Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc

Pull sparc fixes from David Miller:
 "Several build/bug fixes for sparc, including:

  1) Configuring a mix of static vs.  modular sparc64 crypto modules
     didn't work, remove an ill-conceived attempt to only have to build
     the device match table for these drivers once to fix the problem.

     Reported by Meelis Roos.

  2) Make the montgomery multiple/square and mpmul instructions actually
     usable in 32-bit tasks.  Essentially this involves providing 32-bit
     userspace with a way to use a 64-bit stack when it needs to.

  3) Our sparc64 atomic backoffs don't yield cpu strands properly on
     Niagara chips.  Use pause instruction when available to achieve
     this, otherwise use a benign instruction we know blocks the strand
     for some time.

  4) Wire up kcmp

  5) Fix the build of various drivers by removing the unnecessary
     blocking of OF_GPIO when SPARC.

  6) Fix unintended regression wherein of_address_to_resource stopped
     being provided.  Fix from Andreas Larsson.

  7) Fix NULL dereference in leon_handle_ext_irq(), also from Andreas
     Larsson."

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc:
  sparc64: Fix build with mix of modular vs. non-modular crypto drivers.
  sparc: Support atomic64_dec_if_positive properly.
  of/address: sparc: Declare of_address_to_resource() as an extern function for sparc again
  sparc32, leon: Check for existent irq_map entry in leon_handle_ext_irq
  sparc: Add sparc support for platform_get_irq()
  sparc: Allow OF_GPIO on sparc.
  qlogicpti: Fix build warning.
  sparc: Wire up sys_kcmp.
  sparc64: Improvde documentation and readability of atomic backoff code.
  sparc64: Use pause instruction when available.
  sparc64: Fix cpu strand yielding.
  sparc64: Make montmul/montsqr/mpmul usable in 32-bit threads.

arch/sparc/Kconfig

@@ -20,6 +20,7 @@ config SPARC
 	select HAVE_ARCH_TRACEHOOK
 	select SYSCTL_EXCEPTION_TRACE
 	select ARCH_WANT_OPTIONAL_GPIOLIB
+	select ARCH_HAS_ATOMIC64_DEC_IF_POSITIVE
 	select RTC_CLASS
 	select RTC_DRV_M48T59
 	select HAVE_IRQ_WORK

arch/sparc/crypto/Makefile

@@ -13,13 +13,13 @@ obj-$(CONFIG_CRYPTO_DES_SPARC64) += camellia-sparc64.o
 
 obj-$(CONFIG_CRYPTO_CRC32C_SPARC64) += crc32c-sparc64.o
 
-sha1-sparc64-y := sha1_asm.o sha1_glue.o crop_devid.o
-sha256-sparc64-y := sha256_asm.o sha256_glue.o crop_devid.o
-sha512-sparc64-y := sha512_asm.o sha512_glue.o crop_devid.o
-md5-sparc64-y := md5_asm.o md5_glue.o crop_devid.o
+sha1-sparc64-y := sha1_asm.o sha1_glue.o
+sha256-sparc64-y := sha256_asm.o sha256_glue.o
+sha512-sparc64-y := sha512_asm.o sha512_glue.o
+md5-sparc64-y := md5_asm.o md5_glue.o
 
-aes-sparc64-y := aes_asm.o aes_glue.o crop_devid.o
-des-sparc64-y := des_asm.o des_glue.o crop_devid.o
-camellia-sparc64-y := camellia_asm.o camellia_glue.o crop_devid.o
+aes-sparc64-y := aes_asm.o aes_glue.o
+des-sparc64-y := des_asm.o des_glue.o
+camellia-sparc64-y := camellia_asm.o camellia_glue.o
 
-crc32c-sparc64-y := crc32c_asm.o crc32c_glue.o crop_devid.o
+crc32c-sparc64-y := crc32c_asm.o crc32c_glue.o

arch/sparc/crypto/aes_glue.c

@@ -475,3 +475,5 @@ MODULE_LICENSE("GPL");
 MODULE_DESCRIPTION("AES Secure Hash Algorithm, sparc64 aes opcode accelerated");
 
 MODULE_ALIAS("aes");
+
+#include "crop_devid.c"

arch/sparc/crypto/camellia_glue.c

@@ -320,3 +320,5 @@ MODULE_LICENSE("GPL");
 MODULE_DESCRIPTION("Camellia Cipher Algorithm, sparc64 camellia opcode accelerated");
 
 MODULE_ALIAS("aes");
+
+#include "crop_devid.c"

arch/sparc/crypto/crc32c_glue.c

@@ -177,3 +177,5 @@ MODULE_LICENSE("GPL");
 MODULE_DESCRIPTION("CRC32c (Castagnoli), sparc64 crc32c opcode accelerated");
 
 MODULE_ALIAS("crc32c");
+
+#include "crop_devid.c"

arch/sparc/crypto/des_glue.c

@@ -527,3 +527,5 @@ MODULE_LICENSE("GPL");
 MODULE_DESCRIPTION("DES & Triple DES EDE Cipher Algorithms, sparc64 des opcode accelerated");
 
 MODULE_ALIAS("des");
+
+#include "crop_devid.c"

arch/sparc/crypto/md5_glue.c

@@ -186,3 +186,5 @@ MODULE_LICENSE("GPL");
 MODULE_DESCRIPTION("MD5 Secure Hash Algorithm, sparc64 md5 opcode accelerated");
 
 MODULE_ALIAS("md5");
+
+#include "crop_devid.c"

arch/sparc/crypto/sha1_glue.c

@@ -181,3 +181,5 @@ MODULE_LICENSE("GPL");
 MODULE_DESCRIPTION("SHA1 Secure Hash Algorithm, sparc64 sha1 opcode accelerated");
 
 MODULE_ALIAS("sha1");
+
+#include "crop_devid.c"

arch/sparc/crypto/sha256_glue.c

@@ -239,3 +239,5 @@ MODULE_DESCRIPTION("SHA-224 and SHA-256 Secure Hash Algorithm, sparc64 sha256 op
 
 MODULE_ALIAS("sha224");
 MODULE_ALIAS("sha256");
+
+#include "crop_devid.c"

arch/sparc/crypto/sha512_glue.c

@@ -224,3 +224,5 @@ MODULE_DESCRIPTION("SHA-384 and SHA-512 Secure Hash Algorithm, sparc64 sha512 op
 
 MODULE_ALIAS("sha384");
 MODULE_ALIAS("sha512");
+
+#include "crop_devid.c"

arch/sparc/include/asm/atomic_64.h

@@ -1,7 +1,7 @@
 /* atomic.h: Thankfully the V9 is at least reasonable for this
  *           stuff.
  *
- * Copyright (C) 1996, 1997, 2000 David S. Miller (davem@redhat.com)
+ * Copyright (C) 1996, 1997, 2000, 2012 David S. Miller (davem@redhat.com)
  */
 
 #ifndef __ARCH_SPARC64_ATOMIC__
@@ -106,6 +106,8 @@ static inline long atomic64_add_unless(atomic64_t *v, long a, long u)
 
 #define atomic64_inc_not_zero(v) atomic64_add_unless((v), 1, 0)
 
+extern long atomic64_dec_if_positive(atomic64_t *v);
+
 /* Atomic operations are already serializing */
 #define smp_mb__before_atomic_dec()	barrier()
 #define smp_mb__after_atomic_dec()	barrier()

arch/sparc/include/asm/backoff.h

@@ -1,6 +1,46 @@
 #ifndef _SPARC64_BACKOFF_H
 #define _SPARC64_BACKOFF_H
 
+/* The macros in this file implement an exponential backoff facility
+ * for atomic operations.
+ *
+ * When multiple threads compete on an atomic operation, it is
+ * possible for one thread to be continually denied a successful
+ * completion of the compare-and-swap instruction.  Heavily
+ * threaded cpu implementations like Niagara can compound this
+ * problem even further.
+ *
+ * When an atomic operation fails and needs to be retried, we spin a
+ * certain number of times.  At each subsequent failure of the same
+ * operation we double the spin count, realizing an exponential
+ * backoff.
+ *
+ * When we spin, we try to use an operation that will cause the
+ * current cpu strand to block, and therefore make the core fully
+ * available to any other other runnable strands.  There are two
+ * options, based upon cpu capabilities.
+ *
+ * On all cpus prior to SPARC-T4 we do three dummy reads of the
+ * condition code register.  Each read blocks the strand for something
+ * between 40 and 50 cpu cycles.
+ *
+ * For SPARC-T4 and later we have a special "pause" instruction
+ * available.  This is implemented using writes to register %asr27.
+ * The cpu will block the number of cycles written into the register,
+ * unless a disrupting trap happens first.  SPARC-T4 specifically
+ * implements pause with a granularity of 8 cycles.  Each strand has
+ * an internal pause counter which decrements every 8 cycles.  So the
+ * chip shifts the %asr27 value down by 3 bits, and writes the result
+ * into the pause counter.  If a value smaller than 8 is written, the
+ * chip blocks for 1 cycle.
+ *
+ * To achieve the same amount of backoff as the three %ccr reads give
+ * on earlier chips, we shift the backoff value up by 7 bits.  (Three
+ * %ccr reads block for about 128 cycles, 1 << 7 == 128) We write the
+ * whole amount we want to block into the pause register, rather than
+ * loop writing 128 each time.
+ */
+
 #define BACKOFF_LIMIT	(4 * 1024)
 
 #ifdef CONFIG_SMP
@@ -11,16 +51,25 @@
 #define BACKOFF_LABEL(spin_label, continue_label) \
 	spin_label
 
-#define BACKOFF_SPIN(reg, tmp, label)	\
-	mov	reg, tmp; \
-88:	brnz,pt	tmp, 88b; \
-	 sub	tmp, 1, tmp; \
-	set	BACKOFF_LIMIT, tmp; \
-	cmp	reg, tmp; \
-	bg,pn	%xcc, label; \
-	 nop; \
-	ba,pt	%xcc, label; \
-	 sllx	reg, 1, reg;
+#define BACKOFF_SPIN(reg, tmp, label)		\
+	mov		reg, tmp;		\
+88:	rd		%ccr, %g0;		\
+	rd		%ccr, %g0;		\
+	rd		%ccr, %g0;		\
+	.section	.pause_3insn_patch,"ax";\
+	.word		88b;			\
+	sllx		tmp, 7, tmp;		\
+	wr		tmp, 0, %asr27;		\
+	clr		tmp;			\
+	.previous;				\
+	brnz,pt		tmp, 88b;		\
+	 sub		tmp, 1, tmp;		\
+	set		BACKOFF_LIMIT, tmp;	\
+	cmp		reg, tmp;		\
+	bg,pn		%xcc, label;		\
+	 nop;					\
+	ba,pt		%xcc, label;		\
+	 sllx		reg, 1, reg;
 
 #else
 

arch/sparc/include/asm/compat.h

@@ -232,9 +232,10 @@ static inline void __user *arch_compat_alloc_user_space(long len)
 	struct pt_regs *regs = current_thread_info()->kregs;
 	unsigned long usp = regs->u_regs[UREG_I6];
 
-	if (!(test_thread_flag(TIF_32BIT)))
+	if (test_thread_64bit_stack(usp))
 		usp += STACK_BIAS;
-	else
+
+	if (test_thread_flag(TIF_32BIT))
 		usp &= 0xffffffffUL;
 
 	usp -= len;

arch/sparc/include/asm/processor_64.h

@@ -196,7 +196,22 @@ extern unsigned long get_wchan(struct task_struct *task);
 #define KSTK_EIP(tsk)  (task_pt_regs(tsk)->tpc)
 #define KSTK_ESP(tsk)  (task_pt_regs(tsk)->u_regs[UREG_FP])
 
-#define cpu_relax()	barrier()
+/* Please see the commentary in asm/backoff.h for a description of
+ * what these instructions are doing and how they have been choosen.
+ * To make a long story short, we are trying to yield the current cpu
+ * strand during busy loops.
+ */
+#define cpu_relax()	asm volatile("\n99:\n\t"			\
+				     "rd	%%ccr, %%g0\n\t"	\
+				     "rd	%%ccr, %%g0\n\t"	\
+				     "rd	%%ccr, %%g0\n\t"	\
+				     ".section	.pause_3insn_patch,\"ax\"\n\t"\
+				     ".word	99b\n\t"		\
+				     "wr	%%g0, 128, %%asr27\n\t"	\
+				     "nop\n\t"				\
+				     "nop\n\t"				\
+				     ".previous"			\
+				     ::: "memory")
 
 /* Prefetch support.  This is tuned for UltraSPARC-III and later.
  * UltraSPARC-I will treat these as nops, and UltraSPARC-II has

arch/sparc/include/asm/prom.h

@@ -63,5 +63,10 @@ extern char *of_console_options;
 extern void irq_trans_init(struct device_node *dp);
 extern char *build_path_component(struct device_node *dp);
 
+/* SPARC has a local implementation */
+extern int of_address_to_resource(struct device_node *dev, int index,
+				  struct resource *r);
+#define of_address_to_resource of_address_to_resource
+
 #endif /* __KERNEL__ */
 #endif /* _SPARC_PROM_H */

arch/sparc/include/asm/thread_info_64.h

@@ -259,6 +259,11 @@ static inline bool test_and_clear_restore_sigmask(void)
 
 #define tsk_is_polling(t) test_tsk_thread_flag(t, TIF_POLLING_NRFLAG)
 
+#define thread32_stack_is_64bit(__SP) (((__SP) & 0x1) != 0)
+#define test_thread_64bit_stack(__SP) \
+	((test_thread_flag(TIF_32BIT) && !thread32_stack_is_64bit(__SP)) ? \
+	 false : true)
+
 #endif	/* !__ASSEMBLY__ */
 
 #endif /* __KERNEL__ */

arch/sparc/include/asm/ttable.h

@@ -372,7 +372,9 @@ etrap_spill_fixup_64bit:				\
 
 /* Normal 32bit spill */
 #define SPILL_2_GENERIC(ASI)				\
-	srl	%sp, 0, %sp;				\
+	and	%sp, 1, %g3;				\
+	brnz,pn	%g3, (. - (128 + 4));			\
+	 srl	%sp, 0, %sp;				\
 	stwa	%l0, [%sp + %g0] ASI;			\
 	mov	0x04, %g3;				\
 	stwa	%l1, [%sp + %g3] ASI;			\
@@ -398,14 +400,16 @@ etrap_spill_fixup_64bit:				\
 	stwa	%i6, [%g1 + %g0] ASI;			\
 	stwa	%i7, [%g1 + %g3] ASI;			\
 	saved;						\
-        retry; nop; nop;				\
+        retry;						\
 	b,a,pt	%xcc, spill_fixup_dax;			\
 	b,a,pt	%xcc, spill_fixup_mna;			\
 	b,a,pt	%xcc, spill_fixup;
 
 #define SPILL_2_GENERIC_ETRAP		\
 etrap_user_spill_32bit:			\
-	srl	%sp, 0, %sp;		\
+	and	%sp, 1, %g3;		\
+	brnz,pn	%g3, etrap_user_spill_64bit;	\
+	 srl	%sp, 0, %sp;		\
 	stwa	%l0, [%sp + 0x00] %asi;	\
 	stwa	%l1, [%sp + 0x04] %asi;	\
 	stwa	%l2, [%sp + 0x08] %asi;	\
@@ -427,7 +431,7 @@ etrap_user_spill_32bit:			\
 	ba,pt	%xcc, etrap_save;	\
 	 wrpr	%g1, %cwp;		\
 	nop; nop; nop; nop;		\
-	nop; nop; nop; nop;		\
+	nop; nop;			\
 	ba,a,pt	%xcc, etrap_spill_fixup_32bit; \
 	ba,a,pt	%xcc, etrap_spill_fixup_32bit; \
 	ba,a,pt	%xcc, etrap_spill_fixup_32bit;
@@ -592,7 +596,9 @@ user_rtt_fill_64bit:					\
 
 /* Normal 32bit fill */
 #define FILL_2_GENERIC(ASI)				\
-	srl	%sp, 0, %sp;				\
+	and	%sp, 1, %g3;				\
+	brnz,pn	%g3, (. - (128 + 4));			\
+	 srl	%sp, 0, %sp;				\
 	lduwa	[%sp + %g0] ASI, %l0;			\
 	mov	0x04, %g2;				\
 	mov	0x08, %g3;				\
@@ -616,14 +622,16 @@ user_rtt_fill_64bit:					\
 	lduwa	[%g1 + %g3] ASI, %i6;			\
 	lduwa	[%g1 + %g5] ASI, %i7;			\
 	restored;					\
-	retry; nop; nop; nop; nop;			\
+	retry; nop; nop;				\
 	b,a,pt	%xcc, fill_fixup_dax;			\
 	b,a,pt	%xcc, fill_fixup_mna;			\
 	b,a,pt	%xcc, fill_fixup;
 
 #define FILL_2_GENERIC_RTRAP				\
 user_rtt_fill_32bit:					\
-	srl	%sp, 0, %sp;				\
+	and	%sp, 1, %g3;				\
+	brnz,pn	%g3, user_rtt_fill_64bit;		\
+	 srl	%sp, 0, %sp;				\
 	lduwa	[%sp + 0x00] %asi, %l0;			\
 	lduwa	[%sp + 0x04] %asi, %l1;			\
 	lduwa	[%sp + 0x08] %asi, %l2;			\
@@ -643,7 +651,7 @@ user_rtt_fill_32bit:					\
 	ba,pt	%xcc, user_rtt_pre_restore;		\
 	 restored;					\
 	nop; nop; nop; nop; nop;			\
-	nop; nop; nop; nop; nop;			\
+	nop; nop; nop;					\
 	ba,a,pt	%xcc, user_rtt_fill_fixup;		\
 	ba,a,pt	%xcc, user_rtt_fill_fixup;		\
 	ba,a,pt	%xcc, user_rtt_fill_fixup;

arch/sparc/include/uapi/asm/unistd.h

@@ -405,8 +405,13 @@
 #define __NR_setns		337
 #define __NR_process_vm_readv	338
 #define __NR_process_vm_writev	339
+#define __NR_kern_features	340
+#define __NR_kcmp		341
 
-#define NR_syscalls		340
+#define NR_syscalls		342
+
+/* Bitmask values returned from kern_features system call.  */
+#define KERN_FEATURE_MIXED_MODE_STACK	0x00000001
 
 #ifdef __32bit_syscall_numbers__
 /* Sparc 32-bit only has the "setresuid32", "getresuid32" variants,

arch/sparc/kernel/entry.h

@@ -59,6 +59,13 @@ struct popc_6insn_patch_entry {
 extern struct popc_6insn_patch_entry __popc_6insn_patch,
 	__popc_6insn_patch_end;
 
+struct pause_patch_entry {
+	unsigned int	addr;
+	unsigned int	insns[3];
+};
+extern struct pause_patch_entry __pause_3insn_patch,
+	__pause_3insn_patch_end;
+
 extern void __init per_cpu_patch(void);
 extern void sun4v_patch_1insn_range(struct sun4v_1insn_patch_entry *,
 				    struct sun4v_1insn_patch_entry *);

arch/sparc/kernel/leon_kernel.c

@@ -56,11 +56,13 @@ static inline unsigned int leon_eirq_get(int cpu)
 static void leon_handle_ext_irq(unsigned int irq, struct irq_desc *desc)
 {
 	unsigned int eirq;
+	struct irq_bucket *p;
 	int cpu = sparc_leon3_cpuid();
 
 	eirq = leon_eirq_get(cpu);
-	if ((eirq & 0x10) && irq_map[eirq]->irq) /* bit4 tells if IRQ happened */
-		generic_handle_irq(irq_map[eirq]->irq);
+	p = irq_map[eirq];
+	if ((eirq & 0x10) && p && p->irq) /* bit4 tells if IRQ happened */
+		generic_handle_irq(p->irq);
 }
 
 /* The extended IRQ controller has been found, this function registers it */

arch/sparc/kernel/perf_event.c

@@ -1762,15 +1762,25 @@ static void perf_callchain_user_32(struct perf_callchain_entry *entry,
 
 	ufp = regs->u_regs[UREG_I6] & 0xffffffffUL;
 	do {
-		struct sparc_stackf32 *usf, sf;
 		unsigned long pc;
 
-		usf = (struct sparc_stackf32 *) ufp;
-		if (__copy_from_user_inatomic(&sf, usf, sizeof(sf)))
-			break;
+		if (thread32_stack_is_64bit(ufp)) {
+			struct sparc_stackf *usf, sf;
 
-		pc = sf.callers_pc;
-		ufp = (unsigned long)sf.fp;
+			ufp += STACK_BIAS;
+			usf = (struct sparc_stackf *) ufp;
+			if (__copy_from_user_inatomic(&sf, usf, sizeof(sf)))
+				break;
+			pc = sf.callers_pc & 0xffffffff;
+			ufp = ((unsigned long) sf.fp) & 0xffffffff;
+		} else {
+			struct sparc_stackf32 *usf, sf;
+			usf = (struct sparc_stackf32 *) ufp;
+			if (__copy_from_user_inatomic(&sf, usf, sizeof(sf)))
+				break;
+			pc = sf.callers_pc;
+			ufp = (unsigned long)sf.fp;
+		}
 		perf_callchain_store(entry, pc);
 	} while (entry->nr < PERF_MAX_STACK_DEPTH);
 }