6 files changed, 135 insertions, 85 deletions

diff --git a/arch/x86/platform/ce4100/falconfalls.dts b/arch/x86/platform/ce4100/falconfalls.dts
index 2d6d226f2b10..e70be38ce039 100644
--- a/arch/x86/platform/ce4100/falconfalls.dts
+++ b/arch/x86/platform/ce4100/falconfalls.dts
@@ -347,7 +347,7 @@
                                                    "pciclass0c03";
 
                                         reg = <0x16800 0x0 0x0 0x0 0x0>;
-                                        interrupts = <22 3>;
+                                        interrupts = <22 1>;
                                 };
 
                                 usb@d,1 {
@@ -357,7 +357,7 @@
                                                    "pciclass0c03";
 
                                         reg = <0x16900 0x0 0x0 0x0 0x0>;
-                                        interrupts = <22 3>;
+                                        interrupts = <22 1>;
                                 };
 
                                 sata@e,0 {
@@ -367,7 +367,7 @@
                                                    "pciclass0106";
 
                                         reg = <0x17000 0x0 0x0 0x0 0x0>;
-                                        interrupts = <23 3>;
+                                        interrupts = <23 1>;
                                 };
 
                                 flash@f,0 {
diff --git a/arch/x86/platform/efi/efi.c b/arch/x86/platform/efi/efi.c
index 0fe27d7c6258..b30aa26a8df2 100644
--- a/arch/x86/platform/efi/efi.c
+++ b/arch/x86/platform/efi/efi.c
@@ -145,17 +145,6 @@ static void virt_efi_reset_system(int reset_type,
                        data_size, data);
 }
 
-static efi_status_t virt_efi_set_virtual_address_map(
-        unsigned long memory_map_size,
-        unsigned long descriptor_size,
-        u32 descriptor_version,
-        efi_memory_desc_t *virtual_map)
-{
-        return efi_call_virt4(set_virtual_address_map,
-                              memory_map_size, descriptor_size,
-                              descriptor_version, virtual_map);
-}
-
 static efi_status_t __init phys_efi_set_virtual_address_map(
         unsigned long memory_map_size,
         unsigned long descriptor_size,
@@ -468,11 +457,25 @@ void __init efi_init(void)
 #endif
 }
 
+void __init efi_set_executable(efi_memory_desc_t *md, bool executable)
+{
+        u64 addr, npages;
+
+        addr = md->virt_addr;
+        npages = md->num_pages;
+
+        memrange_efi_to_native(&addr, &npages);
+
+        if (executable)
+                set_memory_x(addr, npages);
+        else
+                set_memory_nx(addr, npages);
+}
+
 static void __init runtime_code_page_mkexec(void)
 {
         efi_memory_desc_t *md;
         void *p;
-        u64 addr, npages;
 
         /* Make EFI runtime service code area executable */
         for (p = memmap.map; p < memmap.map_end; p += memmap.desc_size) {
@@ -481,10 +484,7 @@ static void __init runtime_code_page_mkexec(void)
                 if (md->type != EFI_RUNTIME_SERVICES_CODE)
                         continue;
 
-                addr = md->virt_addr;
-                npages = md->num_pages;
-                memrange_efi_to_native(&addr, &npages);
-                set_memory_x(addr, npages);
+                efi_set_executable(md, true);
         }
 }
 
@@ -498,13 +498,42 @@ static void __init runtime_code_page_mkexec(void)
  */
 void __init efi_enter_virtual_mode(void)
 {
-        efi_memory_desc_t *md;
+        efi_memory_desc_t *md, *prev_md = NULL;
         efi_status_t status;
         unsigned long size;
         u64 end, systab, addr, npages, end_pfn;
-        void *p, *va;
+        void *p, *va, *new_memmap = NULL;
+        int count = 0;
 
         efi.systab = NULL;
+
+        /* Merge contiguous regions of the same type and attribute */
+        for (p = memmap.map; p < memmap.map_end; p += memmap.desc_size) {
+                u64 prev_size;
+                md = p;
+
+                if (!prev_md) {
+                        prev_md = md;
+                        continue;
+                }
+
+                if (prev_md->type != md->type ||
+                    prev_md->attribute != md->attribute) {
+                        prev_md = md;
+                        continue;
+                }
+
+                prev_size = prev_md->num_pages << EFI_PAGE_SHIFT;
+
+                if (md->phys_addr == (prev_md->phys_addr + prev_size)) {
+                        prev_md->num_pages += md->num_pages;
+                        md->type = EFI_RESERVED_TYPE;
+                        md->attribute = 0;
+                        continue;
+                }
+                prev_md = md;
+        }
+
         for (p = memmap.map; p < memmap.map_end; p += memmap.desc_size) {
                 md = p;
                 if (!(md->attribute & EFI_MEMORY_RUNTIME))
@@ -541,15 +570,21 @@ void __init efi_enter_virtual_mode(void)
                         systab += md->virt_addr - md->phys_addr;
                         efi.systab = (efi_system_table_t *) (unsigned long) systab;
                 }
+                new_memmap = krealloc(new_memmap,
+                                      (count + 1) * memmap.desc_size,
+                                      GFP_KERNEL);
+                memcpy(new_memmap + (count * memmap.desc_size), md,
+                       memmap.desc_size);
+                count++;
         }
 
         BUG_ON(!efi.systab);
 
         status = phys_efi_set_virtual_address_map(
-                memmap.desc_size * memmap.nr_map,
+                memmap.desc_size * count,
                 memmap.desc_size,
                 memmap.desc_version,
-                memmap.phys_map);
+                (efi_memory_desc_t *)__pa(new_memmap));
 
         if (status != EFI_SUCCESS) {
                 printk(KERN_ALERT "Unable to switch EFI into virtual mode "
@@ -572,11 +607,12 @@ void __init efi_enter_virtual_mode(void)
         efi.set_variable = virt_efi_set_variable;
         efi.get_next_high_mono_count = virt_efi_get_next_high_mono_count;
         efi.reset_system = virt_efi_reset_system;
-        efi.set_virtual_address_map = virt_efi_set_virtual_address_map;
+        efi.set_virtual_address_map = NULL;
         if (__supported_pte_mask & _PAGE_NX)
                 runtime_code_page_mkexec();
         early_iounmap(memmap.map, memmap.nr_map * memmap.desc_size);
         memmap.map = NULL;
+        kfree(new_memmap);
 }
 
 /*
diff --git a/arch/x86/platform/efi/efi_64.c b/arch/x86/platform/efi/efi_64.c
index ac0621a7ac3d..2649426a7905 100644
--- a/arch/x86/platform/efi/efi_64.c
+++ b/arch/x86/platform/efi/efi_64.c
@@ -41,22 +41,7 @@
 static pgd_t save_pgd __initdata;
 static unsigned long efi_flags __initdata;
 
-static void __init early_mapping_set_exec(unsigned long start,
-                                          unsigned long end,
-                                          int executable)
-{
-        unsigned long num_pages;
-
-        start &= PMD_MASK;
-        end = (end + PMD_SIZE - 1) & PMD_MASK;
-        num_pages = (end - start) >> PAGE_SHIFT;
-        if (executable)
-                set_memory_x((unsigned long)__va(start), num_pages);
-        else
-                set_memory_nx((unsigned long)__va(start), num_pages);
-}
-
-static void __init early_runtime_code_mapping_set_exec(int executable)
+static void __init early_code_mapping_set_exec(int executable)
 {
         efi_memory_desc_t *md;
         void *p;
@@ -67,11 +52,8 @@ static void __init early_runtime_code_mapping_set_exec(int executable)
         /* Make EFI runtime service code area executable */
         for (p = memmap.map; p < memmap.map_end; p += memmap.desc_size) {
                 md = p;
-                if (md->type == EFI_RUNTIME_SERVICES_CODE) {
-                        unsigned long end;
-                        end = md->phys_addr + (md->num_pages << EFI_PAGE_SHIFT);
-                        early_mapping_set_exec(md->phys_addr, end, executable);
-                }
+                if (md->type == EFI_RUNTIME_SERVICES_CODE)
+                        efi_set_executable(md, executable);
         }
 }
 
@@ -79,7 +61,7 @@ void __init efi_call_phys_prelog(void)
 {
         unsigned long vaddress;
 
-        early_runtime_code_mapping_set_exec(1);
+        early_code_mapping_set_exec(1);
         local_irq_save(efi_flags);
         vaddress = (unsigned long)__va(0x0UL);
         save_pgd = *pgd_offset_k(0x0UL);
@@ -95,7 +77,7 @@ void __init efi_call_phys_epilog(void)
         set_pgd(pgd_offset_k(0x0UL), save_pgd);
         __flush_tlb_all();
         local_irq_restore(efi_flags);
-        early_runtime_code_mapping_set_exec(0);
+        early_code_mapping_set_exec(0);
 }
 
 void __iomem *__init efi_ioremap(unsigned long phys_addr, unsigned long size,
@@ -107,8 +89,10 @@ void __iomem *__init efi_ioremap(unsigned long phys_addr, unsigned long size,
                 return ioremap(phys_addr, size);
 
         last_map_pfn = init_memory_mapping(phys_addr, phys_addr + size);
-        if ((last_map_pfn << PAGE_SHIFT) < phys_addr + size)
-                return NULL;
+        if ((last_map_pfn << PAGE_SHIFT) < phys_addr + size) {
+                unsigned long top = last_map_pfn << PAGE_SHIFT;
+                efi_ioremap(top, size - (top - phys_addr), type);
+        }
 
         return (void __iomem *)__va(phys_addr);
 }
diff --git a/arch/x86/platform/mrst/mrst.c b/arch/x86/platform/mrst/mrst.c
index 275dbc19e2cf..7000e74b3087 100644
--- a/arch/x86/platform/mrst/mrst.c
+++ b/arch/x86/platform/mrst/mrst.c
@@ -194,7 +194,7 @@ static unsigned long __init mrst_calibrate_tsc(void)
         return 0;
 }
 
-void __init mrst_time_init(void)
+static void __init mrst_time_init(void)
 {
         sfi_table_parse(SFI_SIG_MTMR, NULL, NULL, sfi_parse_mtmr);
         switch (mrst_timer_options) {
@@ -216,7 +216,7 @@ void __init mrst_time_init(void)
         apbt_time_init();
 }
 
-void __cpuinit mrst_arch_setup(void)
+static void __cpuinit mrst_arch_setup(void)
 {
         if (boot_cpu_data.x86 == 6 && boot_cpu_data.x86_model == 0x27)
                 __mrst_cpu_chip = MRST_CPU_CHIP_PENWELL;
diff --git a/arch/x86/platform/uv/tlb_uv.c b/arch/x86/platform/uv/tlb_uv.c
index 7cb6424317f6..c58e0ea39ef5 100644
--- a/arch/x86/platform/uv/tlb_uv.c
+++ b/arch/x86/platform/uv/tlb_uv.c
@@ -699,16 +699,17 @@ const struct cpumask *uv_flush_tlb_others(const struct cpumask *cpumask,
                                           struct mm_struct *mm,
                                           unsigned long va, unsigned int cpu)
 {
-        int tcpu;
-        int uvhub;
         int locals = 0;
         int remotes = 0;
         int hubs = 0;
+        int tcpu;
+        int tpnode;
         struct bau_desc *bau_desc;
         struct cpumask *flush_mask;
         struct ptc_stats *stat;
         struct bau_control *bcp;
         struct bau_control *tbcp;
+        struct hub_and_pnode *hpp;
 
         /* kernel was booted 'nobau' */
         if (nobau)
@@ -750,11 +751,18 @@ const struct cpumask *uv_flush_tlb_others(const struct cpumask *cpumask,
         bau_desc += UV_ITEMS_PER_DESCRIPTOR * bcp->uvhub_cpu;
         bau_uvhubs_clear(&bau_desc->distribution, UV_DISTRIBUTION_SIZE);
 
-        /* cpu statistics */
         for_each_cpu(tcpu, flush_mask) {
-                uvhub = uv_cpu_to_blade_id(tcpu);
-                bau_uvhub_set(uvhub, &bau_desc->distribution);
-                if (uvhub == bcp->uvhub)
+                /*
+                 * The distribution vector is a bit map of pnodes, relative
+                 * to the partition base pnode (and the partition base nasid
+                 * in the header).
+                 * Translate cpu to pnode and hub using an array stored
+                 * in local memory.
+                 */
+                hpp = &bcp->socket_master->target_hub_and_pnode[tcpu];
+                tpnode = hpp->pnode - bcp->partition_base_pnode;
+                bau_uvhub_set(tpnode, &bau_desc->distribution);
+                if (hpp->uvhub == bcp->uvhub)
                         locals++;
                 else
                         remotes++;
@@ -855,7 +863,7 @@ void uv_bau_message_interrupt(struct pt_regs *regs)
  * an interrupt, but causes an error message to be returned to
  * the sender.
  */
-static void uv_enable_timeouts(void)
+static void __init uv_enable_timeouts(void)
 {
         int uvhub;
         int nuvhubs;
@@ -1326,10 +1334,10 @@ static int __init uv_ptc_init(void)
 }
 
 /*
- * initialize the sending side's sending buffers
+ * Initialize the sending side's sending buffers.
  */
 static void
-uv_activation_descriptor_init(int node, int pnode)
+uv_activation_descriptor_init(int node, int pnode, int base_pnode)
 {
         int i;
         int cpu;
@@ -1352,11 +1360,11 @@ uv_activation_descriptor_init(int node, int pnode)
         n = pa >> uv_nshift;
         m = pa & uv_mmask;
 
+        /* the 14-bit pnode */
         uv_write_global_mmr64(pnode, UVH_LB_BAU_SB_DESCRIPTOR_BASE,
                               (n << UV_DESC_BASE_PNODE_SHIFT | m));
-
         /*
-         * initializing all 8 (UV_ITEMS_PER_DESCRIPTOR) descriptors for each
+         * Initializing all 8 (UV_ITEMS_PER_DESCRIPTOR) descriptors for each
          * cpu even though we only use the first one; one descriptor can
          * describe a broadcast to 256 uv hubs.
          */
@@ -1365,12 +1373,13 @@ uv_activation_descriptor_init(int node, int pnode)
                 memset(bd2, 0, sizeof(struct bau_desc));
                 bd2->header.sw_ack_flag = 1;
                 /*
-                 * base_dest_nodeid is the nasid of the first uvhub
-                 * in the partition. The bit map will indicate uvhub numbers,
-                 * which are 0-N in a partition. Pnodes are unique system-wide.
+                 * The base_dest_nasid set in the message header is the nasid
+                 * of the first uvhub in the partition. The bit map will
+                 * indicate destination pnode numbers relative to that base.
+                 * They may not be consecutive if nasid striding is being used.
                  */
-                bd2->header.base_dest_nodeid = UV_PNODE_TO_NASID(uv_partition_base_pnode);
-                bd2->header.dest_subnodeid = 0x10; /* the LB */
+                bd2->header.base_dest_nasid = UV_PNODE_TO_NASID(base_pnode);
+                bd2->header.dest_subnodeid = UV_LB_SUBNODEID;
                 bd2->header.command = UV_NET_ENDPOINT_INTD;
                 bd2->header.int_both = 1;
                 /*
@@ -1442,7 +1451,7 @@ uv_payload_queue_init(int node, int pnode)
 /*
  * Initialization of each UV hub's structures
  */
-static void __init uv_init_uvhub(int uvhub, int vector)
+static void __init uv_init_uvhub(int uvhub, int vector, int base_pnode)
 {
         int node;
         int pnode;
@@ -1450,11 +1459,11 @@ static void __init uv_init_uvhub(int uvhub, int vector)
 
         node = uvhub_to_first_node(uvhub);
         pnode = uv_blade_to_pnode(uvhub);
-        uv_activation_descriptor_init(node, pnode);
+        uv_activation_descriptor_init(node, pnode, base_pnode);
         uv_payload_queue_init(node, pnode);
         /*
-         * the below initialization can't be in firmware because the
-         * messaging IRQ will be determined by the OS
+         * The below initialization can't be in firmware because the
+         * messaging IRQ will be determined by the OS.
          */
         apicid = uvhub_to_first_apicid(uvhub) | uv_apicid_hibits;
         uv_write_global_mmr64(pnode, UVH_BAU_DATA_CONFIG,
@@ -1491,10 +1500,11 @@ calculate_destination_timeout(void)
 /*
  * initialize the bau_control structure for each cpu
  */
-static int __init uv_init_per_cpu(int nuvhubs)
+static int __init uv_init_per_cpu(int nuvhubs, int base_part_pnode)
 {
         int i;
         int cpu;
+        int tcpu;
         int pnode;
         int uvhub;
         int have_hmaster;
@@ -1528,6 +1538,15 @@ static int __init uv_init_per_cpu(int nuvhubs)
                 bcp = &per_cpu(bau_control, cpu);
                 memset(bcp, 0, sizeof(struct bau_control));
                 pnode = uv_cpu_hub_info(cpu)->pnode;
+                if ((pnode - base_part_pnode) >= UV_DISTRIBUTION_SIZE) {
+                        printk(KERN_EMERG
+                                "cpu %d pnode %d-%d beyond %d; BAU disabled\n",
+                                cpu, pnode, base_part_pnode,
+                                UV_DISTRIBUTION_SIZE);
+                        return 1;
+                }
+                bcp->osnode = cpu_to_node(cpu);
+                bcp->partition_base_pnode = uv_partition_base_pnode;
                 uvhub = uv_cpu_hub_info(cpu)->numa_blade_id;
                 *(uvhub_mask + (uvhub/8)) |= (1 << (uvhub%8));
                 bdp = &uvhub_descs[uvhub];
@@ -1536,7 +1555,7 @@ static int __init uv_init_per_cpu(int nuvhubs)
                 bdp->pnode = pnode;
                 /* kludge: 'assuming' one node per socket, and assuming that
                    disabling a socket just leaves a gap in node numbers */
-                socket = (cpu_to_node(cpu) & 1);
+                socket = bcp->osnode & 1;
                 bdp->socket_mask |= (1 << socket);
                 sdp = &bdp->socket[socket];
                 sdp->cpu_number[sdp->num_cpus] = cpu;
@@ -1585,6 +1604,20 @@ static int __init uv_init_per_cpu(int nuvhubs)
 nextsocket:
                         socket++;
                         socket_mask = (socket_mask >> 1);
+                        /* each socket gets a local array of pnodes/hubs */
+                        bcp = smaster;
+                        bcp->target_hub_and_pnode = kmalloc_node(
+                                sizeof(struct hub_and_pnode) *
+                                num_possible_cpus(), GFP_KERNEL, bcp->osnode);
+                        memset(bcp->target_hub_and_pnode, 0,
+                                sizeof(struct hub_and_pnode) *
+                                num_possible_cpus());
+                        for_each_present_cpu(tcpu) {
+                                bcp->target_hub_and_pnode[tcpu].pnode =
+                                        uv_cpu_hub_info(tcpu)->pnode;
+                                bcp->target_hub_and_pnode[tcpu].uvhub =
+                                        uv_cpu_hub_info(tcpu)->numa_blade_id;
+                        }
                 }
         }
         kfree(uvhub_descs);
@@ -1637,21 +1670,22 @@ static int __init uv_bau_init(void)
         spin_lock_init(&disable_lock);
         congested_cycles = microsec_2_cycles(congested_response_us);
 
-        if (uv_init_per_cpu(nuvhubs)) {
-                nobau = 1;
-                return 0;
-        }
-
         uv_partition_base_pnode = 0x7fffffff;
-        for (uvhub = 0; uvhub < nuvhubs; uvhub++)
+        for (uvhub = 0; uvhub < nuvhubs; uvhub++) {
                 if (uv_blade_nr_possible_cpus(uvhub) &&
                         (uv_blade_to_pnode(uvhub) < uv_partition_base_pnode))
                         uv_partition_base_pnode = uv_blade_to_pnode(uvhub);
+        }
+
+        if (uv_init_per_cpu(nuvhubs, uv_partition_base_pnode)) {
+                nobau = 1;
+                return 0;
+        }
 
         vector = UV_BAU_MESSAGE;
         for_each_possible_blade(uvhub)
                 if (uv_blade_nr_possible_cpus(uvhub))
-                        uv_init_uvhub(uvhub, vector);
+                        uv_init_uvhub(uvhub, vector, uv_partition_base_pnode);
 
         uv_enable_timeouts();
         alloc_intr_gate(vector, uv_bau_message_intr1);
diff --git a/arch/x86/platform/uv/uv_time.c b/arch/x86/platform/uv/uv_time.c
index 9daf5d1af9f1..0eb90184515f 100644
--- a/arch/x86/platform/uv/uv_time.c
+++ b/arch/x86/platform/uv/uv_time.c
@@ -40,7 +40,6 @@ static struct clocksource clocksource_uv = {
         .rating         = 400,
         .read           = uv_read_rtc,
         .mask           = (cycle_t)UVH_RTC_REAL_TIME_CLOCK_MASK,
-        .shift          = 10,
         .flags          = CLOCK_SOURCE_IS_CONTINUOUS,
 };
 
@@ -372,14 +371,11 @@ static __init int uv_rtc_setup_clock(void)
         if (!is_uv_system())
                 return -ENODEV;
 
-        clocksource_uv.mult = clocksource_hz2mult(sn_rtc_cycles_per_second,
-                                clocksource_uv.shift);
-
         /* If single blade, prefer tsc */
         if (uv_num_possible_blades() == 1)
                 clocksource_uv.rating = 250;
 
-        rc = clocksource_register(&clocksource_uv);
+        rc = clocksource_register_hz(&clocksource_uv, sn_rtc_cycles_per_second);
         if (rc)
                 printk(KERN_INFO "UV RTC clocksource failed rc %d\n", rc);
         else