Add qemu 2.4.0
[kvmfornfv.git] / qemu / roms / u-boot / arch / powerpc / cpu / mpc86xx / start.S
diff --git a/qemu/roms/u-boot/arch/powerpc/cpu/mpc86xx/start.S b/qemu/roms/u-boot/arch/powerpc/cpu/mpc86xx/start.S
new file mode 100644 (file)
index 0000000..ec5f4a7
--- /dev/null
@@ -0,0 +1,993 @@
+/*
+ * Copyright 2004, 2007, 2011 Freescale Semiconductor.
+ * Srikanth Srinivasan <srikanth.srinivaan@freescale.com>
+ *
+ * SPDX-License-Identifier:    GPL-2.0+
+ */
+
+/*  U-Boot - Startup Code for 86xx PowerPC based Embedded Boards
+ *
+ *
+ *  The processor starts at 0xfff00100 and the code is executed
+ *  from flash. The code is organized to be at an other address
+ *  in memory, but as long we don't jump around before relocating.
+ *  board_init lies at a quite high address and when the cpu has
+ *  jumped there, everything is ok.
+ */
+#include <asm-offsets.h>
+#include <config.h>
+#include <mpc86xx.h>
+#include <version.h>
+
+#include <ppc_asm.tmpl>
+#include <ppc_defs.h>
+
+#include <asm/cache.h>
+#include <asm/mmu.h>
+#include <asm/u-boot.h>
+
+/*
+ * Need MSR_DR | MSR_IR enabled to access I/O (printf) in exceptions
+ */
+
+/*
+ * Set up GOT: Global Offset Table
+ *
+ * Use r12 to access the GOT
+ */
+       START_GOT
+       GOT_ENTRY(_GOT2_TABLE_)
+       GOT_ENTRY(_FIXUP_TABLE_)
+
+       GOT_ENTRY(_start)
+       GOT_ENTRY(_start_of_vectors)
+       GOT_ENTRY(_end_of_vectors)
+       GOT_ENTRY(transfer_to_handler)
+
+       GOT_ENTRY(__init_end)
+       GOT_ENTRY(__bss_end)
+       GOT_ENTRY(__bss_start)
+       END_GOT
+
+/*
+ * r3 - 1st arg to board_init(): IMMP pointer
+ * r4 - 2nd arg to board_init(): boot flag
+ */
+       .text
+       .long   0x27051956              /* U-Boot Magic Number */
+       .globl  version_string
+version_string:
+       .ascii U_BOOT_VERSION_STRING, "\0"
+
+       . = EXC_OFF_SYS_RESET
+       .globl  _start
+_start:
+       b       boot_cold
+
+       /* the boot code is located below the exception table */
+
+       .globl  _start_of_vectors
+_start_of_vectors:
+
+/* Machine check */
+       STD_EXCEPTION(0x200, MachineCheck, MachineCheckException)
+
+/* Data Storage exception. */
+       STD_EXCEPTION(0x300, DataStorage, UnknownException)
+
+/* Instruction Storage exception. */
+       STD_EXCEPTION(0x400, InstStorage, UnknownException)
+
+/* External Interrupt exception. */
+       STD_EXCEPTION(0x500, ExtInterrupt, external_interrupt)
+
+/* Alignment exception. */
+       . = 0x600
+Alignment:
+       EXCEPTION_PROLOG(SRR0, SRR1)
+       mfspr   r4,DAR
+       stw     r4,_DAR(r21)
+       mfspr   r5,DSISR
+       stw     r5,_DSISR(r21)
+       addi    r3,r1,STACK_FRAME_OVERHEAD
+       EXC_XFER_TEMPLATE(Alignment, AlignmentException, MSR_KERNEL, COPY_EE)
+
+/* Program check exception */
+       . = 0x700
+ProgramCheck:
+       EXCEPTION_PROLOG(SRR0, SRR1)
+       addi    r3,r1,STACK_FRAME_OVERHEAD
+       EXC_XFER_TEMPLATE(ProgramCheck, ProgramCheckException,
+               MSR_KERNEL, COPY_EE)
+
+       STD_EXCEPTION(0x800, FPUnavailable, UnknownException)
+
+       /* I guess we could implement decrementer, and may have
+        * to someday for timekeeping.
+        */
+       STD_EXCEPTION(0x900, Decrementer, timer_interrupt)
+       STD_EXCEPTION(0xa00, Trap_0a, UnknownException)
+       STD_EXCEPTION(0xb00, Trap_0b, UnknownException)
+       STD_EXCEPTION(0xc00, SystemCall, UnknownException)
+       STD_EXCEPTION(0xd00, SingleStep, UnknownException)
+       STD_EXCEPTION(0xe00, Trap_0e, UnknownException)
+       STD_EXCEPTION(0xf00, Trap_0f, UnknownException)
+       STD_EXCEPTION(0x1000, SoftEmu, SoftEmuException)
+       STD_EXCEPTION(0x1100, InstructionTLBMiss, UnknownException)
+       STD_EXCEPTION(0x1200, DataTLBMiss, UnknownException)
+       STD_EXCEPTION(0x1300, InstructionTLBError, UnknownException)
+       STD_EXCEPTION(0x1400, DataTLBError, UnknownException)
+       STD_EXCEPTION(0x1500, Reserved5, UnknownException)
+       STD_EXCEPTION(0x1600, Reserved6, UnknownException)
+       STD_EXCEPTION(0x1700, Reserved7, UnknownException)
+       STD_EXCEPTION(0x1800, Reserved8, UnknownException)
+       STD_EXCEPTION(0x1900, Reserved9, UnknownException)
+       STD_EXCEPTION(0x1a00, ReservedA, UnknownException)
+       STD_EXCEPTION(0x1b00, ReservedB, UnknownException)
+       STD_EXCEPTION(0x1c00, DataBreakpoint, UnknownException)
+       STD_EXCEPTION(0x1d00, InstructionBreakpoint, UnknownException)
+       STD_EXCEPTION(0x1e00, PeripheralBreakpoint, UnknownException)
+       STD_EXCEPTION(0x1f00, DevPortBreakpoint, UnknownException)
+
+       .globl  _end_of_vectors
+_end_of_vectors:
+
+       . = 0x2000
+
+boot_cold:
+       /*
+        * NOTE: Only Cpu 0 will ever come here.  Other cores go to an
+        * address specified by the BPTR
+        */
+1:
+#ifdef CONFIG_SYS_RAMBOOT
+       /* disable everything */
+       li      r0, 0
+       mtspr   HID0, r0
+       sync
+       mtmsr   0
+#endif
+
+       /* Invalidate BATs */
+       bl      invalidate_bats
+       sync
+       /* Invalidate all of TLB before MMU turn on */
+       bl      clear_tlbs
+       sync
+
+#ifdef CONFIG_SYS_L2
+       /* init the L2 cache */
+       lis     r3, L2_INIT@h
+       ori     r3, r3, L2_INIT@l
+       mtspr   l2cr, r3
+       /* invalidate the L2 cache */
+       bl      l2cache_invalidate
+       sync
+#endif
+
+       /*
+        * Calculate absolute address in FLASH and jump there
+        *------------------------------------------------------*/
+       lis     r3, CONFIG_SYS_MONITOR_BASE_EARLY@h
+       ori     r3, r3, CONFIG_SYS_MONITOR_BASE_EARLY@l
+       addi    r3, r3, in_flash - _start + EXC_OFF_SYS_RESET
+       mtlr    r3
+       blr
+
+in_flash:
+       /* let the C-code set up the rest                       */
+       /*                                                      */
+       /* Be careful to keep code relocatable !                */
+       /*------------------------------------------------------*/
+       /* perform low-level init */
+
+       /* enable extended addressing */
+       bl      enable_ext_addr
+
+       /* setup the bats */
+       bl      early_bats
+
+       /*
+        * Cache must be enabled here for stack-in-cache trick.
+        * This means we need to enable the BATS.
+        * Cache should be turned on after BATs, since by default
+        * everything is write-through.
+        */
+
+       /* enable address translation */
+       mfmsr   r5
+       ori     r5, r5, (MSR_IR | MSR_DR)
+       lis     r3,addr_trans_enabled@h
+       ori     r3, r3, addr_trans_enabled@l
+       mtspr   SPRN_SRR0,r3
+       mtspr   SPRN_SRR1,r5
+       rfi
+
+addr_trans_enabled:
+       /* enable and invalidate the data cache */
+/*     bl      l1dcache_enable */
+       bl      dcache_enable
+       sync
+
+#if 1
+       bl      icache_enable
+#endif
+
+#ifdef CONFIG_SYS_INIT_RAM_LOCK
+       bl      lock_ram_in_cache
+       sync
+#endif
+
+#if (CONFIG_SYS_CCSRBAR_DEFAULT != CONFIG_SYS_CCSRBAR)
+       bl      setup_ccsrbar
+#endif
+
+       /* set up the stack pointer in our newly created
+        * cache-ram (r1) */
+       lis     r1, (CONFIG_SYS_INIT_RAM_ADDR + CONFIG_SYS_GBL_DATA_OFFSET)@h
+       ori     r1, r1, (CONFIG_SYS_INIT_RAM_ADDR + CONFIG_SYS_GBL_DATA_OFFSET)@l
+
+       li      r0, 0           /* Make room for stack frame header and */
+       stwu    r0, -4(r1)      /* clear final stack frame so that      */
+       stwu    r0, -4(r1)      /* stack backtraces terminate cleanly   */
+
+       GET_GOT                 /* initialize GOT access        */
+
+       /* run low-level CPU init code     (from Flash) */
+       bl      cpu_init_f
+       sync
+
+#ifdef RUN_DIAG
+
+       /* Load PX_AUX register address in r4 */
+       lis     r4, PIXIS_BASE@h
+       ori     r4, r4, 0x6
+       /* Load contents of PX_AUX in r3 bits 24 to 31*/
+       lbz     r3, 0(r4)
+
+       /* Mask and obtain the bit in r3 */
+       rlwinm. r3, r3, 0, 24, 24
+       /* If not zero, jump and continue with u-boot */
+       bne     diag_done
+
+       /* Load back contents of PX_AUX in r3 bits 24 to 31 */
+       lbz     r3, 0(r4)
+       /* Set the MSB of the register value */
+       ori     r3, r3, 0x80
+       /* Write value in r3 back to PX_AUX */
+       stb     r3, 0(r4)
+
+       /* Get the address to jump to in r3*/
+       lis     r3, CONFIG_SYS_DIAG_ADDR@h
+       ori     r3, r3, CONFIG_SYS_DIAG_ADDR@l
+
+       /* Load the LR with the branch address */
+       mtlr    r3
+
+       /* Branch to diagnostic */
+       blr
+
+diag_done:
+#endif
+
+/*     bl      l2cache_enable */
+
+       /* run 1st part of board init code (from Flash)   */
+       li      r3, 0           /* clear boot_flag for calling board_init_f */
+       bl      board_init_f
+       sync
+
+       /* NOTREACHED - board_init_f() does not return */
+
+       .globl  invalidate_bats
+invalidate_bats:
+
+       li      r0, 0
+       /* invalidate BATs */
+       mtspr   IBAT0U, r0
+       mtspr   IBAT1U, r0
+       mtspr   IBAT2U, r0
+       mtspr   IBAT3U, r0
+       mtspr   IBAT4U, r0
+       mtspr   IBAT5U, r0
+       mtspr   IBAT6U, r0
+       mtspr   IBAT7U, r0
+
+       isync
+       mtspr   DBAT0U, r0
+       mtspr   DBAT1U, r0
+       mtspr   DBAT2U, r0
+       mtspr   DBAT3U, r0
+       mtspr   DBAT4U, r0
+       mtspr   DBAT5U, r0
+       mtspr   DBAT6U, r0
+       mtspr   DBAT7U, r0
+
+       isync
+       sync
+       blr
+
+#define CONFIG_BAT_PAIR(n) \
+       lis     r4, CONFIG_SYS_IBAT##n##L@h;            \
+       ori     r4, r4, CONFIG_SYS_IBAT##n##L@l;        \
+       lis     r3, CONFIG_SYS_IBAT##n##U@h;            \
+       ori     r3, r3, CONFIG_SYS_IBAT##n##U@l;        \
+       mtspr   IBAT##n##L, r4;                         \
+       mtspr   IBAT##n##U, r3;                         \
+       lis     r4, CONFIG_SYS_DBAT##n##L@h;            \
+       ori     r4, r4, CONFIG_SYS_DBAT##n##L@l;        \
+       lis     r3, CONFIG_SYS_DBAT##n##U@h;            \
+       ori     r3, r3, CONFIG_SYS_DBAT##n##U@l;        \
+       mtspr   DBAT##n##L, r4;                         \
+       mtspr   DBAT##n##U, r3;
+
+/*
+ * setup_bats:
+ *
+ * Set up the final BAT registers now that setup is done.
+ *
+ * Assumes that:
+ *     1) Address translation is enabled upon entry
+ *     2) The boot rom is still accessible via 1:1 translation
+ */
+       .globl setup_bats
+setup_bats:
+       mflr    r5
+       sync
+
+       /*
+        * When we disable address translation, we will get 1:1 (VA==PA)
+        * translation.  The only place we know for sure is safe for that is
+        * the bootrom where we originally started out.  Pop back into there.
+        */
+       lis     r4, CONFIG_SYS_MONITOR_BASE_EARLY@h
+       ori     r4, r4, CONFIG_SYS_MONITOR_BASE_EARLY@l
+       addi    r4, r4, trans_disabled - _start + EXC_OFF_SYS_RESET
+
+       /* disable address translation */
+       mfmsr   r3
+       rlwinm  r3, r3, 0, 28, 25
+       mtspr   SRR0, r4
+       mtspr   SRR1, r3
+       rfi
+
+trans_disabled:
+#if defined(CONFIG_SYS_DBAT0U) && defined(CONFIG_SYS_DBAT0L) \
+       && defined(CONFIG_SYS_IBAT0U) && defined(CONFIG_SYS_IBAT0L)
+       CONFIG_BAT_PAIR(0)
+#endif
+       CONFIG_BAT_PAIR(1)
+       CONFIG_BAT_PAIR(2)
+       CONFIG_BAT_PAIR(3)
+       CONFIG_BAT_PAIR(4)
+       CONFIG_BAT_PAIR(5)
+       CONFIG_BAT_PAIR(6)
+       CONFIG_BAT_PAIR(7)
+
+       sync
+       isync
+
+       /* Turn translation back on and return */
+       mfmsr   r3
+       ori     r3, r3, (MSR_IR | MSR_DR)
+       mtspr   SPRN_SRR0,r5
+       mtspr   SPRN_SRR1,r3
+       rfi
+
+/*
+ * early_bats:
+ *
+ * Set up bats needed early on - this is usually the BAT for the
+ * stack-in-cache, the Flash, and CCSR space
+ */
+       .globl  early_bats
+early_bats:
+       /* IBAT 3 */
+       lis     r4, CONFIG_SYS_IBAT3L@h
+       ori     r4, r4, CONFIG_SYS_IBAT3L@l
+       lis     r3, CONFIG_SYS_IBAT3U@h
+       ori     r3, r3, CONFIG_SYS_IBAT3U@l
+       mtspr   IBAT3L, r4
+       mtspr   IBAT3U, r3
+       isync
+
+       /* DBAT 3 */
+       lis     r4, CONFIG_SYS_DBAT3L@h
+       ori     r4, r4, CONFIG_SYS_DBAT3L@l
+       lis     r3, CONFIG_SYS_DBAT3U@h
+       ori     r3, r3, CONFIG_SYS_DBAT3U@l
+       mtspr   DBAT3L, r4
+       mtspr   DBAT3U, r3
+       isync
+
+       /* IBAT 5 */
+       lis     r4, CONFIG_SYS_IBAT5L@h
+       ori     r4, r4, CONFIG_SYS_IBAT5L@l
+       lis     r3, CONFIG_SYS_IBAT5U@h
+       ori     r3, r3, CONFIG_SYS_IBAT5U@l
+       mtspr   IBAT5L, r4
+       mtspr   IBAT5U, r3
+       isync
+
+       /* DBAT 5 */
+       lis     r4, CONFIG_SYS_DBAT5L@h
+       ori     r4, r4, CONFIG_SYS_DBAT5L@l
+       lis     r3, CONFIG_SYS_DBAT5U@h
+       ori     r3, r3, CONFIG_SYS_DBAT5U@l
+       mtspr   DBAT5L, r4
+       mtspr   DBAT5U, r3
+       isync
+
+       /* IBAT 6 */
+       lis     r4, CONFIG_SYS_IBAT6L_EARLY@h
+       ori     r4, r4, CONFIG_SYS_IBAT6L_EARLY@l
+       lis     r3, CONFIG_SYS_IBAT6U_EARLY@h
+       ori     r3, r3, CONFIG_SYS_IBAT6U_EARLY@l
+       mtspr   IBAT6L, r4
+       mtspr   IBAT6U, r3
+       isync
+
+       /* DBAT 6 */
+       lis     r4, CONFIG_SYS_DBAT6L_EARLY@h
+       ori     r4, r4, CONFIG_SYS_DBAT6L_EARLY@l
+       lis     r3, CONFIG_SYS_DBAT6U_EARLY@h
+       ori     r3, r3, CONFIG_SYS_DBAT6U_EARLY@l
+       mtspr   DBAT6L, r4
+       mtspr   DBAT6U, r3
+       isync
+
+#if(CONFIG_SYS_CCSRBAR_DEFAULT != CONFIG_SYS_CCSRBAR)
+       /* IBAT 7 */
+       lis     r4, CONFIG_SYS_CCSR_DEFAULT_IBATL@h
+       ori     r4, r4, CONFIG_SYS_CCSR_DEFAULT_IBATL@l
+       lis     r3, CONFIG_SYS_CCSR_DEFAULT_IBATU@h
+       ori     r3, r3, CONFIG_SYS_CCSR_DEFAULT_IBATU@l
+       mtspr   IBAT7L, r4
+       mtspr   IBAT7U, r3
+       isync
+
+       /* DBAT 7 */
+       lis     r4, CONFIG_SYS_CCSR_DEFAULT_DBATL@h
+       ori     r4, r4, CONFIG_SYS_CCSR_DEFAULT_DBATL@l
+       lis     r3, CONFIG_SYS_CCSR_DEFAULT_DBATU@h
+       ori     r3, r3, CONFIG_SYS_CCSR_DEFAULT_DBATU@l
+       mtspr   DBAT7L, r4
+       mtspr   DBAT7U, r3
+       isync
+#endif
+       blr
+
+       .globl clear_tlbs
+clear_tlbs:
+       addis   r3, 0, 0x0000
+       addis   r5, 0, 0x4
+       isync
+tlblp:
+       tlbie   r3
+       sync
+       addi    r3, r3, 0x1000
+       cmp     0, 0, r3, r5
+       blt tlblp
+       blr
+
+       .globl disable_addr_trans
+disable_addr_trans:
+       /* disable address translation */
+       mflr    r4
+       mfmsr   r3
+       andi.   r0, r3, (MSR_IR | MSR_DR)
+       beqlr
+       andc    r3, r3, r0
+       mtspr   SRR0, r4
+       mtspr   SRR1, r3
+       rfi
+
+/*
+ * This code finishes saving the registers to the exception frame
+ * and jumps to the appropriate handler for the exception.
+ * Register r21 is pointer into trap frame, r1 has new stack pointer.
+ */
+       .globl  transfer_to_handler
+transfer_to_handler:
+       stw     r22,_NIP(r21)
+       lis     r22,MSR_POW@h
+       andc    r23,r23,r22
+       stw     r23,_MSR(r21)
+       SAVE_GPR(7, r21)
+       SAVE_4GPRS(8, r21)
+       SAVE_8GPRS(12, r21)
+       SAVE_8GPRS(24, r21)
+       mflr    r23
+       andi.   r24,r23,0x3f00          /* get vector offset */
+       stw     r24,TRAP(r21)
+       li      r22,0
+       stw     r22,RESULT(r21)
+       mtspr   SPRG2,r22               /* r1 is now kernel sp */
+       lwz     r24,0(r23)              /* virtual address of handler */
+       lwz     r23,4(r23)              /* where to go when done */
+       mtspr   SRR0,r24
+       mtspr   SRR1,r20
+       mtlr    r23
+       SYNC
+       rfi                             /* jump to handler, enable MMU */
+
+int_return:
+       mfmsr   r28             /* Disable interrupts */
+       li      r4,0
+       ori     r4,r4,MSR_EE
+       andc    r28,r28,r4
+       SYNC                    /* Some chip revs need this... */
+       mtmsr   r28
+       SYNC
+       lwz     r2,_CTR(r1)
+       lwz     r0,_LINK(r1)
+       mtctr   r2
+       mtlr    r0
+       lwz     r2,_XER(r1)
+       lwz     r0,_CCR(r1)
+       mtspr   XER,r2
+       mtcrf   0xFF,r0
+       REST_10GPRS(3, r1)
+       REST_10GPRS(13, r1)
+       REST_8GPRS(23, r1)
+       REST_GPR(31, r1)
+       lwz     r2,_NIP(r1)     /* Restore environment */
+       lwz     r0,_MSR(r1)
+       mtspr   SRR0,r2
+       mtspr   SRR1,r0
+       lwz     r0,GPR0(r1)
+       lwz     r2,GPR2(r1)
+       lwz     r1,GPR1(r1)
+       SYNC
+       rfi
+
+       .globl  dc_read
+dc_read:
+       blr
+
+       .globl get_pvr
+get_pvr:
+       mfspr   r3, PVR
+       blr
+
+       .globl get_svr
+get_svr:
+       mfspr   r3, SVR
+       blr
+
+
+/*
+ * Function:   in8
+ * Description:        Input 8 bits
+ */
+       .globl  in8
+in8:
+       lbz     r3,0x0000(r3)
+       blr
+
+/*
+ * Function:   out8
+ * Description:        Output 8 bits
+ */
+       .globl  out8
+out8:
+       stb     r4,0x0000(r3)
+       blr
+
+/*
+ * Function:   out16
+ * Description:        Output 16 bits
+ */
+       .globl  out16
+out16:
+       sth     r4,0x0000(r3)
+       blr
+
+/*
+ * Function:   out16r
+ * Description:        Byte reverse and output 16 bits
+ */
+       .globl  out16r
+out16r:
+       sthbrx  r4,r0,r3
+       blr
+
+/*
+ * Function:   out32
+ * Description:        Output 32 bits
+ */
+       .globl  out32
+out32:
+       stw     r4,0x0000(r3)
+       blr
+
+/*
+ * Function:   out32r
+ * Description:        Byte reverse and output 32 bits
+ */
+       .globl  out32r
+out32r:
+       stwbrx  r4,r0,r3
+       blr
+
+/*
+ * Function:   in16
+ * Description:        Input 16 bits
+ */
+       .globl  in16
+in16:
+       lhz     r3,0x0000(r3)
+       blr
+
+/*
+ * Function:   in16r
+ * Description:        Input 16 bits and byte reverse
+ */
+       .globl  in16r
+in16r:
+       lhbrx   r3,r0,r3
+       blr
+
+/*
+ * Function:   in32
+ * Description:        Input 32 bits
+ */
+       .globl  in32
+in32:
+       lwz     3,0x0000(3)
+       blr
+
+/*
+ * Function:   in32r
+ * Description:        Input 32 bits and byte reverse
+ */
+       .globl  in32r
+in32r:
+       lwbrx   r3,r0,r3
+       blr
+
+/*
+ * void relocate_code (addr_sp, gd, addr_moni)
+ *
+ * This "function" does not return, instead it continues in RAM
+ * after relocating the monitor code.
+ *
+ * r3 = dest
+ * r4 = src
+ * r5 = length in bytes
+ * r6 = cachelinesize
+ */
+       .globl  relocate_code
+relocate_code:
+
+       mr      r1,  r3         /* Set new stack pointer                */
+       mr      r9,  r4         /* Save copy of Global Data pointer     */
+       mr      r10, r5         /* Save copy of Destination Address     */
+
+       GET_GOT
+       mr      r3,  r5                         /* Destination Address  */
+       lis     r4, CONFIG_SYS_MONITOR_BASE@h           /* Source      Address  */
+       ori     r4, r4, CONFIG_SYS_MONITOR_BASE@l
+       lwz     r5, GOT(__init_end)
+       sub     r5, r5, r4
+       li      r6, CONFIG_SYS_CACHELINE_SIZE           /* Cache Line Size      */
+
+       /*
+        * Fix GOT pointer:
+        *
+        * New GOT-PTR = (old GOT-PTR - CONFIG_SYS_MONITOR_BASE) + Destination Address
+        *
+        * Offset:
+        */
+       sub     r15, r10, r4
+
+       /* First our own GOT */
+       add     r12, r12, r15
+       /* then the one used by the C code */
+       add     r30, r30, r15
+
+       /*
+        * Now relocate code
+        */
+       cmplw   cr1,r3,r4
+       addi    r0,r5,3
+       srwi.   r0,r0,2
+       beq     cr1,4f          /* In place copy is not necessary       */
+       beq     7f              /* Protect against 0 count              */
+       mtctr   r0
+       bge     cr1,2f
+
+       la      r8,-4(r4)
+       la      r7,-4(r3)
+1:     lwzu    r0,4(r8)
+       stwu    r0,4(r7)
+       bdnz    1b
+       b       4f
+
+2:     slwi    r0,r0,2
+       add     r8,r4,r0
+       add     r7,r3,r0
+3:     lwzu    r0,-4(r8)
+       stwu    r0,-4(r7)
+       bdnz    3b
+/*
+ * Now flush the cache: note that we must start from a cache aligned
+ * address. Otherwise we might miss one cache line.
+ */
+4:     cmpwi   r6,0
+       add     r5,r3,r5
+       beq     7f              /* Always flush prefetch queue in any case */
+       subi    r0,r6,1
+       andc    r3,r3,r0
+       mr      r4,r3
+5:     dcbst   0,r4
+       add     r4,r4,r6
+       cmplw   r4,r5
+       blt     5b
+       sync                    /* Wait for all dcbst to complete on bus */
+       mr      r4,r3
+6:     icbi    0,r4
+       add     r4,r4,r6
+       cmplw   r4,r5
+       blt     6b
+7:     sync                    /* Wait for all icbi to complete on bus */
+       isync
+
+/*
+ * We are done. Do not return, instead branch to second part of board
+ * initialization, now running from RAM.
+ */
+       addi    r0, r10, in_ram - _start + EXC_OFF_SYS_RESET
+       mtlr    r0
+       blr
+
+in_ram:
+       /*
+        * Relocation Function, r12 point to got2+0x8000
+        *
+        * Adjust got2 pointers, no need to check for 0, this code
+        * already puts a few entries in the table.
+        */
+       li      r0,__got2_entries@sectoff@l
+       la      r3,GOT(_GOT2_TABLE_)
+       lwz     r11,GOT(_GOT2_TABLE_)
+       mtctr   r0
+       sub     r11,r3,r11
+       addi    r3,r3,-4
+1:     lwzu    r0,4(r3)
+       cmpwi   r0,0
+       beq-    2f
+       add     r0,r0,r11
+       stw     r0,0(r3)
+2:     bdnz    1b
+
+       /*
+        * Now adjust the fixups and the pointers to the fixups
+        * in case we need to move ourselves again.
+        */
+       li      r0,__fixup_entries@sectoff@l
+       lwz     r3,GOT(_FIXUP_TABLE_)
+       cmpwi   r0,0
+       mtctr   r0
+       addi    r3,r3,-4
+       beq     4f
+3:     lwzu    r4,4(r3)
+       lwzux   r0,r4,r11
+       cmpwi   r0,0
+       add     r0,r0,r11
+       stw     r4,0(r3)
+       beq-    5f
+       stw     r0,0(r4)
+5:     bdnz    3b
+4:
+/* clear_bss: */
+       /*
+        * Now clear BSS segment
+        */
+       lwz     r3,GOT(__bss_start)
+       lwz     r4,GOT(__bss_end)
+
+       cmplw   0, r3, r4
+       beq     6f
+
+       li      r0, 0
+5:
+       stw     r0, 0(r3)
+       addi    r3, r3, 4
+       cmplw   0, r3, r4
+       bne     5b
+6:
+       mr      r3, r9          /* Init Date pointer            */
+       mr      r4, r10         /* Destination Address          */
+       bl      board_init_r
+
+       /* not reached - end relocate_code */
+/*-----------------------------------------------------------------------*/
+
+       /*
+        * Copy exception vector code to low memory
+        *
+        * r3: dest_addr
+        * r7: source address, r8: end address, r9: target address
+        */
+       .globl  trap_init
+trap_init:
+       mflr    r4                      /* save link register           */
+       GET_GOT
+       lwz     r7, GOT(_start)
+       lwz     r8, GOT(_end_of_vectors)
+
+       li      r9, 0x100               /* reset vector always at 0x100 */
+
+       cmplw   0, r7, r8
+       bgelr                           /* return if r7>=r8 - just in case */
+1:
+       lwz     r0, 0(r7)
+       stw     r0, 0(r9)
+       addi    r7, r7, 4
+       addi    r9, r9, 4
+       cmplw   0, r7, r8
+       bne     1b
+
+       /*
+        * relocate `hdlr' and `int_return' entries
+        */
+       li      r7, .L_MachineCheck - _start + EXC_OFF_SYS_RESET
+       li      r8, Alignment - _start + EXC_OFF_SYS_RESET
+2:
+       bl      trap_reloc
+       addi    r7, r7, 0x100           /* next exception vector        */
+       cmplw   0, r7, r8
+       blt     2b
+
+       li      r7, .L_Alignment - _start + EXC_OFF_SYS_RESET
+       bl      trap_reloc
+
+       li      r7, .L_ProgramCheck - _start + EXC_OFF_SYS_RESET
+       bl      trap_reloc
+
+       li      r7, .L_FPUnavailable - _start + EXC_OFF_SYS_RESET
+       li      r8, SystemCall - _start + EXC_OFF_SYS_RESET
+3:
+       bl      trap_reloc
+       addi    r7, r7, 0x100           /* next exception vector        */
+       cmplw   0, r7, r8
+       blt     3b
+
+       li      r7, .L_SingleStep - _start + EXC_OFF_SYS_RESET
+       li      r8, _end_of_vectors - _start + EXC_OFF_SYS_RESET
+4:
+       bl      trap_reloc
+       addi    r7, r7, 0x100           /* next exception vector        */
+       cmplw   0, r7, r8
+       blt     4b
+
+       /* enable execptions from RAM vectors */
+       mfmsr   r7
+       li      r8,MSR_IP
+       andc    r7,r7,r8
+       ori     r7,r7,MSR_ME            /* Enable Machine Check */
+       mtmsr   r7
+
+       mtlr    r4                      /* restore link register        */
+       blr
+
+.globl enable_ext_addr
+enable_ext_addr:
+       mfspr   r0, HID0
+       lis     r0, (HID0_HIGH_BAT_EN | HID0_XBSEN | HID0_XAEN)@h
+       ori     r0, r0, (HID0_HIGH_BAT_EN | HID0_XBSEN | HID0_XAEN)@l
+       mtspr   HID0, r0
+       sync
+       isync
+       blr
+
+#if (CONFIG_SYS_CCSRBAR_DEFAULT != CONFIG_SYS_CCSRBAR)
+.globl setup_ccsrbar
+setup_ccsrbar:
+       /* Special sequence needed to update CCSRBAR itself */
+       lis     r4, CONFIG_SYS_CCSRBAR_DEFAULT@h
+       ori     r4, r4, CONFIG_SYS_CCSRBAR_DEFAULT@l
+
+       lis     r5, CONFIG_SYS_CCSRBAR_PHYS_LOW@h
+       ori     r5, r5, CONFIG_SYS_CCSRBAR_PHYS_LOW@l
+       srwi    r5,r5,12
+       li      r6, CONFIG_SYS_CCSRBAR_PHYS_HIGH@l
+       rlwimi  r5,r6,20,8,11
+       stw     r5, 0(r4) /* Store physical value of CCSR */
+       isync
+
+       lis     r5, CONFIG_SYS_TEXT_BASE@h
+       ori     r5,r5,CONFIG_SYS_TEXT_BASE@l
+       lwz     r5, 0(r5)
+       isync
+
+       /* Use VA of CCSR to do read */
+       lis     r3, CONFIG_SYS_CCSRBAR@h
+       lwz     r5, CONFIG_SYS_CCSRBAR@l(r3)
+       isync
+
+       blr
+#endif
+
+#ifdef CONFIG_SYS_INIT_RAM_LOCK
+lock_ram_in_cache:
+       /* Allocate Initial RAM in data cache.
+        */
+       lis     r3, (CONFIG_SYS_INIT_RAM_ADDR & ~31)@h
+       ori     r3, r3, (CONFIG_SYS_INIT_RAM_ADDR & ~31)@l
+       li      r4, ((CONFIG_SYS_INIT_RAM_SIZE & ~31) + \
+                    (CONFIG_SYS_INIT_RAM_ADDR & 31) + 31) / 32
+       mtctr   r4
+1:
+       dcbz    r0, r3
+       addi    r3, r3, 32
+       bdnz    1b
+#if 1
+/* Lock the data cache */
+       mfspr   r0, HID0
+       ori     r0, r0, 0x1000
+       sync
+       mtspr   HID0, r0
+       sync
+       blr
+#endif
+#if 0
+       /* Lock the first way of the data cache */
+       mfspr   r0, LDSTCR
+       ori     r0, r0, 0x0080
+#if defined(CONFIG_ALTIVEC)
+       dssall
+#endif
+       sync
+       mtspr   LDSTCR, r0
+       sync
+       isync
+       blr
+#endif
+
+.globl unlock_ram_in_cache
+unlock_ram_in_cache:
+       /* invalidate the INIT_RAM section */
+       lis     r3, (CONFIG_SYS_INIT_RAM_ADDR & ~31)@h
+       ori     r3, r3, (CONFIG_SYS_INIT_RAM_ADDR & ~31)@l
+       li      r4, ((CONFIG_SYS_INIT_RAM_SIZE & ~31) + \
+                    (CONFIG_SYS_INIT_RAM_ADDR & 31) + 31) / 32
+       mtctr   r4
+1:     icbi    r0, r3
+       addi    r3, r3, 32
+       bdnz    1b
+       sync                    /* Wait for all icbi to complete on bus */
+       isync
+#if 1
+/* Unlock the data cache and invalidate it */
+       mfspr   r0, HID0
+       li      r3,0x1000
+       andc    r0,r0,r3
+       li      r3,0x0400
+       or      r0,r0,r3
+       sync
+       mtspr   HID0, r0
+       sync
+       blr
+#endif
+#if 0
+       /* Unlock the first way of the data cache */
+       mfspr   r0, LDSTCR
+       li      r3,0x0080
+       andc    r0,r0,r3
+#ifdef CONFIG_ALTIVEC
+       dssall
+#endif
+       sync
+       mtspr   LDSTCR, r0
+       sync
+       isync
+       li      r3,0x0400
+       or      r0,r0,r3
+       sync
+       mtspr   HID0, r0
+       sync
+       blr
+#endif
+#endif