Merge 3.10-rc3 into tty-next

We want these fixes.

Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

12 files changed, 231 insertions, 138 deletions

diff --git a/arch/parisc/Kconfig b/arch/parisc/Kconfig
index cad060f..6507dab 100644
--- a/arch/parisc/Kconfig
+++ b/arch/parisc/Kconfig
@@ -245,7 +245,7 @@ config SMP
 
 config IRQSTACKS
 	bool "Use separate kernel stacks when processing interrupts"
-	default n
+	default y
 	help
 	  If you say Y here the kernel will use separate kernel stacks
 	  for handling hard and soft interrupts.  This can help avoid
diff --git a/arch/parisc/Makefile b/arch/parisc/Makefile
index 2f967cc..1976900 100644
--- a/arch/parisc/Makefile
+++ b/arch/parisc/Makefile
@@ -23,24 +23,21 @@ NM		= sh $(srctree)/arch/parisc/nm
 CHECKFLAGS	+= -D__hppa__=1
 LIBGCC		= $(shell $(CC) $(KBUILD_CFLAGS) -print-libgcc-file-name)
 
-MACHINE		:= $(shell uname -m)
-NATIVE		:= $(if $(filter parisc%,$(MACHINE)),1,0)
-
 ifdef CONFIG_64BIT
 UTS_MACHINE	:= parisc64
 CHECKFLAGS	+= -D__LP64__=1 -m64
-WIDTH		:= 64
+CC_ARCHES	= hppa64
 else # 32-bit
-WIDTH		:=
+CC_ARCHES	= hppa hppa2.0 hppa1.1
 endif
 
-# attempt to help out folks who are cross-compiling
-ifeq ($(NATIVE),1)
-CROSS_COMPILE	:= hppa$(WIDTH)-linux-
-else
- ifeq ($(CROSS_COMPILE),)
- CROSS_COMPILE	:= hppa$(WIDTH)-linux-gnu-
- endif
+ifneq ($(SUBARCH),$(UTS_MACHINE))
+	ifeq ($(CROSS_COMPILE),)
+		CC_SUFFIXES = linux linux-gnu unknown-linux-gnu
+		CROSS_COMPILE := $(call cc-cross-prefix, \
+			$(foreach a,$(CC_ARCHES), \
+			$(foreach s,$(CC_SUFFIXES),$(a)-$(s)-)))
+	endif
 endif
 
 OBJCOPY_FLAGS =-O binary -R .note -R .comment -S
diff --git a/arch/parisc/include/asm/assembly.h b/arch/parisc/include/asm/assembly.h
index 89fb400..0da8482 100644
--- a/arch/parisc/include/asm/assembly.h
+++ b/arch/parisc/include/asm/assembly.h
@@ -438,7 +438,6 @@
 	SAVE_SP  (%sr4, PT_SR4 (\regs))
 	SAVE_SP  (%sr5, PT_SR5 (\regs))
 	SAVE_SP  (%sr6, PT_SR6 (\regs))
-	SAVE_SP  (%sr7, PT_SR7 (\regs))
 
 	SAVE_CR  (%cr17, PT_IASQ0(\regs))
 	mtctl	 %r0,	%cr17
diff --git a/arch/parisc/include/asm/hardirq.h b/arch/parisc/include/asm/hardirq.h
index 12373c4..241c345 100644
--- a/arch/parisc/include/asm/hardirq.h
+++ b/arch/parisc/include/asm/hardirq.h
@@ -11,15 +11,20 @@
 #include <linux/threads.h>
 #include <linux/irq.h>
 
+#ifdef CONFIG_IRQSTACKS
+#define __ARCH_HAS_DO_SOFTIRQ
+#endif
+
 typedef struct {
 	unsigned int __softirq_pending;
-#ifdef CONFIG_DEBUG_STACKOVERFLOW
 	unsigned int kernel_stack_usage;
-#endif
+	unsigned int irq_stack_usage;
 #ifdef CONFIG_SMP
 	unsigned int irq_resched_count;
 	unsigned int irq_call_count;
 #endif
+	unsigned int irq_unaligned_count;
+	unsigned int irq_fpassist_count;
 	unsigned int irq_tlb_count;
 } ____cacheline_aligned irq_cpustat_t;
 
@@ -28,6 +33,7 @@ DECLARE_PER_CPU_SHARED_ALIGNED(irq_cpustat_t, irq_stat);
 #define __ARCH_IRQ_STAT
 #define __IRQ_STAT(cpu, member) (irq_stat[cpu].member)
 #define inc_irq_stat(member)	this_cpu_inc(irq_stat.member)
+#define __inc_irq_stat(member)	__this_cpu_inc(irq_stat.member)
 #define local_softirq_pending()	this_cpu_read(irq_stat.__softirq_pending)
 
 #define __ARCH_SET_SOFTIRQ_PENDING
diff --git a/arch/parisc/include/asm/processor.h b/arch/parisc/include/asm/processor.h
index 0640155..cc2290a 100644
--- a/arch/parisc/include/asm/processor.h
+++ b/arch/parisc/include/asm/processor.h
@@ -17,7 +17,6 @@
 #include <asm/ptrace.h>
 #include <asm/types.h>
 #include <asm/percpu.h>
-
 #endif /* __ASSEMBLY__ */
 
 /*
@@ -59,23 +58,6 @@
 #ifndef __ASSEMBLY__
 
 /*
- * IRQ STACK - used for irq handler
- */
-#ifdef __KERNEL__
-
-#define IRQ_STACK_SIZE      (4096 << 2) /* 16k irq stack size */
-
-union irq_stack_union {
-	unsigned long stack[IRQ_STACK_SIZE/sizeof(unsigned long)];
-};
-
-DECLARE_PER_CPU(union irq_stack_union, irq_stack_union);
-
-void call_on_stack(unsigned long p1, void *func, unsigned long new_stack);
-
-#endif /* __KERNEL__ */
-
-/*
  * Data detected about CPUs at boot time which is the same for all CPU's.
  * HP boxes are SMP - ie identical processors.
  *
diff --git a/arch/parisc/kernel/entry.S b/arch/parisc/kernel/entry.S
index 4bb96ad..e8f07dd 100644
--- a/arch/parisc/kernel/entry.S
+++ b/arch/parisc/kernel/entry.S
@@ -65,15 +65,11 @@
 	rsm	PSW_SM_I, %r0	/* barrier for "Relied upon Translation */
 	mtsp	%r0, %sr4
 	mtsp	%r0, %sr5
-	mfsp	%sr7, %r1
-	or,=    %r0,%r1,%r0	/* Only save sr7 in sr3 if sr7 != 0 */
-	mtsp	%r1, %sr3
+	mtsp	%r0, %sr6
 	tovirt_r1 %r29
 	load32	KERNEL_PSW, %r1
 
 	rsm     PSW_SM_QUIET,%r0	/* second "heavy weight" ctl op */
-	mtsp	%r0, %sr6
-	mtsp	%r0, %sr7
 	mtctl	%r0, %cr17	/* Clear IIASQ tail */
 	mtctl	%r0, %cr17	/* Clear IIASQ head */
 	mtctl	%r1, %ipsw
@@ -119,17 +115,20 @@
 
 	/* we save the registers in the task struct */
 
+	copy	%r30, %r17
 	mfctl   %cr30, %r1
+	ldo	THREAD_SZ_ALGN(%r1), %r30
+	mtsp	%r0,%sr7
+	mtsp	%r16,%sr3
 	tophys  %r1,%r9
 	LDREG	TI_TASK(%r9), %r1	/* thread_info -> task_struct */
 	tophys  %r1,%r9
 	ldo     TASK_REGS(%r9),%r9
-	STREG   %r30, PT_GR30(%r9)
+	STREG   %r17,PT_GR30(%r9)
 	STREG   %r29,PT_GR29(%r9)
 	STREG   %r26,PT_GR26(%r9)
+	STREG	%r16,PT_SR7(%r9)
 	copy    %r9,%r29
-	mfctl   %cr30, %r1
-	ldo	THREAD_SZ_ALGN(%r1), %r30
 	.endm
 
 	.macro  get_stack_use_r30
@@ -137,10 +136,12 @@
 	/* we put a struct pt_regs on the stack and save the registers there */
 
 	tophys  %r30,%r9
-	STREG   %r30,PT_GR30(%r9)
+	copy	%r30,%r1
 	ldo	PT_SZ_ALGN(%r30),%r30
+	STREG   %r1,PT_GR30(%r9)
 	STREG   %r29,PT_GR29(%r9)
 	STREG   %r26,PT_GR26(%r9)
+	STREG	%r16,PT_SR7(%r9)
 	copy    %r9,%r29
 	.endm
 
@@ -452,9 +453,41 @@
 	L2_ptep		\pgd,\pte,\index,\va,\fault
 	.endm
 
+	/* Acquire pa_dbit_lock lock. */
+	.macro		dbit_lock	spc,tmp,tmp1
+#ifdef CONFIG_SMP
+	cmpib,COND(=),n	0,\spc,2f
+	load32		PA(pa_dbit_lock),\tmp
+1:	LDCW		0(\tmp),\tmp1
+	cmpib,COND(=)	0,\tmp1,1b
+	nop
+2:
+#endif
+	.endm
+
+	/* Release pa_dbit_lock lock without reloading lock address. */
+	.macro		dbit_unlock0	spc,tmp
+#ifdef CONFIG_SMP
+	or,COND(=)	%r0,\spc,%r0
+	stw             \spc,0(\tmp)
+#endif
+	.endm
+
+	/* Release pa_dbit_lock lock. */
+	.macro		dbit_unlock1	spc,tmp
+#ifdef CONFIG_SMP
+	load32		PA(pa_dbit_lock),\tmp
+	dbit_unlock0	\spc,\tmp
+#endif
+	.endm
+
 	/* Set the _PAGE_ACCESSED bit of the PTE.  Be clever and
 	 * don't needlessly dirty the cache line if it was already set */
-	.macro		update_ptep	ptep,pte,tmp,tmp1
+	.macro		update_ptep	spc,ptep,pte,tmp,tmp1
+#ifdef CONFIG_SMP
+	or,COND(=)	%r0,\spc,%r0
+	LDREG		0(\ptep),\pte
+#endif
 	ldi		_PAGE_ACCESSED,\tmp1
 	or		\tmp1,\pte,\tmp
 	and,COND(<>)	\tmp1,\pte,%r0
@@ -463,7 +496,11 @@
 
 	/* Set the dirty bit (and accessed bit).  No need to be
 	 * clever, this is only used from the dirty fault */
-	.macro		update_dirty	ptep,pte,tmp
+	.macro		update_dirty	spc,ptep,pte,tmp
+#ifdef CONFIG_SMP
+	or,COND(=)	%r0,\spc,%r0
+	LDREG		0(\ptep),\pte
+#endif
 	ldi		_PAGE_ACCESSED|_PAGE_DIRTY,\tmp
 	or		\tmp,\pte,\pte
 	STREG		\pte,0(\ptep)
@@ -1111,11 +1148,13 @@ dtlb_miss_20w:
 
 	L3_ptep		ptp,pte,t0,va,dtlb_check_alias_20w
 
-	update_ptep	ptp,pte,t0,t1
+	dbit_lock	spc,t0,t1
+	update_ptep	spc,ptp,pte,t0,t1
 
 	make_insert_tlb	spc,pte,prot
 	
 	idtlbt          pte,prot
+	dbit_unlock1	spc,t0
 
 	rfir
 	nop
@@ -1135,11 +1174,13 @@ nadtlb_miss_20w:
 
 	L3_ptep		ptp,pte,t0,va,nadtlb_check_alias_20w
 
-	update_ptep	ptp,pte,t0,t1
+	dbit_lock	spc,t0,t1
+	update_ptep	spc,ptp,pte,t0,t1
 
 	make_insert_tlb	spc,pte,prot
 
 	idtlbt          pte,prot
+	dbit_unlock1	spc,t0
 
 	rfir
 	nop
@@ -1161,7 +1202,8 @@ dtlb_miss_11:
 
 	L2_ptep		ptp,pte,t0,va,dtlb_check_alias_11
 
-	update_ptep	ptp,pte,t0,t1
+	dbit_lock	spc,t0,t1
+	update_ptep	spc,ptp,pte,t0,t1
 
 	make_insert_tlb_11	spc,pte,prot
 
@@ -1172,6 +1214,7 @@ dtlb_miss_11:
 	idtlbp		prot,(%sr1,va)
 
 	mtsp		t0, %sr1	/* Restore sr1 */
+	dbit_unlock1	spc,t0
 
 	rfir
 	nop
@@ -1192,7 +1235,8 @@ nadtlb_miss_11:
 
 	L2_ptep		ptp,pte,t0,va,nadtlb_check_alias_11
 
-	update_ptep	ptp,pte,t0,t1
+	dbit_lock	spc,t0,t1
+	update_ptep	spc,ptp,pte,t0,t1
 
 	make_insert_tlb_11	spc,pte,prot
 
@@ -1204,6 +1248,7 @@ nadtlb_miss_11:
 	idtlbp		prot,(%sr1,va)
 
 	mtsp		t0, %sr1	/* Restore sr1 */
+	dbit_unlock1	spc,t0
 
 	rfir
 	nop
@@ -1224,13 +1269,15 @@ dtlb_miss_20:
 
 	L2_ptep		ptp,pte,t0,va,dtlb_check_alias_20
 
-	update_ptep	ptp,pte,t0,t1
+	dbit_lock	spc,t0,t1
+	update_ptep	spc,ptp,pte,t0,t1
 
 	make_insert_tlb	spc,pte,prot
 
 	f_extend	pte,t0
 
 	idtlbt          pte,prot
+	dbit_unlock1	spc,t0
 
 	rfir
 	nop
@@ -1250,13 +1297,15 @@ nadtlb_miss_20:
 
 	L2_ptep		ptp,pte,t0,va,nadtlb_check_alias_20
 
-	update_ptep	ptp,pte,t0,t1
+	dbit_lock	spc,t0,t1
+	update_ptep	spc,ptp,pte,t0,t1
 
 	make_insert_tlb	spc,pte,prot
 
 	f_extend	pte,t0
 	
         idtlbt          pte,prot
+	dbit_unlock1	spc,t0
 
 	rfir
 	nop
@@ -1357,11 +1406,13 @@ itlb_miss_20w:
 
 	L3_ptep		ptp,pte,t0,va,itlb_fault
 
-	update_ptep	ptp,pte,t0,t1
+	dbit_lock	spc,t0,t1
+	update_ptep	spc,ptp,pte,t0,t1
 
 	make_insert_tlb	spc,pte,prot
 	
 	iitlbt          pte,prot
+	dbit_unlock1	spc,t0
 
 	rfir
 	nop
@@ -1379,11 +1430,13 @@ naitlb_miss_20w:
 
 	L3_ptep		ptp,pte,t0,va,naitlb_check_alias_20w
 
-	update_ptep	ptp,pte,t0,t1
+	dbit_lock	spc,t0,t1
+	update_ptep	spc,ptp,pte,t0,t1
 
 	make_insert_tlb	spc,pte,prot
 
 	iitlbt          pte,prot
+	dbit_unlock1	spc,t0
 
 	rfir
 	nop
@@ -1405,7 +1458,8 @@ itlb_miss_11:
 
 	L2_ptep		ptp,pte,t0,va,itlb_fault
 
-	update_ptep	ptp,pte,t0,t1
+	dbit_lock	spc,t0,t1
+	update_ptep	spc,ptp,pte,t0,t1
 
 	make_insert_tlb_11	spc,pte,prot
 
@@ -1416,6 +1470,7 @@ itlb_miss_11:
 	iitlbp		prot,(%sr1,va)
 
 	mtsp		t0, %sr1	/* Restore sr1 */
+	dbit_unlock1	spc,t0
 
 	rfir
 	nop
@@ -1427,7 +1482,8 @@ naitlb_miss_11:
 
 	L2_ptep		ptp,pte,t0,va,naitlb_check_alias_11
 
-	update_ptep	ptp,pte,t0,t1
+	dbit_lock	spc,t0,t1
+	update_ptep	spc,ptp,pte,t0,t1
 
 	make_insert_tlb_11	spc,pte,prot
 
@@ -1438,6 +1494,7 @@ naitlb_miss_11:
 	iitlbp		prot,(%sr1,va)
 
 	mtsp		t0, %sr1	/* Restore sr1 */
+	dbit_unlock1	spc,t0
 
 	rfir
 	nop
@@ -1459,13 +1516,15 @@ itlb_miss_20:
 
 	L2_ptep		ptp,pte,t0,va,itlb_fault
 
-	update_ptep	ptp,pte,t0,t1
+	dbit_lock	spc,t0,t1
+	update_ptep	spc,ptp,pte,t0,t1
 
 	make_insert_tlb	spc,pte,prot
 
 	f_extend	pte,t0	
 
 	iitlbt          pte,prot
+	dbit_unlock1	spc,t0
 
 	rfir
 	nop
@@ -1477,13 +1536,15 @@ naitlb_miss_20:
 
 	L2_ptep		ptp,pte,t0,va,naitlb_check_alias_20
 
-	update_ptep	ptp,pte,t0,t1
+	dbit_lock	spc,t0,t1
+	update_ptep	spc,ptp,pte,t0,t1
 
 	make_insert_tlb	spc,pte,prot
 
 	f_extend	pte,t0
 
 	iitlbt          pte,prot
+	dbit_unlock1	spc,t0
 
 	rfir
 	nop
@@ -1507,29 +1568,13 @@ dbit_trap_20w:
 
 	L3_ptep		ptp,pte,t0,va,dbit_fault
 
-#ifdef CONFIG_SMP
-	cmpib,COND(=),n        0,spc,dbit_nolock_20w
-	load32		PA(pa_dbit_lock),t0
-
-dbit_spin_20w:
-	LDCW		0(t0),t1
-	cmpib,COND(=)         0,t1,dbit_spin_20w
-	nop
-
-dbit_nolock_20w:
-#endif
-	update_dirty	ptp,pte,t1
+	dbit_lock	spc,t0,t1
+	update_dirty	spc,ptp,pte,t1
 
 	make_insert_tlb	spc,pte,prot
 		
 	idtlbt          pte,prot
-#ifdef CONFIG_SMP
-	cmpib,COND(=),n        0,spc,dbit_nounlock_20w
-	ldi             1,t1
-	stw             t1,0(t0)
-
-dbit_nounlock_20w:
-#endif
+	dbit_unlock0	spc,t0
 
 	rfir
 	nop
@@ -1543,18 +1588,8 @@ dbit_trap_11:
 
 	L2_ptep		ptp,pte,t0,va,dbit_fault
 
-#ifdef CONFIG_SMP
-	cmpib,COND(=),n        0,spc,dbit_nolock_11
-	load32		PA(pa_dbit_lock),t0
-
-dbit_spin_11:
-	LDCW		0(t0),t1
-	cmpib,=         0,t1,dbit_spin_11
-	nop
-
-dbit_nolock_11:
-#endif
-	update_dirty	ptp,pte,t1
+	dbit_lock	spc,t0,t1
+	update_dirty	spc,ptp,pte,t1
 
 	make_insert_tlb_11	spc,pte,prot
 
@@ -1565,13 +1600,7 @@ dbit_nolock_11:
 	idtlbp		prot,(%sr1,va)
 
 	mtsp            t1, %sr1     /* Restore sr1 */
-#ifdef CONFIG_SMP
-	cmpib,COND(=),n        0,spc,dbit_nounlock_11
-	ldi             1,t1
-	stw             t1,0(t0)
-
-dbit_nounlock_11:
-#endif
+	dbit_unlock0	spc,t0
 
 	rfir
 	nop
@@ -1583,32 +1612,15 @@ dbit_trap_20:
 
 	L2_ptep		ptp,pte,t0,va,dbit_fault
 
-#ifdef CONFIG_SMP
-	cmpib,COND(=),n        0,spc,dbit_nolock_20
-	load32		PA(pa_dbit_lock),t0
-
-dbit_spin_20:
-	LDCW		0(t0),t1
-	cmpib,=         0,t1,dbit_spin_20
-	nop
-
-dbit_nolock_20:
-#endif
-	update_dirty	ptp,pte,t1
+	dbit_lock	spc,t0,t1
+	update_dirty	spc,ptp,pte,t1
 
 	make_insert_tlb	spc,pte,prot
 
 	f_extend	pte,t1
 	
         idtlbt          pte,prot
-
-#ifdef CONFIG_SMP
-	cmpib,COND(=),n        0,spc,dbit_nounlock_20
-	ldi             1,t1
-	stw             t1,0(t0)
-
-dbit_nounlock_20:
-#endif
+	dbit_unlock0	spc,t0
 
 	rfir
 	nop
diff --git a/arch/parisc/kernel/hardware.c b/arch/parisc/kernel/hardware.c
index f7752f6..9e2d2e4 100644
--- a/arch/parisc/kernel/hardware.c
+++ b/arch/parisc/kernel/hardware.c
@@ -222,6 +222,7 @@ static struct hp_hardware hp_hardware_list[] = {
 	{HPHW_NPROC,0x5DD,0x4,0x81,"Duet W2"},
 	{HPHW_NPROC,0x5DE,0x4,0x81,"Piccolo W+"},
 	{HPHW_NPROC,0x5DF,0x4,0x81,"Cantata W2"},
+	{HPHW_NPROC,0x5DF,0x0,0x00,"Marcato W+? (rp5470)"},
 	{HPHW_NPROC,0x5E0,0x4,0x91,"Cantata DC- W2"},
 	{HPHW_NPROC,0x5E1,0x4,0x91,"Crescendo DC- W2"},
 	{HPHW_NPROC,0x5E2,0x4,0x91,"Crescendo 650 W2"},
diff --git a/arch/parisc/kernel/irq.c b/arch/parisc/kernel/irq.c
index e255db0..2e6443b 100644
--- a/arch/parisc/kernel/irq.c
+++ b/arch/parisc/kernel/irq.c
@@ -27,11 +27,11 @@
 #include <linux/interrupt.h>
 #include <linux/kernel_stat.h>
 #include <linux/seq_file.h>
-#include <linux/spinlock.h>
 #include <linux/types.h>
 #include <asm/io.h>
 
 #include <asm/smp.h>
+#include <asm/ldcw.h>
 
 #undef PARISC_IRQ_CR16_COUNTS
 
@@ -166,22 +166,36 @@ int arch_show_interrupts(struct seq_file *p, int prec)
 	seq_printf(p, "%*s: ", prec, "STK");
 	for_each_online_cpu(j)
 		seq_printf(p, "%10u ", irq_stats(j)->kernel_stack_usage);
-	seq_printf(p, "  Kernel stack usage\n");
+	seq_puts(p, "  Kernel stack usage\n");
+# ifdef CONFIG_IRQSTACKS
+	seq_printf(p, "%*s: ", prec, "IST");
+	for_each_online_cpu(j)
+		seq_printf(p, "%10u ", irq_stats(j)->irq_stack_usage);
+	seq_puts(p, "  Interrupt stack usage\n");
+# endif
 #endif
 #ifdef CONFIG_SMP
 	seq_printf(p, "%*s: ", prec, "RES");
 	for_each_online_cpu(j)
 		seq_printf(p, "%10u ", irq_stats(j)->irq_resched_count);
-	seq_printf(p, "  Rescheduling interrupts\n");
+	seq_puts(p, "  Rescheduling interrupts\n");
 	seq_printf(p, "%*s: ", prec, "CAL");
 	for_each_online_cpu(j)
 		seq_printf(p, "%10u ", irq_stats(j)->irq_call_count);
-	seq_printf(p, "  Function call interrupts\n");
+	seq_puts(p, "  Function call interrupts\n");
 #endif
+	seq_printf(p, "%*s: ", prec, "UAH");
+	for_each_online_cpu(j)
+		seq_printf(p, "%10u ", irq_stats(j)->irq_unaligned_count);
+	seq_puts(p, "  Unaligned access handler traps\n");
+	seq_printf(p, "%*s: ", prec, "FPA");
+	for_each_online_cpu(j)
+		seq_printf(p, "%10u ", irq_stats(j)->irq_fpassist_count);
+	seq_puts(p, "  Floating point assist traps\n");
 	seq_printf(p, "%*s: ", prec, "TLB");
 	for_each_online_cpu(j)
 		seq_printf(p, "%10u ", irq_stats(j)->irq_tlb_count);
-	seq_printf(p, "  TLB shootdowns\n");
+	seq_puts(p, "  TLB shootdowns\n");
 	return 0;
 }
 
@@ -366,6 +380,24 @@ static inline int eirr_to_irq(unsigned long eirr)
 	return (BITS_PER_LONG - bit) + TIMER_IRQ;
 }
 
+#ifdef CONFIG_IRQSTACKS
+/*
+ * IRQ STACK - used for irq handler
+ */
+#define IRQ_STACK_SIZE      (4096 << 2) /* 16k irq stack size */
+
+union irq_stack_union {
+	unsigned long stack[IRQ_STACK_SIZE/sizeof(unsigned long)];
+	volatile unsigned int slock[4];
+	volatile unsigned int lock[1];
+};
+
+DEFINE_PER_CPU(union irq_stack_union, irq_stack_union) = {
+		.slock = { 1,1,1,1 },
+	};
+#endif
+
+
 int sysctl_panic_on_stackoverflow = 1;
 
 static inline void stack_overflow_check(struct pt_regs *regs)
@@ -378,6 +410,7 @@ static inline void stack_overflow_check(struct pt_regs *regs)
 	unsigned long sp = regs->gr[30];
 	unsigned long stack_usage;
 	unsigned int *last_usage;
+	int cpu = smp_processor_id();
 
 	/* if sr7 != 0, we interrupted a userspace process which we do not want
 	 * to check for stack overflow. We will only check the kernel stack. */
@@ -386,7 +419,31 @@ static inline void stack_overflow_check(struct pt_regs *regs)
 
 	/* calculate kernel stack usage */
 	stack_usage = sp - stack_start;
-	last_usage = &per_cpu(irq_stat.kernel_stack_usage, smp_processor_id());
+#ifdef CONFIG_IRQSTACKS
+	if (likely(stack_usage <= THREAD_SIZE))
+		goto check_kernel_stack; /* found kernel stack */
+
+	/* check irq stack usage */
+	stack_start = (unsigned long) &per_cpu(irq_stack_union, cpu).stack;
+	stack_usage = sp - stack_start;
+
+	last_usage = &per_cpu(irq_stat.irq_stack_usage, cpu);
+	if (unlikely(stack_usage > *last_usage))
+		*last_usage = stack_usage;
+
+	if (likely(stack_usage < (IRQ_STACK_SIZE - STACK_MARGIN)))
+		return;
+
+	pr_emerg("stackcheck: %s will most likely overflow irq stack "
+		 "(sp:%lx, stk bottom-top:%lx-%lx)\n",
+		current->comm, sp, stack_start, stack_start + IRQ_STACK_SIZE);
+	goto panic_check;
+
+check_kernel_stack:
+#endif
+
+	/* check kernel stack usage */
+	last_usage = &per_cpu(irq_stat.kernel_stack_usage, cpu);
 
 	if (unlikely(stack_usage > *last_usage))
 		*last_usage = stack_usage;
@@ -398,31 +455,66 @@ static inline void stack_overflow_check(struct pt_regs *regs)
 		 "(sp:%lx, stk bottom-top:%lx-%lx)\n",
 		current->comm, sp, stack_start, stack_start + THREAD_SIZE);
 
+#ifdef CONFIG_IRQSTACKS
+panic_check:
+#endif
 	if (sysctl_panic_on_stackoverflow)
 		panic("low stack detected by irq handler - check messages\n");
 #endif
 }
 
 #ifdef CONFIG_IRQSTACKS
-DEFINE_PER_CPU(union irq_stack_union, irq_stack_union);
+/* in entry.S: */
+void call_on_stack(unsigned long p1, void *func, unsigned long new_stack);
 
 static void execute_on_irq_stack(void *func, unsigned long param1)
 {
-	unsigned long *irq_stack_start;
+	union irq_stack_union *union_ptr;
 	unsigned long irq_stack;
-	int cpu = smp_processor_id();
+	volatile unsigned int *irq_stack_in_use;
+
+	union_ptr = &per_cpu(irq_stack_union, smp_processor_id());
+	irq_stack = (unsigned long) &union_ptr->stack;
+	irq_stack = ALIGN(irq_stack + sizeof(irq_stack_union.slock),
+			 64); /* align for stack frame usage */
 
-	irq_stack_start = &per_cpu(irq_stack_union, cpu).stack[0];
-	irq_stack = (unsigned long) irq_stack_start;
-	irq_stack = ALIGN(irq_stack, 16); /* align for stack frame usage */
+	/* We may be called recursive. If we are already using the irq stack,
+	 * just continue to use it. Use spinlocks to serialize
+	 * the irq stack usage.
+	 */
+	irq_stack_in_use = (volatile unsigned int *)__ldcw_align(union_ptr);
+	if (!__ldcw(irq_stack_in_use)) {
+		void (*direct_call)(unsigned long p1) = func;
 
-	BUG_ON(*irq_stack_start); /* report bug if we were called recursive. */
-	*irq_stack_start = 1;
+		/* We are using the IRQ stack already.
+		 * Do direct call on current stack. */
+		direct_call(param1);
+		return;
+	}
 
 	/* This is where we switch to the IRQ stack. */
 	call_on_stack(param1, func, irq_stack);
 
-	*irq_stack_start = 0;
+	/* free up irq stack usage. */
+	*irq_stack_in_use = 1;
+}
+
+asmlinkage void do_softirq(void)
+{
+	__u32 pending;
+	unsigned long flags;
+
+	if (in_interrupt())
+		return;
+
+	local_irq_save(flags);
+
+	pending = local_softirq_pending();
+
+	if (pending)
+		execute_on_irq_stack(__do_softirq, 0);
+
+	local_irq_restore(flags);
 }
 #endif /* CONFIG_IRQSTACKS */
 
diff --git a/arch/parisc/kernel/pacache.S b/arch/parisc/kernel/pacache.S
index 5e1de60..36d7f40 100644
--- a/arch/parisc/kernel/pacache.S
+++ b/arch/parisc/kernel/pacache.S
@@ -605,14 +605,14 @@ ENTRY(copy_user_page_asm)
 	convert_phys_for_tlb_insert20 %r26	/* convert phys addr to tlb insert format */
 	convert_phys_for_tlb_insert20 %r23	/* convert phys addr to tlb insert format */
 	depd		%r24,63,22, %r28	/* Form aliased virtual address 'to' */
-	depdi		0, 63,PAGE_SHIFT, %r28  /* Clear any offset bits */
+	depdi		0, 63,PAGE_SHIFT, %r28	/* Clear any offset bits */
 	copy		%r28, %r29
 	depdi		1, 41,1, %r29		/* Form aliased virtual address 'from' */
 #else
 	extrw,u		%r26, 24,25, %r26	/* convert phys addr to tlb insert format */
 	extrw,u		%r23, 24,25, %r23	/* convert phys addr to tlb insert format */
 	depw		%r24, 31,22, %r28	/* Form aliased virtual address 'to' */
-	depwi		0, 31,12, %r28		/* Clear any offset bits */
+	depwi		0, 31,PAGE_SHIFT, %r28	/* Clear any offset bits */
 	copy		%r28, %r29
 	depwi		1, 9,1, %r29		/* Form aliased virtual address 'from' */
 #endif
@@ -762,7 +762,7 @@ ENTRY(clear_user_page_asm)
 #else
 	extrw,u		%r26, 24,25, %r26	/* convert phys addr to tlb insert format */
 	depw		%r25, 31,22, %r28	/* Form aliased virtual address 'to' */
-	depwi		0, 31,12, %r28		/* Clear any offset bits */
+	depwi		0, 31,PAGE_SHIFT, %r28	/* Clear any offset bits */
 #endif
 
 	/* Purge any old translation */
@@ -846,7 +846,7 @@ ENTRY(flush_dcache_page_asm)
 #else
 	extrw,u		%r26, 24,25, %r26	/* convert phys addr to tlb insert format */
 	depw		%r25, 31,22, %r28	/* Form aliased virtual address 'to' */
-	depwi		0, 31,12, %r28		/* Clear any offset bits */
+	depwi		0, 31,PAGE_SHIFT, %r28	/* Clear any offset bits */
 #endif
 
 	/* Purge any old translation */
@@ -918,11 +918,11 @@ ENTRY(flush_icache_page_asm)
 #endif
 	convert_phys_for_tlb_insert20 %r26	/* convert phys addr to tlb insert format */
 	depd		%r25, 63,22, %r28	/* Form aliased virtual address 'to' */
-	depdi		0, 63,PAGE_SHIFT, %r28  /* Clear any offset bits */
+	depdi		0, 63,PAGE_SHIFT, %r28	/* Clear any offset bits */
 #else
 	extrw,u		%r26, 24,25, %r26	/* convert phys addr to tlb insert format */
 	depw		%r25, 31,22, %r28	/* Form aliased virtual address 'to' */
-	depwi		0, 31,12, %r28		/* Clear any offset bits */
+	depwi		0, 31,PAGE_SHIFT, %r28	/* Clear any offset bits */
 #endif
 
 	/* Purge any old translation */
diff --git a/arch/parisc/kernel/traps.c b/arch/parisc/kernel/traps.c
index fe41a98..04e47c6 100644
--- a/arch/parisc/kernel/traps.c
+++ b/arch/parisc/kernel/traps.c
@@ -646,6 +646,7 @@ void notrace handle_interruption(int code, struct pt_regs *regs)
 	case 14:
 		/* Assist Exception Trap, i.e. floating point exception. */
 		die_if_kernel("Floating point exception", regs, 0); /* quiet */
+		__inc_irq_stat(irq_fpassist_count);
 		handle_fpe(regs);
 		return;
 
diff --git a/arch/parisc/kernel/unaligned.c b/arch/parisc/kernel/unaligned.c
index 234e368..d7c0acb 100644
--- a/arch/parisc/kernel/unaligned.c
+++ b/arch/parisc/kernel/unaligned.c
@@ -27,6 +27,7 @@
 #include <linux/signal.h>
 #include <linux/ratelimit.h>
 #include <asm/uaccess.h>
+#include <asm/hardirq.h>
 
 /* #define DEBUG_UNALIGNED 1 */
 
@@ -454,6 +455,8 @@ void handle_unaligned(struct pt_regs *regs)
 	struct siginfo si;
 	register int flop=0;	/* true if this is a flop */
 
+	__inc_irq_stat(irq_unaligned_count);
+
 	/* log a message with pacing */
 	if (user_mode(regs)) {
 		if (current->thread.flags & PARISC_UAC_SIGBUS) {
diff --git a/arch/parisc/mm/init.c b/arch/parisc/mm/init.c
index ce939ac..1c96564 100644
--- a/arch/parisc/mm/init.c
+++ b/arch/parisc/mm/init.c
@@ -1069,7 +1069,7 @@ void flush_tlb_all(void)
 {
 	int do_recycle;
 
-	inc_irq_stat(irq_tlb_count);
+	__inc_irq_stat(irq_tlb_count);
 	do_recycle = 0;
 	spin_lock(&sid_lock);
 	if (dirty_space_ids > RECYCLE_THRESHOLD) {
@@ -1090,7 +1090,7 @@ void flush_tlb_all(void)
 #else
 void flush_tlb_all(void)
 {
-	inc_irq_stat(irq_tlb_count);
+	__inc_irq_stat(irq_tlb_count);
 	spin_lock(&sid_lock);
 	flush_tlb_all_local(NULL);
 	recycle_sids();