Merge branch 'master' of git+ssh://galak@master.kernel.org/pub/scm/linux/kernel/git/galak/powerpc

13 files changed, 108 insertions, 57 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 6147770b4347..e5b14c57eaa0 100644
--- a/arch/x86_64/kernel/apic.c
+++ b/arch/x86_64/kernel/apic.c
@@ -708,7 +708,7 @@ static void setup_APIC_timer(unsigned int clocks)
         local_irq_save(flags);
 
         /* wait for irq slice */
-        if (vxtime.hpet_address) {
+        if (vxtime.hpet_address && hpet_use_timer) {
                 int trigger = hpet_readl(HPET_T0_CMP);
                 while (hpet_readl(HPET_COUNTER) >= trigger)
                         /* do nothing */ ;
@@ -1152,6 +1152,7 @@ __setup("noapicmaintimer", setup_noapicmaintimer);
 static __init int setup_apicpmtimer(char *s)
 {
         apic_calibrate_pmtmr = 1;
+        notsc_setup(NULL);
         return setup_apicmaintimer(NULL);
 }
 __setup("apicpmtimer", setup_apicpmtimer);
diff --git a/arch/x86_64/kernel/entry.S b/arch/x86_64/kernel/entry.S
index b150c87a08c6..7c10e9009d61 100644
--- a/arch/x86_64/kernel/entry.S
+++ b/arch/x86_64/kernel/entry.S
@@ -554,6 +554,7 @@ iret_label:
         /* running with kernel gs */
 bad_iret:
         movq $-9999,%rdi        /* better code? */
+        sti
         jmp do_exit                     
         .previous       
         
diff --git a/arch/x86_64/kernel/head.S b/arch/x86_64/kernel/head.S
index 692c737feddb..02fc7fa0ea28 100644
--- a/arch/x86_64/kernel/head.S
+++ b/arch/x86_64/kernel/head.S
@@ -213,6 +213,11 @@ ENTRY(early_idt_handler)
         cmpl $2,early_recursion_flag(%rip)
         jz  1f
         call dump_stack
+#ifdef CONFIG_KALLSYMS  
+        leaq early_idt_ripmsg(%rip),%rdi
+        movq 8(%rsp),%rsi       # get rip again
+        call __print_symbol
+#endif
 1:      hlt
         jmp 1b
 early_recursion_flag:
@@ -220,6 +225,8 @@ early_recursion_flag:
 
 early_idt_msg:
         .asciz "PANIC: early exception rip %lx error %lx cr2 %lx\n"
+early_idt_ripmsg:
+        .asciz "RIP %s\n"
 
 .code32
 ENTRY(no_long_mode)
diff --git a/arch/x86_64/kernel/io_apic.c b/arch/x86_64/kernel/io_apic.c
index 4282d72b2a26..ffed464e6b12 100644
--- a/arch/x86_64/kernel/io_apic.c
+++ b/arch/x86_64/kernel/io_apic.c
@@ -30,6 +30,9 @@
 #include <linux/mc146818rtc.h>
 #include <linux/acpi.h>
 #include <linux/sysdev.h>
+#ifdef CONFIG_ACPI
+#include <acpi/acpi_bus.h>
+#endif
 
 #include <asm/io.h>
 #include <asm/smp.h>
@@ -47,6 +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 };
 
@@ -248,6 +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>
@@ -260,6 +279,8 @@ __setup("apic", enable_ioapic_setup);
 
    And another hack to disable the IOMMU on VIA chipsets.
 
+   ... and others. Really should move this somewhere else.
+
    Kludge-O-Rama. */
 void __init check_ioapic(void) 
 { 
@@ -304,16 +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;
+                                        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);
@@ -1757,6 +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)
 {
@@ -1779,7 +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);
@@ -1834,7 +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/mpparse.c b/arch/x86_64/kernel/mpparse.c
index dc49bfb6db0a..9013a90b5c2e 100644
--- a/arch/x86_64/kernel/mpparse.c
+++ b/arch/x86_64/kernel/mpparse.c
@@ -288,9 +288,9 @@ static int __init smp_read_mpc(struct mp_config_table *mpc)
 
         memcpy(str,mpc->mpc_productid,12);
         str[12]=0;
-        printk(KERN_INFO "Product ID: %s ",str);
+        printk("Product ID: %s ",str);
 
-        printk(KERN_INFO "APIC at: 0x%X\n",mpc->mpc_lapic);
+        printk("APIC at: 0x%X\n",mpc->mpc_lapic);
 
         /* save the local APIC address, it might be non-default */
         if (!acpi_lapic)
diff --git a/arch/x86_64/kernel/nmi.c b/arch/x86_64/kernel/nmi.c
index 8be407a1f62d..5bf17e41cd2d 100644
--- a/arch/x86_64/kernel/nmi.c
+++ b/arch/x86_64/kernel/nmi.c
@@ -236,6 +236,7 @@ static void enable_lapic_nmi_watchdog(void)
 {
         if (nmi_active < 0) {
                 nmi_watchdog = NMI_LOCAL_APIC;
+                touch_nmi_watchdog();
                 setup_apic_nmi_watchdog();
         }
 }
@@ -456,15 +457,17 @@ static DEFINE_PER_CPU(int, nmi_touch);
 
 void touch_nmi_watchdog (void)
 {
-        int i;
+        if (nmi_watchdog > 0) {
+                unsigned cpu;
 
-        /*
-         * Tell other CPUs to reset their alert counters. We cannot
-         * do it ourselves because the alert count increase is not
-         * atomic.
-         */
-        for (i = 0; i < NR_CPUS; i++)
-                per_cpu(nmi_touch, i) = 1;
+                /*
+                 * Tell other CPUs to reset their alert counters. We cannot
+                 * do it ourselves because the alert count increase is not
+                 * atomic.
+                 */
+                for_each_present_cpu (cpu)
+                        per_cpu(nmi_touch, cpu) = 1;
+        }
 
         touch_softlockup_watchdog();
 }
diff --git a/arch/x86_64/kernel/pci-gart.c b/arch/x86_64/kernel/pci-gart.c
index 2fe23a6c361b..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; 
 }
 
@@ -379,7 +369,7 @@ static int __dma_map_cont(struct scatterlist *sg, int start, int stopat,
                         SET_LEAK(iommu_page);
                         addr += PAGE_SIZE;
                         iommu_page++;
-        } 
+                }
         } 
         BUG_ON(iommu_page - iommu_start != pages);      
         return 0;
@@ -634,17 +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 dba7237be5c1..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);
 
@@ -59,7 +61,7 @@ static int notsc __initdata = 0;
 unsigned int cpu_khz;                                   /* TSC clocks / usec, not used here */
 static unsigned long hpet_period;                       /* fsecs / HPET clock */
 unsigned long hpet_tick;                                /* HPET clocks / interrupt */
-static int hpet_use_timer;                              /* Use counter of hpet for time keeping, otherwise PIT */
+int hpet_use_timer;                             /* Use counter of hpet for time keeping, otherwise PIT */
 unsigned long vxtime_hz = PIT_TICK_RATE;
 int report_lost_ticks;                          /* command line option */
 unsigned long long monotonic_base;
@@ -326,7 +328,10 @@ static noinline void handle_lost_ticks(int lost, struct pt_regs *regs)
             print_symbol("rip %s\n", regs->rip);
             if (vxtime.mode == VXTIME_TSC && vxtime.hpet_address) {
                     printk(KERN_WARNING "Falling back to HPET\n");
-                    vxtime.last = hpet_readl(HPET_T0_CMP) - hpet_tick;
+                    if (hpet_use_timer)
+                        vxtime.last = hpet_readl(HPET_T0_CMP) - hpet_tick;
+                    else
+                        vxtime.last = hpet_readl(HPET_COUNTER);
                     vxtime.mode = VXTIME_HPET;
                     do_gettimeoffset = do_gettimeoffset_hpet;
             }
@@ -898,6 +903,7 @@ static struct irqaction irq0 = {
 void __init time_init(void)
 {
         char *timename;
+        char *gtod;
 
 #ifdef HPET_HACK_ENABLE_DANGEROUS
         if (!vxtime.hpet_address) {
@@ -942,21 +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
 }
 
 /*
@@ -978,9 +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;
 
@@ -988,7 +992,10 @@ void __init time_init_gtod(void)
                 notsc = 1;
         if (vxtime.hpet_address && notsc) {
                 timetype = hpet_use_timer ? "HPET" : "PIT/HPET";
-                vxtime.last = hpet_readl(HPET_T0_CMP) - hpet_tick;
+                if (hpet_use_timer)
+                        vxtime.last = hpet_readl(HPET_T0_CMP) - hpet_tick;
+                else
+                        vxtime.last = hpet_readl(HPET_COUNTER);
                 vxtime.mode = VXTIME_HPET;
                 do_gettimeoffset = do_gettimeoffset_hpet;
 #ifdef CONFIG_X86_PM_TIMER
@@ -1005,8 +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);
@@ -1321,8 +1327,7 @@ static int __init nohpet_setup(char *s)
 
 __setup("nohpet", nohpet_setup);
 
-
-static int __init notsc_setup(char *s)
+int __init notsc_setup(char *s)
 {
         notsc = 1;
         return 0;
diff --git a/arch/x86_64/kernel/traps.c b/arch/x86_64/kernel/traps.c
index ee1b2da9e5e7..28d50dc540e8 100644
--- a/arch/x86_64/kernel/traps.c
+++ b/arch/x86_64/kernel/traps.c
@@ -90,6 +90,20 @@ static inline void conditional_sti(struct pt_regs *regs)
                 local_irq_enable();
 }
 
+static inline void preempt_conditional_sti(struct pt_regs *regs)
+{
+        preempt_disable();
+        if (regs->eflags & X86_EFLAGS_IF)
+                local_irq_enable();
+}
+
+static inline void preempt_conditional_cli(struct pt_regs *regs)
+{
+        if (regs->eflags & X86_EFLAGS_IF)
+                local_irq_disable();
+        preempt_enable_no_resched();
+}
+
 static int kstack_depth_to_print = 10;
 
 #ifdef CONFIG_KALLSYMS
@@ -693,7 +707,7 @@ asmlinkage void __kprobes do_debug(struct pt_regs * regs,
                                                 SIGTRAP) == NOTIFY_STOP)
                 return;
 
-        conditional_sti(regs);
+        preempt_conditional_sti(regs);
 
         /* Mask out spurious debug traps due to lazy DR7 setting */
         if (condition & (DR_TRAP0|DR_TRAP1|DR_TRAP2|DR_TRAP3)) {
@@ -738,11 +752,13 @@ asmlinkage void __kprobes do_debug(struct pt_regs * regs,
 
 clear_dr7:
         set_debugreg(0UL, 7);
+        preempt_conditional_cli(regs);
         return;
 
 clear_TF_reenable:
         set_tsk_thread_flag(tsk, TIF_SINGLESTEP);
         regs->eflags &= ~TF_MASK;
+        preempt_conditional_cli(regs);
 }
 
 static int kernel_math_error(struct pt_regs *regs, const char *str, int trapnr)