significant bug we found was that in ~6 places (IIRC) that the various Makefile and Makefile.settings files were attempting to execute config.py, rather than "python ...config.py". That, plus it was
I ran into eactly the same thing when I pulled in the new buildrom and updated my LinuxBIOS config. This was a LinuxBIOS problem and not a buildrom problem. Its now fixed upstream in linuxBIOS. Those files are autogenerated from our Makefile builder and it wasn't doing the right thing (tm). This however, uncovered another build dependency bug that if you change the target config and adjust ROM_SIZE it dosen't always rebuild linubios.rom so its the correct size. The reason for this is pretty deep in our make chain and I've got to think about it more before I try to fix it. The work around for now is that if you change anything in the target config (Config.1M.lb) or change the target config (see below) you need to run a make clean in the linuxbios build dir to guarentee all those changes are picked up. Also Ron had updated the Config.1M.lb to be the proper size for the SPI flash. But that was going to break everybodys PLCC build when they updated to the tip of LinuxBIOS. So I reverted that change and added a new config file called 'Config.SPI.lb' If youi are gong to build an image thats the correct size for the SPI flash then you need to change 'packages/linuxbios/linuxbios.mk' and edit the config file setting to be correct. Don't forget remove the .configured in the stamps dir and to do a make clean in the work/linuxbios/LinuxbiosV2/targets/olpc/rev_a/rev_a_1M dir. Jordan and Ron: Perhaps the config file needs to be a setting in 'Config.mk'? Buildrom now needs to pull Linuxbios rev 2374 Its way past my bedtime and I'm totally wired on coffee. Lets hope I didn't break anything. -- Richard A. Smith _______________________________________________ Devel mailing list [email protected] http://mailman.laptop.org/mailman/listinfo/devel
