Merge davem@nuts.davemloft.net:/disk1/BK/sparc-2.6

into kernel.bkbits.net:/home/davem/sparc-2.6

arch/sparc/Makefile

@@ -67,7 +67,8 @@ include/asm-$(ARCH)/asm_offsets.h: arch/$(ARCH)/kernel/asm-offsets.s
 	$(call filechk,gen-asm-offsets)
 
 CLEAN_FILES +=	include/asm-$(ARCH)/asm_offsets.h	\
-		arch/$(ARCH)/kernel/asm-offsets.s
+		arch/$(ARCH)/kernel/asm-offsets.s	\
+		arch/$(ARCH)/boot/System.map
 
 # Don't use tabs in echo arguments.
 define archhelp

arch/sparc/boot/Makefile

@@ -32,6 +32,3 @@ $(obj)/tftpboot.img: $(obj)/piggyback $(obj)/System.map $(obj)/image FORCE
 
 $(obj)/btfix.s: $(obj)/btfixupprep vmlinux FORCE
 	$(call if_changed,btfix)
-
-clean:
-	rm $(obj)/System.map

arch/sparc/kernel/entry.S

@@ -32,6 +32,7 @@
 #include <asm/obio.h>
 #include <asm/mxcc.h>
 #include <asm/thread_info.h>
+#include <asm/param.h>
 
 #include <asm/asmmacro.h>
 
@@ -1808,36 +1809,37 @@ fpload:
 	retl
 	 nop
 
-	.globl	ndelay
-ndelay:
+	/* __ndelay and __udelay take two arguments:
+	 * 0 - nsecs or usecs to delay
+	 * 1 - per_cpu udelay_val (loops per jiffy)
+	 *
+	 * Note that ndelay gives HZ times higher resolution but has a 10ms
+	 * limit.  udelay can handle up to 1s.
+	 */
+	.globl	__ndelay
+__ndelay:
 	save	%sp, -STACKFRAME_SZ, %sp
 	mov	%i0, %o0
 	call	.umul
-	 mov	5, %o1
+	 mov	0x1ad, %o1		! 2**32 / (1 000 000 000 / HZ)
+	call	.umul
+	 mov	%i1, %o1		! udelay_val
 	ba	delay_continue
-	 nop
+	 mov	%o1, %o0		! >>32 later for better resolution
 
-	.globl	udelay
-udelay:
+	.globl	__udelay
+__udelay:
 	save	%sp, -STACKFRAME_SZ, %sp
 	mov	%i0, %o0
 	sethi	%hi(0x10c6), %o1
 	call	.umul
-	 or	%o1, %lo(0x10c6), %o1
-delay_continue:
-#ifndef CONFIG_SMP
-	sethi	%hi(loops_per_jiffy), %o3
+	 or	%o1, %lo(0x10c6), %o1	! 2**32 / 1 000 000
 	call	.umul
-	 ld	[%o3 + %lo(loops_per_jiffy)], %o1
-#else
-	GET_PROCESSOR_OFFSET(o4, o2)
-	set	cpu_data, %o3
-	call	.umul
-	 ld	[%o3 + %o4], %o1
-#endif
+	 mov	%i1, %o1		! udelay_val
 	call	.umul
-	 mov	100, %o0
+	 mov	HZ, %o0			! >>32 earlier for wider range
 
+delay_continue:
 	cmp	%o0, 0x0
 1:
 	bne	1b

arch/sparc/kernel/setup.c

@@ -331,6 +331,7 @@ void __init setup_arch(char **cmdline_p)
 		if (highest_paddr < top)
 			highest_paddr = top;
 	}
+	pfn_base = phys_base >> PAGE_SHIFT;
 
 	if (!root_flags)
 		root_mountflags &= ~MS_RDONLY;

arch/sparc/kernel/smp.c

@@ -20,11 +20,11 @@
 #include <linux/fs.h>
 #include <linux/seq_file.h>
 #include <linux/cache.h>
+#include <linux/delay.h>
 
 #include <asm/ptrace.h>
 #include <asm/atomic.h>
 
-#include <asm/delay.h>
 #include <asm/irq.h>
 #include <asm/page.h>
 #include <asm/pgalloc.h>

arch/sparc/kernel/sparc_ksyms.c

@@ -145,6 +145,7 @@ EXPORT_SYMBOL(__down_interruptible);
 
 EXPORT_SYMBOL(sparc_valid_addr_bitmap);
 EXPORT_SYMBOL(phys_base);
+EXPORT_SYMBOL(pfn_base);
 
 /* Atomic operations. */
 EXPORT_SYMBOL(___atomic24_add);
@@ -164,8 +165,8 @@ EXPORT_SYMBOL(__cpu_number_map);
 EXPORT_SYMBOL(__cpu_logical_map);
 #endif
 
-EXPORT_SYMBOL(udelay);
-EXPORT_SYMBOL(ndelay);
+EXPORT_SYMBOL(__udelay);
+EXPORT_SYMBOL(__ndelay);
 EXPORT_SYMBOL(rtc_lock);
 EXPORT_SYMBOL(mostek_lock);
 EXPORT_SYMBOL(mstk48t02_regs);

arch/sparc/kernel/sun4d_smp.c

@@ -49,7 +49,6 @@ static int smp_highest_cpu;
 extern int smp_threads_ready;
 extern volatile unsigned long cpu_callin_map[NR_CPUS];
 extern struct cpuinfo_sparc cpu_data[NR_CPUS];
-extern unsigned long cpu_offset[NR_CPUS];
 extern unsigned char boot_cpu_id;
 extern int smp_activated;
 extern volatile int __cpu_number_map[NR_CPUS];
@@ -171,9 +170,6 @@ void __init smp4d_boot_cpus(void)
 
 	printk("Entering SMP Mode...\n");
 	
-	for (i = 0; i < NR_CPUS; i++)
-		cpu_offset[i] = (char *)&(cpu_data(i)) - (char *)&(cpu_data(0));
-		
 	if (boot_cpu_id)
 		current_set[0] = NULL;
 
@@ -427,9 +423,6 @@ void smp4d_message_pass(int target, int msg, unsigned long data, int wait)
 
 extern void sparc_do_profile(unsigned long pc, unsigned long o7);
 
-#define prof_multiplier(__cpu)		cpu_data(__cpu).multiplier
-#define prof_counter(__cpu)		cpu_data(__cpu).counter
-
 void smp4d_percpu_timer_interrupt(struct pt_regs *regs)
 {
 	int cpu = hard_smp4d_processor_id();

arch/sparc/kernel/sun4m_smp.c

@@ -44,7 +44,6 @@ extern unsigned long cpu_present_map;
 extern int smp_num_cpus;
 extern int smp_threads_ready;
 extern volatile unsigned long cpu_callin_map[NR_CPUS];
-extern unsigned long cpu_offset[NR_CPUS];
 extern unsigned char boot_cpu_id;
 extern int smp_activated;
 extern volatile int __cpu_number_map[NR_CPUS];
@@ -152,9 +151,7 @@ void __init smp4m_boot_cpus(void)
 	for (i = 0; !cpu_find_by_instance(i, NULL, &mid); i++)
 		cpu_present_map |= (1<<mid);
 
-	/* XXX cpu_offset is broken -Keith */
 	for(i=0; i < NR_CPUS; i++) {
-		cpu_offset[i] = (char *)&(cpu_data(i)) - (char *)&(cpu_data(0));
 		__cpu_number_map[i] = -1;
 		__cpu_logical_map[i] = -1;
 	}
@@ -409,9 +406,6 @@ void smp4m_cross_call_irq(void)
 
 extern void sparc_do_profile(unsigned long pc, unsigned long o7);
 
-#define prof_multiplier(__cpu)		cpu_data(__cpu).multiplier
-#define prof_counter(__cpu)		cpu_data(__cpu).counter
-
 void smp4m_percpu_timer_interrupt(struct pt_regs *regs)
 {
 	int cpu = smp_processor_id();

arch/sparc/mm/init.c

@@ -38,6 +38,7 @@ DEFINE_PER_CPU(struct mmu_gather, mmu_gathers);
 unsigned long *sparc_valid_addr_bitmap;
 
 unsigned long phys_base;
+unsigned long pfn_base;
 
 unsigned long page_kernel;
 
@@ -134,7 +135,7 @@ unsigned long calc_highpages(void)
 unsigned long calc_max_low_pfn(void)
 {
 	int i;
-	unsigned long tmp = (SRMMU_MAXMEM >> PAGE_SHIFT);
+	unsigned long tmp = pfn_base + (SRMMU_MAXMEM >> PAGE_SHIFT);
 	unsigned long curr_pfn, last_pfn;
 
 	last_pfn = (sp_banks[0].base_addr + sp_banks[0].num_bytes) >> PAGE_SHIFT;
@@ -189,9 +190,6 @@ unsigned long __init bootmem_init(unsigned long *pages_avail)
 	 */
 	start_pfn  = (unsigned long)__pa(PAGE_ALIGN((unsigned long) &_end));
 
-	/* Adjust up to the physical address where the kernel begins. */
-	start_pfn += phys_base;
-
 	/* Now shift down to get the real physical page frame number. */
 	start_pfn >>= PAGE_SHIFT;
 
@@ -202,8 +200,8 @@ unsigned long __init bootmem_init(unsigned long *pages_avail)
 	max_low_pfn = max_pfn;
 	highstart_pfn = highend_pfn = max_pfn;
 
-	if (max_low_pfn > (SRMMU_MAXMEM >> PAGE_SHIFT)) {
-		highstart_pfn = (SRMMU_MAXMEM >> PAGE_SHIFT);
+	if (max_low_pfn > pfn_base + (SRMMU_MAXMEM >> PAGE_SHIFT)) {
+		highstart_pfn = pfn_base + (SRMMU_MAXMEM >> PAGE_SHIFT);
 		max_low_pfn = calc_max_low_pfn();
 		printk(KERN_NOTICE "%ldMB HIGHMEM available.\n",
 		    calc_highpages() >> (20 - PAGE_SHIFT));
@@ -230,7 +228,8 @@ unsigned long __init bootmem_init(unsigned long *pages_avail)
 	}
 #endif	
 	/* Initialize the boot-time allocator. */
-	bootmap_size = init_bootmem_node(NODE_DATA(0), bootmap_pfn, phys_base>>PAGE_SHIFT, max_low_pfn);
+	bootmap_size = init_bootmem_node(NODE_DATA(0), bootmap_pfn, pfn_base,
+					 max_low_pfn);
 
 	/* Now register the available physical memory with the
 	 * allocator.
@@ -267,8 +266,8 @@ unsigned long __init bootmem_init(unsigned long *pages_avail)
 		reserve_bootmem(initrd_start, size);
 		*pages_avail -= PAGE_ALIGN(size) >> PAGE_SHIFT;
 
-		initrd_start += PAGE_OFFSET;
-		initrd_end += PAGE_OFFSET;
+		initrd_start = (initrd_start - phys_base) + PAGE_OFFSET;
+		initrd_end = (initrd_end - phys_base) + PAGE_OFFSET;		
 	}
 #endif
 	/* Reserve the kernel text/data/bss. */
@@ -432,7 +431,7 @@ void __init mem_init(void)
 
 	taint_real_pages();
 
-	max_mapnr = last_valid_pfn - (phys_base >> PAGE_SHIFT);
+	max_mapnr = last_valid_pfn - pfn_base;
 	high_memory = __va(max_low_pfn << PAGE_SHIFT);
 
 	num_physpages = totalram_pages = free_all_bootmem();
@@ -474,11 +473,9 @@ void free_initmem (void)
 
 	addr = (unsigned long)(&__init_begin);
 	for (; addr < (unsigned long)(&__init_end); addr += PAGE_SIZE) {
-		unsigned long page;
 		struct page *p;
 
-		page = addr + phys_base;
-		p = virt_to_page(page);
+		p = virt_to_page(addr);
 
 		ClearPageReserved(p);
 		set_page_count(p, 1);

arch/sparc/mm/srmmu.c

@@ -213,7 +213,7 @@ static inline pte_t srmmu_pte_mkyoung(pte_t pte)
  * and a page entry and page directory to the page they refer to.
  */
 static pte_t srmmu_mk_pte(struct page *page, pgprot_t pgprot)
-{ return __pte(((page - mem_map) << (PAGE_SHIFT-4)) | pgprot_val(pgprot)); }
+{ return __pte((page_to_pfn(page) << (PAGE_SHIFT-4)) | pgprot_val(pgprot)); }
 
 static pte_t srmmu_mk_pte_phys(unsigned long page, pgprot_t pgprot)
 { return __pte(((page) >> 4) | pgprot_val(pgprot)); }
@@ -245,7 +245,7 @@ static void srmmu_pmd_populate(pmd_t *pmdp, struct page *ptep)
 	unsigned long ptp;	/* Physical address, shifted right by 4 */
 	int i;
 
-	ptp = (ptep - mem_map) << (PAGE_SHIFT-4);	/* watch for overflow */
+	ptp = page_to_pfn(ptep) << (PAGE_SHIFT-4);	/* watch for overflow */
 	for (i = 0; i < SRMMU_PTRS_PER_PTE_SOFT/SRMMU_PTRS_PER_PTE; i++) {
 		srmmu_set_pte((pte_t *)&pmdp->pmdv[i], SRMMU_ET_PTD | ptp);
 		ptp += (SRMMU_PTRS_PER_PTE*sizeof(pte_t) >> 4);
@@ -480,7 +480,7 @@ srmmu_pte_alloc_one(struct mm_struct *mm, unsigned long address)
 
 	if ((pte = (unsigned long)srmmu_pte_alloc_one_kernel(mm, address)) == 0)
 		return NULL;
-	return mem_map + (__nocache_pa(pte) >> PAGE_SHIFT);
+	return pfn_to_page( __nocache_pa(pte) >> PAGE_SHIFT );
 }
 
 static void srmmu_free_pte_fast(pte_t *pte)
@@ -495,7 +495,7 @@ static void srmmu_pte_free(struct page *pte)
 	p = (unsigned long)page_address(pte);	/* Cached address (for test) */
 	if (p == 0)
 		BUG();
-	p = ((pte - mem_map) << PAGE_SHIFT);	/* Physical address */
+	p = page_to_pfn(pte) << PAGE_SHIFT;	/* Physical address */
 	p = (unsigned long) __nocache_va(p);	/* Nocached virtual */
 	srmmu_free_nocache(p, SRMMU_PTE_SZ_SOFT);
 }
@@ -1316,7 +1316,7 @@ void __init srmmu_paging_init(void)
 		for (znum = 0; znum < MAX_NR_ZONES; znum++)
 			zones_size[znum] = zholes_size[znum] = 0;
 
-		npages = max_low_pfn - (phys_base >> PAGE_SHIFT);
+		npages = max_low_pfn - pfn_base;
 
 		zones_size[ZONE_DMA] = npages;
 		zholes_size[ZONE_DMA] = npages - pages_avail;
@@ -1326,13 +1326,9 @@ void __init srmmu_paging_init(void)
 		zholes_size[ZONE_HIGHMEM] = npages - calc_highpages();
 
 		free_area_init_node(0, &contig_page_data, NULL, zones_size,
-				    phys_base >> PAGE_SHIFT, zholes_size);
+				    pfn_base, zholes_size);
 		mem_map = contig_page_data.node_mem_map;
 	}
-
-/* P3: easy to fix, todo. Current code is utterly broken, though. */
-	if (phys_base != 0)
-		panic("phys_base nonzero");
 }
 
 static void srmmu_mmu_info(struct seq_file *m)

arch/sparc/mm/sun4c.c

@@ -2088,7 +2088,7 @@ void __init sun4c_paging_init(void)
 		for (znum = 0; znum < MAX_NR_ZONES; znum++)
 			zones_size[znum] = zholes_size[znum] = 0;
 
-		npages = max_low_pfn - (phys_base >> PAGE_SHIFT);
+		npages = max_low_pfn - pfn_base;
 
 		zones_size[ZONE_DMA] = npages;
 		zholes_size[ZONE_DMA] = npages - pages_avail;
@@ -2098,7 +2098,7 @@ void __init sun4c_paging_init(void)
 		zholes_size[ZONE_HIGHMEM] = npages - calc_highpages();
 
 		free_area_init_node(0, &contig_page_data, NULL, zones_size,
-				    phys_base >> PAGE_SHIFT, zholes_size);
+				    pfn_base, zholes_size);
 		mem_map = contig_page_data.node_mem_map;
 	}
 

fs/Kconfig.binfmt

@@ -44,7 +44,7 @@ config BINFMT_SHARED_FLAT
 
 config BINFMT_AOUT
 	tristate "Kernel support for a.out and ECOFF binaries"
-	depends on (X86 && !X86_64) || ALPHA || ARM || M68K || MIPS || SPARC
+	depends on (X86 && !X86_64) || ALPHA || ARM || M68K || MIPS || SPARC32
 	---help---
 	  A.out (Assembler.OUTput) is a set of formats for libraries and
 	  executables used in the earliest versions of UNIX.  Linux used

include/asm-sparc/asmmacro.h

@@ -29,13 +29,6 @@
 	srl	%reg, 12, %reg; \
 	and	%reg, 3, %reg;
 
-#define GET_PROCESSOR_OFFSET(reg, tmp) \
-	GET_PROCESSOR_ID(reg) \
-	sethi	%hi(cpu_offset), %tmp; \
-	sll	%reg, 2, %reg; \
-	or	%tmp, %lo(cpu_offset), %tmp; \
-	ld	[%tmp + %reg], %reg;
-
 /* All trap entry points _must_ begin with this macro or else you
  * lose.  It makes sure the kernel has a proper window so that
  * c-code can be called.

include/asm-sparc/delay.h

@@ -7,7 +7,8 @@
 #ifndef __SPARC_DELAY_H
 #define __SPARC_DELAY_H
 
-extern unsigned long loops_per_jiffy;
+#include <linux/config.h>
+#include <asm/cpudata.h>
 
 extern __inline__ void __delay(unsigned long loops)
 {
@@ -20,7 +21,15 @@ extern __inline__ void __delay(unsigned long loops)
 }
 
 /* This is too messy with inline asm on the Sparc. */
-extern void udelay(unsigned long usecs);
-extern void ndelay(unsigned long usecs);
+extern void __udelay(unsigned long usecs, unsigned long lpj);
+extern void __ndelay(unsigned long nsecs, unsigned long lpj);
+
+#ifdef CONFIG_SMP
+#define __udelay_val	cpu_data(smp_processor_id()).udelay_val
+#else /* SMP */
+#define __udelay_val	loops_per_jiffy
+#endif /* SMP */
+#define udelay(__usecs)	__udelay(__usecs, __udelay_val)
+#define ndelay(__nsecs)	__ndelay(__nsecs, __udelay_val)
 
 #endif /* defined(__SPARC_DELAY_H) */

include/asm-sparc/page.h

@@ -156,17 +156,22 @@ extern __inline__ int get_order(unsigned long size)
 #define PAGE_ALIGN(addr)  (((addr)+PAGE_SIZE-1)&PAGE_MASK)
 
 #define PAGE_OFFSET	0xf0000000
-#define __pa(x)                 ((unsigned long)(x) - PAGE_OFFSET)
-#define __va(x)                 ((void *)((unsigned long) (x) + PAGE_OFFSET))
+#ifndef __ASSEMBLY__
+extern unsigned long phys_base;
+extern unsigned long pfn_base;
+#endif
+#define __pa(x)			((unsigned long)(x) - PAGE_OFFSET + phys_base)
+#define __va(x)			((void *)((unsigned long) (x) - phys_base + PAGE_OFFSET))
+
+#define virt_to_phys		__pa
+#define phys_to_virt		__va
 
-#define virt_to_phys(x)		__pa((unsigned long)(x))
-#define phys_to_virt(x)		__va((unsigned long)(x))
+#define pfn_to_page(pfn)	(mem_map + ((pfn)-(pfn_base)))
+#define page_to_pfn(page)	((unsigned long)(((page) - mem_map) + pfn_base))
+#define virt_to_page(kaddr)	(mem_map + ((((unsigned long)(kaddr)-PAGE_OFFSET)>>PAGE_SHIFT)))
 
-#define pfn_to_page(pfn)        (mem_map + (pfn))
-#define page_to_pfn(page)       ((unsigned long)((page) - mem_map))
-#define virt_to_page(kaddr)	(mem_map + (__pa(kaddr) >> PAGE_SHIFT))
-#define pfn_valid(pfn)		((pfn) < max_mapnr)
-#define virt_addr_valid(kaddr)	pfn_valid(__pa(kaddr) >> PAGE_SHIFT)
+#define pfn_valid(pfn)		(((pfn) >= (pfn_base)) && (((pfn)-(pfn_base)) < max_mapnr))
+#define virt_addr_valid(kaddr)	((((unsigned long)(kaddr)-PAGE_OFFSET)>>PAGE_SHIFT) < max_mapnr)
 
 #define VM_DATA_DEFAULT_FLAGS	(VM_READ | VM_WRITE | VM_EXEC | \
 				 VM_MAYREAD | VM_MAYWRITE | VM_MAYEXEC)

include/asm-sparc/pgtable.h

@@ -181,6 +181,7 @@ extern int num_contexts;
  * hit for all __pa()/__va() operations.
  */
 extern unsigned long phys_base;
+extern unsigned long pfn_base;
 
 /*
  * BAD_PAGETABLE is used when we need a bogus page-table, while

include/asm-sparc/smp.h

@@ -148,7 +148,10 @@ extern __inline__ int hard_smp_processor_id(void)
 }
 #endif
 
-#define smp_processor_id() hard_smp_processor_id()
+#define smp_processor_id()	(current_thread_info()->cpu)
+
+#define prof_multiplier(__cpu)		cpu_data(__cpu).multiplier
+#define prof_counter(__cpu)		cpu_data(__cpu).counter
 
 #endif /* !(__ASSEMBLY__) */