Hello,
Am 04.11.2010 20:20, schrieb Albert ARIBAUD:
If I read that patch (66_*) correctly, it's the same as when using --warn-shared-textrel in LDFLAGS.
All right. I don't like getting a warning like that, but as I cannot reproduce your toolchain, I cannot avoid it.
I think they enabbled this by default to do some kind of QA, because those textrels are forcing the dynamic linker to resolve some stuff which otherwise (without those textrels) isn't needed. And this results in longer load times.
Sidenote: My toolchain is reproducable, but that means you have to compile a Gentoo system, because Gentoo is a source only distribution. So it isn't as comfortable as using binary distributions (here), but it has other advantages.
Regarding the build you sent me, the fixup for monitor_flash_len is present in the fixup table at the 11th entry at 00752fb8 and the fixup table is correctly accessible using _rel_dyn_start_ofs and _rel_dyn_end_ofs. That means your code should have done the fixup if the relocation loop was run correctly.
Can you do a step-by-step run of the fixup loop under a debugger?
Sorry, no. That thing has 2mm-header for the jtag and I don't have any prefabricated cable for that. Maybe someone else with binutils 2.20.1 could check your V3. I don't know if I will find the time test your v3 in the next days with some other hw where I can attach a jtag.
Regards,
Alexander