6 files changed, 70 insertions, 31 deletions

diff --git a/arch/x86/kernel/amd_iommu.c b/arch/x86/kernel/amd_iommu.c
index e4899e0e8787..5662e226b0c9 100644
--- a/arch/x86/kernel/amd_iommu.c
+++ b/arch/x86/kernel/amd_iommu.c
@@ -187,6 +187,8 @@ static int iommu_queue_command(struct amd_iommu *iommu, struct iommu_cmd *cmd)
 
         spin_lock_irqsave(&iommu->lock, flags);
         ret = __iommu_queue_command(iommu, cmd);
+        if (!ret)
+                iommu->need_sync = 1;
         spin_unlock_irqrestore(&iommu->lock, flags);
 
         return ret;
@@ -210,10 +212,13 @@ static int iommu_completion_wait(struct amd_iommu *iommu)
         cmd.data[0] = CMD_COMPL_WAIT_INT_MASK;
         CMD_SET_TYPE(&cmd, CMD_COMPL_WAIT);
 
-        iommu->need_sync = 0;
-
         spin_lock_irqsave(&iommu->lock, flags);
 
+        if (!iommu->need_sync)
+                goto out;
+
+        iommu->need_sync = 0;
+
         ret = __iommu_queue_command(iommu, &cmd);
 
         if (ret)
@@ -254,8 +259,6 @@ static int iommu_queue_inv_dev_entry(struct amd_iommu *iommu, u16 devid)
 
         ret = iommu_queue_command(iommu, &cmd);
 
-        iommu->need_sync = 1;
-
         return ret;
 }
 
@@ -281,8 +284,6 @@ static int iommu_queue_inv_iommu_pages(struct amd_iommu *iommu,
 
         ret = iommu_queue_command(iommu, &cmd);
 
-        iommu->need_sync = 1;
-
         return ret;
 }
 
@@ -762,8 +763,6 @@ static void set_device_domain(struct amd_iommu *iommu,
         write_unlock_irqrestore(&amd_iommu_devtable_lock, flags);
 
         iommu_queue_inv_dev_entry(iommu, devid);
-
-        iommu->need_sync = 1;
 }
 
 /*****************************************************************************
@@ -858,6 +857,9 @@ static int get_device_resources(struct device *dev,
                 print_devid(_bdf, 1);
         }
 
+        if (domain_for_device(_bdf) == NULL)
+                set_device_domain(*iommu, *domain, _bdf);
+
         return 1;
 }
 
@@ -1031,8 +1033,7 @@ static dma_addr_t map_single(struct device *dev, phys_addr_t paddr,
         if (addr == bad_dma_address)
                 goto out;
 
-        if (unlikely(iommu->need_sync))
-                iommu_completion_wait(iommu);
+        iommu_completion_wait(iommu);
 
 out:
         spin_unlock_irqrestore(&domain->lock, flags);
@@ -1060,8 +1061,7 @@ static void unmap_single(struct device *dev, dma_addr_t dma_addr,
 
         __unmap_single(iommu, domain->priv, dma_addr, size, dir);
 
-        if (unlikely(iommu->need_sync))
-                iommu_completion_wait(iommu);
+        iommu_completion_wait(iommu);
 
         spin_unlock_irqrestore(&domain->lock, flags);
 }
@@ -1127,8 +1127,7 @@ static int map_sg(struct device *dev, struct scatterlist *sglist,
                         goto unmap;
         }
 
-        if (unlikely(iommu->need_sync))
-                iommu_completion_wait(iommu);
+        iommu_completion_wait(iommu);
 
 out:
         spin_unlock_irqrestore(&domain->lock, flags);
@@ -1173,8 +1172,7 @@ static void unmap_sg(struct device *dev, struct scatterlist *sglist,
                 s->dma_address = s->dma_length = 0;
         }
 
-        if (unlikely(iommu->need_sync))
-                iommu_completion_wait(iommu);
+        iommu_completion_wait(iommu);
 
         spin_unlock_irqrestore(&domain->lock, flags);
 }
@@ -1225,8 +1223,7 @@ static void *alloc_coherent(struct device *dev, size_t size,
                 goto out;
         }
 
-        if (unlikely(iommu->need_sync))
-                iommu_completion_wait(iommu);
+        iommu_completion_wait(iommu);
 
 out:
         spin_unlock_irqrestore(&domain->lock, flags);
@@ -1257,8 +1254,7 @@ static void free_coherent(struct device *dev, size_t size,
 
         __unmap_single(iommu, domain->priv, dma_addr, size, DMA_BIDIRECTIONAL);
 
-        if (unlikely(iommu->need_sync))
-                iommu_completion_wait(iommu);
+        iommu_completion_wait(iommu);
 
         spin_unlock_irqrestore(&domain->lock, flags);
 
diff --git a/arch/x86/kernel/apic.c b/arch/x86/kernel/apic.c
index 04a7f960bbc0..16f94879b525 100644
--- a/arch/x86/kernel/apic.c
+++ b/arch/x86/kernel/apic.c
@@ -1315,7 +1315,7 @@ void enable_x2apic(void)
         }
 }
 
-void enable_IR_x2apic(void)
+void __init enable_IR_x2apic(void)
 {
 #ifdef CONFIG_INTR_REMAP
         int ret;
diff --git a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c
index d3dcd58b87cd..7f05f44b97e9 100644
--- a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c
+++ b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c
@@ -115,9 +115,20 @@ static int query_current_values_with_pending_wait(struct powernow_k8_data *data)
         u32 i = 0;
 
         if (cpu_family == CPU_HW_PSTATE) {
-                rdmsr(MSR_PSTATE_STATUS, lo, hi);
-                i = lo & HW_PSTATE_MASK;
-                data->currpstate = i;
+                if (data->currpstate == HW_PSTATE_INVALID) {
+                        /* read (initial) hw pstate if not yet set */
+                        rdmsr(MSR_PSTATE_STATUS, lo, hi);
+                        i = lo & HW_PSTATE_MASK;
+
+                        /*
+                         * a workaround for family 11h erratum 311 might cause
+                         * an "out-of-range Pstate if the core is in Pstate-0
+                         */
+                        if (i >= data->numps)
+                                data->currpstate = HW_PSTATE_0;
+                        else
+                                data->currpstate = i;
+                }
                 return 0;
         }
         do {
@@ -1121,6 +1132,7 @@ static int __cpuinit powernowk8_cpu_init(struct cpufreq_policy *pol)
         }
 
         data->cpu = pol->cpu;
+        data->currpstate = HW_PSTATE_INVALID;
 
         if (powernow_k8_cpu_init_acpi(data)) {
                 /*
diff --git a/arch/x86/kernel/cpu/cpufreq/powernow-k8.h b/arch/x86/kernel/cpu/cpufreq/powernow-k8.h
index ab48cfed4d96..65cfb5d7f77f 100644
--- a/arch/x86/kernel/cpu/cpufreq/powernow-k8.h
+++ b/arch/x86/kernel/cpu/cpufreq/powernow-k8.h
@@ -5,6 +5,19 @@
  *  http://www.gnu.org/licenses/gpl.html
  */
 
+
+enum pstate {
+        HW_PSTATE_INVALID = 0xff,
+        HW_PSTATE_0 = 0,
+        HW_PSTATE_1 = 1,
+        HW_PSTATE_2 = 2,
+        HW_PSTATE_3 = 3,
+        HW_PSTATE_4 = 4,
+        HW_PSTATE_5 = 5,
+        HW_PSTATE_6 = 6,
+        HW_PSTATE_7 = 7,
+};
+
 struct powernow_k8_data {
         unsigned int cpu;
 
@@ -23,7 +36,9 @@ struct powernow_k8_data {
         u32 exttype; /* extended interface = 1 */
 
         /* keep track of the current fid / vid or pstate */
-        u32 currvid, currfid, currpstate;
+        u32 currvid;
+        u32 currfid;
+        enum pstate currpstate;
 
         /* the powernow_table includes all frequency and vid/fid pairings:
          * fid are the lower 8 bits of the index, vid are the upper 8 bits.
diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c
index 1fec0f9b1508..9043251210fb 100644
--- a/arch/x86/kernel/io_apic.c
+++ b/arch/x86/kernel/io_apic.c
@@ -3755,7 +3755,9 @@ int acpi_get_override_irq(int bus_irq, int *trigger, int *polarity)
 void __init setup_ioapic_dest(void)
 {
         int pin, ioapic, irq, irq_entry;
+        struct irq_desc *desc;
         struct irq_cfg *cfg;
+        cpumask_t mask;
 
         if (skip_ioapic_setup == 1)
                 return;
@@ -3772,16 +3774,30 @@ void __init setup_ioapic_dest(void)
                          * cpu is online.
                          */
                         cfg = irq_cfg(irq);
-                        if (!cfg->vector)
+                        if (!cfg->vector) {
                                 setup_IO_APIC_irq(ioapic, pin, irq,
                                                   irq_trigger(irq_entry),
                                                   irq_polarity(irq_entry));
+                                continue;
+
+                        }
+
+                        /*
+                         * Honour affinities which have been set in early boot
+                         */
+                        desc = irq_to_desc(irq);
+                        if (desc->status &
+                            (IRQ_NO_BALANCING | IRQ_AFFINITY_SET))
+                                mask = desc->affinity;
+                        else
+                                mask = TARGET_CPUS;
+
 #ifdef CONFIG_INTR_REMAP
-                        else if (intr_remapping_enabled)
-                                set_ir_ioapic_affinity_irq(irq, TARGET_CPUS);
-#endif
+                        if (intr_remapping_enabled)
+                                set_ir_ioapic_affinity_irq(irq, mask);
                         else
-                                set_ioapic_affinity_irq(irq, TARGET_CPUS);
+#endif
+                                set_ioapic_affinity_irq(irq, mask);
                 }
 
         }
diff --git a/arch/x86/kernel/kvmclock.c b/arch/x86/kernel/kvmclock.c
index 1c9cc431ea4f..e169ae9b6a62 100644
--- a/arch/x86/kernel/kvmclock.c
+++ b/arch/x86/kernel/kvmclock.c
@@ -128,7 +128,7 @@ static int kvm_register_clock(char *txt)
 }
 
 #ifdef CONFIG_X86_LOCAL_APIC
-static void __devinit kvm_setup_secondary_clock(void)
+static void __cpuinit kvm_setup_secondary_clock(void)
 {
         /*
          * Now that the first cpu already had this clocksource initialized,