Merge branch 'next' of git://git.monstr.eu/linux-2.6-microblaze

Pull arch/microblaze fixes from Michal Simek

* 'next' of git://git.monstr.eu/linux-2.6-microblaze:
  microblaze: Handle TLB skip size dynamically
  microblaze: Introduce TLB skip size
  microblaze: Improve TLB calculation for small systems
  microblaze: Extend space for compiled-in FDT to 32kB
  microblaze: Clear all MSR flags on the first kernel instruction
  microblaze: Use node name instead of compatible string
  microblaze: Fix mapin_ram function
  microblaze: Highmem support
  microblaze: Use active regions
  microblaze: Show more detailed information about memory
  microblaze: Introduce fixmap
  microblaze: mm: Fix lowmem max memory size limits
  microblaze: mm: Use ZONE_DMA instead of ZONE_NORMAL
  microblaze: trivial: Fix typo fault in timer.c
  microblaze: Use vsprintf extention %pf with builtin_return_address
  microblaze: Add PVR version string for MB 8.20.b and 8.30.a
  microblaze: Fix makefile to work with latest toolchain
  microblaze: Fix typo in early_printk.c

23 files changed, 710 insertions, 95 deletions

diff --git a/arch/microblaze/Kconfig b/arch/microblaze/Kconfig
index 11060fa..ac22dc7 100644
--- a/arch/microblaze/Kconfig
+++ b/arch/microblaze/Kconfig
@@ -1,6 +1,7 @@
 config MICROBLAZE
 	def_bool y
 	select HAVE_MEMBLOCK
+	select HAVE_MEMBLOCK_NODE_MAP
 	select HAVE_FUNCTION_TRACER
 	select HAVE_FUNCTION_TRACE_MCOUNT_TEST
 	select HAVE_FUNCTION_GRAPH_TRACER
@@ -28,6 +29,12 @@ config SWAP
 config RWSEM_GENERIC_SPINLOCK
 	def_bool y
 
+config ZONE_DMA
+	def_bool y
+
+config ARCH_POPULATES_NODE_MAP
+	def_bool y
+
 config RWSEM_XCHGADD_ALGORITHM
 	bool
 
@@ -153,20 +160,18 @@ config XILINX_UNCACHED_SHADOW
 	  The feature requires the design to define the RAM memory controller
 	  window to be twice as large as the actual physical memory.
 
-config HIGHMEM_START_BOOL
-	bool "Set high memory pool address"
-	depends on ADVANCED_OPTIONS && HIGHMEM
+config HIGHMEM
+	bool "High memory support"
+	depends on MMU
 	help
-	  This option allows you to set the base address of the kernel virtual
-	  area used to map high memory pages.  This can be useful in
-	  optimizing the layout of kernel virtual memory.
+	  The address space of Microblaze processors is only 4 Gigabytes large
+	  and it has to accommodate user address space, kernel address
+	  space as well as some memory mapped IO. That means that, if you
+	  have a large amount of physical memory and/or IO, not all of the
+	  memory can be "permanently mapped" by the kernel. The physical
+	  memory that is not permanently mapped is called "high memory".
 
-	  Say N here unless you know what you are doing.
-
-config HIGHMEM_START
-	hex "Virtual start address of high memory pool" if HIGHMEM_START_BOOL
-	depends on MMU
-	default "0xfe000000"
+	  If unsure, say n.
 
 config LOWMEM_SIZE_BOOL
 	bool "Set maximum low memory"
@@ -255,6 +260,10 @@ config MICROBLAZE_32K_PAGES
 
 endchoice
 
+config KERNEL_PAD
+	hex "Kernel PAD for unpacking" if ADVANCED_OPTIONS
+	default "0x80000" if MMU
+
 endmenu
 
 source "mm/Kconfig"
diff --git a/arch/microblaze/boot/Makefile b/arch/microblaze/boot/Makefile
index 0c796cf..34940c8 100644
--- a/arch/microblaze/boot/Makefile
+++ b/arch/microblaze/boot/Makefile
@@ -8,7 +8,7 @@ obj-y += linked_dtb.o
 
 targets := linux.bin linux.bin.gz simpleImage.%
 
-OBJCOPYFLAGS := -O binary
+OBJCOPYFLAGS := -R .note -R .comment -R .note.gnu.build-id -O binary
 
 # Ensure system.dtb exists
 $(obj)/linked_dtb.o: $(obj)/system.dtb
diff --git a/arch/microblaze/include/asm/fixmap.h b/arch/microblaze/include/asm/fixmap.h
new file mode 100644
index 0000000..f2b312e
--- /dev/null
+++ b/arch/microblaze/include/asm/fixmap.h
@@ -0,0 +1,109 @@
+/*
+ * fixmap.h: compile-time virtual memory allocation
+ *
+ * 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) 1998 Ingo Molnar
+ *
+ * Copyright 2008 Freescale Semiconductor Inc.
+ *   Port to powerpc added by Kumar Gala
+ *
+ * Copyright 2011 Michal Simek <monstr@monstr.eu>
+ * Copyright 2011 PetaLogix Qld Pty Ltd
+ *   Port to Microblaze
+ */
+
+#ifndef _ASM_FIXMAP_H
+#define _ASM_FIXMAP_H
+
+#ifndef __ASSEMBLY__
+#include <linux/kernel.h>
+#include <asm/page.h>
+#ifdef CONFIG_HIGHMEM
+#include <linux/threads.h>
+#include <asm/kmap_types.h>
+#endif
+
+#define FIXADDR_TOP	((unsigned long)(-PAGE_SIZE))
+
+/*
+ * Here we define all the compile-time 'special' virtual
+ * addresses. The point is to have a constant address at
+ * compile time, but to set the physical address only
+ * in the boot process. We allocate these special addresses
+ * from the end of virtual memory (0xfffff000) backwards.
+ * Also this lets us do fail-safe vmalloc(), we
+ * can guarantee that these special addresses and
+ * vmalloc()-ed addresses never overlap.
+ *
+ * these 'compile-time allocated' memory buffers are
+ * fixed-size 4k pages. (or larger if used with an increment
+ * highger than 1) use fixmap_set(idx,phys) to associate
+ * physical memory with fixmap indices.
+ *
+ * TLB entries of such buffers will not be flushed across
+ * task switches.
+ */
+enum fixed_addresses {
+	FIX_HOLE,
+#ifdef CONFIG_HIGHMEM
+	FIX_KMAP_BEGIN,	/* reserved pte's for temporary kernel mappings */
+	FIX_KMAP_END = FIX_KMAP_BEGIN + (KM_TYPE_NR * num_possible_cpus()) - 1,
+#endif
+	__end_of_fixed_addresses
+};
+
+extern void __set_fixmap(enum fixed_addresses idx,
+					phys_addr_t phys, pgprot_t flags);
+
+#define set_fixmap(idx, phys) \
+		__set_fixmap(idx, phys, PAGE_KERNEL)
+/*
+ * Some hardware wants to get fixmapped without caching.
+ */
+#define set_fixmap_nocache(idx, phys) \
+		__set_fixmap(idx, phys, PAGE_KERNEL_CI)
+
+#define clear_fixmap(idx) \
+		__set_fixmap(idx, 0, __pgprot(0))
+
+#define __FIXADDR_SIZE	(__end_of_fixed_addresses << PAGE_SHIFT)
+#define FIXADDR_START		(FIXADDR_TOP - __FIXADDR_SIZE)
+
+#define __fix_to_virt(x)	(FIXADDR_TOP - ((x) << PAGE_SHIFT))
+#define __virt_to_fix(x)	((FIXADDR_TOP - ((x)&PAGE_MASK)) >> PAGE_SHIFT)
+
+extern void __this_fixmap_does_not_exist(void);
+
+/*
+ * 'index to address' translation. If anyone tries to use the idx
+ * directly without tranlation, we catch the bug with a NULL-deference
+ * kernel oops. Illegal ranges of incoming indices are caught too.
+ */
+static __always_inline unsigned long fix_to_virt(const unsigned int idx)
+{
+	/*
+	 * this branch gets completely eliminated after inlining,
+	 * except when someone tries to use fixaddr indices in an
+	 * illegal way. (such as mixing up address types or using
+	 * out-of-range indices).
+	 *
+	 * If it doesn't get removed, the linker will complain
+	 * loudly with a reasonably clear error message..
+	 */
+	if (idx >= __end_of_fixed_addresses)
+		__this_fixmap_does_not_exist();
+
+	return __fix_to_virt(idx);
+}
+
+static inline unsigned long virt_to_fix(const unsigned long vaddr)
+{
+	BUG_ON(vaddr >= FIXADDR_TOP || vaddr < FIXADDR_START);
+	return __virt_to_fix(vaddr);
+}
+
+#endif /* !__ASSEMBLY__ */
+#endif
diff --git a/arch/microblaze/include/asm/highmem.h b/arch/microblaze/include/asm/highmem.h
new file mode 100644
index 0000000..2446a73
--- /dev/null
+++ b/arch/microblaze/include/asm/highmem.h
@@ -0,0 +1,96 @@
+/*
+ * highmem.h: virtual kernel memory mappings for high memory
+ *
+ * Used in CONFIG_HIGHMEM systems for memory pages which
+ * are not addressable by direct kernel virtual addresses.
+ *
+ * Copyright (C) 1999 Gerhard Wichert, Siemens AG
+ *		      Gerhard.Wichert@pdb.siemens.de
+ *
+ *
+ * Redesigned the x86 32-bit VM architecture to deal with
+ * up to 16 Terabyte physical memory. With current x86 CPUs
+ * we now support up to 64 Gigabytes physical RAM.
+ *
+ * Copyright (C) 1999 Ingo Molnar <mingo@redhat.com>
+ */
+#ifndef _ASM_HIGHMEM_H
+#define _ASM_HIGHMEM_H
+
+#ifdef __KERNEL__
+
+#include <linux/init.h>
+#include <linux/interrupt.h>
+#include <linux/uaccess.h>
+#include <asm/fixmap.h>
+
+extern pte_t *kmap_pte;
+extern pgprot_t kmap_prot;
+extern pte_t *pkmap_page_table;
+
+/*
+ * Right now we initialize only a single pte table. It can be extended
+ * easily, subsequent pte tables have to be allocated in one physical
+ * chunk of RAM.
+ */
+/*
+ * We use one full pte table with 4K pages. And with 16K/64K/256K pages pte
+ * table covers enough memory (32MB/512MB/2GB resp.), so that both FIXMAP
+ * and PKMAP can be placed in a single pte table. We use 512 pages for PKMAP
+ * in case of 16K/64K/256K page sizes.
+ */
+
+#define PKMAP_ORDER	PTE_SHIFT
+#define LAST_PKMAP	(1 << PKMAP_ORDER)
+
+#define PKMAP_BASE	((FIXADDR_START - PAGE_SIZE * (LAST_PKMAP + 1)) \
+								& PMD_MASK)
+
+#define LAST_PKMAP_MASK	(LAST_PKMAP - 1)
+#define PKMAP_NR(virt)  ((virt - PKMAP_BASE) >> PAGE_SHIFT)
+#define PKMAP_ADDR(nr)  (PKMAP_BASE + ((nr) << PAGE_SHIFT))
+
+extern void *kmap_high(struct page *page);
+extern void kunmap_high(struct page *page);
+extern void *kmap_atomic_prot(struct page *page, pgprot_t prot);
+extern void __kunmap_atomic(void *kvaddr);
+
+static inline void *kmap(struct page *page)
+{
+	might_sleep();
+	if (!PageHighMem(page))
+		return page_address(page);
+	return kmap_high(page);
+}
+
+static inline void kunmap(struct page *page)
+{
+	BUG_ON(in_interrupt());
+	if (!PageHighMem(page))
+		return;
+	kunmap_high(page);
+}
+
+static inline void *__kmap_atomic(struct page *page)
+{
+	return kmap_atomic_prot(page, kmap_prot);
+}
+
+static inline struct page *kmap_atomic_to_page(void *ptr)
+{
+	unsigned long idx, vaddr = (unsigned long) ptr;
+	pte_t *pte;
+
+	if (vaddr < FIXADDR_START)
+		return virt_to_page(ptr);
+
+	idx = virt_to_fix(vaddr);
+	pte = kmap_pte - (idx - FIX_KMAP_BEGIN);
+	return pte_page(*pte);
+}
+
+#define flush_cache_kmaps()	{ flush_icache(); flush_dcache(); }
+
+#endif /* __KERNEL__ */
+
+#endif /* _ASM_HIGHMEM_H */
diff --git a/arch/microblaze/include/asm/mmu.h b/arch/microblaze/include/asm/mmu.h
index 8d6a654..1f9eddd 100644
--- a/arch/microblaze/include/asm/mmu.h
+++ b/arch/microblaze/include/asm/mmu.h
@@ -56,6 +56,12 @@ typedef struct _SEGREG {
 
 extern void _tlbie(unsigned long va);	/* invalidate a TLB entry */
 extern void _tlbia(void);		/* invalidate all TLB entries */
+
+/*
+ * tlb_skip size stores actual number skipped TLBs from TLB0 - every directy TLB
+ * mapping has to increase tlb_skip size.
+ */
+extern u32 tlb_skip;
 #   endif /* __ASSEMBLY__ */
 
 /*
@@ -69,6 +75,12 @@ extern void _tlbia(void);		/* invalidate all TLB entries */
 
 #  define MICROBLAZE_TLB_SIZE 64
 
+/* For cases when you want to skip some TLB entries */
+#  define MICROBLAZE_TLB_SKIP 0
+
+/* Use the last TLB for temporary access to LMB */
+#  define MICROBLAZE_LMB_TLB_ID 63
+
 /*
  * TLB entries are defined by a "high" tag portion and a "low" data
  * portion. The data portion is 32-bits.
diff --git a/arch/microblaze/include/asm/page.h b/arch/microblaze/include/asm/page.h
index a25e6b5..352cc23 100644
--- a/arch/microblaze/include/asm/page.h
+++ b/arch/microblaze/include/asm/page.h
@@ -135,8 +135,8 @@ extern unsigned long min_low_pfn;
 extern unsigned long max_pfn;
 
 extern unsigned long memory_start;
-extern unsigned long memory_end;
 extern unsigned long memory_size;
+extern unsigned long lowmem_size;
 
 extern int page_is_ram(unsigned long pfn);
 
diff --git a/arch/microblaze/include/asm/pgtable.h b/arch/microblaze/include/asm/pgtable.h
index 44dc67a..3ef7b9c 100644
--- a/arch/microblaze/include/asm/pgtable.h
+++ b/arch/microblaze/include/asm/pgtable.h
@@ -94,8 +94,7 @@ static inline pte_t pte_mkspecial(pte_t pte)	{ return pte; }
 /* Start and end of the vmalloc area. */
 /* Make sure to map the vmalloc area above the pinned kernel memory area
    of 32Mb.  */
-#define VMALLOC_START	(CONFIG_KERNEL_START + \
-				max(32 * 1024 * 1024UL, memory_size))
+#define VMALLOC_START	(CONFIG_KERNEL_START + CONFIG_LOWMEM_SIZE)
 #define VMALLOC_END	ioremap_bot
 
 #endif /* __ASSEMBLY__ */
diff --git a/arch/microblaze/include/asm/setup.h b/arch/microblaze/include/asm/setup.h
index 6c72ed7..9f195c0 100644
--- a/arch/microblaze/include/asm/setup.h
+++ b/arch/microblaze/include/asm/setup.h
@@ -39,7 +39,8 @@ extern void of_platform_reset_gpio_probe(void);
 void time_init(void);
 void init_IRQ(void);
 void machine_early_init(const char *cmdline, unsigned int ram,
-			unsigned int fdt, unsigned int msr);
+		unsigned int fdt, unsigned int msr, unsigned int tlb0,
+		unsigned int tlb1);
 
 void machine_restart(char *cmd);
 void machine_shutdown(void);
diff --git a/arch/microblaze/include/asm/system.h b/arch/microblaze/include/asm/system.h
index 5a433cb..01228d2 100644
--- a/arch/microblaze/include/asm/system.h
+++ b/arch/microblaze/include/asm/system.h
@@ -83,6 +83,7 @@ void default_idle(void);
 void free_init_pages(char *what, unsigned long begin, unsigned long end);
 void free_initmem(void);
 extern char *klimit;
+extern unsigned long kernel_tlb;
 extern void ret_from_fork(void);
 
 extern void *alloc_maybe_bootmem(size_t size, gfp_t mask);
diff --git a/arch/microblaze/include/asm/uaccess.h b/arch/microblaze/include/asm/uaccess.h
index 072b007..ef25f75 100644
--- a/arch/microblaze/include/asm/uaccess.h
+++ b/arch/microblaze/include/asm/uaccess.h
@@ -80,7 +80,7 @@ extern unsigned long search_exception_table(unsigned long);
 static inline int ___range_ok(unsigned long addr, unsigned long size)
 {
 	return ((addr < memory_start) ||
-		((addr + size) > memory_end));
+		((addr + size - 1) > (memory_start + memory_size - 1)));
 }
 
 #define __range_ok(addr, size) \
diff --git a/arch/microblaze/kernel/cpu/cpuinfo.c b/arch/microblaze/kernel/cpu/cpuinfo.c
index 54194b2..eab6abf 100644
--- a/arch/microblaze/kernel/cpu/cpuinfo.c
+++ b/arch/microblaze/kernel/cpu/cpuinfo.c
@@ -35,6 +35,8 @@ const struct cpu_ver_key cpu_ver_lookup[] = {
 	{"8.00.b", 0x13},
 	{"8.10.a", 0x14},
 	{"8.20.a", 0x15},
+	{"8.20.b", 0x16},
+	{"8.30.a", 0x17},
 	{NULL, 0},
 };
 
diff --git a/arch/microblaze/kernel/early_printk.c b/arch/microblaze/kernel/early_printk.c
index 8356e47..ec48587 100644
--- a/arch/microblaze/kernel/early_printk.c
+++ b/arch/microblaze/kernel/early_printk.c
@@ -171,10 +171,24 @@ void __init remap_early_printk(void)
 {
 	if (!early_console_initialized || !early_console)
 		return;
-	printk(KERN_INFO "early_printk_console remaping from 0x%x to ",
+	printk(KERN_INFO "early_printk_console remapping from 0x%x to ",
 								base_addr);
 	base_addr = (u32) ioremap(base_addr, PAGE_SIZE);
 	printk(KERN_CONT "0x%x\n", base_addr);
+
+	/*
+	 * Early console is on the top of skipped TLB entries
+	 * decrease tlb_skip size ensure that hardcoded TLB entry will be
+	 * used by generic algorithm
+	 * FIXME check if early console mapping is on the top by rereading
+	 * TLB entry and compare baseaddr
+	 *  mts  rtlbx, (tlb_skip - 1)
+	 *  nop
+	 *  mfs  rX, rtlblo
+	 *  nop
+	 *  cmp rX, orig_base_addr
+	 */
+	tlb_skip -= 1;
 }
 
 void __init disable_early_printk(void)
diff --git a/arch/microblaze/kernel/head.S b/arch/microblaze/kernel/head.S
index 77320b8..98b17f9 100644
--- a/arch/microblaze/kernel/head.S
+++ b/arch/microblaze/kernel/head.S
@@ -63,9 +63,7 @@ ENTRY(_start)
 real_start:
 #endif
 
-	mfs	r1, rmsr
-	andi	r1, r1, ~2
-	mts	rmsr, r1
+	mts	rmsr, r0
 /*
  * According to Xilinx, msrclr instruction behaves like 'mfs rX,rpc'
  * if the msrclr instruction is not enabled. We use this to detect
@@ -73,6 +71,7 @@ real_start:
  * r8 == 0 - msr instructions are implemented
  * r8 != 0 - msr instructions are not implemented
  */
+	mfs	r1, rmsr
 	msrclr	r8, 0 /* clear nothing - just read msr for test */
 	cmpu	r8, r8, r1 /* r1 must contain msr reg content */
 
@@ -96,7 +95,7 @@ big_endian:
 _prepare_copy_fdt:
 	or	r11, r0, r0 /* incremment */
 	ori	r4, r0, TOPHYS(_fdt_start)
-	ori	r3, r0, (0x4000 - 4)
+	ori	r3, r0, (0x8000 - 4)
 _copy_fdt:
 	lw	r12, r7, r11 /* r12 = r7 + r11 */
 	sw	r12, r4, r11 /* addr[r4 + r11] = r12 */
@@ -150,6 +149,7 @@ _copy_bram:
 _invalidate:
 	mts	rtlbx, r3
 	mts	rtlbhi, r0			/* flush: ensure V is clear   */
+	mts	rtlblo, r0
 	bgtid	r3, _invalidate		/* loop for all entries       */
 	addik	r3, r3, -1
 	/* sync */
@@ -169,6 +169,53 @@ _invalidate:
 	addik	r3,r0, CONFIG_KERNEL_START /* Load the kernel virtual address */
 	tophys(r4,r3)			/* Load the kernel physical address */
 
+	/* start to do TLB calculation */
+	addik	r12, r0, _end
+	rsub	r12, r3, r12
+	addik	r12, r12, CONFIG_KERNEL_PAD /* that's the pad */
+
+	or r9, r0, r0 /* TLB0 = 0 */
+	or r10, r0, r0 /* TLB1 = 0 */
+
+	addik	r11, r12, -0x1000000
+	bgei	r11, GT16 /* size is greater than 16MB */
+	addik	r11, r12, -0x0800000
+	bgei	r11, GT8 /* size is greater than 8MB */
+	addik	r11, r12, -0x0400000
+	bgei	r11, GT4 /* size is greater than 4MB */
+	/* size is less than 4MB */
+	addik	r11, r12, -0x0200000
+	bgei	r11, GT2 /* size is greater than 2MB */
+	addik	r9, r0, 0x0100000 /* TLB0 must be 1MB */
+	addik	r11, r12, -0x0100000
+	bgei	r11, GT1 /* size is greater than 1MB */
+	/* TLB1 is 0 which is setup above */
+	bri tlb_end
+GT4: /* r11 contains the rest - will be either 1 or 4 */
+	ori r9, r0, 0x400000 /* TLB0 is 4MB */
+	bri TLB1
+GT16: /* TLB0 is 16MB */
+	addik	r9, r0, 0x1000000 /* means TLB0 is 16MB */
+TLB1:
+	/* must be used r2 because of substract if failed */
+	addik	r2, r11, -0x0400000
+	bgei	r2, GT20 /* size is greater than 16MB */
+	/* size is >16MB and <20MB */
+	addik	r11, r11, -0x0100000
+	bgei	r11, GT17 /* size is greater than 17MB */
+	/* kernel is >16MB and < 17MB */
+GT1:
+	addik	r10, r0, 0x0100000 /* means TLB1 is 1MB */
+	bri tlb_end
+GT2: /* TLB0 is 0 and TLB1 will be 4MB */
+GT17: /* TLB1 is 4MB - kernel size <20MB */
+	addik	r10, r0, 0x0400000 /* means TLB1 is 4MB */
+	bri tlb_end
+GT8: /* TLB0 is still zero that's why I can use only TLB1 */
+GT20: /* TLB1 is 16MB - kernel size >20MB */
+	addik	r10, r0, 0x1000000 /* means TLB1 is 16MB */
+tlb_end:
+
 	/*
 	 * Configure and load two entries into TLB slots 0 and 1.
 	 * In case we are pinning TLBs, these are reserved in by the
@@ -178,28 +225,81 @@ _invalidate:
 	andi	r4,r4,0xfffffc00	/* Mask off the real page number */
 	ori	r4,r4,(TLB_WR | TLB_EX)	/* Set the write and execute bits */
 
+	/*
+	 * TLB0 is always used - check if is not zero (r9 stores TLB0 value)
+	 * if is use TLB1 value and clear it (r10 stores TLB1 value)
+	 */
+	bnei	r9, tlb0_not_zero
+	add	r9, r10, r0
+	add	r10, r0, r0
+tlb0_not_zero:
+
+	/* look at the code below */
+	ori	r30, r0, 0x200
+	andi	r29, r9, 0x100000
+	bneid	r29, 1f
+	addik	r30, r30, 0x80
+	andi	r29, r9, 0x400000
+	bneid	r29, 1f
+	addik	r30, r30, 0x80
+	andi	r29, r9, 0x1000000
+	bneid	r29, 1f
+	addik	r30, r30, 0x80
+1:
 	andi	r3,r3,0xfffffc00	/* Mask off the effective page number */
-	ori	r3,r3,(TLB_VALID | TLB_PAGESZ(PAGESZ_16M))
+	ori	r3,r3,(TLB_VALID)
+	or	r3, r3, r30
 
-	mts     rtlbx,r0		/* TLB slow 0 */
+	/* Load tlb_skip size value which is index to first unused TLB entry */
+	lwi	r11, r0, TOPHYS(tlb_skip)
+	mts     rtlbx,r11		/* TLB slow 0 */
 
 	mts	rtlblo,r4		/* Load the data portion of the entry */
 	mts	rtlbhi,r3		/* Load the tag portion of the entry */
 
-	addik	r4, r4, 0x01000000	/* Map next 16 M entries */
-	addik	r3, r3, 0x01000000
+	/* Increase tlb_skip size */
+	addik	r11, r11, 1
+	swi	r11, r0, TOPHYS(tlb_skip)
+
+	/* TLB1 can be zeroes that's why we not setup it */
+	beqi	r10, jump_over2
+
+	/* look at the code below */
+	ori	r30, r0, 0x200
+	andi	r29, r10, 0x100000
+	bneid	r29, 1f
+	addik	r30, r30, 0x80
+	andi	r29, r10, 0x400000
+	bneid	r29, 1f
+	addik	r30, r30, 0x80
+	andi	r29, r10, 0x1000000
+	bneid	r29, 1f
+	addik	r30, r30, 0x80
+1:
+	addk	r4, r4, r9	/* previous addr + TLB0 size */
+	addk	r3, r3, r9
 
-	ori	r6,r0,1			/* TLB slot 1 */
-	mts     rtlbx,r6
+	andi	r3,r3,0xfffffc00	/* Mask off the effective page number */
+	ori	r3,r3,(TLB_VALID)
+	or	r3, r3, r30
+
+	lwi	r11, r0, TOPHYS(tlb_skip)
+	mts     rtlbx, r11		/* r11 is used from TLB0 */
 
 	mts	rtlblo,r4		/* Load the data portion of the entry */
 	mts	rtlbhi,r3		/* Load the tag portion of the entry */
 
+	/* Increase tlb_skip size */
+	addik	r11, r11, 1
+	swi	r11, r0, TOPHYS(tlb_skip)
+
+jump_over2:
 	/*
 	 * Load a TLB entry for LMB, since we need access to
 	 * the exception vectors, using a 4k real==virtual mapping.
 	 */
-	ori	r6,r0,3			/* TLB slot 3 */
+	/* Use temporary TLB_ID for LMB - clear this temporary mapping later */
+	ori	r6, r0, MICROBLAZE_LMB_TLB_ID
 	mts     rtlbx,r6
 
 	ori	r4,r0,(TLB_WR | TLB_EX)
@@ -238,8 +338,8 @@ start_here:
 	 * Please see $(ARCH)/mach-$(SUBARCH)/setup.c for
 	 * the function.
 	 */
-	addik	r9, r0, machine_early_init
-	brald	r15, r9
+	addik	r11, r0, machine_early_init
+	brald	r15, r11
 	nop
 
 #ifndef CONFIG_MMU
@@ -268,8 +368,7 @@ start_here:
 
 	/* Load up the kernel context */
 kernel_load_context:
-	# Keep entry 0 and 1 valid. Entry 3 mapped to LMB can go away.
-	ori	r5,r0,3
+	ori	r5, r0, MICROBLAZE_LMB_TLB_ID
 	mts     rtlbx,r5
 	nop
 	mts	rtlbhi,r0
diff --git a/arch/microblaze/kernel/hw_exception_handler.S b/arch/microblaze/kernel/hw_exception_handler.S
index e62be83..aa510f4 100644
--- a/arch/microblaze/kernel/hw_exception_handler.S
+++ b/arch/microblaze/kernel/hw_exception_handler.S
@@ -820,19 +820,26 @@ ex_handler_done:
  *	Upon exit, we reload everything and RFI.
  * A common place to load the TLB.
  */
+.section .data
+.align 4
+.global tlb_skip
+	tlb_skip:
+		.long	MICROBLAZE_TLB_SKIP
 	tlb_index:
-		.long	1 /* MS: storing last used tlb index */
+		/* MS: storing last used tlb index */
+		.long	MICROBLAZE_TLB_SIZE/2
+.previous
 	finish_tlb_load:
 		/* MS: load the last used TLB index. */
 		lwi	r5, r0, TOPHYS(tlb_index)
 		addik	r5, r5, 1 /* MS: inc tlb_index -> use next one */
 
 /* MS: FIXME this is potential fault, because this is mask not count */
-		andi	r5, r5, (MICROBLAZE_TLB_SIZE-1)
+		andi	r5, r5, MICROBLAZE_TLB_SIZE - 1
 		ori	r6, r0, 1
 		cmp	r31, r5, r6
 		blti	r31, ex12
-		addik	r5, r6, 1
+		lwi	r5, r0, TOPHYS(tlb_skip)
 	ex12:
 		/* MS: save back current TLB index */
 		swi	r5, r0, TOPHYS(tlb_index)
diff --git a/arch/microblaze/kernel/intc.c b/arch/microblaze/kernel/intc.c
index ad12067..6c54d4d 100644
--- a/arch/microblaze/kernel/intc.c
+++ b/arch/microblaze/kernel/intc.c
@@ -151,8 +151,8 @@ void __init init_IRQ(void)
 #ifdef CONFIG_SELFMOD_INTC
 	selfmod_function((int *) arr_func, intc_baseaddr);
 #endif
-	printk(KERN_INFO "XPS intc #0 at 0x%08x, num_irq=%d, edge=0x%x\n",
-		intc_baseaddr, nr_irq, intr_mask);
+	printk(KERN_INFO "%s #0 at 0x%08x, num_irq=%d, edge=0x%x\n",
+		intc->name, intc_baseaddr, nr_irq, intr_mask);
 
 	/*
 	 * Disable all external interrupts until they are
diff --git a/arch/microblaze/kernel/misc.S b/arch/microblaze/kernel/misc.S
index 206da3d..1dafdde 100644
--- a/arch/microblaze/kernel/misc.S
+++ b/arch/microblaze/kernel/misc.S
@@ -29,16 +29,16 @@
 .type  _tlbia, @function
 .align 4;
 _tlbia:
-	addik	r12, r0, MICROBLAZE_TLB_SIZE - 1 /* flush all entries (63 - 3) */
+	lwi	r12, r0, tlb_skip;
 	/* isync */
 _tlbia_1:
 	mts	rtlbx, r12
 	nop
 	mts	rtlbhi, r0 /* flush: ensure V is clear */
 	nop
-	addik	r11, r12, -2
+	rsubi	r11, r12, MICROBLAZE_TLB_SIZE - 1
 	bneid	r11, _tlbia_1 /* loop for all entries */
-	addik	r12, r12, -1
+	addik	r12, r12, 1
 	/* sync */
 	rtsd	r15, 8
 	nop
@@ -75,7 +75,7 @@ early_console_reg_tlb_alloc:
 	 * Load a TLB entry for the UART, so that microblaze_progress() can use
 	 * the UARTs nice and early.  We use a 4k real==virtual mapping.
 	 */
-	ori	r4, r0, MICROBLAZE_TLB_SIZE - 1
+	lwi	r4, r0, tlb_skip
 	mts	rtlbx, r4 /* TLB slot 63 */
 
 	or	r4,r5,r0
@@ -89,6 +89,11 @@ early_console_reg_tlb_alloc:
 	nop
 	mts	rtlbhi,r5 /* Load the tag portion of the entry */
 	nop
+
+	lwi	r5, r0, tlb_skip
+	addik	r5, r5, 1
+	swi	r5, r0, tlb_skip
+
 	rtsd	r15, 8
 	nop
 
diff --git a/arch/microblaze/kernel/setup.c b/arch/microblaze/kernel/setup.c
index 70e6d0b..9f79fb3 100644
--- a/arch/microblaze/kernel/setup.c
+++ b/arch/microblaze/kernel/setup.c
@@ -95,8 +95,11 @@ inline unsigned get_romfs_len(unsigned *addr)
 }
 #endif	/* CONFIG_MTD_UCLINUX_EBSS */
 
+unsigned long kernel_tlb;
+
 void __init machine_early_init(const char *cmdline, unsigned int ram,
-		unsigned int fdt, unsigned int msr)
+		unsigned int fdt, unsigned int msr, unsigned int tlb0,
+		unsigned int tlb1)
 {
 	unsigned long *src, *dst;
 	unsigned int offset = 0;
@@ -143,6 +146,12 @@ void __init machine_early_init(const char *cmdline, unsigned int ram,
 	setup_early_printk(NULL);
 #endif
 
+	/* setup kernel_tlb after BSS cleaning
+	 * Maybe worth to move to asm code */
+	kernel_tlb = tlb0 + tlb1;
+	/* printk("TLB1 0x%08x, TLB0 0x%08x, tlb 0x%x\n", tlb0,
+							tlb1, kernel_tlb); */
+
 	printk("Ramdisk addr 0x%08x, ", ram);
 	if (fdt)
 		printk("FDT at 0x%08x\n", fdt);
@@ -197,6 +206,19 @@ static int microblaze_debugfs_init(void)
 	return of_debugfs_root == NULL;
 }
 arch_initcall(microblaze_debugfs_init);
+
+static int __init debugfs_tlb(void)
+{
+	struct dentry *d;
+
+	if (!of_debugfs_root)
+		return -ENODEV;
+
+	d = debugfs_create_u32("tlb_skip", S_IRUGO, of_debugfs_root, &tlb_skip);
+	if (!d)
+		return -ENOMEM;
+}
+device_initcall(debugfs_tlb);
 #endif
 
 static int dflt_bus_notify(struct notifier_block *nb,
diff --git a/arch/microblaze/kernel/timer.c b/arch/microblaze/kernel/timer.c
index 3cb0bf6..cadfd56 100644
--- a/arch/microblaze/kernel/timer.c
+++ b/arch/microblaze/kernel/timer.c
@@ -79,7 +79,7 @@ static inline void microblaze_timer0_start_periodic(unsigned long load_val)
 	 * !PWMA - disable pwm
 	 * TINT - clear interrupt status
 	 * ENT- enable timer itself
-	 * EINT - enable interrupt
+	 * ENIT - enable interrupt
 	 * !LOAD - clear the bit to let go
 	 * ARHT - auto reload
 	 * !CAPT - no external trigger
@@ -274,8 +274,8 @@ void __init time_init(void)
 #ifdef CONFIG_SELFMOD_TIMER
 	selfmod_function((int *) arr_func, timer_baseaddr);
 #endif
-	printk(KERN_INFO "XPS timer #0 at 0x%08x, irq=%d\n",
-		timer_baseaddr, irq);
+	printk(KERN_INFO "%s #0 at 0x%08x, irq=%d\n",
+		timer->name, timer_baseaddr, irq);
 
 	/* If there is clock-frequency property than use it */
 	prop = of_get_property(timer, "clock-frequency", NULL);
diff --git a/arch/microblaze/kernel/vmlinux.lds.S b/arch/microblaze/kernel/vmlinux.lds.S
index ac0e1a5..109e9d8 100644
--- a/arch/microblaze/kernel/vmlinux.lds.S
+++ b/arch/microblaze/kernel/vmlinux.lds.S
@@ -44,7 +44,7 @@ SECTIONS {
 	__fdt_blob : AT(ADDR(__fdt_blob) - LOAD_OFFSET) {
 		_fdt_start = . ;		/* place for fdt blob */
 		*(__fdt_blob) ;			/* Any link-placed DTB */
-	        . = _fdt_start + 0x4000;	/* Pad up to 16kbyte */
+	        . = _fdt_start + 0x8000;	/* Pad up to 32kbyte */
 		_fdt_end = . ;
 	}
 
diff --git a/arch/microblaze/mm/Makefile b/arch/microblaze/mm/Makefile
index 09c49ed..7313bd8 100644
--- a/arch/microblaze/mm/Makefile
+++ b/arch/microblaze/mm/Makefile
@@ -5,3 +5,4 @@
 obj-y := consistent.o init.o
 
 obj-$(CONFIG_MMU) += pgtable.o mmu_context.o fault.o
+obj-$(CONFIG_HIGHMEM) += highmem.o
diff --git a/arch/microblaze/mm/highmem.c b/arch/microblaze/mm/highmem.c
new file mode 100644
index 0000000..7d78838
--- /dev/null
+++ b/arch/microblaze/mm/highmem.c
@@ -0,0 +1,88 @@
+/*
+ * highmem.c: virtual kernel memory mappings for high memory
+ *
+ * PowerPC version, stolen from the i386 version.
+ *
+ * Used in CONFIG_HIGHMEM systems for memory pages which
+ * are not addressable by direct kernel virtual addresses.
+ *
+ * Copyright (C) 1999 Gerhard Wichert, Siemens AG
+ *		      Gerhard.Wichert@pdb.siemens.de
+ *
+ *
+ * Redesigned the x86 32-bit VM architecture to deal with
+ * up to 16 Terrabyte physical memory. With current x86 CPUs
+ * we now support up to 64 Gigabytes physical RAM.
+ *
+ * Copyright (C) 1999 Ingo Molnar <mingo@redhat.com>
+ *
+ * Reworked for PowerPC by various contributors. Moved from
+ * highmem.h by Benjamin Herrenschmidt (c) 2009 IBM Corp.
+ */
+
+#include <linux/highmem.h>
+#include <linux/module.h>
+
+/*
+ * The use of kmap_atomic/kunmap_atomic is discouraged - kmap/kunmap
+ * gives a more generic (and caching) interface. But kmap_atomic can
+ * be used in IRQ contexts, so in some (very limited) cases we need
+ * it.
+ */
+#include <asm/tlbflush.h>
+
+void *kmap_atomic_prot(struct page *page, pgprot_t prot)
+{
+
+	unsigned long vaddr;
+	int idx, type;
+
+	/* even !CONFIG_PREEMPT needs this, for in_atomic in do_page_fault */
+	pagefault_disable();
+	if (!PageHighMem(page))
+		return page_address(page);
+
+
+	type = kmap_atomic_idx_push();
+	idx = type + KM_TYPE_NR*smp_processor_id();
+	vaddr = __fix_to_virt(FIX_KMAP_BEGIN + idx);
+#ifdef CONFIG_DEBUG_HIGHMEM
+	BUG_ON(!pte_none(*(kmap_pte-idx)));
+#endif
+	set_pte_at(&init_mm, vaddr, kmap_pte-idx, mk_pte(page, prot));
+	local_flush_tlb_page(NULL, vaddr);
+
+	return (void *) vaddr;
+}
+EXPORT_SYMBOL(kmap_atomic_prot);
+
+void __kunmap_atomic(void *kvaddr)
+{
+	unsigned long vaddr = (unsigned long) kvaddr & PAGE_MASK;
+	int type;
+
+	if (vaddr < __fix_to_virt(FIX_KMAP_END)) {
+		pagefault_enable();
+		return;
+	}
+
+	type = kmap_atomic_idx();
+#ifdef CONFIG_DEBUG_HIGHMEM
+	{
+		unsigned int idx;
+
+		idx = type + KM_TYPE_NR * smp_processor_id();
+		BUG_ON(vaddr != __fix_to_virt(FIX_KMAP_BEGIN + idx));
+
+		/*
+		 * force other mappings to Oops if they'll try to access
+		 * this pte without first remap it
+		 */
+		pte_clear(&init_mm, vaddr, kmap_pte-idx);
+		local_flush_tlb_page(NULL, vaddr);
+	}
+#endif
+	kmap_atomic_idx_pop();
+	pagefault_enable();
+}
+EXPORT_SYMBOL(__kunmap_atomic);
diff --git a/arch/microblaze/mm/init.c b/arch/microblaze/mm/init.c
index 565d193..ce80823 100644
--- a/arch/microblaze/mm/init.c
+++ b/arch/microblaze/mm/init.c
@@ -24,6 +24,7 @@
 #include <asm/pgalloc.h>
 #include <asm/sections.h>
 #include <asm/tlb.h>
+#include <asm/fixmap.h>
 
 /* Use for MMU and noMMU because of PCI generic code */
 int mem_init_done;
@@ -44,9 +45,56 @@ char *klimit = _end;
  */
 unsigned long memory_start;
 EXPORT_SYMBOL(memory_start);
-unsigned long memory_end; /* due to mm/nommu.c */
 unsigned long memory_size;
 EXPORT_SYMBOL(memory_size);
+unsigned long lowmem_size;
+
+#ifdef CONFIG_HIGHMEM
+pte_t *kmap_pte;
+EXPORT_SYMBOL(kmap_pte);
+pgprot_t kmap_prot;
+EXPORT_SYMBOL(kmap_prot);
+
+static inline pte_t *virt_to_kpte(unsigned long vaddr)
+{
+	return pte_offset_kernel(pmd_offset(pgd_offset_k(vaddr),
+			vaddr), vaddr);
+}
+
+static void __init highmem_init(void)
+{
+	pr_debug("%x\n", (u32)PKMAP_BASE);
+	map_page(PKMAP_BASE, 0, 0);	/* XXX gross */
+	pkmap_page_table = virt_to_kpte(PKMAP_BASE);
+
+	kmap_pte = virt_to_kpte(__fix_to_virt(FIX_KMAP_BEGIN));
+	kmap_prot = PAGE_KERNEL;
+}
+
+static unsigned long highmem_setup(void)
+{
+	unsigned long pfn;
+	unsigned long reservedpages = 0;
+
+	for (pfn = max_low_pfn; pfn < max_pfn; ++pfn) {
+		struct page *page = pfn_to_page(pfn);
+
+		/* FIXME not sure about */
+		if (memblock_is_reserved(pfn << PAGE_SHIFT))
+			continue;
+		ClearPageReserved(page);
+		init_page_count(page);
+		__free_page(page);
+		totalhigh_pages++;
+		reservedpages++;
+	}
+	totalram_pages += totalhigh_pages;
+	printk(KERN_INFO "High memory: %luk\n",
+					totalhigh_pages << (PAGE_SHIFT-10));
+
+	return reservedpages;
+}
+#endif /* CONFIG_HIGHMEM */
 
 /*
  * paging_init() sets up the page tables - in fact we've already done this.
@@ -54,17 +102,28 @@ EXPORT_SYMBOL(memory_size);
 static void __init paging_init(void)
 {
 	unsigned long zones_size[MAX_NR_ZONES];
+#ifdef CONFIG_MMU
+	int idx;
+
+	/* Setup fixmaps */
+	for (idx = 0; idx < __end_of_fixed_addresses; idx++)
+		clear_fixmap(idx);
+#endif
 
 	/* Clean every zones */
 	memset(zones_size, 0, sizeof(zones_size));
 
-	/*
-	 * old: we can DMA to/from any address.put all page into ZONE_DMA
-	 * We use only ZONE_NORMAL
-	 */
-	zones_size[ZONE_NORMAL] = max_mapnr;
+#ifdef CONFIG_HIGHMEM
+	highmem_init();
 
-	free_area_init(zones_size);
+	zones_size[ZONE_DMA] = max_low_pfn;
+	zones_size[ZONE_HIGHMEM] = max_pfn;
+#else
+	zones_size[ZONE_DMA] = max_pfn;
+#endif
+
+	/* We don't have holes in memory map */
+	free_area_init_nodes(zones_size);
 }
 
 void __init setup_memory(void)
@@ -78,32 +137,31 @@ void __init setup_memory(void)
 	/* Find main memory where is the kernel */
 	for_each_memblock(memory, reg) {
 		memory_start = (u32)reg->base;
-		memory_end = (u32) reg->base + reg->size;
+		lowmem_size = reg->size;
 		if ((memory_start <= (u32)_text) &&
-					((u32)_text <= memory_end)) {
-			memory_size = memory_end - memory_start;
+			((u32)_text <= (memory_start + lowmem_size - 1))) {
+			memory_size = lowmem_size;
 			PAGE_OFFSET = memory_start;
-			printk(KERN_INFO "%s: Main mem: 0x%x-0x%x, "
+			printk(KERN_INFO "%s: Main mem: 0x%x, "
 				"size 0x%08x\n", __func__, (u32) memory_start,
-					(u32) memory_end, (u32) memory_size);
+					(u32) memory_size);
 			break;
 		}
 	}
 
-	if (!memory_start || !memory_end) {
-		panic("%s: Missing memory setting 0x%08x-0x%08x\n",
-			__func__, (u32) memory_start, (u32) memory_end);
+	if (!memory_start || !memory_size) {
+		panic("%s: Missing memory setting 0x%08x, size=0x%08x\n",
+			__func__, (u32) memory_start, (u32) memory_size);
 	}
 
 	/* reservation of region where is the kernel */
 	kernel_align_start = PAGE_DOWN((u32)_text);
 	/* ALIGN can be remove because _end in vmlinux.lds.S is align */
 	kernel_align_size = PAGE_UP((u32)klimit) - kernel_align_start;
-	memblock_reserve(kernel_align_start, kernel_align_size);
-	printk(KERN_INFO "%s: kernel addr=0x%08x-0x%08x size=0x%08x\n",
+	printk(KERN_INFO "%s: kernel addr:0x%08x-0x%08x size=0x%08x\n",
 		__func__, kernel_align_start, kernel_align_start
 			+ kernel_align_size, kernel_align_size);
-
+	memblock_reserve(kernel_align_start, kernel_align_size);
 #endif
 	/*
 	 * Kernel:
@@ -120,11 +178,13 @@ void __init setup_memory(void)
 	min_low_pfn = memory_start >> PAGE_SHIFT; /* minimum for allocation */
 	/* RAM is assumed contiguous */
 	num_physpages = max_mapnr = memory_size >> PAGE_SHIFT;
-	max_pfn = max_low_pfn = memory_end >> PAGE_SHIFT;
+	max_low_pfn = ((u64)memory_start + (u64)lowmem_size) >> PAGE_SHIFT;
+	max_pfn = ((u64)memory_start + (u64)memory_size) >> PAGE_SHIFT;
 
 	printk(KERN_INFO "%s: max_mapnr: %#lx\n", __func__, max_mapnr);
 	printk(KERN_INFO "%s: min_low_pfn: %#lx\n", __func__, min_low_pfn);
 	printk(KERN_INFO "%s: max_low_pfn: %#lx\n", __func__, max_low_pfn);
+	printk(KERN_INFO "%s: max_pfn: %#lx\n", __func__, max_pfn);
 
 	/*
 	 * Find an area to use for the bootmem bitmap.
@@ -137,15 +197,39 @@ void __init setup_memory(void)
 		PFN_UP(TOPHYS((u32)klimit)), min_low_pfn, max_low_pfn);
 	memblock_reserve(PFN_UP(TOPHYS((u32)klimit)) << PAGE_SHIFT, map_size);
 
+	/* Add active regions with valid PFNs */
+	for_each_memblock(memory, reg) {
+		unsigned long start_pfn, end_pfn;
+
+		start_pfn = memblock_region_memory_base_pfn(reg);
+		end_pfn = memblock_region_memory_end_pfn(reg);
+		memblock_set_node(start_pfn << PAGE_SHIFT,
+					(end_pfn - start_pfn) << PAGE_SHIFT, 0);
+	}
+
 	/* free bootmem is whole main memory */
-	free_bootmem(memory_start, memory_size);
+	free_bootmem_with_active_regions(0, max_low_pfn);
 
 	/* reserve allocate blocks */
 	for_each_memblock(reserved, reg) {
-		pr_debug("reserved - 0x%08x-0x%08x\n",
-			 (u32) reg->base, (u32) reg->size);
-		reserve_bootmem(reg->base, reg->size, BOOTMEM_DEFAULT);
+		unsigned long top = reg->base + reg->size - 1;
+
+		pr_debug("reserved - 0x%08x-0x%08x, %lx, %lx\n",
+			 (u32) reg->base, (u32) reg->size, top,
+						memory_start + lowmem_size - 1);
+
+		if (top <= (memory_start + lowmem_size - 1)) {
+			reserve_bootmem(reg->base, reg->size, BOOTMEM_DEFAULT);
+		} else if (reg->base < (memory_start + lowmem_size - 1)) {
+			unsigned long trunc_size = memory_start + lowmem_size -
+								reg->base;
+			reserve_bootmem(reg->base, trunc_size, BOOTMEM_DEFAULT);
+		}
 	}
+
+	/* XXX need to clip this if using highmem? */
+	sparse_memory_present_with_active_regions(0);
+
 #ifdef CONFIG_MMU
 	init_bootmem_done = 1;
 #endif
@@ -190,13 +274,58 @@ void free_initmem(void)
 
 void __init mem_init(void)
 {
-	high_memory = (void *)__va(memory_end);
+	pg_data_t *pgdat;
+	unsigned long reservedpages = 0, codesize, initsize, datasize, bsssize;
+
+	high_memory = (void *)__va(memory_start + lowmem_size - 1);
+
 	/* this will put all memory onto the freelists */
 	totalram_pages += free_all_bootmem();
 
-	printk(KERN_INFO "Memory: %luk/%luk available\n",
-	       nr_free_pages() << (PAGE_SHIFT-10),
-	       num_physpages << (PAGE_SHIFT-10));
+	for_each_online_pgdat(pgdat) {
+		unsigned long i;
+		struct page *page;
+
+		for (i = 0; i < pgdat->node_spanned_pages; i++) {
+			if (!pfn_valid(pgdat->node_start_pfn + i))
+				continue;
+			page = pgdat_page_nr(pgdat, i);
+			if (PageReserved(page))
+				reservedpages++;
+		}
+	}
+
+#ifdef CONFIG_HIGHMEM
+	reservedpages -= highmem_setup();
+#endif
+
+	codesize = (unsigned long)&_sdata - (unsigned long)&_stext;
+	datasize = (unsigned long)&_edata - (unsigned long)&_sdata;
+	initsize = (unsigned long)&__init_end - (unsigned long)&__init_begin;
+	bsssize = (unsigned long)&__bss_stop - (unsigned long)&__bss_start;
+
+	pr_info("Memory: %luk/%luk available (%luk kernel code, "
+		"%luk reserved, %luk data, %luk bss, %luk init)\n",
+		nr_free_pages() << (PAGE_SHIFT-10),
+		num_physpages << (PAGE_SHIFT-10),
+		codesize >> 10,
+		reservedpages << (PAGE_SHIFT-10),
+		datasize >> 10,
+		bsssize >> 10,
+		initsize >> 10);
+
+#ifdef CONFIG_MMU
+	pr_info("Kernel virtual memory layout:\n");
+	pr_info("  * 0x%08lx..0x%08lx  : fixmap\n", FIXADDR_START, FIXADDR_TOP);
+#ifdef CONFIG_HIGHMEM
+	pr_info("  * 0x%08lx..0x%08lx  : highmem PTEs\n",
+		PKMAP_BASE, PKMAP_ADDR(LAST_PKMAP));
+#endif /* CONFIG_HIGHMEM */
+	pr_info("  * 0x%08lx..0x%08lx  : early ioremap\n",
+		ioremap_bot, ioremap_base);
+	pr_info("  * 0x%08lx..0x%08lx  : vmalloc & ioremap\n",
+		(unsigned long)VMALLOC_START, VMALLOC_END);
+#endif
 	mem_init_done = 1;
 }
 
@@ -226,7 +355,6 @@ static void mm_cmdline_setup(void)
 		maxmem = memparse(p, &p);
 		if (maxmem && memory_size > maxmem) {
 			memory_size = maxmem;
-			memory_end = memory_start + memory_size;
 			memblock.memory.regions[0].size = memory_size;
 		}
 	}
@@ -270,15 +398,26 @@ asmlinkage void __init mmu_init(void)
 		machine_restart(NULL);
 	}
 
-	if ((u32) memblock.memory.regions[0].size < 0x1000000) {
-		printk(KERN_EMERG "Memory must be greater than 16MB\n");
+	if ((u32) memblock.memory.regions[0].size < 0x400000) {
+		printk(KERN_EMERG "Memory must be greater than 4MB\n");
+		machine_restart(NULL);
+	}
+
+	if ((u32) memblock.memory.regions[0].size < kernel_tlb) {
+		printk(KERN_EMERG "Kernel size is greater than memory node\n");
 		machine_restart(NULL);
 	}
+
 	/* Find main memory where the kernel is */
 	memory_start = (u32) memblock.memory.regions[0].base;
-	memory_end = (u32) memblock.memory.regions[0].base +
-				(u32) memblock.memory.regions[0].size;
-	memory_size = memory_end - memory_start;
+	lowmem_size = memory_size = (u32) memblock.memory.regions[0].size;
+
+	if (lowmem_size > CONFIG_LOWMEM_SIZE) {
+		lowmem_size = CONFIG_LOWMEM_SIZE;
+#ifndef CONFIG_HIGHMEM
+		memory_size = lowmem_size;
+#endif
+	}
 
 	mm_cmdline_setup(); /* FIXME parse args from command line - not used */
 
@@ -305,15 +444,20 @@ asmlinkage void __init mmu_init(void)
 	/* Map in all of RAM starting at CONFIG_KERNEL_START */
 	mapin_ram();
 
-#ifdef CONFIG_HIGHMEM_START_BOOL
-	ioremap_base = CONFIG_HIGHMEM_START;
+	/* Extend vmalloc and ioremap area as big as possible */
+#ifdef CONFIG_HIGHMEM
+	ioremap_base = ioremap_bot = PKMAP_BASE;
 #else
-	ioremap_base = 0xfe000000UL;	/* for now, could be 0xfffff000 */
-#endif /* CONFIG_HIGHMEM_START_BOOL */
-	ioremap_bot = ioremap_base;
+	ioremap_base = ioremap_bot = FIXADDR_START;
+#endif
 
 	/* Initialize the context management stuff */
 	mmu_context_init();
+
+	/* Shortly after that, the entire linear mapping will be available */
+	/* This will also cause that unflatten device tree will be allocated
+	 * inside 768MB limit */
+	memblock_set_current_limit(memory_start + lowmem_size - 1);
 }
 
 /* This is only called until mem_init is done. */
@@ -324,11 +468,11 @@ void __init *early_get_page(void)
 		p = alloc_bootmem_pages(PAGE_SIZE);
 	} else {
 		/*
-		 * Mem start + 32MB -> here is limit
+		 * Mem start + kernel_tlb -> here is limit
 		 * because of mem mapping from head.S
 		 */
 		p = __va(memblock_alloc_base(PAGE_SIZE, PAGE_SIZE,
-					memory_start + 0x2000000));
+					memory_start + kernel_tlb));
 	}
 	return p;
 }
diff --git a/arch/microblaze/mm/pgtable.c b/arch/microblaze/mm/pgtable.c
index 59bf233..d1c06d0 100644
--- a/arch/microblaze/mm/pgtable.c
+++ b/arch/microblaze/mm/pgtable.c
@@ -37,6 +37,7 @@
 #include <linux/io.h>
 #include <asm/mmu.h>
 #include <asm/sections.h>
+#include <asm/fixmap.h>
 
 #define flush_HPTE(X, va, pg)	_tlbie(va)
 
@@ -44,11 +45,6 @@ unsigned long ioremap_base;
 unsigned long ioremap_bot;
 EXPORT_SYMBOL(ioremap_bot);
 
-/* The maximum lowmem defaults to 768Mb, but this can be configured to
- * another value.
- */
-#define MAX_LOW_MEM	CONFIG_LOWMEM_SIZE
-
 #ifndef CONFIG_SMP
 struct pgtable_cache_struct quicklists;
 #endif
@@ -80,7 +76,7 @@ static void __iomem *__ioremap(phys_addr_t addr, unsigned long size,
 		!(p >= virt_to_phys((unsigned long)&__bss_stop) &&
 		p < virt_to_phys((unsigned long)__bss_stop))) {
 		printk(KERN_WARNING "__ioremap(): phys addr "PTE_FMT
-			" is RAM lr %p\n", (unsigned long)p,
+			" is RAM lr %pf\n", (unsigned long)p,
 			__builtin_return_address(0));
 		return NULL;
 	}
@@ -171,7 +167,7 @@ void __init mapin_ram(void)
 
 	v = CONFIG_KERNEL_START;
 	p = memory_start;
-	for (s = 0; s < memory_size; s += PAGE_SIZE) {
+	for (s = 0; s < lowmem_size; s += PAGE_SIZE) {
 		f = _PAGE_PRESENT | _PAGE_ACCESSED |
 				_PAGE_SHARED | _PAGE_HWEXEC;
 		if ((char *) v < _stext || (char *) v >= _etext)
@@ -254,3 +250,13 @@ __init_refok pte_t *pte_alloc_one_kernel(struct mm_struct *mm,
 	}
 	return pte;
 }
+
+void __set_fixmap(enum fixed_addresses idx, phys_addr_t phys, pgprot_t flags)
+{
+	unsigned long address = __fix_to_virt(idx);
+
+	if (idx >= __end_of_fixed_addresses)
+		BUG();
+
+	map_page(address, phys, pgprot_val(flags));
+}