On Sun, Oct 4, 2009 at 8:09 AM, Wolfgang Denk wd@denx.de wrote:

Dear Graeme,

In message d66caabb0910030124x12a9958fl4536b55ed35aaca2@mail.gmail.com you wrote:

...

Did you happen to see this patch set fly past your inbox? (http://thread.gmane.org/gmane.comp.boot-loaders.u-boot/68924)

I've seenit, but had no time yet for a closer look.

...

The patch series is a bit dirty, and I would like to clean it up a bit, but would appreciate some feedback on the approach I have taken first

I'm afraid I don't have the slightest clue how this is supposed to work on x86. I always avoided to do any low-level fork for this architecture.

I don't actually think that the last patch (i.e. the actual relocation patch itself) is specific to x86 (or any other arch) - I actually figured it out from information I found at:

http://www.oit.uci.edu/dcslib/digital_unix/digital-v40d/APS31DTE/DOCU_002.HT...

Which, as you might guess from the URL, is about Digital Unix and not x86 Linux. This was an absolute gold mine of information in relation to ELF relocation. I think as long as U-Boot is linked as an ELF executable, and the necessary symbols are provided in the linker script to survive the strip step, the method I have implemented should be very portable across all architectures. It would be really good if someone(s) could give it a try on other, currently non-relocating, arches

Best regards,

Wolfgang Denk

Regards,

Graeme