On 04.04.2008 18:51, Carl-Daniel Hailfinger wrote:
On 04.04.2008 17:57, Stefan Reinauer wrote:
Carl-Daniel Hailfinger wrote:
So either we get this problem solved differently or I'll have to go ahead and check the patch in. Relying on experimental toolchains is not an option. Even if your toolchain provider uses experimental binutils versions, that's nothing we should enforce.
Can you retest with current HEAD? Thanks!
Great! It reduced the number of errors:
AS build/arch/x86/stage0_i586.o /Users/stepan/Devel/coreboot-v3/arch/x86/stage0_i586.S: Assembler messages: /Users/stepan/Devel/coreboot-v3/arch/x86/stage0_i586.S:352: Error: missing ')' /Users/stepan/Devel/coreboot-v3/arch/x86/stage0_i586.S:352: Error: missing ')' /Users/stepan/Devel/coreboot-v3/arch/x86/stage0_i586.S:352: Error: missing ')' /Users/stepan/Devel/coreboot-v3/arch/x86/stage0_i586.S:352: Error: missing ')' make: *** [/Users/stepan/Devel/coreboot-v3/build/arch/x86/stage0_i586.o] Error 1
I think this is getting there.
Strange. The code now should be identical to what we use for v2 AMD CAR.
Actually, I think that any v2 AMD CAR target should fail as well on a machine having problems with this code in v3. Maybe there is a difference after preprocessing somewhere...
I'll see what I can do. Probably some #define needs to be replaced with .set to fix the rest of the bug. I'll report back ASAP.
By the way, why are you working with the _generic_ x86 CAR code in v3 instead of porting the _AMD_ CAR code from v2 to v3?
Do you have any patch which would help me reproduce the issue? The v3 tree currently lacks a target using the code in question.
Regards, Carl-Daniel