Merge branch 'x86-64'

* x86-64: (83 commits)
  [PATCH] x86_64: x86_64 stack usage debugging
  [PATCH] x86_64: (resend) x86_64 stack overflow debugging
  [PATCH] x86_64: msi_apic.c build fix
  [PATCH] x86_64: i386/x86-64 Add nmi watchdog support for new Intel CPUs
  [PATCH] x86_64: Avoid broadcasting NMI IPIs
  [PATCH] x86_64: fix apic error on bootup
  [PATCH] x86_64: enlarge window for stack growth
  [PATCH] x86_64: Minor string functions optimizations
  [PATCH] x86_64: Move export symbols to their C functions
  [PATCH] x86_64: Standardize i386/x86_64 handling of NMI_VECTOR
  [PATCH] x86_64: Fix modular pc speaker
  [PATCH] x86_64: remove sys32_ni_syscall()
  [PATCH] x86_64: Do not use -ffunction-sections for modules
  [PATCH] x86_64: Add cpu_relax to apic_wait_icr_idle
  [PATCH] x86_64: adjust kstack_depth_to_print default
  [PATCH] i386/x86-64: adjust /proc/interrupts column headings
  [PATCH] x86_64: Fix race in cpu_local_* on preemptible kernels
  [PATCH] x86_64: Fix fast check in safe_smp_processor_id
  [PATCH] x86_64: x86_64 setup.c - printing cmp related boottime information
  [PATCH] i386/x86-64/ia64: Move polling flag into thread_info_status
  ...

Manual resolve of trivial conflict in arch/i386/kernel/Makefile

5 files changed, 40 insertions, 63 deletions

diff --git a/drivers/acpi/processor_idle.c b/drivers/acpi/processor_idle.c
index a5f4f2aa007a..8a74bf3efd8e 100644
--- a/drivers/acpi/processor_idle.c
+++ b/drivers/acpi/processor_idle.c
@@ -206,11 +206,11 @@ acpi_processor_power_activate(struct acpi_processor *pr,
 
 static void acpi_safe_halt(void)
 {
-	clear_thread_flag(TIF_POLLING_NRFLAG);
+	current_thread_info()->status &= ~TS_POLLING;
 	smp_mb__after_clear_bit();
 	if (!need_resched())
 		safe_halt();
-	set_thread_flag(TIF_POLLING_NRFLAG);
+	current_thread_info()->status |= TS_POLLING;
 }
 
 static atomic_t c3_cpu_count;
@@ -330,10 +330,10 @@ static void acpi_processor_idle(void)
 	 * Invoke the current Cx state to put the processor to sleep.
 	 */
 	if (cx->type == ACPI_STATE_C2 || cx->type == ACPI_STATE_C3) {
-		clear_thread_flag(TIF_POLLING_NRFLAG);
+		current_thread_info()->status &= ~TS_POLLING;
 		smp_mb__after_clear_bit();
 		if (need_resched()) {
-			set_thread_flag(TIF_POLLING_NRFLAG);
+			current_thread_info()->status |= TS_POLLING;
 			local_irq_enable();
 			return;
 		}
@@ -376,7 +376,7 @@ static void acpi_processor_idle(void)
 #endif
 		/* Re-enable interrupts */
 		local_irq_enable();
-		set_thread_flag(TIF_POLLING_NRFLAG);
+		current_thread_info()->status |= TS_POLLING;
 		/* Compute time (ticks) that we were actually asleep */
 		sleep_ticks =
 		    ticks_elapsed(t1, t2) - cx->latency_ticks - C2_OVERHEAD;
@@ -420,7 +420,7 @@ static void acpi_processor_idle(void)
 #endif
 		/* Re-enable interrupts */
 		local_irq_enable();
-		set_thread_flag(TIF_POLLING_NRFLAG);
+		current_thread_info()->status |= TS_POLLING;
 		/* Compute time (ticks) that we were actually asleep */
 		sleep_ticks =
 		    ticks_elapsed(t1, t2) - cx->latency_ticks - C3_OVERHEAD;
diff --git a/drivers/char/agp/Kconfig b/drivers/char/agp/Kconfig
index 46685a540772..9826a399fa02 100644
--- a/drivers/char/agp/Kconfig
+++ b/drivers/char/agp/Kconfig
@@ -55,9 +55,9 @@ config AGP_AMD
 	  X on AMD Irongate, 761, and 762 chipsets.
 
 config AGP_AMD64
-	tristate "AMD Opteron/Athlon64 on-CPU GART support" if !GART_IOMMU
+	tristate "AMD Opteron/Athlon64 on-CPU GART support" if !IOMMU
 	depends on AGP && X86
-	default y if GART_IOMMU
+	default y if IOMMU
 	help
 	  This option gives you AGP support for the GLX component of
 	  X using the on-CPU northbridge of the AMD Athlon64/Opteron CPUs.
diff --git a/drivers/char/agp/amd64-agp.c b/drivers/char/agp/amd64-agp.c
index ac3c33a2e37d..f690ee8cb732 100644
--- a/drivers/char/agp/amd64-agp.c
+++ b/drivers/char/agp/amd64-agp.c
@@ -15,11 +15,9 @@
 #include <linux/agp_backend.h>
 #include <linux/mmzone.h>
 #include <asm/page.h>		/* PAGE_SIZE */
+#include <asm/k8.h>
 #include "agp.h"
 
-/* Will need to be increased if AMD64 ever goes >8-way. */
-#define MAX_HAMMER_GARTS   8
-
 /* PTE bits. */
 #define GPTE_VALID	1
 #define GPTE_COHERENT	2
@@ -53,28 +51,12 @@
 #define ULI_X86_64_HTT_FEA_REG		0x50
 #define ULI_X86_64_ENU_SCR_REG		0x54
 
-static int nr_garts;
-static struct pci_dev * hammers[MAX_HAMMER_GARTS];
-
 static struct resource *aperture_resource;
 static int __initdata agp_try_unsupported = 1;
 
-#define for_each_nb() for(gart_iterator=0;gart_iterator<nr_garts;gart_iterator++)
-
-static void flush_amd64_tlb(struct pci_dev *dev)
-{
-	u32 tmp;
-
-	pci_read_config_dword (dev, AMD64_GARTCACHECTL, &tmp);
-	tmp |= INVGART;
-	pci_write_config_dword (dev, AMD64_GARTCACHECTL, tmp);
-}
-
 static void amd64_tlbflush(struct agp_memory *temp)
 {
-	int gart_iterator;
-	for_each_nb()
-		flush_amd64_tlb(hammers[gart_iterator]);
+	k8_flush_garts();
 }
 
 static int amd64_insert_memory(struct agp_memory *mem, off_t pg_start, int type)
@@ -153,7 +135,7 @@ static int amd64_fetch_size(void)
 	u32 temp;
 	struct aper_size_info_32 *values;
 
-	dev = hammers[0];
+	dev = k8_northbridges[0];
 	if (dev==NULL)
 		return 0;
 
@@ -201,9 +183,6 @@ static u64 amd64_configure (struct pci_dev *hammer, u64 gatt_table)
 	tmp &= ~(DISGARTCPU | DISGARTIO);
 	pci_write_config_dword(hammer, AMD64_GARTAPERTURECTL, tmp);
 
-	/* keep CPU's coherent. */
-	flush_amd64_tlb (hammer);
-
 	return aper_base;
 }
 
@@ -222,13 +201,14 @@ static struct aper_size_info_32 amd_8151_sizes[7] =
 static int amd_8151_configure(void)
 {
 	unsigned long gatt_bus = virt_to_gart(agp_bridge->gatt_table_real);
-	int gart_iterator;
+	int i;
 
 	/* Configure AGP regs in each x86-64 host bridge. */
-	for_each_nb() {
+        for (i = 0; i < num_k8_northbridges; i++) {
 		agp_bridge->gart_bus_addr =
-				amd64_configure(hammers[gart_iterator],gatt_bus);
+				amd64_configure(k8_northbridges[i], gatt_bus);
 	}
+	k8_flush_garts();
 	return 0;
 }
 
@@ -236,12 +216,13 @@ static int amd_8151_configure(void)
 static void amd64_cleanup(void)
 {
 	u32 tmp;
-	int gart_iterator;
-	for_each_nb() {
+	int i;
+        for (i = 0; i < num_k8_northbridges; i++) {
+		struct pci_dev *dev = k8_northbridges[i];
 		/* disable gart translation */
-		pci_read_config_dword (hammers[gart_iterator], AMD64_GARTAPERTURECTL, &tmp);
+		pci_read_config_dword (dev, AMD64_GARTAPERTURECTL, &tmp);
 		tmp &= ~AMD64_GARTEN;
-		pci_write_config_dword (hammers[gart_iterator], AMD64_GARTAPERTURECTL, tmp);
+		pci_write_config_dword (dev, AMD64_GARTAPERTURECTL, tmp);
 	}
 }
 
@@ -311,7 +292,7 @@ static int __devinit aperture_valid(u64 aper, u32 size)
 /*
  * W*s centric BIOS sometimes only set up the aperture in the AGP
  * bridge, not the northbridge. On AMD64 this is handled early
- * in aperture.c, but when GART_IOMMU is not enabled or we run
+ * in aperture.c, but when IOMMU is not enabled or we run
  * on a 32bit kernel this needs to be redone.
  * Unfortunately it is impossible to fix the aperture here because it's too late
  * to allocate that much memory. But at least error out cleanly instead of
@@ -361,17 +342,15 @@ static __devinit int fix_northbridge(struct pci_dev *nb, struct pci_dev *agp,
 
 static __devinit int cache_nbs (struct pci_dev *pdev, u32 cap_ptr)
 {
-	struct pci_dev *loop_dev = NULL;
-	int i = 0;
-
-	/* cache pci_devs of northbridges. */
-	while ((loop_dev = pci_get_device(PCI_VENDOR_ID_AMD, 0x1103, loop_dev))
-			!= NULL) {
-		if (i == MAX_HAMMER_GARTS) {
-			printk(KERN_ERR PFX "Too many northbridges for AGP\n");
-			return -1;
-		}
-		if (fix_northbridge(loop_dev, pdev, cap_ptr) < 0) {
+	int i;
+
+	if (cache_k8_northbridges() < 0)
+		return -ENODEV;
+
+	i = 0;
+	for (i = 0; i < num_k8_northbridges; i++) {
+		struct pci_dev *dev = k8_northbridges[i];
+		if (fix_northbridge(dev, pdev, cap_ptr) < 0) {
 			printk(KERN_ERR PFX "No usable aperture found.\n");
 #ifdef __x86_64__
 			/* should port this to i386 */
@@ -379,10 +358,8 @@ static __devinit int cache_nbs (struct pci_dev *pdev, u32 cap_ptr)
 #endif
 			return -1;
 		}
-		hammers[i++] = loop_dev;
 	}
-		nr_garts = i;
-	return i == 0 ? -1 : 0;
+	return 0;
 }
 
 /* Handle AMD 8151 quirks */
@@ -450,7 +427,7 @@ static int __devinit uli_agp_init(struct pci_dev *pdev)
 	}
 
 	/* shadow x86-64 registers into ULi registers */
-	pci_read_config_dword (hammers[0], AMD64_GARTAPERTUREBASE, &httfea);
+	pci_read_config_dword (k8_northbridges[0], AMD64_GARTAPERTUREBASE, &httfea);
 
 	/* if x86-64 aperture base is beyond 4G, exit here */
 	if ((httfea & 0x7fff) >> (32 - 25))
@@ -513,7 +490,7 @@ static int __devinit nforce3_agp_init(struct pci_dev *pdev)
 	pci_write_config_dword(dev1, NVIDIA_X86_64_1_APSIZE, tmp);
 
 	/* shadow x86-64 registers into NVIDIA registers */
-	pci_read_config_dword (hammers[0], AMD64_GARTAPERTUREBASE, &apbase);
+	pci_read_config_dword (k8_northbridges[0], AMD64_GARTAPERTUREBASE, &apbase);
 
 	/* if x86-64 aperture base is beyond 4G, exit here */
 	if ( (apbase & 0x7fff) >> (32 - 25) ) {
@@ -754,10 +731,6 @@ static struct pci_driver agp_amd64_pci_driver = {
 int __init agp_amd64_init(void)
 {
 	int err = 0;
-	static struct pci_device_id amd64nb[] = {
-		{ PCI_DEVICE(PCI_VENDOR_ID_AMD, 0x1103) },
-		{ },
-	};
 
 	if (agp_off)
 		return -EINVAL;
@@ -774,7 +747,7 @@ int __init agp_amd64_init(void)
 		}
 
 		/* First check that we have at least one AMD64 NB */
-		if (!pci_dev_present(amd64nb))
+		if (!pci_dev_present(k8_nb_ids))
 			return -ENODEV;
 
 		/* Look for any AGP bridge */
@@ -802,7 +775,7 @@ static void __exit agp_amd64_cleanup(void)
 
 /* On AMD64 the PCI driver needs to initialize this driver early
    for the IOMMU, so it has to be called via a backdoor. */
-#ifndef CONFIG_GART_IOMMU
+#ifndef CONFIG_IOMMU
 module_init(agp_amd64_init);
 module_exit(agp_amd64_cleanup);
 #endif
diff --git a/drivers/pci/msi-apic.c b/drivers/pci/msi-apic.c
index 0eb5fe9003a2..5ed798b319c7 100644
--- a/drivers/pci/msi-apic.c
+++ b/drivers/pci/msi-apic.c
@@ -4,6 +4,7 @@
 
 #include <linux/pci.h>
 #include <linux/irq.h>
+#include <asm/smp.h>
 
 #include "msi.h"
 
diff --git a/drivers/scsi/aacraid/comminit.c b/drivers/scsi/aacraid/comminit.c
index 35b0a6ebd3f5..7cea514e810a 100644
--- a/drivers/scsi/aacraid/comminit.c
+++ b/drivers/scsi/aacraid/comminit.c
@@ -104,8 +104,11 @@ static int aac_alloc_comm(struct aac_dev *dev, void **commaddr, unsigned long co
 	 * always true on real computers. It also has some slight problems
 	 * with the GART on x86-64. I've btw never tried DMA from PCI space
 	 * on this platform but don't be surprised if its problematic.
+	 * [AK: something is very very wrong when a driver tests this symbol.
+ 	 *  Someone should figure out what the comment writer really meant here and fix
+	 *  the code. Or just remove that bad code. ]
 	 */
-#ifndef CONFIG_GART_IOMMU
+#ifndef CONFIG_IOMMU
 	if ((num_physpages << (PAGE_SHIFT - 12)) <= AAC_MAX_HOSTPHYSMEMPAGES) {
 		init->HostPhysMemPages = 
 			cpu_to_le32(num_physpages << (PAGE_SHIFT-12));