On 11-08-17 13:00, Fabian Arrotin wrote: > On 07/08/17 12:16, Jacco Ligthart wrote: >> Hi Fabian, >> >> >> Here is what I did to make devtools-gcc build on armv5: >> >> https://github.com/redsleeve-linux/el7/blob/master/sclo/devtoolset-6-gcc/0001-patched-for-arm32.patch >> >> I have not tried that yet. please let this list know if you find a >> resolution. >> >> >> Jacco > Hi Jacco, > > I applied your patches , but it still fails (after several hours, it's a > loooooong build). > Any hints ? mock log files are temporary here : > https://armv7.dev.centos.org/repodir/c71708-devtools-6/devtools-6-gcc.armv7hl/ > > Cheers, > > PS : devtoolset-6-binutils and devtoolset-6-gdb build fine, just not > devtoolset-6-gcc (last one needed to be able to build webkitgtk4 and so > unblock failing builds ) > I know it's a long build. I have done it many a time to get to my patch set. The original error you mentioned (7E thingy) is fixed by setting "gcc_target_platform" in the spec file. Everything else in my patch set is to get armv5 compatibility in this gcc version. It is based on work we done earlier on other gcc versions. I think that the errors you see are specific to patches I made, so maybe you need less of this patch. I know there are some items not (completely) implemented for armv5, while they do work for armv7. which might also explain why you need *less* of a patch. I'm guessing though, that all of the places where I added "defined __arm__" should stay in the patch set. Most of them are a discriminator between 32 and 64 bits. I don't have any clearer answer :( Jacco