Xtensa patchset for v3.11-rc1

-----BEGIN PGP SIGNATURE-----
 Version: GnuPG v1.4.12 (GNU/Linux)
 
 iQIcBAABAgAGBQJR3gHDAAoJEI9vqH3mFV2slu4QAIRGqonB3fbRi8OnBJDwrpDY
 4U8MG6uvz1tReBI+BdSXBCxF2CSCINxt2pDzEmQo+1suBfXRDOa5haCUHoLWEMHS
 za2i5bOdsN1U8fio1OFRFn6hdP7L5Xy2YBTZN5ufj9pYA1hkaECdMufWUkM7hklY
 H21zomT+NLiOEAgB5W4JpHZDzjmjwONPebwBiphi0wnWJY0TttK2huZQ/iAK79Fl
 nu0c8cQ2YG++NSnAgTQ0mmXrZNLp4zb40CJwMdOR1tWllYayhuypJdymfwFX7LjJ
 YU5MrAPWNjDAzuGYPnFiR9dli26B7aODpBWTanyWYq1HwKxk3RGQjzty/D1izmWh
 te4qtAhB/sYCgpdMpiMH1bGg6QqkvSJL8raxktFQ1N8ZwuiR3x1Zj6DGI0U3kGLT
 7P194njxIV1+4lCuAfVLTzXiYWV5nrtp2mrT6y1hIzWIolvXqDmJvnJzDTMavihL
 lD2AaLwrkF8rx8kmAFLLA4KqdrmD0IQp79UOyy9KMKhhk9SAD4FqyfuwjsYUrSsN
 vd2QqBxjE8GAc4MYIqqzIqTIvhfDtNk3cFXGn/lR26AW7H5LWT8DKkB0q67WMAJi
 /gvGpfnYED+bD2y0DfkCopGnpNWWXyXf1/mifbZH3jNvaWYAplKpy7eVuvPLgg1E
 7eyVBIVnS2iNYkFAy5jG
 =qScd
 -----END PGP SIGNATURE-----

Merge tag 'xtensa-next-20130710' of git://github.com/czankel/xtensa-linux

Pull Xtensa updates from Chris Zankel.

* tag 'xtensa-next-20130710' of git://github.com/czankel/xtensa-linux: (22 commits)
  xtensa: remove the second argument of __bio_kmap_atomic()
  xtensa: add static function tracer support
  xtensa: Flat DeviceTree copy not future-safe
  xtensa: check TLB sanity on return to userspace
  xtensa: adjust boot parameters address when INITIALIZE_XTENSA_MMU_INSIDE_VMLINUX is selected
  xtensa: bootparams: fix typo
  xtensa: tell git to ignore generated .dtb files
  xtensa: ccount based sched_clock
  xtensa: ccount based clockevent implementation
  xtensa: consolidate ccount access routines
  xtensa: cleanup ccount frequency tracking
  xtensa: timex.h: remove unused symbols
  xtensa: tell git to ignore copied zlib source files
  xtensa: fix section mismatch in pcibios_fixup_bus
  xtensa: ISS: fix section mismatch in iss_net_setup
  arch: xtensa: include: asm: compiling issue, need cmpxchg64() defined.
  xtensa: xtfpga: fix section mismatch
  xtensa: remove unused platform_init_irq()
  xtensa: tell git to ignore generated files
  xtensa: flush TLB entries for pages of non-current mm correctly
  ...

arch/xtensa/Kconfig

@@ -6,10 +6,12 @@ config XTENSA
 	select ARCH_WANT_FRAME_POINTERS
 	select HAVE_IDE
 	select GENERIC_ATOMIC64
+	select GENERIC_CLOCKEVENTS
 	select HAVE_GENERIC_HARDIRQS
 	select VIRT_TO_BUS
 	select GENERIC_IRQ_SHOW
 	select GENERIC_CPU_DEVICES
+	select GENERIC_SCHED_CLOCK
 	select MODULES_USE_ELF_RELA
 	select GENERIC_PCI_IOMAP
 	select ARCH_WANT_IPC_PARSE_VERSION
@@ -17,6 +19,7 @@ config XTENSA
 	select CLONE_BACKWARDS
 	select IRQ_DOMAIN
 	select HAVE_OPROFILE
+	select HAVE_FUNCTION_TRACER
 	help
 	  Xtensa processors are 32-bit RISC machines designed by Tensilica
 	  primarily for embedded systems.  These processors are both

arch/xtensa/Kconfig.debug

@@ -2,6 +2,16 @@ menu "Kernel hacking"
 
 source "lib/Kconfig.debug"
 
+config DEBUG_TLB_SANITY
+	bool "Debug TLB sanity"
+	depends on DEBUG_KERNEL
+	help
+	  Enable this to turn on TLB sanity check on each entry to userspace.
+	  This check can spot missing TLB invalidation/wrong PTE permissions/
+	  premature page freeing.
+
+	  If unsure, say N.
+
 config LD_NO_RELAX
 	bool "Disable linker relaxation"
 	default n

arch/xtensa/boot/.gitignore

@@ -0,0 +1,3 @@
+uImage
+zImage.redboot
+*.dtb

arch/xtensa/boot/boot-elf/.gitignore

@@ -0,0 +1 @@
+boot.lds

arch/xtensa/boot/lib/.gitignore

@@ -0,0 +1,3 @@
+inffast.c
+inflate.c
+inftrees.c

arch/xtensa/boot/lib/Makefile

@@ -7,6 +7,13 @@ zlib	:= inffast.c inflate.c inftrees.c
 lib-y	+= $(zlib:.c=.o) zmem.o
 
 ccflags-y	:= -Ilib/zlib_inflate
+ifdef CONFIG_FUNCTION_TRACER
+CFLAGS_REMOVE_inflate.o = -pg
+CFLAGS_REMOVE_zmem.o = -pg
+CFLAGS_REMOVE_inftrees.o = -pg
+CFLAGS_REMOVE_inffast.o = -pg
+endif
+
 
 quiet_cmd_copy_zlib = COPY    $@
       cmd_copy_zlib = cat $< > $@

arch/xtensa/include/asm/bootparam.h

@@ -20,7 +20,7 @@
 #define BP_TAG_COMMAND_LINE	0x1001	/* command line (0-terminated string)*/
 #define BP_TAG_INITRD		0x1002	/* ramdisk addr and size (bp_meminfo) */
 #define BP_TAG_MEMORY		0x1003	/* memory addr and size (bp_meminfo) */
-#define BP_TAG_SERIAL_BAUSRATE	0x1004	/* baud rate of current console. */
+#define BP_TAG_SERIAL_BAUDRATE	0x1004	/* baud rate of current console. */
 #define BP_TAG_SERIAL_PORT	0x1005	/* serial device of current console */
 #define BP_TAG_FDT		0x1006	/* flat device tree addr */
 

arch/xtensa/include/asm/cmpxchg.h

@@ -93,6 +93,7 @@ static inline unsigned long __cmpxchg_local(volatile void *ptr,
 	((__typeof__(*(ptr)))__cmpxchg_local_generic((ptr), (unsigned long)(o),\
 			(unsigned long)(n), sizeof(*(ptr))))
 #define cmpxchg64_local(ptr, o, n) __cmpxchg64_local_generic((ptr), (o), (n))
+#define cmpxchg64(ptr, o, n)    cmpxchg64_local((ptr), (o), (n))
 
 /*
  * xchg_u32

arch/xtensa/include/asm/delay.h

@@ -12,7 +12,7 @@
 #ifndef _XTENSA_DELAY_H
 #define _XTENSA_DELAY_H
 
-#include <asm/processor.h>
+#include <asm/timex.h>
 #include <asm/param.h>
 
 extern unsigned long loops_per_jiffy;
@@ -24,24 +24,17 @@ static inline void __delay(unsigned long loops)
 			      : "=r" (loops) : "0" (loops));
 }
 
-static __inline__ u32 xtensa_get_ccount(void)
-{
-	u32 ccount;
-	asm volatile ("rsr %0, ccount\n" : "=r" (ccount));
-	return ccount;
-}
-
 /* For SMP/NUMA systems, change boot_cpu_data to something like
  * local_cpu_data->... where local_cpu_data points to the current
  * cpu. */
 
 static __inline__ void udelay (unsigned long usecs)
 {
-	unsigned long start = xtensa_get_ccount();
+	unsigned long start = get_ccount();
 	unsigned long cycles = usecs * (loops_per_jiffy / (1000000UL / HZ));
 
 	/* Note: all variables are unsigned (can wrap around)! */
-	while (((unsigned long)xtensa_get_ccount()) - start < cycles)
+	while (((unsigned long)get_ccount()) - start < cycles)
 		;
 }
 

arch/xtensa/include/asm/ftrace.h

@@ -13,6 +13,7 @@
 #include <asm/processor.h>
 
 #define HAVE_ARCH_CALLER_ADDR
+#ifndef __ASSEMBLY__
 #define CALLER_ADDR0 ({ unsigned long a0, a1; \
 		__asm__ __volatile__ ( \
 			"mov %0, a0\n" \
@@ -24,10 +25,22 @@ extern unsigned long return_address(unsigned level);
 #define CALLER_ADDR1 return_address(1)
 #define CALLER_ADDR2 return_address(2)
 #define CALLER_ADDR3 return_address(3)
-#else
+#else /* CONFIG_FRAME_POINTER */
 #define CALLER_ADDR1 (0)
 #define CALLER_ADDR2 (0)
 #define CALLER_ADDR3 (0)
-#endif
+#endif /* CONFIG_FRAME_POINTER */
+#endif /* __ASSEMBLY__ */
+
+#ifdef CONFIG_FUNCTION_TRACER
+
+#define MCOUNT_ADDR ((unsigned long)(_mcount))
+#define MCOUNT_INSN_SIZE 3
+
+#ifndef __ASSEMBLY__
+extern void _mcount(void);
+#define mcount _mcount
+#endif /* __ASSEMBLY__ */
+#endif /* CONFIG_FUNCTION_TRACER */
 
 #endif /* _XTENSA_FTRACE_H */

arch/xtensa/include/asm/pgtable.h

@@ -5,7 +5,7 @@
  * it under the terms of the GNU General Public License version 2 as
  * published by the Free Software Foundation.
  *
- * Copyright (C) 2001 - 2007 Tensilica Inc.
+ * Copyright (C) 2001 - 2013 Tensilica Inc.
  */
 
 #ifndef _XTENSA_PGTABLE_H
@@ -64,41 +64,82 @@
  * Virtual memory area. We keep a distance to other memory regions to be
  * on the safe side. We also use this area for cache aliasing.
  */
-
 #define VMALLOC_START		0xC0000000
 #define VMALLOC_END		0xC7FEFFFF
 #define TLBTEMP_BASE_1		0xC7FF0000
 #define TLBTEMP_BASE_2		0xC7FF8000
 
 /*
- * Xtensa Linux config PTE layout (when present):
+ * For the Xtensa architecture, the PTE layout is as follows:
- *	31-12:	PPN
- *	11-6:	Software
- *	5-4:	RING
- *	3-0:	CA
  *
- * Similar to the Alpha and MIPS ports, we need to keep track of the ref
+ *		31------12  11  10-9   8-6  5-4  3-2  1-0
- * and mod bits in software.  We have a software "you can read
+ *		+-----------------------------------------+
- * from this page" bit, and a hardware one which actually lets the
+ *		|           |   Software   |   HARDWARE   |
- * process read from the page.  On the same token we have a software
+ *		|    PPN    |          ADW | RI |Attribute|
- * writable bit and the real hardware one which actually lets the
+ *		+-----------------------------------------+
- * process write to the page.
+ *   pte_none	|             MBZ          | 01 | 11 | 00 |
+ *		+-----------------------------------------+
+ *   present	|    PPN    | 0 | 00 | ADW | RI | CA | wx |
+ *		+- - - - - - - - - - - - - - - - - - - - -+
+ *   (PAGE_NONE)|    PPN    | 0 | 00 | ADW | 01 | 11 | 11 |
+ *		+-----------------------------------------+
+ *   swap	|     index     |   type   | 01 | 11 | 00 |
+ *		+- - - - - - - - - - - - - - - - - - - - -+
+ *   file	|        file offset       | 01 | 11 | 10 |
+ *		+-----------------------------------------+
  *
- * See further below for PTE layout for swapped-out pages.
+ * For T1050 hardware and earlier the layout differs for present and (PAGE_NONE)
+ *		+-----------------------------------------+
+ *   present	|    PPN    | 0 | 00 | ADW | RI | CA | w1 |
+ *		+-----------------------------------------+
+ *   (PAGE_NONE)|    PPN    | 0 | 00 | ADW | 01 | 01 | 00 |
+ *		+-----------------------------------------+
+ *
+ *  Legend:
+ *   PPN        Physical Page Number
+ *   ADW	software: accessed (young) / dirty / writable
+ *   RI         ring (0=privileged, 1=user, 2 and 3 are unused)
+ *   CA		cache attribute: 00 bypass, 01 writeback, 10 writethrough
+ *		(11 is invalid and used to mark pages that are not present)
+ *   w		page is writable (hw)
+ *   x		page is executable (hw)
+ *   index      swap offset / PAGE_SIZE (bit 11-31: 21 bits -> 8 GB)
+ *		(note that the index is always non-zero)
+ *   type       swap type (5 bits -> 32 types)
+ *   file offset 26-bit offset into the file, in increments of PAGE_SIZE
+ *
+ *  Notes:
+ *   - (PROT_NONE) is a special case of 'present' but causes an exception for
+ *     any access (read, write, and execute).
+ *   - 'multihit-exception' has the highest priority of all MMU exceptions,
+ *     so the ring must be set to 'RING_USER' even for 'non-present' pages.
+ *   - on older hardware, the exectuable flag was not supported and
+ *     used as a 'valid' flag, so it needs to be always set.
+ *   - we need to keep track of certain flags in software (dirty and young)
+ *     to do this, we use write exceptions and have a separate software w-flag.
+ *   - attribute value 1101 (and 1111 on T1050 and earlier) is reserved
  */
 
+#define _PAGE_ATTRIB_MASK	0xf
+
 #define _PAGE_HW_EXEC		(1<<0)	/* hardware: page is executable */
 #define _PAGE_HW_WRITE		(1<<1)	/* hardware: page is writable */
 
-#define _PAGE_FILE		(1<<1)	/* non-linear mapping, if !present */
-#define _PAGE_PROTNONE		(3<<0)	/* special case for VM_PROT_NONE */
-
-/* None of these cache modes include MP coherency:  */
 #define _PAGE_CA_BYPASS		(0<<2)	/* bypass, non-speculative */
 #define _PAGE_CA_WB		(1<<2)	/* write-back */
 #define _PAGE_CA_WT		(2<<2)	/* write-through */
 #define _PAGE_CA_MASK		(3<<2)
-#define _PAGE_INVALID		(3<<2)
+#define _PAGE_CA_INVALID	(3<<2)
+
+/* We use invalid attribute values to distinguish special pte entries */
+#if XCHAL_HW_VERSION_MAJOR < 2000
+#define _PAGE_HW_VALID		0x01	/* older HW needed this bit set */
+#define _PAGE_NONE		0x04
+#else
+#define _PAGE_HW_VALID		0x00
+#define _PAGE_NONE		0x0f
+#endif
+#define _PAGE_FILE		(1<<1)	/* file mapped page, only if !present */
 
 #define _PAGE_USER		(1<<4)	/* user access (ring=1) */
 
@@ -108,19 +149,12 @@
 #define _PAGE_DIRTY		(1<<7)	/* software: page dirty */
 #define _PAGE_ACCESSED		(1<<8)	/* software: page accessed (read) */
 
-/* On older HW revisions, we always have to set bit 0 */
-#if XCHAL_HW_VERSION_MAJOR < 2000
-# define _PAGE_VALID		(1<<0)
-#else
-# define _PAGE_VALID		0
-#endif
-
-#define _PAGE_CHG_MASK	(PAGE_MASK | _PAGE_ACCESSED | _PAGE_DIRTY)
-#define _PAGE_PRESENT	(_PAGE_VALID | _PAGE_CA_WB | _PAGE_ACCESSED)
-
 #ifdef CONFIG_MMU
 
-#define PAGE_NONE	   __pgprot(_PAGE_INVALID | _PAGE_USER | _PAGE_PROTNONE)
+#define _PAGE_CHG_MASK	   (PAGE_MASK | _PAGE_ACCESSED | _PAGE_DIRTY)
+#define _PAGE_PRESENT	   (_PAGE_HW_VALID | _PAGE_CA_WB | _PAGE_ACCESSED)
+
+#define PAGE_NONE	   __pgprot(_PAGE_NONE | _PAGE_USER)
 #define PAGE_COPY	   __pgprot(_PAGE_PRESENT | _PAGE_USER)
 #define PAGE_COPY_EXEC	   __pgprot(_PAGE_PRESENT | _PAGE_USER | _PAGE_HW_EXEC)
 #define PAGE_READONLY	   __pgprot(_PAGE_PRESENT | _PAGE_USER)
@@ -132,9 +166,9 @@
 #define PAGE_KERNEL_EXEC   __pgprot(_PAGE_PRESENT|_PAGE_HW_WRITE|_PAGE_HW_EXEC)
 
 #if (DCACHE_WAY_SIZE > PAGE_SIZE)
-# define _PAGE_DIRECTORY (_PAGE_VALID | _PAGE_ACCESSED)
+# define _PAGE_DIRECTORY   (_PAGE_HW_VALID | _PAGE_ACCESSED | _PAGE_CA_BYPASS)
 #else
-# define _PAGE_DIRECTORY (_PAGE_VALID | _PAGE_ACCESSED | _PAGE_CA_WB)
+# define _PAGE_DIRECTORY   (_PAGE_HW_VALID | _PAGE_ACCESSED | _PAGE_CA_WB)
 #endif
 
 #else /* no mmu */
@@ -202,12 +236,16 @@ static inline void pgtable_cache_init(void) { }
 /*
  * pte status.
  */
-#define pte_none(pte)	 (pte_val(pte) == _PAGE_INVALID)
+# define pte_none(pte)	 (pte_val(pte) == (_PAGE_CA_INVALID | _PAGE_USER))
-#define pte_present(pte)						\
+#if XCHAL_HW_VERSION_MAJOR < 2000
-	(((pte_val(pte) & _PAGE_CA_MASK) != _PAGE_INVALID)		\
+# define pte_present(pte) ((pte_val(pte) & _PAGE_CA_MASK) != _PAGE_CA_INVALID)
-	 || ((pte_val(pte) & _PAGE_PROTNONE) == _PAGE_PROTNONE))
+#else
+# define pte_present(pte)						\
+	(((pte_val(pte) & _PAGE_CA_MASK) != _PAGE_CA_INVALID)		\
+	 || ((pte_val(pte) & _PAGE_ATTRIB_MASK) == _PAGE_NONE))
+#endif
 #define pte_clear(mm,addr,ptep)						\
-	do { update_pte(ptep, __pte(_PAGE_INVALID)); } while(0)
+	do { update_pte(ptep, __pte(_PAGE_CA_INVALID | _PAGE_USER)); } while (0)
 
 #define pmd_none(pmd)	 (!pmd_val(pmd))
 #define pmd_present(pmd) (pmd_val(pmd) & PAGE_MASK)
@@ -328,35 +366,23 @@ ptep_set_wrprotect(struct mm_struct *mm, unsigned long addr, pte_t *ptep)
 
 
 /*
- * Encode and decode a swap entry.
+ * Encode and decode a swap and file entry.
- *
- * Format of swap pte:
- *  bit	   0	   MBZ
- *  bit	   1	   page-file (must be zero)
- *  bits   2 -  3  page hw access mode (must be 11: _PAGE_INVALID)
- *  bits   4 -  5  ring protection (must be 01: _PAGE_USER)
- *  bits   6 - 10  swap type (5 bits -> 32 types)
- *  bits  11 - 31  swap offset / PAGE_SIZE (21 bits -> 8GB)
- 
- * Format of file pte:
- *  bit	   0	   MBZ
- *  bit	   1	   page-file (must be one: _PAGE_FILE)
- *  bits   2 -  3  page hw access mode (must be 11: _PAGE_INVALID)
- *  bits   4 -  5  ring protection (must be 01: _PAGE_USER)
- *  bits   6 - 31  file offset / PAGE_SIZE
  */
+#define SWP_TYPE_BITS		5
+#define MAX_SWAPFILES_CHECK() BUILD_BUG_ON(MAX_SWAPFILES_SHIFT > SWP_TYPE_BITS)
 
 #define __swp_type(entry)	(((entry).val >> 6) & 0x1f)
 #define __swp_offset(entry)	((entry).val >> 11)
 #define __swp_entry(type,offs)	\
-	((swp_entry_t) {((type) << 6) | ((offs) << 11) | _PAGE_INVALID})
+	((swp_entry_t){((type) << 6) | ((offs) << 11) | \
+	 _PAGE_CA_INVALID | _PAGE_USER})
 #define __pte_to_swp_entry(pte)	((swp_entry_t) { pte_val(pte) })
 #define __swp_entry_to_pte(x)	((pte_t) { (x).val })
 
-#define PTE_FILE_MAX_BITS	28
+#define PTE_FILE_MAX_BITS	26
-#define pte_to_pgoff(pte)	(pte_val(pte) >> 4)
+#define pte_to_pgoff(pte)	(pte_val(pte) >> 6)
 #define pgoff_to_pte(off)	\
-	((pte_t) { ((off) << 4) | _PAGE_INVALID | _PAGE_FILE })
+	((pte_t) { ((off) << 6) | _PAGE_CA_INVALID | _PAGE_FILE | _PAGE_USER })
 
 #endif /*  !defined (__ASSEMBLY__) */
 

arch/xtensa/include/asm/platform.h

@@ -29,11 +29,6 @@ extern void platform_init(bp_tag_t*);
  */
 extern void platform_setup (char **);
 
-/*
- * platform_init_irq is called from init_IRQ.
- */
-extern void platform_init_irq (void);
-
 /*
  * platform_restart is called to restart the system.
  */

arch/xtensa/include/asm/timex.h

@@ -35,19 +35,11 @@
 # error "Bad timer number for Linux configurations!"
 #endif
 
-#define LINUX_TIMER_MASK        (1L << LINUX_TIMER_INT)
-
-#define CLOCK_TICK_RATE 	1193180	/* (everyone is using this value) */
-#define CLOCK_TICK_FACTOR       20 /* Factor of both 10^6 and CLOCK_TICK_RATE */
-
 #ifdef CONFIG_XTENSA_CALIBRATE_CCOUNT
-extern unsigned long ccount_per_jiffy;
+extern unsigned long ccount_freq;
-extern unsigned long nsec_per_ccount;
+#define CCOUNT_PER_JIFFY (ccount_freq / HZ)
-#define CCOUNT_PER_JIFFY ccount_per_jiffy
-#define NSEC_PER_CCOUNT  nsec_per_ccount
 #else
 #define CCOUNT_PER_JIFFY (CONFIG_XTENSA_CPU_CLOCK*(1000000UL/HZ))
-#define NSEC_PER_CCOUNT (1000UL / CONFIG_XTENSA_CPU_CLOCK)
 #endif
 
 

arch/xtensa/kernel/.gitignore

@@ -0,0 +1 @@
+vmlinux.lds

arch/xtensa/kernel/Makefile

@@ -11,6 +11,7 @@ obj-y := align.o coprocessor.o entry.o irq.o pci-dma.o platform.o process.o \
 obj-$(CONFIG_KGDB) += xtensa-stub.o
 obj-$(CONFIG_PCI) += pci.o
 obj-$(CONFIG_MODULES) += xtensa_ksyms.o module.o
+obj-$(CONFIG_FUNCTION_TRACER) += mcount.o
 
 AFLAGS_head.o += -mtext-section-literals
 

arch/xtensa/kernel/entry.S

@@ -458,7 +458,7 @@ common_exception_return:
 
 	_bbsi.l	a4, TIF_NEED_RESCHED, 3f
 	_bbsi.l	a4, TIF_NOTIFY_RESUME, 2f
-	_bbci.l	a4, TIF_SIGPENDING, 4f
+	_bbci.l	a4, TIF_SIGPENDING, 5f
 
 2:	l32i	a4, a1, PT_DEPC
 	bgeui	a4, VALID_DOUBLE_EXCEPTION_ADDRESS, 4f
@@ -476,6 +476,13 @@ common_exception_return:
 	callx4	a4
 	j	1b
 
+5:
+#ifdef CONFIG_DEBUG_TLB_SANITY
+	l32i	a4, a1, PT_DEPC
+	bgeui	a4, VALID_DOUBLE_EXCEPTION_ADDRESS, 4f
+	movi	a4, check_tlb_sanity
+	callx4	a4
+#endif
 4:	/* Restore optional registers. */
 
 	load_xtregs_opt a1 a2 a4 a5 a6 a7 PT_XTREGS_OPT
@@ -1792,10 +1799,15 @@ ENTRY(fast_store_prohibited)
 	l32i	a0, a0, 0
 	beqz	a0, 2f
 
-	/* Note that we assume _PAGE_WRITABLE_BIT is only set if pte is valid.*/
+	/*
+	 * Note that we test _PAGE_WRITABLE_BIT only if PTE is present
+	 * and is not PAGE_NONE. See pgtable.h for possible PTE layouts.
+	 */
 
 	_PTE_OFFSET(a0, a1, a4)
 	l32i	a4, a0, 0		# read pteval
+	movi	a1, _PAGE_CA_INVALID
+	ball	a4, a1, 2f
 	bbci.l	a4, _PAGE_WRITABLE_BIT, 2f
 
 	movi	a1, _PAGE_ACCESSED | _PAGE_DIRTY | _PAGE_HW_WRITE

arch/xtensa/kernel/head.S

@@ -68,6 +68,15 @@ _SetupMMU:
 
 #ifdef CONFIG_INITIALIZE_XTENSA_MMU_INSIDE_VMLINUX
 	initialize_mmu
+#if defined(CONFIG_MMU) && XCHAL_HAVE_PTP_MMU && XCHAL_HAVE_SPANNING_WAY
+	rsr	a2, excsave1
+	movi	a3, 0x08000000
+	bgeu	a2, a3, 1f
+	movi	a3, 0xd0000000
+	add	a2, a2, a3
+	wsr	a2, excsave1
+1:
+#endif
 #endif
 	.end	no-absolute-literals
 

arch/xtensa/kernel/mcount.S

@@ -0,0 +1,50 @@
+/*
+ * arch/xtensa/kernel/mcount.S
+ *
+ * Xtensa specific mcount support
+ *
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file "COPYING" in the main directory of this archive
+ * for more details.
+ *
+ * Copyright (C) 2013 Tensilica Inc.
+ */
+
+#include <linux/linkage.h>
+#include <asm/ftrace.h>
+
+/*
+ * Entry condition:
+ *
+ *   a2:	a0 of the caller
+ */
+
+ENTRY(_mcount)
+
+	entry	a1, 16
+
+	movi	a4, ftrace_trace_function
+	l32i	a4, a4, 0
+	movi	a3, ftrace_stub
+	bne	a3, a4, 1f
+	retw
+
+1: 	xor	a7, a2, a1
+	movi	a3, 0x3fffffff
+	and	a7, a7, a3
+	xor	a7, a7, a1
+
+	xor	a6, a0, a1
+	and	a6, a6, a3
+	xor	a6, a6, a1
+	addi	a6, a6, -MCOUNT_INSN_SIZE
+	callx4	a4
+
+	retw
+
+ENDPROC(_mcount)
+
+ENTRY(ftrace_stub)
+	entry	a1, 16
+	retw
+ENDPROC(ftrace_stub)

arch/xtensa/kernel/pci.c

@@ -77,9 +77,9 @@ pcibios_align_resource(void *data, const struct resource *res,
 
 	if (res->flags & IORESOURCE_IO) {
 		if (size > 0x100) {
-			printk(KERN_ERR "PCI: I/O Region %s/%d too large"
+			pr_err("PCI: I/O Region %s/%d too large (%u bytes)\n",
-			       " (%ld bytes)\n", pci_name(dev),
+					pci_name(dev), dev->resource - res,
-			       dev->resource - res, size);
+					size);
 		}
 
 		if (start & 0x300)
@@ -174,7 +174,7 @@ static int __init pcibios_init(void)
 	struct pci_controller *pci_ctrl;
 	struct list_head resources;
 	struct pci_bus *bus;
-	int next_busno = 0, i;
+	int next_busno = 0;
 
 	printk("PCI: Probing PCI hardware\n");
 
@@ -197,7 +197,7 @@ static int __init pcibios_init(void)
 
 subsys_initcall(pcibios_init);
 
-void __init pcibios_fixup_bus(struct pci_bus *bus)
+void pcibios_fixup_bus(struct pci_bus *bus)
 {
 	if (bus->parent) {
 		/* This is a subordinate bridge */

arch/xtensa/kernel/platform.c

@@ -29,7 +29,6 @@
  */
 
 _F(void, setup, (char** cmd), { });
-_F(void, init_irq, (void), { });
 _F(void, restart, (void), { while(1); });
 _F(void, halt, (void), { while(1); });
 _F(void, power_off, (void), { while(1); });
@@ -42,6 +41,6 @@ _F(void, pcibios_init, (void), { });
 _F(void, calibrate_ccount, (void),
 {
 	pr_err("ERROR: Cannot calibrate cpu frequency! Assuming 10MHz.\n");
-	ccount_per_jiffy = 10 * (1000000UL/HZ);
+	ccount_freq = 10 * 1000000UL;
 });
 #endif

arch/xtensa/kernel/setup.c

@@ -152,8 +152,8 @@ static int __init parse_tag_initrd(const bp_tag_t* tag)
 {
 	meminfo_t* mi;
 	mi = (meminfo_t*)(tag->data);
-	initrd_start = (void*)(mi->start);
+	initrd_start = __va(mi->start);
-	initrd_end = (void*)(mi->end);
+	initrd_end = __va(mi->end);
 
 	return 0;
 }
@@ -164,7 +164,7 @@ __tagtable(BP_TAG_INITRD, parse_tag_initrd);
 
 static int __init parse_tag_fdt(const bp_tag_t *tag)
 {
-	dtb_start = (void *)(tag->data[0]);
+	dtb_start = __va(tag->data[0]);
 	return 0;
 }
 
@@ -256,7 +256,7 @@ void __init early_init_devtree(void *params)
 static void __init copy_devtree(void)
 {
 	void *alloc = early_init_dt_alloc_memory_arch(
-			be32_to_cpu(initial_boot_params->totalsize), 0);
+			be32_to_cpu(initial_boot_params->totalsize), 8);
 	if (alloc) {
 		memcpy(alloc, initial_boot_params,
 				be32_to_cpu(initial_boot_params->totalsize));

arch/xtensa/kernel/time.c

@@ -16,6 +16,7 @@
 #include <linux/sched.h>
 #include <linux/time.h>
 #include <linux/clocksource.h>
+#include <linux/clockchips.h>
 #include <linux/interrupt.h>
 #include <linux/module.h>
 #include <linux/init.h>
@@ -23,13 +24,13 @@
 #include <linux/profile.h>
 #include <linux/delay.h>
 #include <linux/irqdomain.h>
+#include <linux/sched_clock.h>
 
 #include <asm/timex.h>
 #include <asm/platform.h>
 
 #ifdef CONFIG_XTENSA_CALIBRATE_CCOUNT
-unsigned long ccount_per_jiffy;		/* per 1/HZ */
+unsigned long ccount_freq;		/* ccount Hz */
-unsigned long nsec_per_ccount;		/* nsec per ccount increment */
 #endif
 
 static cycle_t ccount_read(struct clocksource *cs)
@@ -37,6 +38,11 @@ static cycle_t ccount_read(struct clocksource *cs)
 	return (cycle_t)get_ccount();
 }
 
+static u32 notrace ccount_sched_clock_read(void)
+{
+	return get_ccount();
+}
+
 static struct clocksource ccount_clocksource = {
 	.name = "ccount",
 	.rating = 200,
@@ -44,29 +50,98 @@ static struct clocksource ccount_clocksource = {
 	.mask = CLOCKSOURCE_MASK(32),
 };
 
+static int ccount_timer_set_next_event(unsigned long delta,
+		struct clock_event_device *dev);
+static void ccount_timer_set_mode(enum clock_event_mode mode,
+		struct clock_event_device *evt);
+static struct ccount_timer_t {
+	struct clock_event_device evt;
+	int irq_enabled;
+} ccount_timer = {
+	.evt = {
+		.name		= "ccount_clockevent",
+		.features	= CLOCK_EVT_FEAT_ONESHOT,
+		.rating		= 300,
+		.set_next_event	= ccount_timer_set_next_event,
+		.set_mode	= ccount_timer_set_mode,
+	},
+};
+
+static int ccount_timer_set_next_event(unsigned long delta,
+		struct clock_event_device *dev)
+{
+	unsigned long flags, next;
+	int ret = 0;
+
+	local_irq_save(flags);
+	next = get_ccount() + delta;
+	set_linux_timer(next);
+	if (next - get_ccount() > delta)
+		ret = -ETIME;
+	local_irq_restore(flags);
+
+	return ret;
+}
+
+static void ccount_timer_set_mode(enum clock_event_mode mode,
+		struct clock_event_device *evt)
+{
+	struct ccount_timer_t *timer =
+		container_of(evt, struct ccount_timer_t, evt);
+
+	/*
+	 * There is no way to disable the timer interrupt at the device level,
+	 * only at the intenable register itself. Since enable_irq/disable_irq
+	 * calls are nested, we need to make sure that these calls are
+	 * balanced.
+	 */
+	switch (mode) {
+	case CLOCK_EVT_MODE_SHUTDOWN:
+	case CLOCK_EVT_MODE_UNUSED:
+		if (timer->irq_enabled) {
+			disable_irq(evt->irq);
+			timer->irq_enabled = 0;
+		}
+		break;
+	case CLOCK_EVT_MODE_RESUME:
+	case CLOCK_EVT_MODE_ONESHOT:
+		if (!timer->irq_enabled) {
+			enable_irq(evt->irq);
+			timer->irq_enabled = 1;
+		}
+	default:
+		break;
+	}
+}
+
 static irqreturn_t timer_interrupt(int irq, void *dev_id);
 static struct irqaction timer_irqaction = {
 	.handler =	timer_interrupt,
-	.flags =	IRQF_DISABLED,
+	.flags =	IRQF_TIMER,
 	.name =		"timer",
+	.dev_id =	&ccount_timer,
 };
 
 void __init time_init(void)
 {
-	unsigned int irq;
 #ifdef CONFIG_XTENSA_CALIBRATE_CCOUNT
 	printk("Calibrating CPU frequency ");
 	platform_calibrate_ccount();
-	printk("%d.%02d MHz\n", (int)ccount_per_jiffy/(1000000/HZ),
+	printk("%d.%02d MHz\n", (int)ccount_freq/1000000,
-			(int)(ccount_per_jiffy/(10000/HZ))%100);
+			(int)(ccount_freq/10000)%100);
 #endif
 	clocksource_register_hz(&ccount_clocksource, CCOUNT_PER_JIFFY * HZ);
 
-	/* Initialize the linux timer interrupt. */
+	ccount_timer.evt.cpumask = cpumask_of(0);
+	ccount_timer.evt.irq = irq_create_mapping(NULL, LINUX_TIMER_INT);
+	if (WARN(!ccount_timer.evt.irq, "error: can't map timer irq"))
+		return;
+	clockevents_config_and_register(&ccount_timer.evt, ccount_freq, 0xf,
+			0xffffffff);
+	setup_irq(ccount_timer.evt.irq, &timer_irqaction);
+	ccount_timer.irq_enabled = 1;
 
-	irq = irq_create_mapping(NULL, LINUX_TIMER_INT);
+	setup_sched_clock(ccount_sched_clock_read, 32, ccount_freq);
-	setup_irq(irq, &timer_irqaction);
-	set_linux_timer(get_ccount() + CCOUNT_PER_JIFFY);
 }
 
 /*
@@ -75,36 +150,14 @@ void __init time_init(void)
 
 irqreturn_t timer_interrupt (int irq, void *dev_id)
 {
+	struct ccount_timer_t *timer = dev_id;
+	struct clock_event_device *evt = &timer->evt;
 
-	unsigned long next;
+	evt->event_handler(evt);
-
-	next = get_linux_timer();
-
-again:
-	while ((signed long)(get_ccount() - next) > 0) {
-
-		profile_tick(CPU_PROFILING);
-#ifndef CONFIG_SMP
-		update_process_times(user_mode(get_irq_regs()));
-#endif
-
-		xtime_update(1); /* Linux handler in kernel/time/timekeeping */
-
-		/* Note that writing CCOMPARE clears the interrupt. */
-
-		next += CCOUNT_PER_JIFFY;
-		set_linux_timer(next);
-	}
 
 	/* Allow platform to do something useful (Wdog). */
-
 	platform_heartbeat();
 
-	/* Make sure we didn't miss any tick... */
-
-	if ((signed long)(get_ccount() - next) > 0)
-		goto again;
-
 	return IRQ_HANDLED;
 }
 

arch/xtensa/kernel/xtensa_ksyms.c

@@ -124,3 +124,7 @@ extern long common_exception_return;
 extern long _spill_registers;
 EXPORT_SYMBOL(common_exception_return);
 EXPORT_SYMBOL(_spill_registers);
+
+#ifdef CONFIG_FUNCTION_TRACER
+EXPORT_SYMBOL(_mcount);
+#endif

arch/xtensa/mm/tlb.c

@@ -64,7 +64,7 @@ void flush_tlb_mm(struct mm_struct *mm)
 {
 	if (mm == current->active_mm) {
 		unsigned long flags;
-		local_save_flags(flags);
+		local_irq_save(flags);
 		__get_new_mmu_context(mm);
 		__load_mmu_context(mm);
 		local_irq_restore(flags);
@@ -94,7 +94,7 @@ void flush_tlb_range (struct vm_area_struct *vma,
 	printk("[tlbrange<%02lx,%08lx,%08lx>]\n",
 			(unsigned long)mm->context, start, end);
 #endif
-	local_save_flags(flags);
+	local_irq_save(flags);
 
 	if (end-start + (PAGE_SIZE-1) <= _TLB_ENTRIES << PAGE_SHIFT) {
 		int oldpid = get_rasid_register();
@@ -128,9 +128,10 @@ void flush_tlb_page (struct vm_area_struct *vma, unsigned long page)
 	if(mm->context == NO_CONTEXT)
 		return;
 
-	local_save_flags(flags);
+	local_irq_save(flags);
 
 	oldpid = get_rasid_register();
+	set_rasid_register(ASID_INSERT(mm->context));
 
 	if (vma->vm_flags & VM_EXEC)
 		invalidate_itlb_mapping(page);
@@ -140,3 +141,116 @@ void flush_tlb_page (struct vm_area_struct *vma, unsigned long page)
 
 	local_irq_restore(flags);
 }
+
+#ifdef CONFIG_DEBUG_TLB_SANITY
+
+static unsigned get_pte_for_vaddr(unsigned vaddr)
+{
+	struct task_struct *task = get_current();
+	struct mm_struct *mm = task->mm;
+	pgd_t *pgd;
+	pmd_t *pmd;
+	pte_t *pte;
+
+	if (!mm)
+		mm = task->active_mm;
+	pgd = pgd_offset(mm, vaddr);
+	if (pgd_none_or_clear_bad(pgd))
+		return 0;
+	pmd = pmd_offset(pgd, vaddr);
+	if (pmd_none_or_clear_bad(pmd))
+		return 0;
+	pte = pte_offset_map(pmd, vaddr);
+	if (!pte)
+		return 0;
+	return pte_val(*pte);
+}
+
+enum {
+	TLB_SUSPICIOUS	= 1,
+	TLB_INSANE	= 2,
+};
+
+static void tlb_insane(void)
+{
+	BUG_ON(1);
+}
+
+static void tlb_suspicious(void)
+{
+	WARN_ON(1);
+}
+
+/*
+ * Check that TLB entries with kernel ASID (1) have kernel VMA (>= TASK_SIZE),
+ * and TLB entries with user ASID (>=4) have VMA < TASK_SIZE.
+ *
+ * Check that valid TLB entries either have the same PA as the PTE, or PTE is
+ * marked as non-present. Non-present PTE and the page with non-zero refcount
+ * and zero mapcount is normal for batched TLB flush operation. Zero refcount
+ * means that the page was freed prematurely. Non-zero mapcount is unusual,
+ * but does not necessary means an error, thus marked as suspicious.
+ */
+static int check_tlb_entry(unsigned w, unsigned e, bool dtlb)
+{
+	unsigned tlbidx = w | (e << PAGE_SHIFT);
+	unsigned r0 = dtlb ?
+		read_dtlb_virtual(tlbidx) : read_itlb_virtual(tlbidx);
+	unsigned vpn = (r0 & PAGE_MASK) | (e << PAGE_SHIFT);
+	unsigned pte = get_pte_for_vaddr(vpn);
+	unsigned mm_asid = (get_rasid_register() >> 8) & ASID_MASK;
+	unsigned tlb_asid = r0 & ASID_MASK;
+	bool kernel = tlb_asid == 1;
+	int rc = 0;
+
+	if (tlb_asid > 0 && ((vpn < TASK_SIZE) == kernel)) {
+		pr_err("%cTLB: way: %u, entry: %u, VPN %08x in %s PTE\n",
+				dtlb ? 'D' : 'I', w, e, vpn,
+				kernel ? "kernel" : "user");
+		rc |= TLB_INSANE;
+	}
+
+	if (tlb_asid == mm_asid) {
+		unsigned r1 = dtlb ? read_dtlb_translation(tlbidx) :
+			read_itlb_translation(tlbidx);
+		if ((pte ^ r1) & PAGE_MASK) {
+			pr_err("%cTLB: way: %u, entry: %u, mapping: %08x->%08x, PTE: %08x\n",
+					dtlb ? 'D' : 'I', w, e, r0, r1, pte);
+			if (pte == 0 || !pte_present(__pte(pte))) {
+				struct page *p = pfn_to_page(r1 >> PAGE_SHIFT);
+				pr_err("page refcount: %d, mapcount: %d\n",
+						page_count(p),
+						page_mapcount(p));
+				if (!page_count(p))
+					rc |= TLB_INSANE;
+				else if (page_mapped(p))
+					rc |= TLB_SUSPICIOUS;
+			} else {
+				rc |= TLB_INSANE;
+			}
+		}
+	}
+	return rc;
+}
+
+void check_tlb_sanity(void)
+{
+	unsigned long flags;
+	unsigned w, e;
+	int bug = 0;
+
+	local_irq_save(flags);
+	for (w = 0; w < DTLB_ARF_WAYS; ++w)
+		for (e = 0; e < (1 << XCHAL_DTLB_ARF_ENTRIES_LOG2); ++e)
+			bug |= check_tlb_entry(w, e, true);
+	for (w = 0; w < ITLB_ARF_WAYS; ++w)
+		for (e = 0; e < (1 << XCHAL_ITLB_ARF_ENTRIES_LOG2); ++e)
+			bug |= check_tlb_entry(w, e, false);
+	if (bug & TLB_INSANE)
+		tlb_insane();
+	if (bug & TLB_SUSPICIOUS)
+		tlb_suspicious();
+	local_irq_restore(flags);
+}
+
+#endif /* CONFIG_DEBUG_TLB_SANITY */

arch/xtensa/platforms/iss/network.c

@@ -700,7 +700,7 @@ struct iss_net_init {
 
 #define ERR KERN_ERR "iss_net_setup: "
 
-static int iss_net_setup(char *str)
+static int __init iss_net_setup(char *str)
 {
 	struct iss_net_private *device = NULL;
 	struct iss_net_init *new;

arch/xtensa/platforms/iss/simdisk.c

@@ -108,13 +108,13 @@ static int simdisk_xfer_bio(struct simdisk *dev, struct bio *bio)
 	sector_t sector = bio->bi_sector;
 
 	bio_for_each_segment(bvec, bio, i) {
-		char *buffer = __bio_kmap_atomic(bio, i, KM_USER0);
+		char *buffer = __bio_kmap_atomic(bio, i);
 		unsigned len = bvec->bv_len >> SECTOR_SHIFT;
 
 		simdisk_transfer(dev, sector, len, buffer,
 				bio_data_dir(bio) == WRITE);
 		sector += len;
-		__bio_kunmap_atomic(bio, KM_USER0);
+		__bio_kunmap_atomic(bio);
 	}
 	return 0;
 }

arch/xtensa/platforms/xtfpga/setup.c

@@ -163,7 +163,7 @@ void platform_heartbeat(void)
 
 #ifdef CONFIG_XTENSA_CALIBRATE_CCOUNT
 
-void platform_calibrate_ccount(void)
+void __init platform_calibrate_ccount(void)
 {
 	long clk_freq = 0;
 #ifdef CONFIG_OF
@@ -179,8 +179,7 @@ void platform_calibrate_ccount(void)
 	if (!clk_freq)
 		clk_freq = *(long *)XTFPGA_CLKFRQ_VADDR;
 
-	ccount_per_jiffy = clk_freq / HZ;
+	ccount_freq = clk_freq;
-	nsec_per_ccount = 1000000000UL / clk_freq;
 }
 
 #endif

arch/xtensa/variants/s6000/delay.c

@@ -1,4 +1,3 @@
-#include <asm/delay.h>
 #include <asm/timex.h>
 #include <asm/io.h>
 #include <variant/hardware.h>
@@ -17,11 +16,10 @@ void platform_calibrate_ccount(void)
 		"1:	l32i %0, %2, 0 ;"
 		"	beq %0, %1, 1b ;"
 		: "=&a"(u) : "a"(t), "a"(tstamp));
-		b = xtensa_get_ccount();
+		b = get_ccount();
 		if (i == LOOPS)
 			a = b;
 	} while (--i >= 0);
 	b -= a;
-	nsec_per_ccount = (LOOPS * 10000) / b;
+	ccount_freq = b * (100000UL / LOOPS);
-	ccount_per_jiffy = b * (100000UL / (LOOPS * HZ));
 }

include/linux/bio.h

@@ -97,11 +97,11 @@ static inline void *bio_data(struct bio *bio)
  * permanent PIO fall back, user is probably better off disabling highmem
  * I/O completely on that queue (see ide-dma for example)
  */
-#define __bio_kmap_atomic(bio, idx, kmtype)				\
+#define __bio_kmap_atomic(bio, idx)				\
 	(kmap_atomic(bio_iovec_idx((bio), (idx))->bv_page) +	\
 		bio_iovec_idx((bio), (idx))->bv_offset)
 
-#define __bio_kunmap_atomic(addr, kmtype) kunmap_atomic(addr)
+#define __bio_kunmap_atomic(addr) kunmap_atomic(addr)
 
 /*
  * merge helpers etc