Re: Porting NetBSD to SGI Fuel?

In article <[email protected]>,
Swift Griggs  <[email protected]> wrote:
>On Wed, 19 Sep 2018, Christos Zoulas wrote:
>>> What is the cause of this problem.
>>> 1) ldscript
>>> 2) compile option
>>> 3) mips64 toolchain
>> I don't know, but objdump --disassemble and readelf are your friends :-)
>
>Christos, do you think it's something that could be done in a MIPS64 
>emulator like qmeu, or would the lack of SGI ROMs blow it?

I would first try to compare the two boot images:
1. compare the two images for sections and load addresses
2. see how the high level disassembly of the two boot images differ

For our image:
1. see if the disassembly of our image matches the assembly of the
relevant code.

>I just wonder how you guys do even attempt such a thing without a 
>debugger/profiler. Sounds like you are saying to use objdump --disassemble 
>to dump some of the ASM and then start through it manually. Wouldn't that 
>require some MIPS ASM wizardry to manually interpret?

Typically I add instructions to print stuff in the screen so I know how
far I got.

>I barely remembered the existence of readelf (I've only used it once). It 
>looks like that would help dump out different parts of the binary 
>metadata, but it doesn't appear to do disassembly. What would one look for 
>in the readelf session? I noticed it's got a way to list out internal 
>symbols and such, which is interesting, but I wonder how it could be used 
>to help with a problem like this.
>
>Basically, I'm an acolyte wondering how the real wizards use these tools.

I have not used a debugger to debug a boot block before... Most of the
stuff I've done manually (printfs and looking at the code). But then again
I did not have access to an emulator...

christos