From: Wolfgang Denk [mailto:[EMAIL PROTECTED] > > NAK. > > Sorry, this doesn't work. We cannot have one board compile > this way, and another one another way - one working with that > tool chain and the other with another tool chain only.
Is there a list of the toolchains that -mrelocatable must be tested against for each board? > The problem you're addressing is a bug, and not only on this > single board, but everywhere. So it must be fixed everywhere. > > Either we add manual relocation to these pointers like > we do elsewhere (that would be my recommendation for a > quick workaround aka bug fix for this release), or we should > find a clean way to get real relocation working (that would > be much better, but probably it is too late for this release). I know of pointers in structures in cmd_mii.c and image.c that need relocation, but how many more are there? And where should the manual relocation code reside? I think the effort should be applied to making sure the toolchain can work with relocatable instead. In my testing of this patch, I found a crash with the Radeon driver because the x86emu code forces entries into got2 - which get double relocated when using -mrelocatedable. See: [PATCH 1/2] Make mpc8572ds, mpc8544ds, mpc8536ds relocatable [PATCH 2/2] Leave x86emu op code tables in default section Here: http://thread.gmane.org/gmane.comp.boot-loaders.u-boot/48102 For a new version of this patch. -Ed _______________________________________________ U-Boot mailing list U-Boot@lists.denx.de http://lists.denx.de/mailman/listinfo/u-boot
