.h
+++ b/arch/x86/include/asm/io_apic.h
@@ -159,7 +159,6 @@ struct io_apic_irq_attr;
 extern int io_apic_set_pci_routing(struct device *dev, int irq,
 		 struct io_apic_irq_attr *irq_attr);
 void setup_IO_APIC_irq_extra(u32 gsi);
-extern int (*ioapic_renumber_irq)(int ioapic, int irq);
 extern void ioapic_init_mappings(void);
 extern void ioapic_insert_resources(void);
 
diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
index c4b4bd9..614f3a7 100644
--- a/arch/x86/kernel/acpi/boot.c
+++ b/arch/x86/kernel/acpi/boot.c
@@ -1116,11 +1116,6 @@ int mp_register_gsi(struct device *dev, u32 gsi, int trigger, int polarity)
 
 	ioapic_pin = mp_find_ioapic_pin(ioapic, gsi);
 
-#ifdef CONFIG_X86_32
-	if (ioapic_renumber_irq)
-		gsi = ioapic_renumber_irq(ioapic, gsi);
-#endif
-
 	if (ioapic_pin &gt; MP_MAX_IOAPIC_PIN) {
 		printk(KERN_ERR &quot;Invalid reference to IOAPIC pin &quot;
 		       &quot;%d-%d\n&quot;, mp_ioapics[ioapic].apicid,
diff --git a/arch/x86/kernel/apic/es7000_32.c b/arch/x86/kernel/apic/es7000_32.c
index dd2b5f2..7a63d47 100644
--- a/arch/x86/kernel/apic/es7000_32.c
+++ b/arch/x86/kernel/apic/es7000_32.c
@@ -130,24 +130,6 @@ int					es7000_plat;
 
 static unsigned int			base;
 
-static int
-es7000_rename_gsi(int ioapic, int gsi)
-{
-	if (es7000_plat == ES7000_ZORRO)
-		return gsi;
-
-	if (!base) {
-		int i;
-		for ...[ message continues ]

[view original message]

From: tip-bot for Eric W. Biederman

Subject: [tip:x86/irq] x86, irq: Kill io_apic_renumber_irq

Date: Tuesday, May 4, 2010 - 7:10 pm

Commit-ID:  7b20bd5fb902088579af4e70f7f802b93181a628
Gitweb:     http://git.kernel.org/tip/7b20bd5fb902088579af4e70f7f802b93181a628
Author:     Eric W. Biederman &lt;ebiederm@xmission.com&gt;
AuthorDate: Tue, 30 Mar 2010 01:07:16 -0700
Committer:  H. Peter Anvin &lt;hpa@zytor.com&gt;
CommitDate: Tue, 4 May 2010 13:35:20 -0700

x86, irq: Kill io_apic_renumber_irq

Now that the generic irq layer is performing the exact same remapping as
io_apic_renumber_irq we can kill this weird  es7000 specific function.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
LKML-Reference: &lt;1269936436-7039-15-git-send-email-ebiederm@xmission.com&gt;
Signed-off-by: H. Peter Anvin &lt;hpa@zytor.com&gt;
---
 arch/x86/include/asm/io_apic.h   |    1 -
 arch/x86/kernel/acpi/boot.c      |    5 -----
 arch/x86/kernel/apic/es7000_32.c |   19 -------------------
 arch/x86/kernel/apic/io_apic.c   |    6 ------
 4 files changed, 0 insertions(+), 31 deletions(-)

diff --git a/arch/x86/include/asm/io_apic.h b/arch/x86/include/asm/io_apic.h
index 37b0f2b..9da192a 100644
--- a/arch/x86/include/asm/io_apic.h
+++ b/arch/x86/include/asm/io_apic.h
@@ -159,7 +159,6 @@ struct io_apic_irq_attr;
 extern int io_apic_set_pci_routing(struct device *dev, int irq,
 		 struct io_apic_irq_attr *irq_attr);
 void setup_IO_APIC_irq_extra(u32 gsi);
-extern int (*ioapic_renumber_irq)(int ioapic, int irq);
 extern void ioapic_init_mappings(void);
 extern void ioapic_insert_resources(void);
 
diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
index 325fbba..9a5ed58 100644
--- a/arch/x86/kernel/acpi/boot.c
+++ b/arch/x86/kernel/acpi/boot.c
@@ -1117,11 +1117,6 @@ int mp_register_gsi(struct device *dev, u32 gsi, int trigger, int polarity)
 
 	ioapic_pin = mp_find_ioapic_pin(ioapic, gsi);
 
-#ifdef CONFIG_X86_32
-	if (ioapic_renumber_irq)
-		gsi = ioapic_renumber_irq(ioapic, gsi);
-#endif
-
 	if (ioapic_pin &gt; MP_MAX_IOAPIC_PIN) {
 		printk(KERN_ERR &quot;Invalid reference to IOAPIC pin &quot;
 		       &quot;%d-%d\n&quot;, ...[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: [PATCH 12/15] x86 ioapic: Optimize pin_2_irq

Date: Tuesday, March 30, 2010 - 1:07 am

From: Eric W. Biederman &lt;ebiederm@xmission.com&gt;

Now that all ioapics have valid gsi_base values use this to
accellerate pin_2_irq.  In the case of acpi this also ensures
that pin_2_irq will compute the same irq value for an ioapic
pin as acpi will.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
---
 arch/x86/kernel/apic/io_apic.c |   13 ++++---------
 1 files changed, 4 insertions(+), 9 deletions(-)

diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
index c505fe0..764c74c 100644
--- a/arch/x86/kernel/apic/io_apic.c
+++ b/arch/x86/kernel/apic/io_apic.c
@@ -1018,7 +1018,7 @@ static inline int irq_trigger(int idx)
 int (*ioapic_renumber_irq)(int ioapic, int irq);
 static int pin_2_irq(int idx, int apic, int pin)
 {
-	int irq, i;
+	int irq;
 	int bus = mp_irqs[idx].srcbus;
 
 	/*
@@ -1030,18 +1030,13 @@ static int pin_2_irq(int idx, int apic, int pin)
 	if (test_bit(bus, mp_bus_not_pci)) {
 		irq = mp_irqs[idx].srcbusirq;
 	} else {
-		/*
-		 * PCI IRQs are mapped in order
-		 */
-		i = irq = 0;
-		while (i &lt; apic)
-			irq += nr_ioapic_registers[i++];
-		irq += pin;
+		u32 gsi = mp_gsi_routing[apic].gsi_base + pin;
 		/*
                  * For MPS mode, so far only needed by ES7000 platform
                  */
 		if (ioapic_renumber_irq)
-			irq = ioapic_renumber_irq(apic, irq);
+			gsi = ioapic_renumber_irq(apic, gsi);
+		irq = gsi;
 	}
 
 #ifdef CONFIG_X86_32
-- 
1.6.5.2.143.g8cc62

--

[ message continues ]

[view original message]

From: tip-bot for Eric W. Biederman

Subject: [tip:x86/irq] x86, ioapic: Optimize pin_2_irq

Date: Tuesday, May 4, 2010 - 7:09 pm

Commit-ID:  d464207c4fdd70c2a0febd4f9c58206fa915bb36
Gitweb:     http://git.kernel.org/tip/d464207c4fdd70c2a0febd4f9c58206fa915bb36
Author:     Eric W. Biederman &lt;ebiederm@xmission.com&gt;
AuthorDate: Tue, 30 Mar 2010 01:07:13 -0700
Committer:  H. Peter Anvin &lt;hpa@zytor.com&gt;
CommitDate: Tue, 4 May 2010 13:35:08 -0700

x86, ioapic: Optimize pin_2_irq

Now that all ioapics have valid gsi_base values use this to
accellerate pin_2_irq.  In the case of acpi this also ensures
that pin_2_irq will compute the same irq value for an ioapic
pin as acpi will.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
LKML-Reference: &lt;1269936436-7039-12-git-send-email-ebiederm@xmission.com&gt;
Signed-off-by: H. Peter Anvin &lt;hpa@zytor.com&gt;
---
 arch/x86/kernel/apic/io_apic.c |   13 ++++---------
 1 files changed, 4 insertions(+), 9 deletions(-)

diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
index dae9240..0d35f46 100644
--- a/arch/x86/kernel/apic/io_apic.c
+++ b/arch/x86/kernel/apic/io_apic.c
@@ -1019,7 +1019,7 @@ static inline int irq_trigger(int idx)
 int (*ioapic_renumber_irq)(int ioapic, int irq);
 static int pin_2_irq(int idx, int apic, int pin)
 {
-	int irq, i;
+	int irq;
 	int bus = mp_irqs[idx].srcbus;
 
 	/*
@@ -1031,18 +1031,13 @@ static int pin_2_irq(int idx, int apic, int pin)
 	if (test_bit(bus, mp_bus_not_pci)) {
 		irq = mp_irqs[idx].srcbusirq;
 	} else {
-		/*
-		 * PCI IRQs are mapped in order
-		 */
-		i = irq = 0;
-		while (i &lt; apic)
-			irq += nr_ioapic_registers[i++];
-		irq += pin;
+		u32 gsi = mp_gsi_routing[apic].gsi_base + pin;
 		/*
                  * For MPS mode, so far only needed by ES7000 platform
                  */
 		if (ioapic_renumber_irq)
-			irq = ioapic_renumber_irq(apic, irq);
+			gsi = ioapic_renumber_irq(apic, gsi);
+		irq = gsi;
 	}
 
 #ifdef CONFIG_X86_32
--

[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: [PATCH 14/15] x86 acpi/irq: Handle isa irqs that are not identity mapped to gsi's.

Date: Tuesday, March 30, 2010 - 1:07 am

From: Eric W. Biederman &lt;ebiederm@xmission.com&gt;

ACPI irq source overrides are allowed for the 16 isa irqs and are
allowed to map any gsi to any isa irq.  A few motherboards have been
seen to take advantage of this and put the isa irqs on the 2nd or
3rd ioapic.  This causes some problems, most notably the fact
that we can not use any gsi &lt; 16.

To correct this move the gsis that are not isa irqs and have
a gsi number &lt; 16 into the linux irq space just past gsi_end.
This is what the es7000 platform is doing today.  Moving only the
low 16 gsis above the rest of the gsi's only penalizes weird
platforms, leaving sane acpi implementations with a 1-1 mapping
of gsis and irqs.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
---
 arch/x86/kernel/acpi/boot.c    |   57 +++++++++++++++++++++++++++++++++++++---
 arch/x86/kernel/apic/io_apic.c |    8 ++++-
 2 files changed, 59 insertions(+), 6 deletions(-)

diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
index 0e514a1..c4b4bd9 100644
--- a/arch/x86/kernel/acpi/boot.c
+++ b/arch/x86/kernel/acpi/boot.c
@@ -92,6 +92,53 @@ static u64 acpi_lapic_addr __initdata = APIC_DEFAULT_PHYS_BASE;
 enum acpi_irq_model_id acpi_irq_model = ACPI_IRQ_MODEL_PIC;
 
 
+/* 
+ * ISA irqs by default are the first 16 gsis but can be
+ * any gsi as specified by an interrupt source override.
+ */
+static u32 isa_irq_to_gsi[NR_IRQS_LEGACY] __read_mostly = {
+	0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15
+};
+
+static unsigned int gsi_to_irq(unsigned int gsi)
+{
+	unsigned int irq = gsi + NR_IRQS_LEGACY;
+	unsigned int i;
+
+	for (i = 0; i &lt; NR_IRQS_LEGACY; i++) {
+		if (isa_irq_to_gsi[i] == gsi) {
+			return i;
+		}
+	}
+
+	/* Provide an identity mapping of gsi == irq
+	 * except on truly weird platforms that have
+	 * non isa irqs in the first 16 gsis.
+	 */
+	if (gsi &gt;= NR_IRQS_LEGACY)
+		irq = gsi;
+	else
+		irq = gsi_end + 1 + gsi;
+
+	return irq;
+}
+
+static u32 irq_to_gsi(int ...[ message continues ]

[view original message]

From: tip-bot for Eric W. Biederman

Subject: [tip:x86/irq] x86, acpi/irq: Handle isa irqs that are not identity mapped to gsi's.

Date: Tuesday, May 4, 2010 - 7:10 pm

Commit-ID:  988856ee1623bd37e384105f7bb2b7fe44c009f6
Gitweb:     http://git.kernel.org/tip/988856ee1623bd37e384105f7bb2b7fe44c009f6
Author:     Eric W. Biederman &lt;ebiederm@xmission.com&gt;
AuthorDate: Tue, 30 Mar 2010 01:07:15 -0700
Committer:  H. Peter Anvin &lt;hpa@zytor.com&gt;
CommitDate: Tue, 4 May 2010 13:35:17 -0700

x86, acpi/irq: Handle isa irqs that are not identity mapped to gsi's.

ACPI irq source overrides are allowed for the 16 isa irqs and are
allowed to map any gsi to any isa irq.  A few motherboards have been
seen to take advantage of this and put the isa irqs on the 2nd or
3rd ioapic.  This causes some problems, most notably the fact
that we can not use any gsi &lt; 16.

To correct this move the gsis that are not isa irqs and have
a gsi number &lt; 16 into the linux irq space just past gsi_end.
This is what the es7000 platform is doing today.  Moving only the
low 16 gsis above the rest of the gsi's only penalizes weird
platforms, leaving sane acpi implementations with a 1-1 mapping
of gsis and irqs.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
LKML-Reference: &lt;1269936436-7039-14-git-send-email-ebiederm@xmission.com&gt;
Signed-off-by: H. Peter Anvin &lt;hpa@zytor.com&gt;
---
 arch/x86/kernel/acpi/boot.c    |   57 +++++++++++++++++++++++++++++++++++++---
 arch/x86/kernel/apic/io_apic.c |    8 ++++-
 2 files changed, 59 insertions(+), 6 deletions(-)

diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
index 07a63ce..325fbba 100644
--- a/arch/x86/kernel/acpi/boot.c
+++ b/arch/x86/kernel/acpi/boot.c
@@ -94,6 +94,53 @@ enum acpi_irq_model_id acpi_irq_model = ACPI_IRQ_MODEL_PIC;
 
 
 /*
+ * ISA irqs by default are the first 16 gsis but can be
+ * any gsi as specified by an interrupt source override.
+ */
+static u32 isa_irq_to_gsi[NR_IRQS_LEGACY] __read_mostly = {
+	0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15
+};
+
+static unsigned int gsi_to_irq(unsigned int gsi)
+{
+	unsigned int irq = gsi + NR_IRQS_LEGACY;
+	unsigned int ...[ message continues ]

[view original message]

From: Yinghai

Subject: Re: [tip:x86/irq] x86, acpi/irq: Handle isa irqs that are not identity mapped to gsi's.

Date: Wednesday, May 5, 2010 - 12:49 am

here should be

can you use legacy_irq-&gt;nr_legacy_irqs instead of NR_IRQS_LEGACY ?

YH
--

[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: [PATCH] x86 acpi/irq: Fix harmless typo.

Date: Wednesday, May 5, 2010 - 1:53 am

YH noticed that the function irq_to_gsi has an off by one error
when translating high irq numbers to gsis.  Today this bug is harmless
because all of the callers restrict their input to the first 16 irqs
so this bug does not matter, but we should fix it to avoid confusion
and later.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
---
 arch/x86/kernel/acpi/boot.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
index c9a5d3f..78222c8 100644
--- a/arch/x86/kernel/acpi/boot.c
+++ b/arch/x86/kernel/acpi/boot.c
@@ -132,7 +132,7 @@ static u32 irq_to_gsi(int irq)
 	else if (irq &lt;= gsi_end)
 		gsi = irq;
 	else if (irq &lt;= (gsi_end + NR_IRQS_LEGACY))
-		gsi = irq - gsi_end;
+		gsi = irq - (gsi_end + 1);
 	else
 		gsi = 0xffffffff;
 
-- 
1.6.5.2.143.g8cc62

--

[ message continues ]

[view original message]

From: Ingo Molnar

Subject: Re: [PATCH] x86 acpi/irq: Fix harmless typo.

Date: Wednesday, May 5, 2010 - 1:58 am

Please use the Reported-by tag.

Thanks,

	Ingo
--

[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: Re: [tip:x86/irq] x86, acpi/irq: Handle isa irqs that are not identity mapped to gsi's.

Date: Wednesday, May 5, 2010 - 2:32 am

No. legacy_irq-&gt;nr_legacy_irqs is just for dealing with hardware that
does not have i8259 pics.

Here I am reserving 16 irqs above the gsi range in case we get a
weird gsi to isa irq mapping.

Currently the sfi spec doesn't mention anything explicitly.  I
took a quick read through it and sfi and I presume Moorestown
is talking about non pci irqs.  So I would not be surprised
if we didn't have isa device drivers running on platforms
without i8259s.

Currently the code had a small bug dealing with multiple ioapics and
the spec really is incomplete about how you map irqs from pci devices
to interrupts.  So it looks like SFI and Moorestown are more a lab
experiment than real hardware at the moment.  So I expect when
it gets completely fleshed out we will have to jump through a
similar set of hoops as we do with acpi to ensure the first 16
irqs are not used for anything except isa irqs.

Eric
--

[ message continues ]

[view original message]

From: H. Peter Anvin

Subject: Re: [tip:x86/irq] x86, acpi/irq: Handle isa irqs that are not identity mapped to gsi's.

Date: Monday, June 7, 2010 - 2:05 pm

I don't know if it is &quot;just&quot; for this purpose.  Quite on the contrary.
In fact, the fact that NR_IRQS_LEGACY is showing up at all looks like a

But that 16 is exactly because those are the (platform-specific) legacy

I'm confused about the above.  Where do you see any mention to fixed

Moorestown hardware certainly exists, although not all the support is
upstream yet.

Either which way, anything which perpetuates NR_IRQS_LEGACY as a
compile-time constant is begging for people to get things wrong, and it
really needs to go away.

	-hpa
--

[ message continues ]

[view original message]

From: Yinghai Lu

Subject: Re: [tip:x86/irq] x86, acpi/irq: Handle isa irqs that are not identity mapped to gsi's.

Date: Tuesday, June 8, 2010 - 3:20 pm

yes, it should be replaced legacy_irq-&gt;nr_legacy_irs to make Moorestown to have sane irq index.

otherwise gsi_to_irq will mapping all [0, 15] to start from nr_irqs_gsi.

Thanks

Yinghai
--

[ message continues ]

[view original message]

From: Ingo Molnar

Subject: Re: [tip:x86/irq] x86, acpi/irq: Handle isa irqs that are not identity mapped to gsi's.

Date: Wednesday, May 5, 2010 - 1:56 am

FYI, the patchset causes this build error with certain configs:

arch/x86/kernel/acpi/boot.c:122: error: 'gsi_end' undeclared (first use in this function)
arch/x86/kernel/acpi/boot.c:122: error: (Each undeclared identifier is reported only once
arch/x86/kernel/acpi/boot.c:122: error: for each function it appears in.)
arch/x86/kernel/acpi/boot.c:133: error: 'gsi_end' undeclared (first use in this function)

	Ingo
--

[ message continues ]

[view original message]

From: Eric Biederman

Subject: Re: [tip:x86/irq] x86, acpi/irq: Handle isa irqs that are not identity mapped to gsi's.

Date: Wednesday, May 5, 2010 - 2:36 am

What configuration supports acpi and doesn't support ioapics?

Eric
--

[ message continues ]

[view original message]

From: Ingo Molnar

Subject: Re: [tip:x86/irq] x86, acpi/irq: Handle isa irqs that are not identity mapped to gsi's.

Date: Wednesday, May 5, 2010 - 3:05 am

an UP kernel with ACPI turned on for example.

	Ingo
--

[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: [PATCH] x86 acpi/irq: Define gsi_end when X86_IO_APIC is undefined

Date: Wednesday, May 5, 2010 - 1:22 pm

Oops.  My recent changes introducing a global gsi_end variable failed
to take into account the case of using acpi on a system not built to
support IO_APICs, causing the build to fail.  Define gsi_end to 15 when
CONFIG_X86_IO_APIC is not set to avoid compile errors.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
---
 arch/x86/include/asm/io_apic.h |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)

diff --git a/arch/x86/include/asm/io_apic.h b/arch/x86/include/asm/io_apic.h
index 9da192a..63cb409 100644
--- a/arch/x86/include/asm/io_apic.h
+++ b/arch/x86/include/asm/io_apic.h
@@ -197,6 +197,7 @@ static const int timer_through_8259 = 0;
 static inline void ioapic_init_mappings(void)	{ }
 static inline void ioapic_insert_resources(void) { }
 static inline void probe_nr_irqs_gsi(void)	{ }
+#define gsi_end (NR_IRQS_LEGACY - 1)
 static inline int mp_find_ioapic(u32 gsi) { return 0; }
 
 struct io_apic_irq_attr;
-- 
1.6.5.2.143.g8cc62

--

[ message continues ]

[view original message]

From: Ingo Molnar

Subject: Re: [PATCH] x86 acpi/irq: Define gsi_end when X86_IO_APIC is undefined

Date: Wednesday, May 5, 2010 - 11:18 pm

Applied to tip:x86/irq, thanks Eric!

	Ingo
--

[ message continues ]

[view original message]

From: tip-bot for Eric W. Biederman

Subject: [tip:x86/irq] x86, acpi/irq: Define gsi_end when X86_IO_APIC is undefined

Date: Thursday, May 6, 2010 - 3:07 am

Commit-ID:  4f47b4c9f0b711bf84adb8c27774ae80d346b628
Gitweb:     http://git.kernel.org/tip/4f47b4c9f0b711bf84adb8c27774ae80d346b628
Author:     Eric W. Biederman &lt;ebiederm@xmission.com&gt;
AuthorDate: Wed, 5 May 2010 13:22:25 -0700
Committer:  Ingo Molnar &lt;mingo@elte.hu&gt;
CommitDate: Thu, 6 May 2010 08:17:51 +0200

x86, acpi/irq: Define gsi_end when X86_IO_APIC is undefined

My recent changes introducing a global gsi_end variable
failed to take into account the case of using acpi on a system
not built to support IO_APICs, causing the build to fail.

Define gsi_end to 15 when CONFIG_X86_IO_APIC is not set to avoid
compile errors.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
Cc: Yinghai Lu &lt;yinghai@kernel.org&gt;
LKML-Reference: &lt;m1tyqm14la.fsf_-_@fess.ebiederm.org&gt;
Signed-off-by: Ingo Molnar &lt;mingo@elte.hu&gt;
---
 arch/x86/include/asm/io_apic.h |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)

diff --git a/arch/x86/include/asm/io_apic.h b/arch/x86/include/asm/io_apic.h
index 9da192a..63cb409 100644
--- a/arch/x86/include/asm/io_apic.h
+++ b/arch/x86/include/asm/io_apic.h
@@ -197,6 +197,7 @@ static const int timer_through_8259 = 0;
 static inline void ioapic_init_mappings(void)	{ }
 static inline void ioapic_insert_resources(void) { }
 static inline void probe_nr_irqs_gsi(void)	{ }
+#define gsi_end (NR_IRQS_LEGACY - 1)
 static inline int mp_find_ioapic(u32 gsi) { return 0; }
 
 struct io_apic_irq_attr;
--

[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: [PATCH 09/15] x86 ioapic: Teach mp_register_ioapic to compute a global gsi_end

Date: Tuesday, March 30, 2010 - 1:07 am

From: Eric W. Biederman &lt;ebiederm@xmission.com&gt;

Add the global variable gsi_end and teach mp_register_ioapic
to keep it uptodate as we add more ioapics into the system.

ioapics can only be added early in boot so the code that
runs later can treat gsi_end as a constant.

Remove the have hacks in sfi.c to second guess mp_register_ioapic
by keeping t's own running total of how many gsi's have been seen,
and instead use the gsi_end.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
---
 arch/x86/include/asm/io_apic.h |    1 +
 arch/x86/kernel/apic/io_apic.c |    6 ++++++
 arch/x86/kernel/sfi.c          |    4 +---
 3 files changed, 8 insertions(+), 3 deletions(-)

diff --git a/arch/x86/include/asm/io_apic.h b/arch/x86/include/asm/io_apic.h
index feeaf0d..37b0f2b 100644
--- a/arch/x86/include/asm/io_apic.h
+++ b/arch/x86/include/asm/io_apic.h
@@ -184,6 +184,7 @@ struct mp_ioapic_gsi{
 	u32 gsi_end;
 };
 extern struct mp_ioapic_gsi  mp_gsi_routing[];
+extern u32 gsi_end;
 int mp_find_ioapic(u32 gsi);
 int mp_find_ioapic_pin(int ioapic, u32 gsi);
 void __init mp_register_ioapic(int id, u32 address, u32 gsi_base);
diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
index 4611c43..71988c5 100644
--- a/arch/x86/kernel/apic/io_apic.c
+++ b/arch/x86/kernel/apic/io_apic.c
@@ -88,6 +88,9 @@ int nr_ioapics;
 /* IO APIC gsi routing info */
 struct mp_ioapic_gsi  mp_gsi_routing[MAX_IO_APICS];
 
+/* The last gsi number used */
+u32 gsi_end;
+
 /* MP IRQ source entries */
 struct mpc_intsrc mp_irqs[MAX_IRQ_SOURCES];
 
@@ -4303,6 +4306,9 @@ void __init mp_register_ioapic(int id, u32 address, u32 gsi_base)
 	mp_gsi_routing[idx].gsi_end = gsi_base +
 	    io_apic_get_redir_entries(idx) - 1;
 
+	if (mp_gsi_routing[idx].gsi_end &gt; gsi_end)
+		gsi_end = mp_gsi_routing[idx].gsi_end;
+
 	printk(KERN_INFO &quot;IOAPIC[%d]: apic_id %d, version %d, address 0x%x, &quot;
 	       &quot;GSI %d-%d\n&quot;, idx, mp_ioapics[idx].apicid,
 	       mp_ioapics[idx].apicver, ...[ message continues ]

[view original message]

From: tip-bot for Eric W. Biederman

Subject: [tip:x86/irq] x86, ioapic: Teach mp_register_ioapic to compute a global gsi_end

Date: Tuesday, May 4, 2010 - 7:09 pm

Commit-ID:  5777372af5c929b8f3c706ed7b295b7279537c88
Gitweb:     http://git.kernel.org/tip/5777372af5c929b8f3c706ed7b295b7279537c88
Author:     Eric W. Biederman &lt;ebiederm@xmission.com&gt;
AuthorDate: Tue, 30 Mar 2010 01:07:10 -0700
Committer:  H. Peter Anvin &lt;hpa@zytor.com&gt;
CommitDate: Tue, 4 May 2010 13:34:56 -0700

x86, ioapic: Teach mp_register_ioapic to compute a global gsi_end

Add the global variable gsi_end and teach mp_register_ioapic
to keep it uptodate as we add more ioapics into the system.

ioapics can only be added early in boot so the code that
runs later can treat gsi_end as a constant.

Remove the have hacks in sfi.c to second guess mp_register_ioapic
by keeping t's own running total of how many gsi's have been seen,
and instead use the gsi_end.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
LKML-Reference: &lt;1269936436-7039-9-git-send-email-ebiederm@xmission.com&gt;
Signed-off-by: H. Peter Anvin &lt;hpa@zytor.com&gt;
---
 arch/x86/include/asm/io_apic.h |    1 +
 arch/x86/kernel/apic/io_apic.c |    6 ++++++
 arch/x86/kernel/sfi.c          |    4 +---
 3 files changed, 8 insertions(+), 3 deletions(-)

diff --git a/arch/x86/include/asm/io_apic.h b/arch/x86/include/asm/io_apic.h
index feeaf0d..37b0f2b 100644
--- a/arch/x86/include/asm/io_apic.h
+++ b/arch/x86/include/asm/io_apic.h
@@ -184,6 +184,7 @@ struct mp_ioapic_gsi{
 	u32 gsi_end;
 };
 extern struct mp_ioapic_gsi  mp_gsi_routing[];
+extern u32 gsi_end;
 int mp_find_ioapic(u32 gsi);
 int mp_find_ioapic_pin(int ioapic, u32 gsi);
 void __init mp_register_ioapic(int id, u32 address, u32 gsi_base);
diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
index 9ab9762..f807255 100644
--- a/arch/x86/kernel/apic/io_apic.c
+++ b/arch/x86/kernel/apic/io_apic.c
@@ -89,6 +89,9 @@ int nr_ioapics;
 /* IO APIC gsi routing info */
 struct mp_ioapic_gsi  mp_gsi_routing[MAX_IO_APICS];
 
+/* The last gsi number used */
+u32 gsi_end;
+
 /* MP IRQ source entries */
 struct mpc_intsrc ...[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: [PATCH 11/15] x86 ioapic: Move nr_ioapic_registers calculation to mp_register_ioapic.

Date: Tuesday, March 30, 2010 - 1:07 am

From: Eric W. Biederman &lt;ebiederm@xmission.com&gt;

Now that all ioapic registration happens in mp_register_ioapic we can
move the calculation of nr_ioapic_registers there from enable_IO_APIC.
The number of ioapic registers is already calucated in mp_register_ioapic
so all that really needs to be done is to save the caluclated value
in nr_ioapic_registers.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
---
 arch/x86/kernel/apic/io_apic.c |   22 ++++++++--------------
 1 files changed, 8 insertions(+), 14 deletions(-)

diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
index 71988c5..c505fe0 100644
--- a/arch/x86/kernel/apic/io_apic.c
+++ b/arch/x86/kernel/apic/io_apic.c
@@ -1944,20 +1944,8 @@ static struct { int pin, apic; } ioapic_i8259 = { -1, -1 };
 
 void __init enable_IO_APIC(void)
 {
-	union IO_APIC_reg_01 reg_01;
 	int i8259_apic, i8259_pin;
 	int apic;
-	unsigned long flags;
-
-	/*
-	 * The number of IO-APIC IRQ registers (== #pins):
-	 */
-	for (apic = 0; apic &lt; nr_ioapics; apic++) {
-		raw_spin_lock_irqsave(&amp;ioapic_lock, flags);
-		reg_01.raw = io_apic_read(apic, 1);
-		raw_spin_unlock_irqrestore(&amp;ioapic_lock, flags);
-		nr_ioapic_registers[apic] = reg_01.bits.entries+1;
-	}
 
 	if (!legacy_pic-&gt;nr_legacy_irqs)
 		return;
@@ -4284,6 +4272,7 @@ static int bad_ioapic(unsigned long address)
 void __init mp_register_ioapic(int id, u32 address, u32 gsi_base)
 {
 	int idx = 0;
+	int entries;
 
 	if (bad_ioapic(address))
 		return;
@@ -4302,9 +4291,14 @@ void __init mp_register_ioapic(int id, u32 address, u32 gsi_base)
 	 * Build basic GSI lookup table to facilitate gsi-&gt;io_apic lookups
 	 * and to prevent reprogramming of IOAPIC pins (PCI GSIs).
 	 */
+	entries = io_apic_get_redir_entries(idx);
 	mp_gsi_routing[idx].gsi_base = gsi_base;
-	mp_gsi_routing[idx].gsi_end = gsi_base +
-	    io_apic_get_redir_entries(idx) - 1;
+	mp_gsi_routing[idx].gsi_end = gsi_base + entries - 1;
+
+	/*
+	 * The number of IO-APIC IRQ ...[ message continues ]

[view original message]

From: tip-bot for Eric W. Biederman

Subject: [tip:x86/irq] x86, ioapic: Move nr_ioapic_registers calculation to mp_register_ioapic.

Date: Tuesday, May 4, 2010 - 7:09 pm

Commit-ID:  7716a5c4ff5f1f3dc5e9edcab125cbf7fceef0af
Gitweb:     http://git.kernel.org/tip/7716a5c4ff5f1f3dc5e9edcab125cbf7fceef0af
Author:     Eric W. Biederman &lt;ebiederm@xmission.com&gt;
AuthorDate: Tue, 30 Mar 2010 01:07:12 -0700
Committer:  H. Peter Anvin &lt;hpa@zytor.com&gt;
CommitDate: Tue, 4 May 2010 13:35:03 -0700

x86, ioapic: Move nr_ioapic_registers calculation to mp_register_ioapic.

Now that all ioapic registration happens in mp_register_ioapic we can
move the calculation of nr_ioapic_registers there from enable_IO_APIC.
The number of ioapic registers is already calucated in mp_register_ioapic
so all that really needs to be done is to save the caluclated value
in nr_ioapic_registers.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
LKML-Reference: &lt;1269936436-7039-11-git-send-email-ebiederm@xmission.com&gt;
Signed-off-by: H. Peter Anvin &lt;hpa@zytor.com&gt;
---
 arch/x86/kernel/apic/io_apic.c |   22 ++++++++--------------
 1 files changed, 8 insertions(+), 14 deletions(-)

diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
index f807255..dae9240 100644
--- a/arch/x86/kernel/apic/io_apic.c
+++ b/arch/x86/kernel/apic/io_apic.c
@@ -1953,20 +1953,8 @@ static struct { int pin, apic; } ioapic_i8259 = { -1, -1 };
 
 void __init enable_IO_APIC(void)
 {
-	union IO_APIC_reg_01 reg_01;
 	int i8259_apic, i8259_pin;
 	int apic;
-	unsigned long flags;
-
-	/*
-	 * The number of IO-APIC IRQ registers (== #pins):
-	 */
-	for (apic = 0; apic &lt; nr_ioapics; apic++) {
-		raw_spin_lock_irqsave(&amp;ioapic_lock, flags);
-		reg_01.raw = io_apic_read(apic, 1);
-		raw_spin_unlock_irqrestore(&amp;ioapic_lock, flags);
-		nr_ioapic_registers[apic] = reg_01.bits.entries+1;
-	}
 
 	if (!legacy_pic-&gt;nr_legacy_irqs)
 		return;
@@ -4293,6 +4281,7 @@ static int bad_ioapic(unsigned long address)
 void __init mp_register_ioapic(int id, u32 address, u32 gsi_base)
 {
 	int idx = 0;
+	int entries;
 
 	if (bad_ioapic(address))
 		return;
@@ -4311,9 +4300,14 @@ void __init ...[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: [PATCH 01/15] x86 acpi/irq: Introduce apci_isa_irq_to_gsi

Date: Tuesday, March 30, 2010 - 1:07 am

From: Eric W. Biederman &lt;ebiederm@xmission.com&gt;

There are a number of cases where the current code makes the assumption
that isa irqs identity map to the first 16 acpi global system intereupts.
In most instances that assumption is correct as that is the required
behaviour in dual i8259 mode and the default behavior in ioapic mode.

However there are some systems out there that take advantage of acpis
interrupt remapping  for the isa irqs to have a completely different
mapping of isa_irq to gsi.

Introduce acpi_isa_irq_to_gsi to perform this mapping explicitly in the
code that needs it.  Initially this will be just the current assumed
identity mapping to ensure it's introduction does not cause regressions.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
---
 arch/ia64/kernel/acpi.c     |    8 ++++++++
 arch/x86/kernel/acpi/boot.c |    8 ++++++++
 include/linux/acpi.h        |    1 +
 3 files changed, 17 insertions(+), 0 deletions(-)

diff --git a/arch/ia64/kernel/acpi.c b/arch/ia64/kernel/acpi.c
index f1c9f70..16921bd 100644
--- a/arch/ia64/kernel/acpi.c
+++ b/arch/ia64/kernel/acpi.c
@@ -784,6 +784,14 @@ int acpi_gsi_to_irq(u32 gsi, unsigned int *irq)
 	return 0;
 }
 
+int acpi_isa_irq_to_gsi(unsigned isa_irq, u32 *gsi)
+{
+	if (isa_irq &gt;= 16)
+		return -1;
+	*gsi = isa_irq;
+	return 0;
+}
+
 /*
  *  ACPI based hotplug CPU support
  */
diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
index 0061ea2..f96060d 100644
--- a/arch/x86/kernel/acpi/boot.c
+++ b/arch/x86/kernel/acpi/boot.c
@@ -457,6 +457,14 @@ int acpi_gsi_to_irq(u32 gsi, unsigned int *irq)
 	return 0;
 }
 
+int acpi_isa_irq_to_gsi(unsigned isa_irq, u32 *gsi)
+{
+	if (isa_irq &gt;= 16)
+		return -1;
+	*gsi = isa_irq;
+	return 0;
+}
+
 /*
  * success: return IRQ number (&gt;=0)
  * failure: return &lt; 0
diff --git a/include/linux/acpi.h b/include/linux/acpi.h
index b926afe..7a937da 100644
--- a/include/linux/acpi.h
+++ b/include/linux/acpi.h
@@ -116,6 +116,7 @@ ...[ message continues ]

[view original message]

From: tip-bot for Eric W. Biederman

Subject: [tip:x86/irq] x86, acpi/irq: Introduce apci_isa_irq_to_gsi

Date: Tuesday, May 4, 2010 - 7:06 pm

Commit-ID:  2c2df8418ac7908eec4558407b83f16739006c54
Gitweb:     http://git.kernel.org/tip/2c2df8418ac7908eec4558407b83f16739006c54
Author:     Eric W. Biederman &lt;ebiederm@xmission.com&gt;
AuthorDate: Tue, 30 Mar 2010 01:07:02 -0700
Committer:  H. Peter Anvin &lt;hpa@zytor.com&gt;
CommitDate: Tue, 4 May 2010 13:34:23 -0700

x86, acpi/irq: Introduce apci_isa_irq_to_gsi

There are a number of cases where the current code makes the assumption
that isa irqs identity map to the first 16 acpi global system intereupts.
In most instances that assumption is correct as that is the required
behaviour in dual i8259 mode and the default behavior in ioapic mode.

However there are some systems out there that take advantage of acpis
interrupt remapping  for the isa irqs to have a completely different
mapping of isa_irq to gsi.

Introduce acpi_isa_irq_to_gsi to perform this mapping explicitly in the
code that needs it.  Initially this will be just the current assumed
identity mapping to ensure it's introduction does not cause regressions.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
LKML-Reference: &lt;1269936436-7039-1-git-send-email-ebiederm@xmission.com&gt;
Signed-off-by: H. Peter Anvin &lt;hpa@zytor.com&gt;
---
 arch/ia64/kernel/acpi.c     |    8 ++++++++
 arch/x86/kernel/acpi/boot.c |    8 ++++++++
 include/linux/acpi.h        |    1 +
 3 files changed, 17 insertions(+), 0 deletions(-)

diff --git a/arch/ia64/kernel/acpi.c b/arch/ia64/kernel/acpi.c
index 4d1a7e9..c6c90f3 100644
--- a/arch/ia64/kernel/acpi.c
+++ b/arch/ia64/kernel/acpi.c
@@ -785,6 +785,14 @@ int acpi_gsi_to_irq(u32 gsi, unsigned int *irq)
 	return 0;
 }
 
+int acpi_isa_irq_to_gsi(unsigned isa_irq, u32 *gsi)
+{
+	if (isa_irq &gt;= 16)
+		return -1;
+	*gsi = isa_irq;
+	return 0;
+}
+
 /*
  *  ACPI based hotplug CPU support
  */
diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
index cd40aba..da718d6 100644
--- a/arch/x86/kernel/acpi/boot.c
+++ b/arch/x86/kernel/acpi/boot.c
@@ -458,6 +458,14 @@ ...[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: [PATCH 02/15] x86 acpi/irq: Teach acpi_get_override_irq to take a gsi not an isa_irq

Date: Tuesday, March 30, 2010 - 1:07 am

From: Eric W. Biederman &lt;ebiederm@xmission.com&gt;

In perverse acpi implementations the isa irqs are not identity mapped
to the first 16 gsi.  Furthermore at least the extended interrupt
resource capability may return gsi's and not isa irqs.  So since
what we get from acpi is a gsi teach acpi_get_overrride_irq to
operate on a gsi instead of an isa_irq.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
---
 arch/x86/kernel/apic/io_apic.c |   23 ++++++++++++++---------
 include/linux/acpi.h           |    4 ++--
 2 files changed, 16 insertions(+), 11 deletions(-)

diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
index e4e0ddc..54ba1f7 100644
--- a/arch/x86/kernel/apic/io_apic.c
+++ b/arch/x86/kernel/apic/io_apic.c
@@ -4073,22 +4073,27 @@ int __init io_apic_get_version(int ioapic)
 	return reg_01.bits.version;
 }
 
-int acpi_get_override_irq(int bus_irq, int *trigger, int *polarity)
+int acpi_get_override_irq(u32 gsi, int *trigger, int *polarity)
 {
-	int i;
+	int ioapic, pin, idx;
 
 	if (skip_ioapic_setup)
 		return -1;
 
-	for (i = 0; i &lt; mp_irq_entries; i++)
-		if (mp_irqs[i].irqtype == mp_INT &amp;&amp;
-		    mp_irqs[i].srcbusirq == bus_irq)
-			break;
-	if (i &gt;= mp_irq_entries)
+	ioapic = mp_find_ioapic(gsi);
+	if (ioapic &lt; 0)
+		return -1;
+
+	pin = mp_find_ioapic_pin(ioapic, gsi);
+	if (pin &lt; 0)
+		return -1;
+
+	idx = find_irq_entry(ioapic, pin, mp_INT);
+	if (idx &lt; 0)
 		return -1;
 
-	*trigger = irq_trigger(i);
-	*polarity = irq_polarity(i);
+	*trigger = irq_trigger(idx);
+	*polarity = irq_polarity(idx);
 	return 0;
 }
 
diff --git a/include/linux/acpi.h b/include/linux/acpi.h
index 7a937da..3da73f5 100644
--- a/include/linux/acpi.h
+++ b/include/linux/acpi.h
@@ -119,9 +119,9 @@ int acpi_gsi_to_irq (u32 gsi, unsigned int *irq);
 int acpi_isa_irq_to_gsi (unsigned isa_irq, u32 *gsi);
 
 #ifdef CONFIG_X86_IO_APIC
-extern int acpi_get_override_irq(int bus_irq, int *trigger, int *polarity);
+extern int ...[ message continues ]

[view original message]

From: tip-bot for Eric W. Biederman

Subject: [tip:x86/irq] x86, acpi/irq: Teach acpi_get_override_irq to take a gsi not an isa_irq

Date: Tuesday, May 4, 2010 - 7:07 pm

Commit-ID:  9a0a91bb56d2915cdb8585717de38376ad20fef9
Gitweb:     http://git.kernel.org/tip/9a0a91bb56d2915cdb8585717de38376ad20fef9
Author:     Eric W. Biederman &lt;ebiederm@xmission.com&gt;
AuthorDate: Tue, 30 Mar 2010 01:07:03 -0700
Committer:  H. Peter Anvin &lt;hpa@zytor.com&gt;
CommitDate: Tue, 4 May 2010 13:34:27 -0700

x86, acpi/irq: Teach acpi_get_override_irq to take a gsi not an isa_irq

In perverse acpi implementations the isa irqs are not identity mapped
to the first 16 gsi.  Furthermore at least the extended interrupt
resource capability may return gsi's and not isa irqs.  So since
what we get from acpi is a gsi teach acpi_get_overrride_irq to
operate on a gsi instead of an isa_irq.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
LKML-Reference: &lt;1269936436-7039-2-git-send-email-ebiederm@xmission.com&gt;
Signed-off-by: H. Peter Anvin &lt;hpa@zytor.com&gt;
---
 arch/x86/kernel/apic/io_apic.c |   23 ++++++++++++++---------
 include/linux/acpi.h           |    4 ++--
 2 files changed, 16 insertions(+), 11 deletions(-)

diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
index 127b871..73ec928 100644
--- a/arch/x86/kernel/apic/io_apic.c
+++ b/arch/x86/kernel/apic/io_apic.c
@@ -4082,22 +4082,27 @@ int __init io_apic_get_version(int ioapic)
 	return reg_01.bits.version;
 }
 
-int acpi_get_override_irq(int bus_irq, int *trigger, int *polarity)
+int acpi_get_override_irq(u32 gsi, int *trigger, int *polarity)
 {
-	int i;
+	int ioapic, pin, idx;
 
 	if (skip_ioapic_setup)
 		return -1;
 
-	for (i = 0; i &lt; mp_irq_entries; i++)
-		if (mp_irqs[i].irqtype == mp_INT &amp;&amp;
-		    mp_irqs[i].srcbusirq == bus_irq)
-			break;
-	if (i &gt;= mp_irq_entries)
+	ioapic = mp_find_ioapic(gsi);
+	if (ioapic &lt; 0)
+		return -1;
+
+	pin = mp_find_ioapic_pin(ioapic, gsi);
+	if (pin &lt; 0)
+		return -1;
+
+	idx = find_irq_entry(ioapic, pin, mp_INT);
+	if (idx &lt; 0)
 		return -1;
 
-	*trigger = irq_trigger(i);
-	*polarity = irq_polarity(i);
+	*trigger = ...[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: [PATCH 03/15] x86 acpi/irq: pci device dev->irq is an isa irq not a gsi

Date: Tuesday, March 30, 2010 - 1:07 am

From: Eric W. Biederman &lt;ebiederm@xmission.com&gt;

Strictly speaking on x86 (where acpi is used) dev-&gt;irq must be
a dual i8259 irq input aka an isa irq.  Therefore we should translate
that isa irq into a gsi before passing it to a function that
takes a gsi.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
---
 drivers/acpi/pci_irq.c |    8 +++++---
 1 files changed, 5 insertions(+), 3 deletions(-)

diff --git a/drivers/acpi/pci_irq.c b/drivers/acpi/pci_irq.c
index 843699e..703541d 100644
--- a/drivers/acpi/pci_irq.c
+++ b/drivers/acpi/pci_irq.c
@@ -400,11 +400,13 @@ int acpi_pci_irq_enable(struct pci_dev *dev)
 	 * driver reported one, then use it. Exit in any case.
 	 */
 	if (gsi &lt; 0) {
+		u32 dev_gsi;
 		dev_warn(&amp;dev-&gt;dev, &quot;PCI INT %c: no GSI&quot;, pin_name(pin));
 		/* Interrupt Line values above 0xF are forbidden */
-		if (dev-&gt;irq &gt; 0 &amp;&amp; (dev-&gt;irq &lt;= 0xF)) {
-			printk(&quot; - using IRQ %d\n&quot;, dev-&gt;irq);
-			acpi_register_gsi(&amp;dev-&gt;dev, dev-&gt;irq,
+		if (dev-&gt;irq &gt; 0 &amp;&amp; (dev-&gt;irq &lt;= 0xF) &amp;&amp;
+		    (acpi_isa_irq_to_gsi(dev-&gt;irq, &amp;dev_gsi) == 0)) {
+			printk(&quot; - using ISA IRQ %d\n&quot;, dev-&gt;irq);
+			acpi_register_gsi(&amp;dev-&gt;dev, dev_gsi,
 					  ACPI_LEVEL_SENSITIVE,
 					  ACPI_ACTIVE_LOW);
 			return 0;
-- 
1.6.5.2.143.g8cc62

--

[ message continues ]

[view original message]

From: tip-bot for Eric W. Biederman

Subject: [tip:x86/irq] x86, acpi/irq: pci device dev->irq is an isa irq not a gsi

Date: Tuesday, May 4, 2010 - 7:07 pm

Commit-ID:  414d3448dbcb40807a1265ace64b2576ef919fbe
Gitweb:     http://git.kernel.org/tip/414d3448dbcb40807a1265ace64b2576ef919fbe
Author:     Eric W. Biederman &lt;ebiederm@xmission.com&gt;
AuthorDate: Tue, 30 Mar 2010 01:07:04 -0700
Committer:  H. Peter Anvin &lt;hpa@zytor.com&gt;
CommitDate: Tue, 4 May 2010 13:34:30 -0700

x86, acpi/irq: pci device dev-&gt;irq is an isa irq not a gsi

Strictly speaking on x86 (where acpi is used) dev-&gt;irq must be
a dual i8259 irq input aka an isa irq.  Therefore we should translate
that isa irq into a gsi before passing it to a function that
takes a gsi.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
LKML-Reference: &lt;1269936436-7039-3-git-send-email-ebiederm@xmission.com&gt;
Signed-off-by: H. Peter Anvin &lt;hpa@zytor.com&gt;
---
 drivers/acpi/pci_irq.c |    8 +++++---
 1 files changed, 5 insertions(+), 3 deletions(-)

diff --git a/drivers/acpi/pci_irq.c b/drivers/acpi/pci_irq.c
index b0a71ec..e4804fb 100644
--- a/drivers/acpi/pci_irq.c
+++ b/drivers/acpi/pci_irq.c
@@ -401,11 +401,13 @@ int acpi_pci_irq_enable(struct pci_dev *dev)
 	 * driver reported one, then use it. Exit in any case.
 	 */
 	if (gsi &lt; 0) {
+		u32 dev_gsi;
 		dev_warn(&amp;dev-&gt;dev, &quot;PCI INT %c: no GSI&quot;, pin_name(pin));
 		/* Interrupt Line values above 0xF are forbidden */
-		if (dev-&gt;irq &gt; 0 &amp;&amp; (dev-&gt;irq &lt;= 0xF)) {
-			printk(&quot; - using IRQ %d\n&quot;, dev-&gt;irq);
-			acpi_register_gsi(&amp;dev-&gt;dev, dev-&gt;irq,
+		if (dev-&gt;irq &gt; 0 &amp;&amp; (dev-&gt;irq &lt;= 0xF) &amp;&amp;
+		    (acpi_isa_irq_to_gsi(dev-&gt;irq, &amp;dev_gsi) == 0)) {
+			printk(&quot; - using ISA IRQ %d\n&quot;, dev-&gt;irq);
+			acpi_register_gsi(&amp;dev-&gt;dev, dev_gsi,
 					  ACPI_LEVEL_SENSITIVE,
 					  ACPI_ACTIVE_LOW);
 			return 0;
--

[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: [PATCH 08/15] x86 ioapic: Fix the types of gsi values

Date: Tuesday, March 30, 2010 - 1:07 am

From: Eric W. Biederman &lt;ebiederm@xmission.com&gt;

This patches fixes the types of gsi_base and gsi_end values in
struct mp_ioapic_gsi, and the gsi parameter of mp_find_ioapic
and mp_find_ioapic_pin

A gsi is cannonically a u32, not an int.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
---
 arch/x86/include/asm/io_apic.h |   10 +++++-----
 arch/x86/kernel/apic/io_apic.c |    4 ++--
 2 files changed, 7 insertions(+), 7 deletions(-)

diff --git a/arch/x86/include/asm/io_apic.h b/arch/x86/include/asm/io_apic.h
index 35832a0..feeaf0d 100644
--- a/arch/x86/include/asm/io_apic.h
+++ b/arch/x86/include/asm/io_apic.h
@@ -180,12 +180,12 @@ extern void ioapic_write_entry(int apic, int pin,
 extern void setup_ioapic_ids_from_mpc(void);
 
 struct mp_ioapic_gsi{
-	int gsi_base;
-	int gsi_end;
+	u32 gsi_base;
+	u32 gsi_end;
 };
 extern struct mp_ioapic_gsi  mp_gsi_routing[];
-int mp_find_ioapic(int gsi);
-int mp_find_ioapic_pin(int ioapic, int gsi);
+int mp_find_ioapic(u32 gsi);
+int mp_find_ioapic_pin(int ioapic, u32 gsi);
 void __init mp_register_ioapic(int id, u32 address, u32 gsi_base);
 extern void __init pre_init_apic_IRQ0(void);
 
@@ -197,7 +197,7 @@ static const int timer_through_8259 = 0;
 static inline void ioapic_init_mappings(void)	{ }
 static inline void ioapic_insert_resources(void) { }
 static inline void probe_nr_irqs_gsi(void)	{ }
-static inline int mp_find_ioapic(int gsi) { return 0; }
+static inline int mp_find_ioapic(u32 gsi) { return 0; }
 
 struct io_apic_irq_attr;
 static inline int io_apic_set_pci_routing(struct device *dev, int irq,
diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
index 7e139bb..4611c43 100644
--- a/arch/x86/kernel/apic/io_apic.c
+++ b/arch/x86/kernel/apic/io_apic.c
@@ -4238,7 +4238,7 @@ void __init ioapic_insert_resources(void)
 	}
 }
 
-int mp_find_ioapic(int gsi)
+int mp_find_ioapic(u32 gsi)
 {
 	int i = 0;
 
@@ -4253,7 +4253,7 @@ int mp_find_ioapic(int gsi)
 	return -1;
 }
 
-int ...[ message continues ]

[view original message]

From: tip-bot for Eric W. Biederman

Subject: [tip:x86/irq] x86, ioapic: Fix the types of gsi values

Date: Tuesday, May 4, 2010 - 7:08 pm

Commit-ID:  eddb0c55a14074d6bac8c2ef169aefd7e2c6f139
Gitweb:     http://git.kernel.org/tip/eddb0c55a14074d6bac8c2ef169aefd7e2c6f139
Author:     Eric W. Biederman &lt;ebiederm@xmission.com&gt;
AuthorDate: Tue, 30 Mar 2010 01:07:09 -0700
Committer:  H. Peter Anvin &lt;hpa@zytor.com&gt;
CommitDate: Tue, 4 May 2010 13:34:52 -0700

x86, ioapic: Fix the types of gsi values

This patches fixes the types of gsi_base and gsi_end values in
struct mp_ioapic_gsi, and the gsi parameter of mp_find_ioapic
and mp_find_ioapic_pin

A gsi is cannonically a u32, not an int.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
LKML-Reference: &lt;1269936436-7039-8-git-send-email-ebiederm@xmission.com&gt;
Signed-off-by: H. Peter Anvin &lt;hpa@zytor.com&gt;
---
 arch/x86/include/asm/io_apic.h |   10 +++++-----
 arch/x86/kernel/apic/io_apic.c |    4 ++--
 2 files changed, 7 insertions(+), 7 deletions(-)

diff --git a/arch/x86/include/asm/io_apic.h b/arch/x86/include/asm/io_apic.h
index 35832a0..feeaf0d 100644
--- a/arch/x86/include/asm/io_apic.h
+++ b/arch/x86/include/asm/io_apic.h
@@ -180,12 +180,12 @@ extern void ioapic_write_entry(int apic, int pin,
 extern void setup_ioapic_ids_from_mpc(void);
 
 struct mp_ioapic_gsi{
-	int gsi_base;
-	int gsi_end;
+	u32 gsi_base;
+	u32 gsi_end;
 };
 extern struct mp_ioapic_gsi  mp_gsi_routing[];
-int mp_find_ioapic(int gsi);
-int mp_find_ioapic_pin(int ioapic, int gsi);
+int mp_find_ioapic(u32 gsi);
+int mp_find_ioapic_pin(int ioapic, u32 gsi);
 void __init mp_register_ioapic(int id, u32 address, u32 gsi_base);
 extern void __init pre_init_apic_IRQ0(void);
 
@@ -197,7 +197,7 @@ static const int timer_through_8259 = 0;
 static inline void ioapic_init_mappings(void)	{ }
 static inline void ioapic_insert_resources(void) { }
 static inline void probe_nr_irqs_gsi(void)	{ }
-static inline int mp_find_ioapic(int gsi) { return 0; }
+static inline int mp_find_ioapic(u32 gsi) { return 0; }
 
 struct io_apic_irq_attr;
 static inline int io_apic_set_pci_routing(struct ...[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: [PATCH 04/15] x86 acpi/irq: Fix acpi_sci_ioapic_setup so it has both bus_irq and gsi

Date: Tuesday, March 30, 2010 - 1:07 am

From: Eric W. Biederman &lt;ebiederm@xmission.com&gt;

Currently acpi_sci_ioapic_setup calls mp_override_legacy_irq with
bus_irq == gsi, which is wrong if we are comming from an override
Instead pass the bus_irq into acpi_sci_ioapic_setup.

This fix was inspired by a similar fix from:
Yinghai Lu &lt;yinghai@kernel.org&gt;

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
---
 arch/x86/kernel/acpi/boot.c |   12 +++++++-----
 1 files changed, 7 insertions(+), 5 deletions(-)

diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
index f96060d..1742888 100644
--- a/arch/x86/kernel/acpi/boot.c
+++ b/arch/x86/kernel/acpi/boot.c
@@ -312,7 +312,7 @@ acpi_parse_ioapic(struct acpi_subtable_header * header, const unsigned long end)
 /*
  * Parse Interrupt Source Override for the ACPI SCI
  */
-static void __init acpi_sci_ioapic_setup(u32 gsi, u16 polarity, u16 trigger)
+static void __init acpi_sci_ioapic_setup(u8 bus_irq, u16 polarity, u16 trigger, u32 gsi)
 {
 	if (trigger == 0)	/* compatible SCI trigger is level */
 		trigger = 3;
@@ -332,7 +332,7 @@ static void __init acpi_sci_ioapic_setup(u32 gsi, u16 polarity, u16 trigger)
 	 * If GSI is &lt; 16, this will update its flags,
 	 * else it will create a new mp_irqs[] entry.
 	 */
-	mp_override_legacy_irq(gsi, polarity, trigger, gsi);
+	mp_override_legacy_irq(bus_irq, polarity, trigger, gsi);
 
 	/*
 	 * stash over-ride to indicate we've been here
@@ -356,9 +356,10 @@ acpi_parse_int_src_ovr(struct acpi_subtable_header * header,
 	acpi_table_print_madt_entry(header);
 
 	if (intsrc-&gt;source_irq == acpi_gbl_FADT.sci_interrupt) {
-		acpi_sci_ioapic_setup(intsrc-&gt;global_irq,
+		acpi_sci_ioapic_setup(intsrc-&gt;source_irq,
 				      intsrc-&gt;inti_flags &amp; ACPI_MADT_POLARITY_MASK,
-				      (intsrc-&gt;inti_flags &amp; ACPI_MADT_TRIGGER_MASK) &gt;&gt; 2);
+				      (intsrc-&gt;inti_flags &amp; ACPI_MADT_TRIGGER_MASK) &gt;&gt; 2,
+				      intsrc-&gt;global_irq);
 		return 0;
 	}
 
@@ -1161,7 +1162,8 @@ static int __init ...[ message continues ]

[view original message]

From: tip-bot for Eric W. Biederman

Subject: [tip:x86/irq] x86, acpi/irq: Fix acpi_sci_ioapic_setup so it has both bus_irq and gsi

Date: Tuesday, May 4, 2010 - 7:07 pm

Commit-ID:  9d2062b879495649bb525cf7979126da2e45d288
Gitweb:     http://git.kernel.org/tip/9d2062b879495649bb525cf7979126da2e45d288
Author:     Eric W. Biederman &lt;ebiederm@xmission.com&gt;
AuthorDate: Tue, 30 Mar 2010 01:07:05 -0700
Committer:  H. Peter Anvin &lt;hpa@zytor.com&gt;
CommitDate: Tue, 4 May 2010 13:34:34 -0700

x86, acpi/irq: Fix acpi_sci_ioapic_setup so it has both bus_irq and gsi

Currently acpi_sci_ioapic_setup calls mp_override_legacy_irq with
bus_irq == gsi, which is wrong if we are comming from an override
Instead pass the bus_irq into acpi_sci_ioapic_setup.

This fix was inspired by a similar fix from:
Yinghai Lu &lt;yinghai@kernel.org&gt;

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
LKML-Reference: &lt;1269936436-7039-4-git-send-email-ebiederm@xmission.com&gt;
Signed-off-by: H. Peter Anvin &lt;hpa@zytor.com&gt;
---
 arch/x86/kernel/acpi/boot.c |   12 +++++++-----
 1 files changed, 7 insertions(+), 5 deletions(-)

diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c
index da718d6..0a036dc 100644
--- a/arch/x86/kernel/acpi/boot.c
+++ b/arch/x86/kernel/acpi/boot.c
@@ -313,7 +313,7 @@ acpi_parse_ioapic(struct acpi_subtable_header * header, const unsigned long end)
 /*
  * Parse Interrupt Source Override for the ACPI SCI
  */
-static void __init acpi_sci_ioapic_setup(u32 gsi, u16 polarity, u16 trigger)
+static void __init acpi_sci_ioapic_setup(u8 bus_irq, u16 polarity, u16 trigger, u32 gsi)
 {
 	if (trigger == 0)	/* compatible SCI trigger is level */
 		trigger = 3;
@@ -333,7 +333,7 @@ static void __init acpi_sci_ioapic_setup(u32 gsi, u16 polarity, u16 trigger)
 	 * If GSI is &lt; 16, this will update its flags,
 	 * else it will create a new mp_irqs[] entry.
 	 */
-	mp_override_legacy_irq(gsi, polarity, trigger, gsi);
+	mp_override_legacy_irq(bus_irq, polarity, trigger, gsi);
 
 	/*
 	 * stash over-ride to indicate we've been here
@@ -357,9 +357,10 @@ acpi_parse_int_src_ovr(struct acpi_subtable_header * header,
 ...[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: [PATCH 06/15] x86 ioapic: Only export mp_find_ioapic and mp_find_ioapic_pin in io_apic.h

Date: Tuesday, March 30, 2010 - 1:07 am

From: Eric W. Biederman &lt;ebiederm@xmission.com&gt;

Multiple declarations of the same function in different headers
is a pain to maintain.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
---
 arch/x86/include/asm/mpspec.h |    4 ----
 1 files changed, 0 insertions(+), 4 deletions(-)

diff --git a/arch/x86/include/asm/mpspec.h b/arch/x86/include/asm/mpspec.h
index d8bf23a..29994f0 100644
--- a/arch/x86/include/asm/mpspec.h
+++ b/arch/x86/include/asm/mpspec.h
@@ -106,10 +106,6 @@ struct device;
 extern int mp_register_gsi(struct device *dev, u32 gsi, int edge_level,
 				 int active_high_low);
 extern int acpi_probe_gsi(void);
-#ifdef CONFIG_X86_IO_APIC
-extern int mp_find_ioapic(int gsi);
-extern int mp_find_ioapic_pin(int ioapic, int gsi);
-#endif
 #else /* !CONFIG_ACPI: */
 static inline int acpi_probe_gsi(void)
 {
-- 
1.6.5.2.143.g8cc62

--

[ message continues ]

[view original message]

From: tip-bot for Eric W. Biederman

Subject: [tip:x86/irq] x86, ioapic: Only export mp_find_ioapic and mp_find_ioapic_pin in io_apic.h

Date: Tuesday, May 4, 2010 - 7:08 pm

Commit-ID:  9638fa521e42c9281c874c6b5a382b1ced4ee496
Gitweb:     http://git.kernel.org/tip/9638fa521e42c9281c874c6b5a382b1ced4ee496
Author:     Eric W. Biederman &lt;ebiederm@xmission.com&gt;
AuthorDate: Tue, 30 Mar 2010 01:07:07 -0700
Committer:  H. Peter Anvin &lt;hpa@zytor.com&gt;
CommitDate: Tue, 4 May 2010 13:34:44 -0700

x86, ioapic: Only export mp_find_ioapic and mp_find_ioapic_pin in io_apic.h

Multiple declarations of the same function in different headers
is a pain to maintain.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
LKML-Reference: &lt;1269936436-7039-6-git-send-email-ebiederm@xmission.com&gt;
Signed-off-by: H. Peter Anvin &lt;hpa@zytor.com&gt;
---
 arch/x86/include/asm/mpspec.h |    4 ----
 1 files changed, 0 insertions(+), 4 deletions(-)

diff --git a/arch/x86/include/asm/mpspec.h b/arch/x86/include/asm/mpspec.h
index d8bf23a..29994f0 100644
--- a/arch/x86/include/asm/mpspec.h
+++ b/arch/x86/include/asm/mpspec.h
@@ -106,10 +106,6 @@ struct device;
 extern int mp_register_gsi(struct device *dev, u32 gsi, int edge_level,
 				 int active_high_low);
 extern int acpi_probe_gsi(void);
-#ifdef CONFIG_X86_IO_APIC
-extern int mp_find_ioapic(int gsi);
-extern int mp_find_ioapic_pin(int ioapic, int gsi);
-#endif
 #else /* !CONFIG_ACPI: */
 static inline int acpi_probe_gsi(void)
 {
--

[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: [PATCH 07/15] x86 ioapic: Fix io_apic_redir_entries to return the number of entries.

Date: Tuesday, March 30, 2010 - 1:07 am

From: Eric W. Biederman &lt;ebiederm@xmission.com&gt;

io_apic_redir_entries has a huge conceptual bug.  It returns the maximum
redirection entry not the number of redirection entries.  Which simply
does not match what the name of the function.  This just caught me
and it caught  Feng Tang, and  Len Brown when they wrote sfi_parse_ioapic.

Modify io_apic_redir_entries to actually return the number of redirection
entries, and fix the callers so that they properly handle receiving the
number of the number of redirection table entries, instead of the
number of redirection table entries less one.

While the usage in sfi.c does not show up in this patch it is fixed
by virtue of the fact that io_apic_redir_entries now has the semantics
sfi_parse_ioapic most reasonably expects.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
---
 arch/x86/kernel/apic/io_apic.c |   10 +++++++---
 1 files changed, 7 insertions(+), 3 deletions(-)

diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
index 54ba1f7..7e139bb 100644
--- a/arch/x86/kernel/apic/io_apic.c
+++ b/arch/x86/kernel/apic/io_apic.c
@@ -3846,7 +3846,11 @@ int __init io_apic_get_redir_entries (int ioapic)
 	reg_01.raw = io_apic_read(ioapic, 1);
 	raw_spin_unlock_irqrestore(&amp;ioapic_lock, flags);
 
-	return reg_01.bits.entries;
+	/* The register returns the maximum index redir index
+	 * supported, which is one less than the total number of redir
+	 * entries.
+	 */
+	return reg_01.bits.entries + 1;
 }
 
 void __init probe_nr_irqs_gsi(void)
@@ -3862,7 +3866,7 @@ void __init probe_nr_irqs_gsi(void)
 
 		nr = 0;
 		for (idx = 0; idx &lt; nr_ioapics; idx++)
-			nr += io_apic_get_redir_entries(idx) + 1;
+			nr += io_apic_get_redir_entries(idx);
 
 		if (nr &gt; nr_irqs_gsi)
 			nr_irqs_gsi = nr;
@@ -4297,7 +4301,7 @@ void __init mp_register_ioapic(int id, u32 address, u32 gsi_base)
 	 */
 	mp_gsi_routing[idx].gsi_base = gsi_base;
 	mp_gsi_routing[idx].gsi_end = gsi_base +
-	    ...[ message continues ]

[view original message]

From: tip-bot for Eric W. Biederman

Subject: [tip:x86/irq] x86, ioapic: Fix io_apic_redir_entries to return the number of entries.

Date: Tuesday, May 4, 2010 - 7:08 pm

Commit-ID:  4b6b19a1c7302477653d799a53d48063dd53d555
Gitweb:     http://git.kernel.org/tip/4b6b19a1c7302477653d799a53d48063dd53d555
Author:     Eric W. Biederman &lt;ebiederm@xmission.com&gt;
AuthorDate: Tue, 30 Mar 2010 01:07:08 -0700
Committer:  H. Peter Anvin &lt;hpa@zytor.com&gt;
CommitDate: Tue, 4 May 2010 13:34:48 -0700

x86, ioapic: Fix io_apic_redir_entries to return the number of entries.

io_apic_redir_entries has a huge conceptual bug.  It returns the maximum
redirection entry not the number of redirection entries.  Which simply
does not match what the name of the function.  This just caught me
and it caught  Feng Tang, and  Len Brown when they wrote sfi_parse_ioapic.

Modify io_apic_redir_entries to actually return the number of redirection
entries, and fix the callers so that they properly handle receiving the
number of the number of redirection table entries, instead of the
number of redirection table entries less one.

While the usage in sfi.c does not show up in this patch it is fixed
by virtue of the fact that io_apic_redir_entries now has the semantics
sfi_parse_ioapic most reasonably expects.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
LKML-Reference: &lt;1269936436-7039-7-git-send-email-ebiederm@xmission.com&gt;
Signed-off-by: H. Peter Anvin &lt;hpa@zytor.com&gt;
---
 arch/x86/kernel/apic/io_apic.c |   10 +++++++---
 1 files changed, 7 insertions(+), 3 deletions(-)

diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
index 73ec928..0a053e6 100644
--- a/arch/x86/kernel/apic/io_apic.c
+++ b/arch/x86/kernel/apic/io_apic.c
@@ -3855,7 +3855,11 @@ int __init io_apic_get_redir_entries (int ioapic)
 	reg_01.raw = io_apic_read(ioapic, 1);
 	raw_spin_unlock_irqrestore(&amp;ioapic_lock, flags);
 
-	return reg_01.bits.entries;
+	/* The register returns the maximum index redir index
+	 * supported, which is one less than the total number of redir
+	 * entries.
+	 */
+	return reg_01.bits.entries + 1;
 }
 
 void __init probe_nr_irqs_gsi(void)
@@ ...[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: [PATCH 10/15] x86 ioapic: In mpparse use mp_register_ioapic

Date: Tuesday, March 30, 2010 - 1:07 am

From: Eric W. Biederman &lt;ebiederm@xmission.com&gt;

Long ago MP_ioapic_info was the primary way of setting up our
ioapic data structures and mp_register_ioapic was a compatibility
shim for acpi code.  Now the situation is reversed and
and mp_register_ioapic is the primary way of setting up our
ioapic data structures.

Keep the setting up of ioapic data structures uniform by
having mp_register_ioapic call mp_register_ioapic.

This changes a few fields:

- type: is now hardset to MP_IOAPIC but type had to
  bey MP_IOAPIC or MP_ioapic_info would not have been called.

- flags: is now hard coded to MPC_APIC_USABLE.
  We require flags to contain at least MPC_APIC_USEBLE in
  MP_ioapic_info and we don't ever examine flags so dropping
  a few flags that might possibly exist that we have never
  used is harmless.

- apicaddr: Unchanged

- apicver: Read from the ioapic instead of using the cached
  hardware value in the MP table.  The real hardware value
  will be more accurate.

- apicid: Now verified to be unique and changed if it is not.
  If the BIOS got this right this is a noop.  If the BIOS did
  not fixing things appears to be the better solution.

This adds gsi_base and gsi_end values to our ioapics defined with
the mpatable, which will make our lives simpler later since
we can always assume gsi_base and gsi_end are valid.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
---
 arch/x86/kernel/mpparse.c |   25 +------------------------
 1 files changed, 1 insertions(+), 24 deletions(-)

diff --git a/arch/x86/kernel/mpparse.c b/arch/x86/kernel/mpparse.c
index a2c1edd..22ba4f3 100644
--- a/arch/x86/kernel/mpparse.c
+++ b/arch/x86/kernel/mpparse.c
@@ -115,21 +115,6 @@ static void __init MP_bus_info(struct mpc_bus *m)
 		printk(KERN_WARNING &quot;Unknown bustype %s - ignoring\n&quot;, str);
 }
 
-static int bad_ioapic(unsigned long address)
-{
-	if (nr_ioapics &gt;= MAX_IO_APICS) {
-		printk(KERN_ERR &quot;ERROR: Max # of I/O APICs (%d) exceeded &quot;
-		       &quot;(found %d)\n&quot;, ...[ message continues ]

[view original message]

From: tip-bot for Eric W. Biederman

Subject: [tip:x86/irq] x86, ioapic: In mpparse use mp_register_ioapic

Date: Tuesday, May 4, 2010 - 7:09 pm

Commit-ID:  cf7500c0ea133d66f8449d86392d83f840102632
Gitweb:     http://git.kernel.org/tip/cf7500c0ea133d66f8449d86392d83f840102632
Author:     Eric W. Biederman &lt;ebiederm@xmission.com&gt;
AuthorDate: Tue, 30 Mar 2010 01:07:11 -0700
Committer:  H. Peter Anvin &lt;hpa@zytor.com&gt;
CommitDate: Tue, 4 May 2010 13:34:59 -0700

x86, ioapic: In mpparse use mp_register_ioapic

Long ago MP_ioapic_info was the primary way of setting up our
ioapic data structures and mp_register_ioapic was a compatibility
shim for acpi code.  Now the situation is reversed and
and mp_register_ioapic is the primary way of setting up our
ioapic data structures.

Keep the setting up of ioapic data structures uniform by
having mp_register_ioapic call mp_register_ioapic.

This changes a few fields:

- type: is now hardset to MP_IOAPIC but type had to
  bey MP_IOAPIC or MP_ioapic_info would not have been called.

- flags: is now hard coded to MPC_APIC_USABLE.
  We require flags to contain at least MPC_APIC_USEBLE in
  MP_ioapic_info and we don't ever examine flags so dropping
  a few flags that might possibly exist that we have never
  used is harmless.

- apicaddr: Unchanged

- apicver: Read from the ioapic instead of using the cached
  hardware value in the MP table.  The real hardware value
  will be more accurate.

- apicid: Now verified to be unique and changed if it is not.
  If the BIOS got this right this is a noop.  If the BIOS did
  not fixing things appears to be the better solution.

This adds gsi_base and gsi_end values to our ioapics defined with
the mpatable, which will make our lives simpler later since
we can always assume gsi_base and gsi_end are valid.

Signed-off-by: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
LKML-Reference: &lt;1269936436-7039-10-git-send-email-ebiederm@xmission.com&gt;
Signed-off-by: H. Peter Anvin &lt;hpa@zytor.com&gt;
---
 arch/x86/kernel/mpparse.c |   25 +------------------------
 1 files changed, 1 insertions(+), 24 deletions(-)

diff --git a/arch/x86/kernel/mpparse.c ...[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: Re: [PATCH 0/15] Start coping gsis < 16 that are not isa irqs. v2

Date: Monday, May 3, 2010 - 4:21 pm

ping

Where are we at with this patchset?

Eric

--

[ message continues ]

[view original message]

From: Len Brown

Subject: Re: [PATCH 0/14] Start coping gsis < 16 that are not isa irqs.

Date: Wednesday, March 31, 2010 - 7:02 pm

You might run into trouble here on the ES7000 -- though I don't know
if anybody is booting a modern kernel on one of those these days.

IIR, ES7000 treated the bottom 16 as a special case.
When there was an irq shortage, I think they used overrides
to map higher PCI irqs into the empty spots below 16,
but I think to make room for them they may have mapped
some of the ISA irqs to high numbers.  Fuzzy memory
on this at the moment...

cheers,
Len Brown, Intel Open Source Technology Center

--

[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: Re: [PATCH 0/14] Start coping gsis < 16 that are not isa irqs.

Date: Wednesday, March 31, 2010 - 8:31 pm

A couple of things.

The ES7000 change is trivially safe because despite differences in how
the numbers are computed I have made that transform that the es7000
does always apply.

I also talked to Natalie about this.  I can't be certain about this
but from your description, from Natalies memories and from looking
at the code.  I believe the problem on the ES7000 is exactly what
I am fixing in the code.  Natalie has threatened to test this on
an ES7000.

The case of interest is this:
Weird but valid platforms where
GSI 0-15 are not ISA irqs.

Some high numbered GSIs are ISA irqs and use interrupt source
overrides to describe them.

That is the case on the IBM platform that regressed a bit ago
and Iranna D Ankad bisected a failure on. On that IBM platform
GSI 0-15 (because the are not ISA irqs) are unusable today.

Everything points to this problem I an fixing being the problem with
early ES7000 machines, including the memory of their boot firmware
developers complaint that they were in spec.

Eric

--

[ message continues ]

[view original message]

From: Yinghai Lu

Subject: [PATCH 05/10] x86: use vector_desc instead of vector_irq

Date: Sunday, March 21, 2010 - 6:36 pm

Eric pointed out that radix tree version of irq_to_desc will magnify delay on
the path of handle_irq.

use vector_desc to reduce the calling of irq_to_desc.

next step: need to change all ack, mask, umask, eoi for all irq_chip to take irq_desc

-v2: irq should be unsigned in 32bit handle_irq according to Eric
     also reset vector_desc for lguest in setup_irq
-v3: keep irq in x+execute_on_irq_stack() ...
-v4: update after legacy_pic

Signed-off-by: Yinghai Lu &lt;yinghai@kernel.org&gt;
Acked-by: &quot;Eric W. Biederman&quot; &lt;ebiederm@xmission.com&gt;
---
 arch/x86/include/asm/desc.h    |    2 +-
 arch/x86/include/asm/hw_irq.h  |   13 ++++---
 arch/x86/include/asm/irq.h     |    3 +-
 arch/x86/kernel/apic/io_apic.c |   77 +++++++++++++++++++++-------------------
 arch/x86/kernel/irq.c          |   15 ++++----
 arch/x86/kernel/irq_32.c       |   13 ++++---
 arch/x86/kernel/irq_64.c       |    7 +---
 arch/x86/kernel/irqinit.c      |   14 +++----
 arch/x86/kernel/smpboot.c      |    2 +-
 arch/x86/kernel/uv_irq.c       |    2 +-
 arch/x86/kernel/vmiclock_32.c  |    2 +-
 arch/x86/lguest/boot.c         |    7 +++-
 12 files changed, 82 insertions(+), 75 deletions(-)

diff --git a/arch/x86/include/asm/desc.h b/arch/x86/include/asm/desc.h
index 617bd56..25c5635 100644
--- a/arch/x86/include/asm/desc.h
+++ b/arch/x86/include/asm/desc.h
@@ -334,7 +334,7 @@ static inline void set_intr_gate(unsigned int n, void *addr)
 }
 
 extern int first_system_vector;
-/* used_vectors is BITMAP for irq is not managed by percpu vector_irq */
+/* used_vectors is BITMAP for irq is not managed by percpu vector_desc */
 extern unsigned long used_vectors[];
 
 static inline void alloc_system_vector(int vector)
diff --git a/arch/x86/include/asm/hw_irq.h b/arch/x86/include/asm/hw_irq.h
index 767d3f8..d23cf94 100644
--- a/arch/x86/include/asm/hw_irq.h
+++ b/arch/x86/include/asm/hw_irq.h
@@ -97,7 +97,8 @@ struct irq_cfg {
 };
 
 extern struct irq_cfg *irq_cfg(unsigned int);
-extern int assign_irq_vector(int, ...[ message continues ]

[view original message]

From: Thomas Gleixner

Subject: Re: [PATCH 05/10] x86: use vector_desc instead of vector_irq

Date: Monday, March 22, 2010 - 6:58 am

__clear_irq_vector(desc, desc-&gt;chip_data);




  That printk is confusing. It's not lacking an irq handler. The
  vector is simply not assigned.

Thanks,

	tglx
--

[ message continues ]

[view original message]

From: Eric W. Biederman

Subject: Re: [PATCH 05/10] x86: use vector_desc instead of vector_irq

Date: Monday, March 22, 2010 - 7:04 am

Well at least originally DECLARE_PER_CPU chocked when given a complex
type. Does:
DECLARE_PER_CPU(struct irq_desc *[NR_VECTORS], vector_desc);


Long evolution.  Do you have a suggestion of better wording?

Eric
--

[ message continues ]

[view original message]

From: Thomas Gleixner

Subject: Re: [PATCH 05/10] x86: use vector_desc instead of vector_irq

Date: Monday, March 22, 2010 - 7:16 am

I really have a hard time to see how assigning a void pointer to a
struct irq_cfg pointer is anymore type safe than using the void

You mean hysterical raisins. Ok, how about:

    pr_emerg(&quot;irq: %d.d irq vector not assigned\n&quot;, ...);

Thanks,

	tglx
--

[ message continues ]

[view original message]

From: Yinghai Lu

Subject: [PATCH 01/10] irq: move some interrupt arch_* functions into struct irq_chip.

Date: Sunday, March 21, 2010 - 6:36 pm

From: Ian Campbell &lt;ian.campbell@citrix.com&gt;

Move arch_init_copy_chip_data and arch_free_chip_data into function
pointers in struct irq_chip since they operate on irq_desc-&gt;chip_data.

arch_init_chip_data cannot be moved into struct irq_chip because
irq_desc-&gt;chip is not known at the time the irq_desc is setup. Instead
rename arch_init_chip_data to arch_init_irq_desc for PowerPC, the
only other user, whose usage better matches the new name.

To replace the x86 arch_init_chip_data functionality
irq_to_desc_alloc_node now takes a pointer to a function to allocate
the chip data. This is necessary to ensure the allocation happens
under the correct locking at the core level. On PowerPC and SH
architectures (the other users of irq_to_desc_alloc_node) pass in NULL
which retains existing chip_data behaviour.

I've retained the chip_data behaviour for uv_irq although it isn't
clear to me if these interrupt types support migration or how closely
related to the APIC modes they really are. If it weren't for this the
x86_{init,copy,free}_chip_data functions could be static to
io_apic.c.

I've tested by booting on an 64 bit x86 system with sparse IRQ enabled
and 32 bit without, but it's not clear to me what actions I need to
take to actually exercise some of these code paths.

-v4: yinghai add irq_to_desc_alloc_node_x...
     so could leave default path not changed...

Signed-off-by: Ian Campbell &lt;ian.campbell@citrix.com&gt;
Acked-by: Michael Ellerman &lt;michael@ellerman.id.au&gt; [PowerPC rename portion]
Cc: Thomas Gleixner &lt;tglx@linutronix.de&gt;
Cc: Ingo Molnar &lt;mingo@redhat.com&gt;
Cc: H. Peter Anvin &lt;hpa@zytor.com&gt;
Cc: Eric W. Biederman &lt;ebiederm@xmission.com&gt;
Cc: Yinghai Lu &lt;yinghai@kernel.org&gt;
Cc: Jeremy Fitzhardinge &lt;jeremy@goop.org&gt;
Cc: Benjamin Herrenschmidt &lt;benh@kernel.crashing.org&gt;
Cc: Paul Mackerras &lt;paulus@samba.org&gt;
Cc: x86@kernel.org
Cc: linuxppc-dev@ozlabs.org
Cc: linux-kernel@vger.kernel.org
Cc: Rusty Russell &lt;rusty@rustcorp.com.au&gt;
Cc: lguest@ozlabs.org
Cc: Paul Mundt ...[ message continues ]

[view original message]

From: Michael Ellerman

Subject: Re: [PATCH 01/10] irq: move some interrupt arch_* functions into struct irq_chip.

Date: Sunday, March 21, 2010 - 6:56 pm

This is a bit feral, that is the init_chip_data_fn.


Which is really just a hack to avoid an if (xen) check isn't it?

It looks to me like this should just be done via a current machine
vector or platform routine, in the same way as powerpc and (I think)

cheers

[ message continues ]

[view original message]

From: Yinghai Lu

Subject: Re: [PATCH 01/10] irq: move some interrupt arch_* functions into struct irq_chip.

Date: Sunday, March 21, 2010 - 8:32 pm

arch/powerpc/kernel/irq.c      |    2 +-

diff --git a/arch/powerpc/kernel/irq.c b/arch/powerpc/kernel/irq.c
index 64f6f20..cafd378 100644
--- a/arch/powerpc/kernel/irq.c
+++ b/arch/powerpc/kernel/irq.c
@@ -1088,7 +1088,7 @@ int arch_early_irq_init(void)
 	return 0;
 }
 
-int arch_init_chip_data(struct irq_desc *desc, int node)
+int arch_init_irq_desc(struct irq_desc *desc, int node, init_chip_data_fn fn)
 {
 	desc-&gt;status |= IRQ_NOREQUEST;
 	return 0;

looks like xen in same run time, some irqs need x86_init_chip_data,
and some may need xen_init_chip_data later.

Thanks

Yinghai
--

[ message continues ]

[view original message]

From: Paul Mundt

Subject: Re: [PATCH 01/10] irq: move some interrupt arch_* functions into struct irq_chip.

Date: Tuesday, March 23, 2010 - 12:10 am

This API seems to be going from bad to worse. Initially we had
arch_init_chip_data(), which had precisely nothing to do with chip_data
and only concerned itself with irq_desc, and now you're renaming it to
something sensible but also trying to bolt on some ad-hoc chip_data
relation at the same time thereby nullifying any benefit obtained from
renaming the function in the first place.

Renaming to xxx_irq_desc() while preserving the existing prototypes would
I'm afraid I am unable to grasp the meaning of this sentence, or what
precisely this has to do with not being able to utilize platform ops to
get this sorted out on x86.

You're effectively trying to have the hardirq code pass around a function
pointer for you that ultimately only serves to bail out on certain code
paths if you're running under xen, which is a concern for how the
platform chooses to initialize the irq desc, none of this has any value
or relevance to the hardirq code outside of that. The fact that the
hardirq code doesn't do anything with this information other than pass it
around for your platform should already be a clear indicator that this is
the wrong way to go.

From a cursory look at the x86 code, this looks like precisely the sort
of thing that arch/x86/include/asm/x86_init.h is well suited for, and
indeed you already have a x86_init_irqs to expand on as needed.

The function pointer thing itself is also a bit unorthodox to say the
least. You're introducing and passing around an opaque type just so you
can get to a 'return 0' in the xen case as far as I can tell, so you
could also just make arch_init_chip_data() a weak symbol and clobber it
in the xen case, no?
--

[ message continues ]

[view original message]

From: Ian Campbell

Subject: Re: [PATCH 01/10] irq: move some interrupt arch_* functions into struct irq_chip.

Date: Wednesday, March 24, 2010 - 6:33 am

The ultimate aim is to have Xen use the chip data to store the event
channel information relating to each IRQ instead of keeping it in a
static NR_IRQs array, the new function only returns 0 as a placeholder
until this can be put in place (which depends on these changes), it
could as well have been left out for the time being (e.g. passing NULL

A single kernel is able to boot native or Xen so a weak symbol doesn't
really work, having the function pointer at the arch level is one
similar option, I've replied to Thomas' similar suggestion.

Ian.

--

[ message continues ]

[view original message]

From: Thomas Gleixner

Subject: Re: [PATCH 01/10] irq: move some interrupt arch_* functions into struct irq_chip.

Date: Monday, March 22, 2010 - 3:19 am

Not sure about that. These functions are solely used by x86 and there
is really no need to generalize them. The problem you try to solve is
x86/xen specific and can be solved by x86_platform_ops as well w/o

Err, that argument is totally bogus. The calling convention of
irq_to_desc_alloc_node and arch_init_chip_data/arch_init_irq_desc is
still the same. It does not explain why the heck we need that function
pointer at all.

AFAICT the function pointer to irq_to_desc_alloc_node is completely
pointless. It just solves a Xen/x86 specific problem which can be

Aside of the general objection against this, please use descriptive
function names and do not distinguish functions by adding random
characters which tell us absolutely nothing about the purpose.

Thanks,

	tglx
--

[ message continues ]

[view original message]

From: Ian Campbell

Subject: Re: [PATCH 01/10] irq: move some interrupt arch_* functions into struct irq_chip.

Date: Wednesday, March 24, 2010 - 6:32 am

I thought the idea of struct irq_chip was to allow the potential for
multiple IRQ controllers in a system? Given that it seems that struct
irq_desc-&gt;chip_data ought to be available for use by whichever struct
irq_chip is managing a given interrupt. At the moment this is not
possible because we step around the abstraction using these arch_*
methods.

Although this might be unusual on x86 I think it is not uncommon in the
embedded world to have an architectural interrupt controller cascading
through to various different IRQ controllers/multiplexors, from random
FPGA based things, to GPIO controllers and things like superio chips
etc.

Currently the set of architectures which typically have this sort of
thing are disjoint from the ones which make use of struct
irq_desc-&gt;chip_data but with the growing use of embedded-x86 is this not
something worth considering? (Genuine question, I've been out of the
embedded space for a while now so maybe my experiences are out of date
or I'm overestimating the role of embedded-x86 etc).

Xen is a bit more of a specialised case than the above since it would
like to replace the architectural interrupt handling but I think the
broad requirements on the irq_chip interface are the same. Going forward
it is possible/likely that we would like to be able to make Xen event
channels available via a cascade model as well -- demultiplexing one (or
more?) x86 architectural interrupts into event channels would be part of

I have no problem with that if that is the x86/irq maintainer's
preference, I just thought it would be nicer to solve what I saw as an

I agree on this one. More generally I would say that the number of
existing users of this interface is small enough that _if_ we decide we
need to modify it then we should just bite the bullet and do that
instead of building compatibility layers around stuff. For this reason I
think my original patch was preferable to this version (general
objections not withstanding).

Ian.

--

[ message continues ]

[view original message]

From: Thomas Gleixner

Subject: Re: [PATCH 01/10] irq: move some interrupt arch_* functions into struct irq_chip.

Date: Wednesday, March 24, 2010 - 10:44 am

Right, but you have exactly _ONE_ irq_chip associated to an irq_desc,
but that same irq_chip can be associated to several irq_descs. So
irq_desc-&gt;data is there to provide data for the irq_chip functions
depending on what irq they handle (e.g. base_address ...). 

irq_desc-&gt;chip_data is set when the irq_chip is assigned to the
irq_desc.

So there is no point in having functions in irq_chip to set
irq_desc-&gt;chip_data.

Thanks,

	tglx
--

[ message continues ]

[view original message]

From: Ian Campbell

Subject: Re: [PATCH 01/10] irq: move some interrupt arch_* functions into struct irq_chip.

Date: Wednesday, March 24, 2010 - 12:16 pm

So how do you know which is the appropriate irq_chip specific function
to call given an irq_desc that you want to copy/free/migrate? The
contents of the chip_data pointer will take different forms for
different irq_chips. The way the generic code is currently structured it
appears you can't (or at least don't) just do a shallow copy by copying
the irq_desc-&gt;chip_data pointer itself -- you need to do a deep copy
using a function which understands that type of chip_data.

How is this operation different to having pointers in irq_chip for
enabling/disabling/masking interrupts for each irq_chip?

Ian.

--

[ message continues ]

[view original message]

From: Thomas Gleixner

Subject: Re: [PATCH 01/10] irq: move some interrupt arch_* functions into struct irq_chip.

Date: Wednesday, March 24, 2010 - 2:25 pm

The design of sparse_irq or to be honest the lack of design grew that
crap and it's not only this detail which is a nightmare. That pointer
should probably be simply copied. Either that or if the chip data
require to be node bound we need something along the line:

struct sparse_irq_chip_data {
       void *data;
       void (*copy)(...);
       void (*free)(...);
};

and a corresponding field in irq_desc.

I'm looking into sparse_irq right now anyway because it has other way

Because that's the purpose of the irq_chip perhaps ?

Thanks,

	tglx
--

[ message continues ]

Navigation

Popular discussions


linux-kernel:
Greg KH	Og dreams of kernels
Jens Axboe	[PATCH 31/33] Fusion: sg chaining support
Arnd Bergmann	Re: finding your own dead "CONFIG_" variables
Mark Brown	[PATCH 2/2] Subject: natsemi: Allow users to disable workaround for DspCfg reset
Tony Breeds	[LGUEST] Look in object dir for .config
git:
Brian Downing	Re: Git in a Nutshell guide
John Benes	Re: master has some toys
Matthias Lederhofer	[PATCH 4/7] introduce GIT_WORK_TREE to specify the work tree
Alexander Sulfrian	[RFC/PATCH] RE: git calls SSH_ASKPASS even if DISPLAY is not set
Junio C Hamano	Re: Rss produced by git is not valid xml?
git-commits-head:
Linux Kernel Mailing List	iSeries: fix section mismatch in iseries_veth
Linux Kernel Mailing List	ixbge: remove TX lock and redo TX accounting.
Linux Kernel Mailing List	ixgbe: fix several counter register errata
Linux Kernel Mailing List	b43: fix build with CONFIG_SSB_PCIHOST=n
Linux Kernel Mailing List	9p: block-based virtio client
linux-netdev:
Michael Breuer	Re: [PATCH] af_packet: Don't use skb after dev_queue_xmit()
Michael Breuer	Re: [PATCH] af_packet: Don't use skb after dev_queue_xmit()
David Daney	[PATCH 5/7] Staging: Octeon Ethernet: Convert to NAPI.
Wolfgang Grandegger	[PATCH net-next v4 1/3] can: mscan: fix improper return if dlc < 8 in start_xmi...
Amit Kumar Salecha	[PATCHv3 NEXT 2/2] NET: Add Qlogic ethernet driver for CNA devices
openbsd-misc:
Theo de Raadt