--- Log opened Thu Jul 16 00:00:26 2009 03:30 -!- koan [n=koan@unaffiliated/koan] has quit ["Lost terminal"] 03:31 -!- koan [n=koan@unaffiliated/koan] has joined #t2 03:55 -!- koan [n=koan@unaffiliated/koan] has quit ["Lost terminal"] 03:56 -!- koan [n=koan@unaffiliated/koan] has joined #t2 09:11 < rxr> http://rene.rebe.name/2009-07-16/itunes-821-verification-of-apple-devices/ 09:21 < CIA-22> rene * r33622 /trunk/package/base/sudo/sudo.desc: * updated sudo (1.7.1 -> 1.7.2) 09:21 < CIA-22> rene * r33623 /trunk/package/xorg/libxcb/libxcb.desc: * updated libxcb (1.3 -> 1.4) 09:21 < CIA-22> rene * r33624 /trunk/package/xorg/libdrm/libdrm.desc: * updated libdrm (2.4.11 -> 2.4.12) 09:21 < CIA-22> rene * r33625 /trunk/package/x11/emelfm2/emelfm2.desc: * updated emelfm2 (0.6.0 -> 0.6.1) 09:21 < CIA-22> rene * r33626 /trunk/package/x11/xdaliclock/xdaliclock.desc: * updated xdaliclock (2.28 -> 2.29) 09:22 < CIA-22> rene * r33628 /trunk/package/scientific/fftw3/fftw3.desc: * updated fftw3 (3.2.1 -> 3.2.2) 09:22 < CIA-22> rene * r33627 /trunk/package/network/ser2net/ser2net.desc: * updated ser2net (2.5 -> 2.6) 09:23 < CIA-22> rene * r33629 /trunk/package/filesystem/e2fsprogs/e2fsprogs.desc: * updated e2fsprogs (1.41.7 -> 1.41.8) 09:28 < CIA-22> rene * r33630 /trunk/package/x11/dwm/ (X11R7-lib64.patch dwm.desc): * updated dwm (5.0.1 -> 5.6) 15:32 -!- michael-i [n=michael-@141.41.40.153] has joined #t2 15:35 < michael-i> hi, rxr. After concentrating on the actual Linux port of AskoziaPBX we're finally on to the blackfin specific bits. 15:38 < michael-i> The toolchain is working well after updating: uclibc r1633 -> r3342, gcc r1617 -> r3342 and binutils r1613 -> r3342. The only problem remaining is a linking failure where "-melf32bfinfd" ld flags are missing from the kernel (85% sure...) 15:38 < michael-i> Once the kernel is compiling successfully, I could provide a patch which updates the t2-blackfin toolchain with these new versions. 15:44 < rxr> sounds great 15:44 < rxr> hi michael-i 15:46 < michael-i> so, that's what I have to offer :) now I have a question....where can I build in that "-melf32bfinfd" flag? 15:47 < rxr> build in ? 15:47 < michael-i> I've tried for a day with no luck....wanted to surprise you with a patch but I'm stuck on this ld flag. 15:47 < rxr> you mean to add it in the t2 kernel build / or remove it ? 15:48 < rxr> ah - missing 15:48 < rxr> so you want to add it 15:48 < rxr> for one thing, you could patch the kernel to add it to their Makefile 15:48 < rxr> (arch/bfin/Makefile I guess) 15:48 < michael-i> I need the kernel to be linked properly. So far it comes back with: "cannot link fdpic object file into non-fdpic executable" when the kernel is done being compiled and a linking starts 15:49 < rxr> t2 has compiler wrappers, I'm not sure if we have a ld wrapper, maybe we have, one could auto-inject the argument this whose wrappers 15:49 < rxr> var_append GCC_WRAPPER_APPEND (or _INSERT) ' ' "-melf32bfinfd" 15:49 < rxr> for the GCC wrapper we have 15:49 < rxr> if we have an ld wrapper that would be 15:50 < rxr> var_append LD_WRAPPER_APPEND (or _INSERT) ' ' "-melf32bfinfd" 15:50 < rxr> append appends at the end, insert insert the arg at the front of the program commandline 15:50 < michael-i> I haven't found an ld wrapper but have tried that with the GCC wrapper as an experiment. :) 15:50 < rxr> I would need to look up if we have an ld wrapper 15:50 < rxr> ok - I feared we have no ld wrapper :-) 15:50 < rxr> we should add one, I was in the need for one sometime in the past :-) 15:51 < rxr> anyway, the most smooth and upstream communicatable way would be the simple kernel Makefile patch 15:51 < rxr> just add it at the appropriated place in (my guess) linux-.../arch/bfin/Makefile 15:51 < michael-i> I'll look at that first. Once it works, I'll see if I can polish this patch off with a LD_WRAPPER.... 15:52 < michael-i> Thanks for the pointers....that error message "fdpic..." was totally new to me. 15:52 < rxr> why is that needed at all 15:52 < rxr> aren't the ADI guys carefully maintaining their "ever changing" toolchain set including linux kernel 15:52 < rxr> shouldn't that combination just build (tm)? 15:53 < michael-i> I wish that were the case! However, I think they're getting it to "just build" with a few helper scripts outside of the normal make process 15:54 < rxr> autsch! 15:55 < michael-i> It's about as nice as how they manage user programs. They just shove the entire source into svn...no patches, no version numbers....nothing. 15:56 < rxr> yeah - I know why I started to use ROCK Linux some 10y+ which envolved into T2 15:56 < rxr> the situation only changed marginally 15:56 < rxr> in some areas (like this, or the Palm Pre / Google Android source graves) even to the worse 15:57 < michael-i> "Hmmm, we need to support PHP. Ok, extract it into SVN, patch until it works and we're done" :( A nightmare to maintain 16:06 < michael-i> Adding -melf32bfinfd to LDFLAGS_vmlinux in the arch/blackfin/Makefile didn't work. This KBUILD_CFLAGS line makes me think it's being set to -mno-fdpic: 16:06 < michael-i> KBUILD_CFLAGS += $(call cc-option,-mno-fdpic) 16:07 < rxr> maybe exchange that exrpession ? 16:08 < michael-i> Not sure I know what you mean 16:12 < michael-i> new run...everything forced to -mfdpic 16:31 < michael-i> nope...some new error messages but most of them are the same 16:31 < michael-i> yuck 16:43 -!- koan [n=koan@unaffiliated/koan] has quit ["Lost terminal"] 16:43 -!- koan [n=koan@unaffiliated/koan] has joined #t2 16:47 < michael-i> I guess this has to wait until after vacation. I'd hoped for a lucky Thursday :) 17:01 < michael-i> thanks for the hints, rxr 17:02 < michael-i> I'm off for now...be back in 8-10 days to kill this last problem and give you a patch 17:02 -!- michael-i [n=michael-@141.41.40.153] has quit [] 17:10 -!- koan [n=koan@unaffiliated/koan] has quit ["Lost terminal"] 17:11 -!- koan [n=koan@unaffiliated/koan] has joined #t2 20:27 -!- koan [n=koan@unaffiliated/koan] has quit ["Lost terminal"] 20:28 -!- koan [n=koan@unaffiliated/koan] has joined #t2 --- Log closed Fri Jul 17 00:00:26 2009