Per our discussion, i've ported the 2.6 nforce skip timer override (and
early PCI access) code to 2.4. This fixes an issue whereupon nforce
systems have incorrect override values for irq0. Architectures affected
are i386 and x86_64
Signed-off-by: Zwane Mwaikambo <[email protected]>
# This is a BitKeeper generated diff -Nru style patch.
#
# ChangeSet
# 2005/02/18 07:53:21-07:00 [email protected]
# ACPI skip_timer_override backport from 2.6 including early PCI bridge detection.
#
# include/asm-x86_64/acpi.h
# 2005/02/18 07:53:18-07:00 [email protected] +1 -0
# ACPI skip_timer_override backport from 2.6 including early PCI bridge detection.
#
# include/asm-i386/pci-direct.h
# 2005/02/18 07:53:18-07:00 [email protected] +1 -0
# ACPI skip_timer_override backport from 2.6 including early PCI bridge detection.
#
# include/asm-i386/acpi.h
# 2005/02/18 07:53:18-07:00 [email protected] +2 -0
# ACPI skip_timer_override backport from 2.6 including early PCI bridge detection.
#
# arch/x86_64/kernel/io_apic.c
# 2005/02/18 07:53:18-07:00 [email protected] +7 -3
# ACPI skip_timer_override backport from 2.6 including early PCI bridge detection.
#
# arch/x86_64/kernel/acpi.c
# 2005/02/18 07:53:18-07:00 [email protected] +7 -0
# ACPI skip_timer_override backport from 2.6 including early PCI bridge detection.
#
# arch/i386/kernel/earlyquirk.c
# 2005/02/18 07:53:18-07:00 [email protected] +53 -0
# ACPI skip_timer_override backport from 2.6 including early PCI bridge detection.
#
# arch/i386/kernel/acpi.c
# 2005/02/18 07:53:18-07:00 [email protected] +9 -0
# ACPI skip_timer_override backport from 2.6 including early PCI bridge detection.
#
# arch/i386/kernel/Makefile
# 2005/02/18 07:53:18-07:00 [email protected] +1 -1
# ACPI skip_timer_override backport from 2.6 including early PCI bridge detection.
#
diff -Nru a/arch/i386/kernel/Makefile b/arch/i386/kernel/Makefile
--- a/arch/i386/kernel/Makefile 2005-02-18 07:53:58 -07:00
+++ b/arch/i386/kernel/Makefile 2005-02-18 07:53:58 -07:00
@@ -40,7 +40,7 @@
obj-$(CONFIG_ACPI_SLEEP) += acpi_wakeup.o
obj-$(CONFIG_SMP) += smp.o smpboot.o trampoline.o
obj-$(CONFIG_X86_LOCAL_APIC) += mpparse.o apic.o nmi.o
-obj-$(CONFIG_X86_IO_APIC) += io_apic.o
+obj-$(CONFIG_X86_IO_APIC) += io_apic.o earlyquirk.o
obj-$(CONFIG_X86_VISWS_APIC) += visws_apic.o
obj-$(CONFIG_EDD) += edd.o
diff -Nru a/arch/i386/kernel/acpi.c b/arch/i386/kernel/acpi.c
--- a/arch/i386/kernel/acpi.c 2005-02-18 07:53:58 -07:00
+++ b/arch/i386/kernel/acpi.c 2005-02-18 07:53:58 -07:00
@@ -55,6 +55,7 @@
acpi_interrupt_flags acpi_sci_flags __initdata;
int acpi_sci_override_gsi __initdata;
+int acpi_skip_timer_override __initdata;
/* --------------------------------------------------------------------------
Boot-time Configuration
-------------------------------------------------------------------------- */
@@ -320,6 +321,12 @@
return 0;
}
+ if (acpi_skip_timer_override &&
+ intsrc->bus_irq == 0 && intsrc->global_irq == 2) {
+ printk(PREFIX "BIOS IRQ0 pin2 override ignored.\n");
+ return 0;
+ }
+
mp_override_legacy_irq (
intsrc->bus_irq,
intsrc->flags.polarity,
@@ -433,6 +440,8 @@
return result;
}
+ check_acpi_pci();
+
result = acpi_blacklisted();
if (result) {
printk(KERN_NOTICE PREFIX "BIOS listed in blacklist, disabling ACPI support\n");
diff -Nru a/arch/i386/kernel/earlyquirk.c b/arch/i386/kernel/earlyquirk.c
--- a/arch/i386/kernel/earlyquirk.c 2005-02-18 07:53:58 -07:00
+++ b/arch/i386/kernel/earlyquirk.c 2005-02-18 07:53:58 -07:00
@@ -0,0 +1,53 @@
+/*
+ * Do early PCI probing for bug detection when the main PCI subsystem is
+ * not up yet.
+ */
+#include <linux/init.h>
+#include <linux/kernel.h>
+#include <linux/pci.h>
+#include <asm/pci-direct.h>
+#include <asm/acpi.h>
+
+#ifdef CONFIG_ACPI
+static int __init check_bridge(int vendor, int device)
+{
+ /* According to Nvidia all timer overrides are bogus. Just ignore
+ them all. */
+ if (vendor == PCI_VENDOR_ID_NVIDIA) {
+ acpi_skip_timer_override = 1;
+ }
+ return 0;
+}
+
+void __init check_acpi_pci(void)
+{
+ int num,slot,func;
+
+ /* Assume the machine supports type 1. If not it will
+ always read ffffffff and should not have any side effect. */
+
+ /* Poor man's PCI discovery */
+ for (num = 0; num < 32; num++) {
+ for (slot = 0; slot < 32; slot++) {
+ for (func = 0; func < 8; func++) {
+ u32 class;
+ u32 vendor;
+ class = read_pci_config(num,slot,func,
+ PCI_CLASS_REVISION);
+ if (class == 0xffffffff)
+ break;
+
+ if ((class >> 16) != PCI_CLASS_BRIDGE_PCI)
+ continue;
+
+ vendor = read_pci_config(num, slot, func,
+ PCI_VENDOR_ID);
+
+ if (check_bridge(vendor&0xffff, vendor >> 16))
+ return;
+ }
+
+ }
+ }
+}
+#endif /* CONFIG_ACPI */
diff -Nru a/arch/x86_64/kernel/acpi.c b/arch/x86_64/kernel/acpi.c
--- a/arch/x86_64/kernel/acpi.c 2005-02-18 07:53:58 -07:00
+++ b/arch/x86_64/kernel/acpi.c 2005-02-18 07:53:58 -07:00
@@ -53,6 +53,7 @@
acpi_interrupt_flags acpi_sci_flags __initdata;
int acpi_sci_override_gsi __initdata;
+int acpi_skip_timer_override __initdata;
/* --------------------------------------------------------------------------
Boot-time Configuration
-------------------------------------------------------------------------- */
@@ -330,6 +331,12 @@
if (intsrc->bus_irq == acpi_fadt.sci_int) {
acpi_sci_ioapic_setup(intsrc->global_irq,
intsrc->flags.polarity, intsrc->flags.trigger);
+ return 0;
+ }
+
+ if (acpi_skip_timer_override &&
+ intsrc->bus_irq == 0 && intsrc->global_irq == 2) {
+ printk(PREFIX "BIOS IRQ0 pin2 override ignored.\n");
return 0;
}
diff -Nru a/arch/x86_64/kernel/io_apic.c b/arch/x86_64/kernel/io_apic.c
--- a/arch/x86_64/kernel/io_apic.c 2005-02-18 07:53:58 -07:00
+++ b/arch/x86_64/kernel/io_apic.c 2005-02-18 07:53:58 -07:00
@@ -259,10 +259,14 @@
case PCI_VENDOR_ID_VIA:
return;
case PCI_VENDOR_ID_NVIDIA:
+#ifdef CONFIG_ACPI
+ /* All timer overrides on Nvidia
+ seem to be wrong. Skip them. */
+ acpi_skip_timer_override = 1;
printk(KERN_INFO
- "PCI bridge %02x:%02x from %x found. Setting \"noapic\". Overwrite with \"apic\"\n",
- num,slot,vendor);
- skip_ioapic_setup = 1;
+ "Nvidia board detected. Ignoring ACPI timer override.\n");
+#endif
+ /* RED-PEN skip them on mptables too? */
return;
}
diff -Nru a/include/asm-i386/acpi.h b/include/asm-i386/acpi.h
--- a/include/asm-i386/acpi.h 2005-02-18 07:53:58 -07:00
+++ b/include/asm-i386/acpi.h 2005-02-18 07:53:58 -07:00
@@ -121,6 +121,8 @@
extern int acpi_strict;
extern int acpi_disabled;
extern int acpi_ht;
+extern int acpi_skip_timer_override;
+void __init check_acpi_pci(void);
static inline void disable_acpi(void)
{
acpi_disabled = 1;
diff -Nru a/include/asm-i386/pci-direct.h b/include/asm-i386/pci-direct.h
--- a/include/asm-i386/pci-direct.h 2005-02-18 07:53:58 -07:00
+++ b/include/asm-i386/pci-direct.h 2005-02-18 07:53:58 -07:00
@@ -0,0 +1 @@
+#include "asm-x86_64/pci-direct.h"
diff -Nru a/include/asm-x86_64/acpi.h b/include/asm-x86_64/acpi.h
--- a/include/asm-x86_64/acpi.h 2005-02-18 07:53:58 -07:00
+++ b/include/asm-x86_64/acpi.h 2005-02-18 07:53:58 -07:00
@@ -118,6 +118,7 @@
extern int acpi_strict;
extern int acpi_disabled;
extern int acpi_ht;
+extern int acpi_skip_timer_override;
static inline void disable_acpi(void)
{
acpi_disabled = 1;
On Wed, Feb 23, 2005 at 06:40:55PM +0000, Edward Miller wrote:
> Package: kernel-source-2.4.27
> Version: 2.4.27-8
> Severity: normal
> Tags: patch
>
> The 2.6 kernel series has, since 2.6.5, had a fix for an erroneous timer
> override present in many BIOSes in nforce[23] chipsets. The 2.4 series
> is missing this, resulting in an XT_PIC timer on systems that have an
> APIC-enabled kernel. This is believed to cause system instability,
> including hard lock-ups (with no ssh). At my request, Zwane Mwaikambo
> has kindly backported the fix for 2.4.30 and Ihave found that this patch
> works almost unaltered on Debian's kernel-source-2.4.27. Bearing in mind
> the proximity of Sarge's release and especially d-i rc3, I thought I
> should send you the patch for review now.
>
> Maybe this will have to be a post-Sarge item but there is a lot of
> cheap nforce2 out there and 2.6.8 may not be suitable for everyone so I
> hope you can consider this patch for inclusion in Sarge's 2.4.27.
Thanks, looks good, though I think the Makefile portion of the patch
should be as follows. I am in the process of some rebuilds to confirm
this. I have CCed the Zwane Mwakikamo and LKML so this reaches the right
eyes.
--
Horms
diff -Nru a/arch/i386/kernel/Makefile b/arch/i386/kernel/Makefile
--- a/arch/i386/kernel/Makefile 2005-02-24 20:01:29.000000000 +0900
+++ b/arch/i386/kernel/Makefile.noedit 2005-02-24 20:03:40.000000000 +0900
@@ -36,7 +36,7 @@
obj-$(CONFIG_X86_CPUID) += cpuid.o
obj-$(CONFIG_MICROCODE) += microcode.o
obj-$(CONFIG_APM) += apm.o
-obj-$(CONFIG_ACPI_BOOT) += acpi.o earlyquirk.o
+obj-$(CONFIG_ACPI_BOOT) += acpi.o
obj-$(CONFIG_ACPI_SLEEP) += acpi_wakeup.o
obj-$(CONFIG_SMP) += smp.o smpboot.o trampoline.o
obj-$(CONFIG_X86_LOCAL_APIC) += mpparse.o apic.o nmi.o