ia64: More 2.5.xx syncing.

arch/ia64/config.in

@@ -245,7 +245,14 @@ if [ "$CONFIG_DEBUG_KERNEL" != "n" ]; then
    bool '  Disable VHPT' CONFIG_DISABLE_VHPT
    bool '  Magic SysRq key' CONFIG_MAGIC_SYSRQ
 
-   bool '  Early printk support (requires VGA!)' CONFIG_IA64_EARLY_PRINTK
+   bool '  Early printk support' CONFIG_IA64_EARLY_PRINTK
+   if [ "$CONFIG_IA64_EARLY_PRINTK" != "n" ]; then
+      bool '    Early printk on MMIO serial port' CONFIG_IA64_EARLY_PRINTK_UART
+      if [ "$CONFIG_IA64_EARLY_PRINTK_UART" != "n" ]; then
+         hex '      UART MMIO base address' CONFIG_IA64_EARLY_PRINTK_UART_BASE ff5e0000
+      fi
+      bool '    Early printk on VGA' CONFIG_IA64_EARLY_PRINTK_VGA
+   fi
    bool '  Debug memory allocations' CONFIG_DEBUG_SLAB
    bool '  Spinlock debugging' CONFIG_DEBUG_SPINLOCK
    bool '  Turn on compare-and-exchange bug checking (slow!)' CONFIG_IA64_DEBUG_CMPXCHG

arch/ia64/hp/common/sba_iommu.c

@@ -450,7 +450,7 @@ sba_search_bitmap(struct ioc *ioc, unsigned long bits_wanted)
 		** We need the alignment to invalidate I/O TLB using
 		** SBA HW features in the unmap path.
 		*/
-		unsigned long o = 1 << get_order(bits_wanted << IOVP_SHIFT);
+		unsigned long o = 1UL << get_order(bits_wanted << IOVP_SHIFT);
 		uint bitshiftcnt = ROUNDUP(ioc->res_bitshift, o);
 		unsigned long mask;
 
@@ -1005,7 +1005,7 @@ sba_coalesce_chunks(struct ioc *ioc, struct scatterlist *startsg,
 		** Prepare for first/next DMA stream
 		*/
 		dma_len = sba_sg_len(startsg);
-		dma_offset = sba_sg_address(startsg);
+		dma_offset = (unsigned long) sba_sg_address(startsg);
 		startsg++;
 		nents--;
 
@@ -1016,7 +1016,7 @@ sba_coalesce_chunks(struct ioc *ioc, struct scatterlist *startsg,
 		** to take advantage of the block IO TLB flush.
 		*/
 		while (nents) {
-			unsigned int end_offset = dma_offset + dma_len;
+			unsigned long end_offset = dma_offset + dma_len;
 
 			/* prev entry must end on a page boundary */
 			if (end_offset & IOVP_MASK)
@@ -1114,8 +1114,8 @@ int sba_map_sg(struct pci_dev *dev, struct scatterlist *sglist, int nents,
 #endif
 	/* Fast path single entry scatterlists. */
 	if (nents == 1) {
-		sba_sg_iova(sglist) = (char *)sba_map_single(dev,
-						sba_sg_iova(sglist),
+		sba_sg_iova(sglist) = sba_map_single(dev,
+						     (void *) sba_sg_iova(sglist),
 						     sba_sg_len(sglist), direction);
 		sba_sg_iova_len(sglist) = sba_sg_len(sglist);
 #ifdef CONFIG_PROC_FS
@@ -1455,7 +1455,7 @@ sba_common_init(struct sba_device *sba_dev)
 			sba_dev->ioc[i].pdir_base[0] = 0x8000badbadc0ffeeULL;
 
 			for (reserved_iov = 0xA0000 ; reserved_iov < 0xC0000 ; reserved_iov += IOVP_SIZE) {
-				u64 *res_ptr = sba_dev->ioc[i].res_map;
+				u64 *res_ptr = (u64 *) sba_dev->ioc[i].res_map;
 				int index = PDIR_INDEX(reserved_iov);
 				int res_word;
 				u64 mask;
@@ -1586,7 +1586,7 @@ void __init sba_init(void)
 
 	for (i = 0; i < PCI_NUM_RESOURCES; i++) {
 		if (pci_resource_flags(device, i) == IORESOURCE_MEM) {
-			hpa = ioremap(pci_resource_start(device, i),
+			hpa = (u64) ioremap(pci_resource_start(device, i),
 					    pci_resource_len(device, i));
 			break;
 		}
@@ -1595,7 +1595,7 @@ void __init sba_init(void)
 	func_id = READ_REG(hpa + SBA_FUNC_ID);
 
 	if (func_id == ZX1_FUNC_ID_VALUE) {
-		(void)strcpy(sba_rev, "zx1");
+		strcpy(sba_rev, "zx1");
 		func_offset = zx1_func_offsets;
 	} else {
 		return;

arch/ia64/kernel/acpi.c

@@ -474,7 +474,7 @@ acpi_find_rsdp (void)
 }
 
 
-#ifdef CONFIG_SERIAL_ACPI
+#ifdef CONFIG_SERIAL_8250_ACPI
 
 #include <linux/acpi_serial.h>
 
@@ -529,7 +529,7 @@ acpi_parse_spcr (unsigned long phys_addr, unsigned long size)
 	return 0;
 }
 
-#endif /* CONFIG_SERIAL_ACPI */
+#endif /* CONFIG_SERIAL_8250_ACPI */
 
 
 int __init
@@ -587,7 +587,7 @@ acpi_boot_init (char *cmdline)
 	if (acpi_table_parse(ACPI_FACP, acpi_parse_fadt) < 1)
 		printk(KERN_ERR PREFIX "Can't find FADT\n");
 
-#ifdef CONFIG_SERIAL_ACPI
+#ifdef CONFIG_SERIAL_8250_ACPI
 	/*
 	 * TBD: Need phased approach to table parsing (only do those absolutely
 	 *      required during boot-up).  Recommend expanding concept of fix-

arch/ia64/kernel/efi.c

@@ -446,6 +446,9 @@ efi_init (void)
 		} else if (efi_guidcmp(config_tables[i].guid, SAL_SYSTEM_TABLE_GUID) == 0) {
 			efi.sal_systab = __va(config_tables[i].table);
 			printk(" SALsystab=0x%lx", config_tables[i].table);
+		} else if (efi_guidcmp(config_tables[i].guid, HCDP_TABLE_GUID) == 0) {
+			efi.hcdp = __va(config_tables[i].table);
+			printk(" HCDP=0x%lx", config_tables[i].table);
 		}
 	}
 	printk("\n");

arch/ia64/kernel/iosapic.c

@@ -160,6 +160,10 @@ set_rte (unsigned int vector, unsigned long dest)
 	int pin;
 	char redir;
 
+#ifdef DEBUG_IRQ_ROUTING
+	printk(KERN_DEBUG "set_rte: routing vector 0x%02x to 0x%lx\n", vector, dest);
+#endif
+
 	pin = iosapic_irq[vector].pin;
 	if (pin < 0)
 		return;		/* not an IOSAPIC interrupt */
@@ -406,7 +410,7 @@ iosapic_reassign_vector (int vector)
 	    || iosapic_irq[vector].polarity || iosapic_irq[vector].trigger)
 	{
 		new_vector = ia64_alloc_irq();
-		printk("Reassigning Vector 0x%x to 0x%x\n", vector, new_vector);
+		printk("Reassigning vector 0x%x to 0x%x\n", vector, new_vector);
 		memcpy (&iosapic_irq[new_vector], &iosapic_irq[vector],
 			sizeof(struct iosapic_irq));
 		memset (&iosapic_irq[vector], 0, sizeof(struct iosapic_irq));
@@ -757,10 +761,11 @@ iosapic_pci_fixup (int phase)
 				if (!(smp_int_redirect & SMP_IRQ_REDIRECTION)) {
 					static int cpu_index = 0;
 
-					set_rte(vector, cpu_physical_id(cpu_index) & 0xffff);
+					while (!cpu_online(cpu_index))
+						if (++cpu_index >= NR_CPUS)
+							cpu_index = 0;
 
-					for (cpu_index++; !cpu_online(cpu_index % NR_CPUS); cpu_index++);
-                                        cpu_index %= NR_CPUS;
+					set_rte(vector, cpu_physical_id(cpu_index) & 0xffff);
 				} else {
 					/*
 					 * Direct the interrupt vector to the current cpu,

arch/ia64/kernel/irq.c

@@ -368,7 +368,7 @@ unsigned int do_IRQ(unsigned long irq, struct pt_regs *regs)
 		 * use the action we have.
 		 */
 		action = NULL;
-		if (!(status & (IRQ_DISABLED | IRQ_INPROGRESS))) {
+		if (likely(!(status & (IRQ_DISABLED | IRQ_INPROGRESS)))) {
 			action = desc->action;
 			status &= ~IRQ_PENDING; /* we commit to handling */
 			status |= IRQ_INPROGRESS; /* we are handling it */
@@ -381,7 +381,7 @@ unsigned int do_IRQ(unsigned long irq, struct pt_regs *regs)
 		 * a different instance of this same irq, the other processor
 		 * will take care of it.
 		 */
-		if (!action)
+		if (unlikely(!action))
 			goto out;
 
 		/*
@@ -403,8 +403,8 @@ unsigned int do_IRQ(unsigned long irq, struct pt_regs *regs)
 				break;
 			desc->status &= ~IRQ_PENDING;
 		}
-		desc->status &= ~IRQ_INPROGRESS;
 	  out:
+		desc->status &= ~IRQ_INPROGRESS;
 		/*
 		 * The ->end() handler has to deal with interrupts which got
 		 * disabled while the handler was running.

arch/ia64/kernel/setup.c

@@ -347,6 +347,14 @@ setup_arch (char **cmdline_p)
 #ifdef CONFIG_ACPI_BOOT
 	acpi_boot_init(*cmdline_p);
 #endif
+#ifdef CONFIG_SERIAL_HCDP
+	if (efi.hcdp) {
+		void setup_serial_hcdp(void *);
+
+		/* Setup the serial ports described by HCDP */
+		setup_serial_hcdp(efi.hcdp);
+	}
+#endif
 #ifdef CONFIG_VT
 # if defined(CONFIG_DUMMY_CONSOLE)
 	conswitchp = &dummy_con;

arch/ia64/kernel/traps.c

@@ -130,6 +130,8 @@ ia64_bad_break (unsigned long break_num, struct pt_regs *regs)
 	siginfo_t siginfo;
 	int sig, code;
 
+	die_if_kernel("bad break", regs, break_num);
+
 	/* SIGILL, SIGFPE, SIGSEGV, and SIGBUS want these field initialized: */
 	siginfo.si_addr = (void *) (regs->cr_iip + ia64_psr(regs)->ri);
 	siginfo.si_imm = break_num;

arch/ia64/mm/init.c

@@ -108,7 +108,7 @@ free_initmem (void)
 
 	addr = (unsigned long) &__init_begin;
 	for (; addr < (unsigned long) &__init_end; addr += PAGE_SIZE) {
-		clear_bit(PG_reserved, &virt_to_page(addr)->flags);
+		ClearPageReserved(virt_to_page(addr));
 		set_page_count(virt_to_page(addr), 1);
 		free_page(addr);
 		++totalram_pages;
@@ -162,9 +162,9 @@ free_initrd_mem (unsigned long start, unsigned long end)
 		if (!virt_addr_valid(start))
 			continue;
 		page = virt_to_page(start);
-		clear_bit(PG_reserved, &page->flags);
+		ClearPageReserved(page);
 		set_page_count(page, 1);
-		__free_page(page);
+		free_page(page);
 		++totalram_pages;
 	}
 }

include/asm-ia64/agp.h

-#ifndef AGP_H
-#define AGP_H 1
+#ifndef _ASM_IA64_AGP_H
+#define _ASM_IA64_AGP_H
 
-/* dummy for now */
+/*
+ * IA-64 specific AGP definitions.
+ *
+ * Copyright (C) 2002 Hewlett-Packard Co
+ *	David Mosberger-Tang <davidm@hpl.hp.com>
+ */
 
-#define map_page_into_agp(page) 
-#define unmap_page_from_agp(page) 
-#define flush_agp_mappings() 
+/*
+ * To avoid memory-attribute aliasing issues, we require that the AGPGART engine operate
+ * in coherent mode, which lets us map the AGP memory as normal (write-back) memory
+ * (unlike x86, where it gets mapped "write-coalescing").
+ */
+#define map_page_into_agp(page)		/* nothing */
+#define unmap_page_from_agp(page)	/* nothing */
+#define flush_agp_mappings()		/* nothing */
 #define flush_agp_cache()		mb()
 
-#endif
+#endif /* _ASM_IA64_AGP_H */

include/asm-ia64/efi.h

@@ -190,6 +190,9 @@ typedef void efi_reset_system_t (int reset_type, efi_status_t status,
 #define SAL_SYSTEM_TABLE_GUID    \
     EFI_GUID(  0xeb9d2d32, 0x2d88, 0x11d3, 0x9a, 0x16, 0x0, 0x90, 0x27, 0x3f, 0xc1, 0x4d )
 
+#define HCDP_TABLE_GUID	\
+    EFI_GUID(  0xf951938d, 0x620b, 0x42ef, 0x82, 0x79, 0xa8, 0x4b, 0x79, 0x61, 0x78, 0x98 )
+
 typedef struct {
 	efi_guid_t guid;
 	u64 table;
@@ -225,6 +228,7 @@ extern struct efi {
 	void *smbios;			/* SM BIOS table */
 	void *sal_systab;		/* SAL system table */
 	void *boot_info;		/* boot info table */
+	void *hcdp;			/* HCDP table */
 	efi_get_time_t *get_time;
 	efi_set_time_t *set_time;
 	efi_get_wakeup_time_t *get_wakeup_time;

include/asm-ia64/elf.h

@@ -2,7 +2,7 @@
 #define _ASM_IA64_ELF_H
 
 /*
- * ELF archtecture specific definitions.
+ * ELF-specific definitions.
  *
  * Copyright (C) 1998, 1999, 2002 Hewlett-Packard Co
  *	David Mosberger-Tang <davidm@hpl.hp.com>

include/asm-ia64/page.h

@@ -87,7 +87,12 @@ typedef union ia64_va {
 #define REGION_SIZE		REGION_NUMBER(1)
 #define REGION_KERNEL		7
 
-#define BUG() do { printk("kernel BUG at %s:%d!\n", __FILE__, __LINE__); *(int *)0=0; } while (0)
+#if (__GNUC__ > 3) || (__GNUC__ == 3 && __GNUC_MINOR__ >= 1)
+# define ia64_abort()	__builtin_trap()
+#else
+# define ia64_abort()	(*(volatile int *) 0 = 0)
+#endif
+#define BUG() do { printk("kernel BUG at %s:%d!\n", __FILE__, __LINE__); ia64_abort(); } while (0)
 #define PAGE_BUG(page) do { BUG(); } while (0)
 
 static __inline__ int

include/asm-ia64/serial.h

@@ -115,21 +115,8 @@
 #define HUB6_SERIAL_PORT_DFNS
 #endif
 
-#ifdef CONFIG_MCA
-#define MCA_SERIAL_PORT_DFNS			\
-	{ 0, BASE_BAUD, 0x3220, 3, STD_COM_FLAGS },	\
-	{ 0, BASE_BAUD, 0x3228, 3, STD_COM_FLAGS },	\
-	{ 0, BASE_BAUD, 0x4220, 3, STD_COM_FLAGS },	\
-	{ 0, BASE_BAUD, 0x4228, 3, STD_COM_FLAGS },	\
-	{ 0, BASE_BAUD, 0x5220, 3, STD_COM_FLAGS },	\
-	{ 0, BASE_BAUD, 0x5228, 3, STD_COM_FLAGS },
-#else
-#define MCA_SERIAL_PORT_DFNS
-#endif
-
 #define SERIAL_PORT_DFNS		\
 	STD_SERIAL_PORT_DEFNS		\
 	EXTRA_SERIAL_PORT_DEFNS		\
-	HUB6_SERIAL_PORT_DFNS		\
-	MCA_SERIAL_PORT_DFNS
+	HUB6_SERIAL_PORT_DFNS
 

include/asm-ia64/tlb.h

@@ -123,13 +123,20 @@ tlb_gather_mmu (struct mm_struct *mm, unsigned int full_mm_flush)
 	mmu_gather_t *tlb = &mmu_gathers[smp_processor_id()];
 
 	tlb->mm = mm;
-	tlb->nr = 0;
-	if (full_mm_flush || num_online_cpus() == 1)
 	/*
-		 * Use fast mode if only 1 CPU is online or if we're tearing down the
-		 * entire address space.
+	 * Use fast mode if only 1 CPU is online.
+	 *
+	 * It would be tempting to turn on fast-mode for full_mm_flush as well.  But this
+	 * doesn't work because of speculative accesses and software prefetching: the page
+	 * table of "mm" may (and usually is) the currently active page table and even
+	 * though the kernel won't do any user-space accesses during the TLB shoot down, a
+	 * compiler might use speculation or lfetch.fault on what happens to be a valid
+	 * user-space address.  This in turn could trigger a TLB miss fault (or a VHPT
+	 * walk) and re-insert a TLB entry we just removed.  Slow mode avoids such
+	 * problems.  (We could make fast-mode work by switching the current task to a
+	 * different "mm" during the shootdown.) --davidm 08/02/2002
 	 */
-		tlb->nr =  ~0U;
+	tlb->nr = (num_online_cpus() == 1) ? ~0U : 0;
 	tlb->fullmm = full_mm_flush;
 	tlb->freed = 0;
 	tlb->start_addr = ~0UL;

include/asm-ia64/tlbflush.h

@@ -60,6 +60,8 @@ flush_tlb_page (struct vm_area_struct *vma, unsigned long addr)
 #else
 	if (vma->vm_mm == current->active_mm)
 		asm volatile ("ptc.l %0,%1" :: "r"(addr), "r"(PAGE_SHIFT << 2) : "memory");
+	else
+		vma->vm_mm->context = 0;
 #endif
 }