diff options
author | Michal Simek <monstr@monstr.eu> | 2010-11-15 09:54:43 +0000 |
---|---|---|
committer | Michal Simek <monstr@monstr.eu> | 2011-10-03 08:01:56 +0200 |
commit | f3090fce77202a867ae23d0df91a2b54ed9ee6e4 (patch) | |
tree | c6a54f4c8d2e64c762ec006d17c0438c818ec12f /arch | |
parent | 86c1b2a86b59f3050413e78bfe40196ac3334c13 (diff) |
microblaze: Save and restore first unused vector
Use one memory space to detect little/big endian platforms.
The first unused address(0x28) is used instead 0x0 address (reset vectors).
Detection rewrited reset vector setup from first stage bootloader.
Workflow:
1. Store 0x28 to r7
2. Do little/big endian test
3. Restore r7 to 0x28
Signed-off-by: Michal Simek <monstr@monstr.eu>
Diffstat (limited to 'arch')
-rw-r--r-- | arch/microblaze/cpu/start.S | 6 |
1 files changed, 4 insertions, 2 deletions
diff --git a/arch/microblaze/cpu/start.S b/arch/microblaze/cpu/start.S index 17c0e287b1b..42104faf9ec 100644 --- a/arch/microblaze/cpu/start.S +++ b/arch/microblaze/cpu/start.S @@ -51,8 +51,10 @@ _start: * 4b) BIG endian - r10 contains 0x0 because 0x2 offset is on addr 0x3 */ addik r6, r0, 0x2 /* BIG/LITTLE endian offset */ - swi r6, r0, 0 - lbui r10, r0, 0 + lwi r7, r0, 0x28 + swi r6, r0, 0x28 /* used first unused MB vector */ + lbui r10, r0, 0x28 /* used first unused MB vector */ + swi r7, r0, 0x28 /* add opcode instruction for 32bit jump - 2 instruction imm & brai */ addi r2, r0, 0xb0000000 /* hex b000 opcode imm */ |