8 files changed, 63 insertions, 52 deletions

diff --git a/arch/x86_64/kernel/Makefile b/arch/x86_64/kernel/Makefile
index 72fe60c20d39..a098a11e7755 100644
--- a/arch/x86_64/kernel/Makefile
+++ b/arch/x86_64/kernel/Makefile
@@ -43,7 +43,7 @@ CFLAGS_vsyscall.o		:= $(PROFILING) -g0
 
 bootflag-y                      += ../../i386/kernel/bootflag.o
 cpuid-$(subst m,y,$(CONFIG_X86_CPUID))  += ../../i386/kernel/cpuid.o
-topology-y                     += ../../i386/mach-default/topology.o
+topology-y                     += ../../i386/kernel/topology.o
 microcode-$(subst m,y,$(CONFIG_MICROCODE))  += ../../i386/kernel/microcode.o
 intel_cacheinfo-y               += ../../i386/kernel/cpu/intel_cacheinfo.o
 quirks-y                        += ../../i386/kernel/quirks.o
diff --git a/arch/x86_64/kernel/aperture.c b/arch/x86_64/kernel/aperture.c
index e4e2b7d01f89..a0f955b9995f 100644
--- a/arch/x86_64/kernel/aperture.c
+++ b/arch/x86_64/kernel/aperture.c
@@ -248,7 +248,7 @@ void __init iommu_hole_init(void)
                 /* Got the aperture from the AGP bridge */
         } else if (swiotlb && !valid_agp) {
                 /* Do nothing */
-        } else if ((!no_iommu && end_pfn >= MAX_DMA32_PFN) ||
+        } else if ((!no_iommu && end_pfn > MAX_DMA32_PFN) ||
                    force_iommu ||
                    valid_agp ||
                    fallback_aper_force) { 
diff --git a/arch/x86_64/kernel/apic.c b/arch/x86_64/kernel/apic.c
index e5b14c57eaa0..d70605eda333 100644
--- a/arch/x86_64/kernel/apic.c
+++ b/arch/x86_64/kernel/apic.c
@@ -962,12 +962,14 @@ void smp_apic_timer_interrupt(struct pt_regs *regs)
         irq_exit();
 }
 
+int __initdata unsync_tsc_on_multicluster;
+
 /*
  * oem_force_hpet_timer -- force HPET mode for some boxes.
  *
  * Thus far, the major user of this is IBM's Summit2 series:
  *
- * Clustered boxes may have unsynced TSC problems if they are
+ * Some clustered boxes may have unsynced TSC problems if they are
  * multi-chassis. Use available data to take a good guess.
  * If in doubt, go HPET.
  */
@@ -977,6 +979,11 @@ __cpuinit int oem_force_hpet_timer(void)
         unsigned id;
         DECLARE_BITMAP(clustermap, NUM_APIC_CLUSTERS);
 
+        /* Only do this check on IBM machines - big Unisys systems
+           use multiple clusters too, but have synchronized TSC */
+        if (!unsync_tsc_on_multicluster)
+                return 0;
+
         bitmap_zero(clustermap, NUM_APIC_CLUSTERS);
 
         for (i = 0; i < NR_CPUS; i++) {
diff --git a/arch/x86_64/kernel/io_apic.c b/arch/x86_64/kernel/io_apic.c
index 2585c1d92b26..ffed464e6b12 100644
--- a/arch/x86_64/kernel/io_apic.c
+++ b/arch/x86_64/kernel/io_apic.c
@@ -50,6 +50,8 @@ static int no_timer_check;
 
 int disable_timer_pin_1 __initdata;
 
+int timer_over_8254 __initdata = 1;
+
 /* Where if anywhere is the i8259 connect in external int mode */
 static struct { int pin, apic; } ioapic_i8259 = { -1, -1 };
 
@@ -251,6 +253,20 @@ static int __init enable_ioapic_setup(char *str)
 __setup("noapic", disable_ioapic_setup);
 __setup("apic", enable_ioapic_setup);
 
+static int __init setup_disable_8254_timer(char *s)
+{
+        timer_over_8254 = -1;
+        return 1;
+}
+static int __init setup_enable_8254_timer(char *s)
+{
+        timer_over_8254 = 2;
+        return 1;
+}
+
+__setup("disable_8254_timer", setup_disable_8254_timer);
+__setup("enable_8254_timer", setup_enable_8254_timer);
+
 #include <asm/pci-direct.h>
 #include <linux/pci_ids.h>
 #include <linux/pci.h>
@@ -309,27 +325,20 @@ void __init check_ioapic(void)
 #endif
                                         /* RED-PEN skip them on mptables too? */
                                         return;
+
+                                /* This should be actually default, but
+                                   for 2.6.16 let's do it for ATI only where
+                                   it's really needed. */
                                 case PCI_VENDOR_ID_ATI:
-                                        if (apic_runs_main_timer != 0)
-                                                break;
-#ifdef CONFIG_ACPI
-                                        /* Don't do this for laptops right
-                                           right now because their timer
-                                           doesn't necessarily tick in C2/3 */
-                                        if (acpi_fadt.revision >= 3 &&
-                        (acpi_fadt.plvl2_lat + acpi_fadt.plvl3_lat) < 1100) {
-                                                printk(KERN_INFO
-"ATI board detected, but seems to be a laptop. Timer might be shakey, sorry\n");
-                                                break;
-                                        }
-#endif                                  
+                                        if (timer_over_8254 == 1) {     
+                                                timer_over_8254 = 0;    
                                         printk(KERN_INFO
-             "ATI board detected. Using APIC/PM timer.\n");
-                                        apic_runs_main_timer = 1;
-                                        nohpet = 1;
+                "ATI board detected. Disabling timer routing over 8254.\n");
+                                        }       
                                         return;
                                 } 
 
+
                                 /* No multi-function device? */
                                 type = read_pci_config_byte(num,slot,func,
                                                             PCI_HEADER_TYPE);
@@ -1773,6 +1782,8 @@ static inline void unlock_ExtINT_logic(void)
  * a wide range of boards and BIOS bugs.  Fortunately only the timer IRQ
  * is so screwy.  Thanks to Brian Perkins for testing/hacking this beast
  * fanatically on his truly buggy board.
+ *
+ * FIXME: really need to revamp this for modern platforms only.
  */
 static inline void check_timer(void)
 {
@@ -1795,7 +1806,8 @@ static inline void check_timer(void)
          */
         apic_write(APIC_LVT0, APIC_LVT_MASKED | APIC_DM_EXTINT);
         init_8259A(1);
-        enable_8259A_irq(0);
+        if (timer_over_8254 > 0)
+                enable_8259A_irq(0);
 
         pin1  = find_isa_irq_pin(0, mp_INT);
         apic1 = find_isa_irq_apic(0, mp_INT);
@@ -1850,7 +1862,7 @@ static inline void check_timer(void)
         }
         printk(" failed.\n");
 
-        if (nmi_watchdog) {
+        if (nmi_watchdog == NMI_IO_APIC) {
                 printk(KERN_WARNING "timer doesn't work through the IO-APIC - disabling NMI Watchdog!\n");
                 nmi_watchdog = 0;
         }
diff --git a/arch/x86_64/kernel/pci-gart.c b/arch/x86_64/kernel/pci-gart.c
index dd0718dc178b..0c3f052ba6ce 100644
--- a/arch/x86_64/kernel/pci-gart.c
+++ b/arch/x86_64/kernel/pci-gart.c
@@ -228,11 +228,6 @@ static inline int need_iommu(struct device *dev, unsigned long addr, size_t size
         int mmu = high;
         if (force_iommu) 
                 mmu = 1; 
-        if (no_iommu) { 
-                if (high) 
-                        panic("PCI-DMA: high address but no IOMMU.\n"); 
-                mmu = 0; 
-        }       
         return mmu; 
 }
 
@@ -241,11 +236,6 @@ static inline int nonforced_iommu(struct device *dev, unsigned long addr, size_t
         u64 mask = *dev->dma_mask;
         int high = addr + size >= mask;
         int mmu = high;
-        if (no_iommu) { 
-                if (high) 
-                        panic("PCI-DMA: high address but no IOMMU.\n"); 
-                mmu = 0; 
-        }       
         return mmu; 
 }
 
@@ -310,7 +300,7 @@ void gart_unmap_sg(struct device *dev, struct scatterlist *sg, int nents, int di
 
         for (i = 0; i < nents; i++) {
                 struct scatterlist *s = &sg[i];
-                if (!s->dma_length)
+                if (!s->dma_length || !s->length)
                         break;
                 dma_unmap_single(dev, s->dma_address, s->dma_length, dir);
         }
@@ -364,6 +354,7 @@ static int __dma_map_cont(struct scatterlist *sg, int start, int stopat,
                 
                 BUG_ON(i > start && s->offset);
                 if (i == start) {
+                        *sout = *s; 
                         sout->dma_address = iommu_bus_base;
                         sout->dma_address += iommu_page*PAGE_SIZE + s->offset;
                         sout->dma_length = s->length;
@@ -390,6 +381,7 @@ static inline int dma_map_cont(struct scatterlist *sg, int start, int stopat,
 {
         if (!need) { 
                 BUG_ON(stopat - start != 1);
+                *sout = sg[start]; 
                 sout->dma_length = sg[start].length; 
                 return 0;
         } 
@@ -632,17 +624,13 @@ static int __init pci_iommu_init(void)
                 (agp_copy_info(agp_bridge, &info) < 0);
 #endif  
 
-        if (swiotlb) { 
-                no_iommu = 1;
+        if (swiotlb)
                 return -1; 
-        } 
-        
+
         if (no_iommu ||
             (!force_iommu && end_pfn <= MAX_DMA32_PFN) ||
             !iommu_aperture ||
             (no_agp && init_k8_gatt(&info) < 0)) {
-                no_iommu = 1;
-                no_iommu_init();
                 printk(KERN_INFO "PCI-DMA: Disabling IOMMU.\n");
                 if (end_pfn > MAX_DMA32_PFN) {
                         printk(KERN_ERR "WARNING more than 4GB of memory "
diff --git a/arch/x86_64/kernel/setup.c b/arch/x86_64/kernel/setup.c
index 9435ab7d6fb8..aa55e3cec665 100644
--- a/arch/x86_64/kernel/setup.c
+++ b/arch/x86_64/kernel/setup.c
@@ -423,6 +423,12 @@ static __init void parse_cmdline_early (char ** cmdline_p)
                 else if(!memcmp(from, "elfcorehdr=", 11))
                         elfcorehdr_addr = memparse(from+11, &from);
 #endif
+
+#ifdef CONFIG_HOTPLUG_CPU
+                else if (!memcmp(from, "additional_cpus=", 16))
+                        setup_additional_cpus(from+16);
+#endif
+
         next_char:
                 c = *(from++);
                 if (!c)
diff --git a/arch/x86_64/kernel/smpboot.c b/arch/x86_64/kernel/smpboot.c
index 67e4e28f4df8..66e98659d077 100644
--- a/arch/x86_64/kernel/smpboot.c
+++ b/arch/x86_64/kernel/smpboot.c
@@ -1152,8 +1152,6 @@ void __init smp_cpus_done(unsigned int max_cpus)
         setup_ioapic_dest();
 #endif
 
-        time_init_gtod();
-
         check_nmi_watchdog();
 }
 
@@ -1244,7 +1242,7 @@ void __cpu_die(unsigned int cpu)
         printk(KERN_ERR "CPU %u didn't die...\n", cpu);
 }
 
-static __init int setup_additional_cpus(char *s)
+__init int setup_additional_cpus(char *s)
 {
         return get_option(&s, &additional_cpus);
 }
diff --git a/arch/x86_64/kernel/time.c b/arch/x86_64/kernel/time.c
index 67841d11ed1f..3080f84bf7b7 100644
--- a/arch/x86_64/kernel/time.c
+++ b/arch/x86_64/kernel/time.c
@@ -48,6 +48,8 @@ static void cpufreq_delayed_get(void);
 extern void i8254_timer_resume(void);
 extern int using_apic_timer;
 
+static char *time_init_gtod(void);
+
 DEFINE_SPINLOCK(rtc_lock);
 DEFINE_SPINLOCK(i8253_lock);
 
@@ -901,6 +903,7 @@ static struct irqaction irq0 = {
 void __init time_init(void)
 {
         char *timename;
+        char *gtod;
 
 #ifdef HPET_HACK_ENABLE_DANGEROUS
         if (!vxtime.hpet_address) {
@@ -945,21 +948,19 @@ void __init time_init(void)
                 timename = "PIT";
         }
 
-        printk(KERN_INFO "time.c: Using %ld.%06ld MHz %s timer.\n",
-               vxtime_hz / 1000000, vxtime_hz % 1000000, timename);
+        vxtime.mode = VXTIME_TSC;
+        gtod = time_init_gtod();
+
+        printk(KERN_INFO "time.c: Using %ld.%06ld MHz WALL %s GTOD %s timer.\n",
+               vxtime_hz / 1000000, vxtime_hz % 1000000, timename, gtod);
         printk(KERN_INFO "time.c: Detected %d.%03d MHz processor.\n",
                 cpu_khz / 1000, cpu_khz % 1000);
-        vxtime.mode = VXTIME_TSC;
         vxtime.quot = (1000000L << 32) / vxtime_hz;
         vxtime.tsc_quot = (1000L << 32) / cpu_khz;
         vxtime.last_tsc = get_cycles_sync();
         setup_irq(0, &irq0);
 
         set_cyc2ns_scale(cpu_khz);
-
-#ifndef CONFIG_SMP
-        time_init_gtod();
-#endif
 }
 
 /*
@@ -981,9 +982,9 @@ __cpuinit int unsynchronized_tsc(void)
 }
 
 /*
- * Decide after all CPUs are booted what mode gettimeofday should use.
+ * Decide what mode gettimeofday should use.
  */
-void __init time_init_gtod(void)
+__init static char *time_init_gtod(void)
 {
         char *timetype;
 
@@ -1011,8 +1012,7 @@ void __init time_init_gtod(void)
                 timetype = hpet_use_timer ? "HPET/TSC" : "PIT/TSC";
                 vxtime.mode = VXTIME_TSC;
         }
-
-        printk(KERN_INFO "time.c: Using %s based timekeeping.\n", timetype);
+        return timetype;
 }
 
 __setup("report_lost_ticks", time_setup);