//
//  Copyright (c) 2011-2013, ARM Limited. All rights reserved.
//  Copyright (c) 2015, Linaro Limited. All rights reserved.
//
//  This program and the accompanying materials
//  are licensed and made available under the terms and conditions of the BSD License
//  which accompanies this distribution.  The full text of the license may be found at
//  http://opensource.org/licenses/bsd-license.php
//
//  THE PROGRAM IS DISTRIBUTED UNDER THE BSD LICENSE ON AN "AS IS" BASIS,
//  WITHOUT WARRANTIES OR REPRESENTATIONS OF ANY KIND, EITHER EXPRESS OR IMPLIED.
//
//

#include <AsmMacroIoLibV8.h>
#include <Base.h>
#include <Library/PcdLib.h>
#include <AutoGen.h>

.text
.align 3

GCC_ASM_IMPORT(ArmPlatformIsPrimaryCore)
GCC_ASM_IMPORT(ArmReadMpidr)
GCC_ASM_IMPORT(ArmPlatformPeiBootAction)
GCC_ASM_IMPORT(ArmPlatformStackSet)
GCC_ASM_EXPORT(_ModuleEntryPoint)
ASM_GLOBAL ASM_PFX(mSystemMemoryEnd)

StartupAddr:                  .8byte ASM_PFX(CEntryPoint)
ASM_PFX(mSystemMemoryEnd):    .8byte 0

ASM_PFX(_ModuleEntryPoint):
  //
  // We are built as a ET_DYN PIE executable, so we need to process all
  // relative relocations regardless of whether or not we are executing from
  // the same offset we were linked at. This is only possible if we are
  // running from RAM.
  //
  adr   x8, __reloc_base
  adr   x9, __reloc_start
  adr   x10, __reloc_end

.Lreloc_loop:
  cmp   x9, x10
  bhs   .Lreloc_done

  //
  // AArch64 uses the ELF64 RELA format, which means each entry in the
  // relocation table consists of
  //
  //   UINT64 offset          : the relative offset of the value that needs to
  //                            be relocated
  //   UINT64 info            : relocation type and symbol index (the latter is
  //                            not used for R_AARCH64_RELATIVE relocations)
  //   UINT64 addend          : value to be added to the value being relocated
  //
  ldp   x11, x12, [x9], #24   // read offset into x11 and info into x12
  cmp   x12, #0x403           // check info == R_AARCH64_RELATIVE?
  bne   .Lreloc_loop          // not a relative relocation? then skip

  ldr   x12, [x9, #-8]        // read addend into x12
  add   x12, x12, x8          // add reloc base to addend to get relocated value
  str   x12, [x11, x8]        // write relocated value at offset
  b     .Lreloc_loop
.Lreloc_done:

  // Do early platform specific actions
  bl    ASM_PFX(ArmPlatformPeiBootAction)

  // Get ID of this CPU in Multicore system
  bl    ASM_PFX(ArmReadMpidr)
  // Keep a copy of the MpId register value
  mov   x20, x0

// Check if we can install the stack at the top of the System Memory or if we need
// to install the stacks at the bottom of the Firmware Device (case the FD is located
// at the top of the DRAM)
_SetupStackPosition:
  // Compute Top of System Memory
  ldr   x1, PcdGet64 (PcdSystemMemoryBase)
  ldr   x2, PcdGet64 (PcdSystemMemorySize)
  sub   x2, x2, #1
  add   x1, x1, x2      // x1 = SystemMemoryTop = PcdSystemMemoryBase + PcdSystemMemorySize
  adr   x2, mSystemMemoryEnd
  str   x1, [x2]

  // Calculate Top of the Firmware Device
  ldr   x2, PcdGet64 (PcdFdBaseAddress)
  ldr   w3, PcdGet32 (PcdFdSize)
  sub   x3, x3, #1
  add   x3, x3, x2      // x3 = FdTop = PcdFdBaseAddress + PcdFdSize

  // UEFI Memory Size (stacks are allocated in this region)
  LoadConstantToReg (FixedPcdGet32(PcdSystemMemoryUefiRegionSize), x4)

  //
  // Reserve the memory for the UEFI region (contain stacks on its top)
  //

  // Calculate how much space there is between the top of the Firmware and the Top of the System Memory
  subs  x0, x1, x3   // x0 = SystemMemoryTop - FdTop
  b.mi  _SetupStack  // Jump if negative (FdTop > SystemMemoryTop). Case when the PrePi is in XIP memory outside of the DRAM
  cmp   x0, x4
  b.ge  _SetupStack

  // Case the top of stacks is the FdBaseAddress
  mov   x1, x2

_SetupStack:
  // x1 contains the top of the stack (and the UEFI Memory)

  // Because the 'push' instruction is equivalent to 'stmdb' (decrement before), we need to increment
  // one to the top of the stack. We check if incrementing one does not overflow (case of DRAM at the
  // top of the memory space)
  adds  x21, x1, #1
  b.cs  _SetupOverflowStack

_SetupAlignedStack:
  mov   x1, x21
  b     _GetBaseUefiMemory

_SetupOverflowStack:
  // Case memory at the top of the address space. Ensure the top of the stack is EFI_PAGE_SIZE
  // aligned (4KB)
  LoadConstantToReg (EFI_PAGE_MASK, x21)
  and   x21, x21, x1
  sub   x1, x1, x21

_GetBaseUefiMemory:
  // Calculate the Base of the UEFI Memory
  sub   x21, x1, x4

_GetStackBase:
  // r1 = The top of the Mpcore Stacks
  // Stack for the primary core = PrimaryCoreStack
  LoadConstantToReg (FixedPcdGet32(PcdCPUCorePrimaryStackSize), x2)
  sub   x22, x1, x2

  // Stack for the secondary core = Number of Cores - 1
  LoadConstantToReg (FixedPcdGet32(PcdCoreCount), x0)
  sub   x0, x0, #1
  LoadConstantToReg (FixedPcdGet32(PcdCPUCoreSecondaryStackSize), x1)
  mul   x1, x1, x0
  sub   x22, x22, x1

  // x22 = The base of the MpCore Stacks (primary stack & secondary stacks)
  mov   x0, x22
  mov   x1, x20
  //ArmPlatformStackSet(StackBase, MpId, PrimaryStackSize, SecondaryStackSize)
  LoadConstantToReg (FixedPcdGet32(PcdCPUCorePrimaryStackSize), x2)
  LoadConstantToReg (FixedPcdGet32(PcdCPUCoreSecondaryStackSize), x3)
  bl    ASM_PFX(ArmPlatformStackSet)

  // Is it the Primary Core ?
  mov   x0, x10
  bl    ASM_PFX(ArmPlatformIsPrimaryCore)
  cmp   x0, #1
  bne   _PrepareArguments

_PrepareArguments:
  mov   x0, x20
  mov   x1, x21
  mov   x2, x22

  // Move sec startup address into a data register
  // Ensure we're jumping to FV version of the code (not boot remapped alias)
  ldr   x4, StartupAddr

  // Jump to PrePiCore C code
  //    x0 = MpId
  //    x1 = UefiMemoryBase
  //    x2 = StacksBase
  blr   x4

_NeverReturn:
  b _NeverReturn