SPARC: More work to get sparc32 working in 2.5.x

- page-size PTE directory with 16-word pmd_t as suggested by RMK and Riel
- support for 2.5.x softirq infrastructure
- other miscellanea

arch/sparc/kernel/ebus.c

@@ -24,9 +24,6 @@
 
 struct linux_ebus *ebus_chain = 0;
 
-#ifdef CONFIG_SUN_AUXIO
-extern void auxio_probe(void);
-#endif
 extern void rs_init(void);
 
 /* We are together with pcic.c under CONFIG_PCI. */
@@ -366,7 +363,4 @@ void __init ebus_init(void)
 	}
 
 	rs_init();
-#ifdef CONFIG_SUN_AUXIO
-	auxio_probe();
-#endif
 }

arch/sparc/kernel/irq.c

@@ -415,7 +415,7 @@ void handler_irq(int irq, struct pt_regs * regs)
 	extern void smp4m_irq_rotate(int cpu);
 #endif
 
-	irq_enter(cpu, irq);
+	irq_enter();
 	disable_pil_irq(irq);
 #ifdef CONFIG_SMP
 	/* Only rotate on lower priority IRQ's (scsi, ethernet, etc.). */
@@ -431,9 +431,7 @@ void handler_irq(int irq, struct pt_regs * regs)
 		action = action->next;
 	} while (action);
 	enable_pil_irq(irq);
-	irq_exit(cpu, irq);
-	if (softirq_pending(cpu))
-		do_softirq();
+	irq_exit();
 }
 
 #ifdef CONFIG_BLK_DEV_FD
@@ -444,13 +442,14 @@ void sparc_floppy_irq(int irq, void *dev_id, struct pt_regs *regs)
 	int cpu = smp_processor_id();
 
 	disable_pil_irq(irq);
-	irq_enter(cpu, irq);
+	irq_enter();
 	kstat.irqs[cpu][irq]++;
 	floppy_interrupt(irq, dev_id, regs);
-	irq_exit(cpu, irq);
+	irq_exit();
 	enable_pil_irq(irq);
-	if (softirq_pending(cpu))
-		do_softirq();
+	// XXX Eek, it's totally changed with preempt_count() and such
+	// if (softirq_pending(cpu))
+	//	do_softirq();
 }
 #endif
 
@@ -644,7 +643,7 @@ void __init init_IRQ(void)
 	extern void sun4c_init_IRQ( void );
 	extern void sun4m_init_IRQ( void );
 	extern void sun4d_init_IRQ( void );
-    
+
 	switch(sparc_cpu_model) {
 	case sun4c:
 	case sun4:

arch/sparc/kernel/pcic.c

@@ -55,10 +55,6 @@ asmlinkage int sys_pciconfig_write(unsigned long bus,
 
 #else
 
-#ifdef CONFIG_SUN_JSFLASH
-extern int jsflash_init(void);
-#endif
-
 struct pci_fixup pcibios_fixups[] = {
 	{ 0 }
 };
@@ -435,7 +431,7 @@ static void __init pcic_pbm_scan_bus(struct linux_pcic *pcic)
 /*
  * Main entry point from the PCI subsystem.
  */
-void __init pcibios_init(void)
+static int __init pcibios_init(void)
 {
 	struct linux_pcic *pcic;
 
@@ -444,7 +440,7 @@ void __init pcibios_init(void)
 	 * So, here we report the presence of PCIC and do some magic passes.
 	 */
 	if(!pcic0_up)
-		return;
+		return 0;
 	pcic = &pcic0;
 
 	/*
@@ -465,9 +461,7 @@ void __init pcibios_init(void)
 	pcic_pbm_scan_bus(pcic);
 
 	ebus_init();
-#ifdef CONFIG_SUN_JSFLASH
-	jsflash_init();
-#endif
+	return 0;
 }
 
 int pcic_present(void)
@@ -1037,3 +1031,5 @@ void insl(unsigned long addr, void *dst, unsigned long count) {
 }
 
 #endif
+
+subsys_initcall(pcibios_init);

arch/sparc/kernel/rtrap.S

@@ -86,7 +86,8 @@ ret_trap_continue:
 	wr	%t_psr, 0x0, %psr
 	WRITE_PAUSE
 
-	ld	[%curptr + AOFF_task_thread + AOFF_thread_w_saved], %twin_tmp1
+	ld	[%curptr + TI_TASK], %o5
+	ld	[%o5 + AOFF_task_thread + AOFF_thread_w_saved], %twin_tmp1
 	orcc	%g0, %twin_tmp1, %g0
 	be	ret_trap_nobufwins
 	 nop

arch/sparc/kernel/sun4d_irq.c

@@ -198,7 +198,7 @@ void sun4d_handler_irq(int irq, struct pt_regs * regs)
 	
 	cc_set_iclr(1 << irq);
 	
-	irq_enter(cpu, irq);
+	irq_enter();
 	kstat.irqs[cpu][irq]++;
 	if (!sbusl) {
 		action = *(irq + irq_action);
@@ -239,9 +239,7 @@ void sun4d_handler_irq(int irq, struct pt_regs * regs)
 					}
 			}
 	}
-	irq_exit(cpu, irq);
-	if (softirq_pending(cpu))
-		do_softirq();
+	irq_exit();
 }
 
 unsigned int sun4d_build_irq(struct sbus_dev *sdev, int irq)

arch/sparc/kernel/sun4d_smp.c

@@ -458,9 +458,9 @@ void smp4d_percpu_timer_interrupt(struct pt_regs *regs)
 	if(!--prof_counter[cpu]) {
 		int user = user_mode(regs);
 
-		irq_enter(cpu, 0);
+		irq_enter();
 		update_process_times(user);
-		irq_exit(cpu, 0);
+		irq_exit();
 
 		prof_counter[cpu] = prof_multiplier[cpu];
 	}

arch/sparc/kernel/sun4m_smp.c

@@ -445,9 +445,9 @@ void smp4m_percpu_timer_interrupt(struct pt_regs *regs)
 	if(!--prof_counter[cpu]) {
 		int user = user_mode(regs);
 
-		irq_enter(cpu, 0);
+		irq_enter();
 		update_process_times(user);
-		irq_exit(cpu, 0);
+		irq_exit();
 
 		prof_counter[cpu] = prof_multiplier[cpu];
 	}

arch/sparc/mm/sun4c.c

@@ -1710,13 +1710,13 @@ static void sun4c_pgd_set(pgd_t * pgdp, pmd_t * pmdp)
 
 static void sun4c_pmd_set(pmd_t * pmdp, pte_t * ptep)
 {
-	*pmdp = __pmd(PGD_TABLE | (unsigned long) ptep);
+	pmdp->pmdv[0] = PGD_TABLE | (unsigned long) ptep;
 }
 
 static void sun4c_pmd_populate(pmd_t * pmdp, struct page * ptep)
 {
 	if (page_address(ptep) == NULL) BUG();	/* No highmem on sun4c */
-	*pmdp = __pmd(PGD_TABLE | (unsigned long) page_address(ptep));
+	pmdp->pmdv[0] = PGD_TABLE | (unsigned long) page_address(ptep);
 }
 
 static int sun4c_pte_present(pte_t pte)
@@ -1735,7 +1735,14 @@ static int sun4c_pmd_present(pmd_t pmd)
 {
 	return ((pmd_val(pmd) & PGD_PRESENT) != 0);
 }
+
+#if 0 /* if PMD takes one word */
 static void sun4c_pmd_clear(pmd_t *pmdp)	{ *pmdp = __pmd(0); }
+#else /* if pmd_t is a longish aggregate */
+static void sun4c_pmd_clear(pmd_t *pmdp) {
+	memset((void *)pmdp, 0, sizeof(pmd_t));
+}
+#endif
 
 static int sun4c_pgd_none(pgd_t pgd)		{ return 0; }
 static int sun4c_pgd_bad(pgd_t pgd)		{ return 0; }
@@ -1913,7 +1920,7 @@ static void sun4c_pte_free(struct page *pte)
  * allocating and freeing a pmd is trivial: the 1-entry pmd is
  * inside the pgd, so has no extra memory associated with it.
  */
-static pmd_t *sun4c_pmd_alloc_one_fast(struct mm_struct *mm, unsigned long address)
+static pmd_t *sun4c_pmd_alloc_one(struct mm_struct *mm, unsigned long address)
 {
 	BUG();
 	return NULL;
@@ -2176,7 +2183,7 @@ void __init ld_mmu_sun4c(void)
 	BTFIXUPSET_CALL(pte_alloc_one_kernel, sun4c_pte_alloc_one_kernel, BTFIXUPCALL_NORM);
 	BTFIXUPSET_CALL(pte_alloc_one, sun4c_pte_alloc_one, BTFIXUPCALL_NORM);
 	BTFIXUPSET_CALL(free_pmd_fast, sun4c_free_pmd_fast, BTFIXUPCALL_NOP);
-	BTFIXUPSET_CALL(pmd_alloc_one_fast, sun4c_pmd_alloc_one_fast, BTFIXUPCALL_RETO0);
+	BTFIXUPSET_CALL(pmd_alloc_one, sun4c_pmd_alloc_one, BTFIXUPCALL_RETO0);
 	BTFIXUPSET_CALL(free_pgd_fast, sun4c_free_pgd_fast, BTFIXUPCALL_NORM);
 	BTFIXUPSET_CALL(get_pgd_fast, sun4c_get_pgd_fast, BTFIXUPCALL_NORM);
 

drivers/sbus/char/jsflash.c

@@ -674,10 +674,9 @@ int jsfd_init(void) {
 	return 0;
 }
 
-#ifdef MODULE
 MODULE_LICENSE("GPL");
 
-int init_module(void) {
+static int __init jsflash_init_module(void) {
 	int rc;
 
 	if ((rc = jsflash_init()) == 0) {
@@ -687,7 +686,7 @@ int init_module(void) {
 	return rc;
 }
 
-void cleanup_module(void) {
+static void __exit jsflash_cleanup_module(void) {
 
 	/* for (all probed units) {  } */
 	if (jsf0.busy)
@@ -700,4 +699,6 @@ void cleanup_module(void) {
 		printk("jsfd: cleanup_module failed\n");
 	blk_cleanup_queue(BLK_DEFAULT_QUEUE(MAJOR_NR));
 }
-#endif
+
+module_init(jsflash_init_module);
+module_exit(jsflash_cleanup_module);

include/asm-sparc/hardirq.h

@@ -12,47 +12,108 @@
 #include <linux/brlock.h>
 #include <linux/spinlock.h>
 
-/* entry.S is sensitive to the offsets of these fields */
+/* entry.S is sensitive to the offsets of these fields */ /* XXX P3 Is it? */
 typedef struct {
 	unsigned int __softirq_pending;
 	unsigned int __unused_1;
 #ifndef CONFIG_SMP
-	unsigned int __local_irq_count;
+	unsigned int WAS__local_irq_count;
 #else
 	unsigned int __unused_on_SMP;	/* DaveM says use brlock for SMP irq. KAO */
 #endif
-	unsigned int __local_bh_count;
+	unsigned int WAS__local_bh_count;
 	unsigned int __syscall_count;
         struct task_struct * __ksoftirqd_task;
 } ____cacheline_aligned irq_cpustat_t;
 
 #include <linux/irq_cpustat.h>	/* Standard mappings for irq_cpustat_t above */
-/* Note that local_irq_count() is replaced by sparc64 specific version for SMP */
 
-#ifndef CONFIG_SMP
-#define irq_enter(cpu, irq)	((void)(irq), local_irq_count(cpu)++)
-#define irq_exit(cpu, irq)	((void)(irq), local_irq_count(cpu)--)
-#else
-#undef local_irq_count
-#define local_irq_count(cpu)	(__brlock_array[cpu][BR_GLOBALIRQ_LOCK])
-#define irq_enter(cpu, irq)	br_read_lock(BR_GLOBALIRQ_LOCK)
-#define irq_exit(cpu, irq)	br_read_unlock(BR_GLOBALIRQ_LOCK)
+/*
+ * We put the hardirq and softirq counter into the preemption
+ * counter. The bitmask has the following meaning:
+ *
+ * - bits 0-7 are the preemption count (max preemption depth: 256)
+ * - bits 8-15 are the softirq count (max # of softirqs: 256)
+ * - bits 16-23 are the hardirq count (max # of hardirqs: 256)
+ *
+ * - ( bit 26 is the PREEMPT_ACTIVE flag. )
+ *
+ * PREEMPT_MASK: 0x000000ff
+ * HARDIRQ_MASK: 0x0000ff00
+ * SOFTIRQ_MASK: 0x00ff0000
+ */
+
+#define PREEMPT_BITS    8
+#define SOFTIRQ_BITS    8
+#define HARDIRQ_BITS    8
+
+#define PREEMPT_SHIFT   0
+#define SOFTIRQ_SHIFT   (PREEMPT_SHIFT + PREEMPT_BITS)
+#define HARDIRQ_SHIFT   (SOFTIRQ_SHIFT + SOFTIRQ_BITS)
+
+#define __MASK(x)       ((1UL << (x))-1)
+
+#define PREEMPT_MASK    (__MASK(PREEMPT_BITS) << PREEMPT_SHIFT)
+#define HARDIRQ_MASK    (__MASK(HARDIRQ_BITS) << HARDIRQ_SHIFT)
+#define SOFTIRQ_MASK    (__MASK(SOFTIRQ_BITS) << SOFTIRQ_SHIFT)
+
+#define hardirq_count() (preempt_count() & HARDIRQ_MASK)
+#define softirq_count() (preempt_count() & SOFTIRQ_MASK)
+#define irq_count()     (preempt_count() & (HARDIRQ_MASK | SOFTIRQ_MASK))
+
+#define PREEMPT_OFFSET  (1UL << PREEMPT_SHIFT)
+#define SOFTIRQ_OFFSET  (1UL << SOFTIRQ_SHIFT)
+#define HARDIRQ_OFFSET  (1UL << HARDIRQ_SHIFT)
+
+/*
+ * The hardirq mask has to be large enough to have
+ * space for potentially all IRQ sources in the system
+ * nesting on a single CPU:
+ */
+#if (1 << HARDIRQ_BITS) < NR_IRQS
+# error HARDIRQ_BITS is too low!
 #endif
 
 /*
- * Are we in an interrupt context? Either doing bottom half
- * or hardware interrupt processing?
+ * Are we doing bottom half or hardware interrupt processing?
+ * Are we in a softirq context? Interrupt context?
  */
-#define in_interrupt() ((local_irq_count(smp_processor_id()) + \
-		         local_bh_count(smp_processor_id())) != 0)
+#define in_irq()                (hardirq_count())
+#define in_softirq()            (softirq_count())
+#define in_interrupt()          (irq_count())
+
 
-/* This tests only the local processors hw IRQ context disposition.  */
-#define in_irq() (local_irq_count(smp_processor_id()) != 0)
+#define hardirq_trylock()       (!in_interrupt())
+#define hardirq_endlock()       do { } while (0)
 
 #ifndef CONFIG_SMP
+#define irq_enter()             (preempt_count() += HARDIRQ_OFFSET)
 
-#define hardirq_trylock(cpu)	((void)(cpu), local_irq_count(smp_processor_id()) == 0)
-#define hardirq_endlock(cpu)	do { (void)(cpu); } while(0)
+#if CONFIG_PREEMPT
+# define IRQ_EXIT_OFFSET (HARDIRQ_OFFSET-1)
+#else
+# define IRQ_EXIT_OFFSET HARDIRQ_OFFSET
+#endif
+#define irq_exit()                                                      \
+do {                                                                    \
+                preempt_count() -= IRQ_EXIT_OFFSET;                     \
+                if (!in_interrupt() && softirq_pending(smp_processor_id())) \
+                        do_softirq();                                   \
+                preempt_enable_no_resched();                            \
+} while (0)
+
+#else
+
+/* Note that local_irq_count() is replaced by sparc64 specific version for SMP */
+
+/* XXX This is likely to be broken by the above preempt-based IRQs */
+#define irq_enter()		br_read_lock(BR_GLOBALIRQ_LOCK)
+#undef local_irq_count
+#define local_irq_count(cpu)	(__brlock_array[cpu][BR_GLOBALIRQ_LOCK])
+#define irq_exit()		br_read_unlock(BR_GLOBALIRQ_LOCK)
+#endif
+
+#ifndef CONFIG_SMP
 
 #define synchronize_irq()	barrier()
 
@@ -79,17 +140,19 @@ static inline void release_irqlock(int cpu)
 	}
 }
 
+#if 0
 static inline int hardirq_trylock(int cpu)
 {
 	spinlock_t *lock = &__br_write_locks[BR_GLOBALIRQ_LOCK].lock;
 
 	return (!local_irq_count(cpu) && !spin_is_locked(lock));
 }
+#endif
 
-#define hardirq_endlock(cpu)	do { (void)(cpu); } while (0)
-
-extern void synchronize_irq(void);
+extern void synchronize_irq(unsigned int irq);
 
 #endif /* CONFIG_SMP */
 
+// extern void show_stack(unsigned long * esp);
+
 #endif /* __SPARC_HARDIRQ_H */

include/asm-sparc/page.h

@@ -24,7 +24,7 @@
 
 #ifdef __KERNEL__
 
-#include <asm/head.h>       /* for KERNBASE */
+/* #include <asm/head.h> XXX */      /* for KERNBASE */
 #include <asm/btfixup.h>
 
 #ifndef __ASSEMBLY__
@@ -89,7 +89,7 @@ extern struct cache_palias *sparc_aliases;
  */
 typedef struct { unsigned long pte; } pte_t;
 typedef struct { unsigned long iopte; } iopte_t;
-typedef struct { unsigned long pmd; } pmd_t;
+typedef struct { unsigned long pmdv[16]; } pmd_t;
 typedef struct { unsigned long pgd; } pgd_t;
 typedef struct { unsigned long ctxd; } ctxd_t;
 typedef struct { unsigned long pgprot; } pgprot_t;
@@ -97,7 +97,7 @@ typedef struct { unsigned long iopgprot; } iopgprot_t;
 
 #define pte_val(x)	((x).pte)
 #define iopte_val(x)	((x).iopte)
-#define pmd_val(x)      ((x).pmd)
+#define pmd_val(x)      ((x).pmdv[0])
 #define pgd_val(x)	((x).pgd)
 #define ctxd_val(x)	((x).ctxd)
 #define pgprot_val(x)	((x).pgprot)
@@ -105,7 +105,7 @@ typedef struct { unsigned long iopgprot; } iopgprot_t;
 
 #define __pte(x)	((pte_t) { (x) } )
 #define __iopte(x)	((iopte_t) { (x) } )
-#define __pmd(x)        ((pmd_t) { (x) } )
+/* #define __pmd(x)        ((pmd_t) { (x) } ) */ /* XXX procedure with loop */
 #define __pgd(x)	((pgd_t) { (x) } )
 #define __ctxd(x)	((ctxd_t) { (x) } )
 #define __pgprot(x)	((pgprot_t) { (x) } )
@@ -117,7 +117,7 @@ typedef struct { unsigned long iopgprot; } iopgprot_t;
  */
 typedef unsigned long pte_t;
 typedef unsigned long iopte_t;
-typedef unsigned long pmd_t;
+typedef struct { unsigned long pmdv[16]; } pmd_t;
 typedef unsigned long pgd_t;
 typedef unsigned long ctxd_t;
 typedef unsigned long pgprot_t;
@@ -125,7 +125,7 @@ typedef unsigned long iopgprot_t;
 
 #define pte_val(x)	(x)
 #define iopte_val(x)	(x)
-#define pmd_val(x)      (x)
+#define pmd_val(x)      ((x).pmdv[0])
 #define pgd_val(x)	(x)
 #define ctxd_val(x)	(x)
 #define pgprot_val(x)	(x)
@@ -133,7 +133,7 @@ typedef unsigned long iopgprot_t;
 
 #define __pte(x)	(x)
 #define __iopte(x)	(x)
-#define __pmd(x)        (x)
+/* #define __pmd(x)        (x) */ /* XXX later */
 #define __pgd(x)	(x)
 #define __ctxd(x)	(x)
 #define __pgprot(x)	(x)

include/asm-sparc/pgalloc.h

@@ -40,13 +40,8 @@ BTFIXUPDEF_CALL(void, pgd_set, pgd_t *, pmd_t *)
 #define pgd_set(pgdp,pmdp) BTFIXUP_CALL(pgd_set)(pgdp,pmdp)
 #define pgd_populate(MM, PGD, PMD)      pgd_set(PGD, PMD)
 
-static __inline__ pmd_t *pmd_alloc_one(struct mm_struct *mm, unsigned long address)
-{
-	return 0;
-}
-
-BTFIXUPDEF_CALL(pmd_t *, pmd_alloc_one_fast, struct mm_struct *, unsigned long)
-#define pmd_alloc_one_fast(mm, address)	BTFIXUP_CALL(pmd_alloc_one_fast)(mm, address)
+BTFIXUPDEF_CALL(pmd_t *, pmd_alloc_one, struct mm_struct *, unsigned long)
+#define pmd_alloc_one(mm, address)	BTFIXUP_CALL(pmd_alloc_one)(mm, address)
 
 BTFIXUPDEF_CALL(void, free_pmd_fast, pmd_t *)
 #define free_pmd_fast(pmd)	BTFIXUP_CALL(free_pmd_fast)(pmd)
@@ -65,8 +60,7 @@ BTFIXUPDEF_CALL(pte_t *, pte_alloc_one_kernel, struct mm_struct *, unsigned long
 #define pte_alloc_one_kernel(mm, addr)	BTFIXUP_CALL(pte_alloc_one_kernel)(mm, addr)
 
 BTFIXUPDEF_CALL(void, free_pte_fast, pte_t *)
-#define free_pte_fast(pte)	BTFIXUP_CALL(free_pte_fast)(pte)
-#define pte_free_kernel(pte)	free_pte_fast(pte)
+#define pte_free_kernel(pte)	BTFIXUP_CALL(free_pte_fast)(pte)
 
 BTFIXUPDEF_CALL(void, pte_free, struct page *)
 #define pte_free(pte)		BTFIXUP_CALL(pte_free)(pte)

include/asm-sparc/pgtable.h

@@ -11,9 +11,7 @@
 
 #include <linux/config.h>
 #include <linux/spinlock.h>
-/* XXX This creates many nasty warnings. */
-/* #include <linux/highmem.h> */	/* kmap_atomic in pte_offset_map */
-#include <asm/asi.h>
+/* #include <asm/asi.h> */	/* doesn't seem like being used XXX */
 #ifdef CONFIG_SUN4
 #include <asm/pgtsun4.h>
 #else
@@ -32,10 +30,6 @@ struct vm_area_struct;
 
 extern void load_mmu(void);
 extern unsigned long calc_highpages(void);
-			       
-BTFIXUPDEF_CALL(void, quick_kernel_fault, unsigned long)
-
-#define quick_kernel_fault(addr) BTFIXUP_CALL(quick_kernel_fault)(addr)
 
 /* Routines for data transfer buffers. */
 BTFIXUPDEF_CALL(char *, mmu_lockarea, char *, unsigned long)
@@ -189,14 +183,8 @@ extern unsigned long empty_zero_page;
 #define BAD_PAGE __bad_page()
 #define ZERO_PAGE(vaddr) (virt_to_page(empty_zero_page))
 
-/* number of bits that fit into a memory pointer */
-#define BITS_PER_PTR      (8*sizeof(unsigned long))
-
-/* to align the pointer to a pointer address */
-#define PTR_MASK          (~(sizeof(void*)-1))
-
-#define SIZEOF_PTR_LOG2   2
-
+/*
+ */
 BTFIXUPDEF_CALL_CONST(struct page *, pmd_page, pmd_t)
 BTFIXUPDEF_CALL_CONST(unsigned long, pgd_page, pgd_t)
 
@@ -340,20 +328,15 @@ BTFIXUPDEF_CALL(pmd_t *, pmd_offset, pgd_t *, unsigned long)
 BTFIXUPDEF_CALL(pte_t *, pte_offset_kernel, pmd_t *, unsigned long)
 #define pte_offset_kernel(dir,addr) BTFIXUP_CALL(pte_offset_kernel)(dir,addr)
 
-/* __pte_offset is not BTFIXUP-ed, but PTRS_PER_PTE is, so it's ok. */
-#define __pte_offset(address) \
-	(((address) >> PAGE_SHIFT) & (PTRS_PER_PTE - 1))
-#if 0 /* XXX Should we expose pmd_page_kernel? */
-#define pte_offset_kernel(dir, addr) \
-	((pte_t *) pmd_page_kernel(*(dir)) + __pte_offset(addr))
-#endif
-#define pte_offset_map(dir, addr) \
-	((pte_t *) kmap_atomic(pmd_page(*(dir)), KM_PTE0) + __pte_offset(addr))
-#define pte_offset_map_nested(dir, addr) \
-	((pte_t *) kmap_atomic(pmd_page(*(dir)), KM_PTE1) + __pte_offset(addr))
+/*
+ * This shortcut works on sun4m (and sun4d) because the nocache area is static,
+ * and sun4c is guaranteed to have no highmem anyway.
+ */
+#define pte_offset_map(d, a)		pte_offset_kernel(d,a)
+#define pte_offset_map_nested(d, a)	pte_offset_kernel(d,a)
 
-#define pte_unmap(pte)		kunmap_atomic(pte, KM_PTE0)
-#define pte_unmap_nested(pte)	kunmap_atomic(pte, KM_PTE1)
+#define pte_unmap(pte)		do{}while(0)
+#define pte_unmap_nested(pte)	do{}while(0)
 
 /* The permissions for pgprot_val to make a page mapped on the obio space */
 extern unsigned int pg_iobits;

include/asm-sparc/pgtsrmmu.h

@@ -10,11 +10,11 @@
 #include <asm/page.h>
 #include <asm/thread_info.h>	/* TI_UWINMASK for WINDOW_FLUSH */
 
-/* PMD_SHIFT determines the size of the area a second-level page table can map */
+/* PMD_SHIFT determines the size of the area a second-level page table entry can map */
 #define SRMMU_PMD_SHIFT         18
 #define SRMMU_PMD_SIZE          (1UL << SRMMU_PMD_SHIFT)
 #define SRMMU_PMD_MASK          (~(SRMMU_PMD_SIZE-1))
-#define SRMMU_PMD_ALIGN(addr)   (((addr)+SRMMU_PMD_SIZE-1)&SRMMU_PMD_MASK)
+/* #define SRMMU_PMD_ALIGN(addr)   (((addr)+SRMMU_PMD_SIZE-1)&SRMMU_PMD_MASK) */
 
 /* PGDIR_SHIFT determines what a third-level page table entry can map */
 #define SRMMU_PGDIR_SHIFT       24

include/asm-sparc/softirq.h

@@ -7,21 +7,25 @@
 #ifndef __SPARC_SOFTIRQ_H
 #define __SPARC_SOFTIRQ_H
 
-#include <linux/threads.h>	/* For NR_CPUS */
+// #include <linux/threads.h>	/* For NR_CPUS */
 
-#include <asm/atomic.h>
+// #include <asm/atomic.h>
 #include <asm/smp.h>
 #include <asm/hardirq.h>
 
-#define local_bh_disable()	(local_bh_count(smp_processor_id())++)
-#define __local_bh_enable()	(local_bh_count(smp_processor_id())--)
+#define local_bh_disable() \
+		do { preempt_count() += SOFTIRQ_OFFSET; barrier(); } while (0)
+#define __local_bh_enable() \
+		do { barrier(); preempt_count() -= SOFTIRQ_OFFSET; } while (0)
+
 #define local_bh_enable()			  \
-do { if (!--local_bh_count(smp_processor_id()) && \
+do {						\
+	__local_bh_enable();			\
+	if (!in_interrupt() &&			\
 	 softirq_pending(smp_processor_id())) {   \
 		do_softirq();			  \
-		local_irq_enable();			  \
      }						  \
+	preempt_check_resched();		  \
 } while (0)
-#define in_softirq() (local_bh_count(smp_processor_id()) != 0)
 
 #endif	/* __SPARC_SOFTIRQ_H */

include/asm-sparc/system.h

@@ -104,7 +104,7 @@ extern void fpsave(unsigned long *fpregs, unsigned long *fsr,
  * XXX prepare_arch_switch() is much smarter than this in sparc64, are we sure?
  * XXX Cosider if doing it the flush_user_windows way is faster (by uwinmask).
  */
-#define prepare_arch_switch(rq) do { \
+#define prepare_arch_switch(rq, next) do { \
 	__asm__ __volatile__( \
 	".globl\tflush_patch_switch\nflush_patch_switch:\n\t" \
 	"save %sp, -0x40, %sp; save %sp, -0x40, %sp; save %sp, -0x40, %sp\n\t" \
@@ -112,7 +112,7 @@ extern void fpsave(unsigned long *fpregs, unsigned long *fsr,
 	"save %sp, -0x40, %sp\n\t" \
 	"restore; restore; restore; restore; restore; restore; restore"); \
 } while(0)
-#define finish_arch_switch(rq)	do{ }while(0)
+#define finish_arch_switch(rq, next)	do{ }while(0)
 
 	/* Much care has gone into this code, do not touch it.
 	 *