# 17:34:53 |
ctngbot |
joins #crosstool-ng |
# 17:35:00 |
y_morin |
bhundven: Done! |
# 17:37:04 |
oskude |
quits : Quit: Leaving |
# 18:03:23 |
smartin |
quits : Ping timeout: 246 seconds |
# 18:11:36 |
mingwandroid_ |
quits : Remote host closed the connection |
# 18:12:48 |
smartin |
joins #crosstool-ng |
# 18:31:54 |
memleak |
boom boom dudge... bah doom bah doom DUDGE DUDGE |
# 18:39:06 |
smartin_ |
joins #crosstool-ng |
# 18:51:42 |
y_morin |
quits : Read error: Operation timed out |
# 18:55:14 |
y_morin |
joins #crosstool-ng |
# 18:58:52 |
alan_o |
joins #crosstool-ng |
# 19:16:43 |
mingwandroid |
joins #crosstool-ng |
# 19:27:05 |
codyps |
bhundven: do we not have an override for /bin/sh to always use bash? (Or am I thinking of some other project) |
# 19:29:20 |
oskude |
joins #crosstool-ng |
# 19:29:47 |
mingwandroid |
codyps: There's CT_CONFIG_SHELL |
# 19:31:27 |
mingwandroid |
I had to add two companion libs, gettext for Windows and Darwin and iconv for Windows .. but they take for-ever to compile .. Is there an accepted way to say "use the system version please"? |
# 19:32:18 |
mingwandroid |
All I've got ATM is CT_GETTEXT=y/n .. I was thinking CT_GETTEXT=system, thoughts? |
# 19:54:42 |
bhundven |
y_morin: ? |
# 19:56:47 |
bhundven |
oh, ibot :) |
# 19:58:20 |
memleak |
y_morin is channel op so he can ban people from channel faster |
# 19:58:52 |
bhundven |
yea, I keep running into this problem with "Installing C library headers & start files": configure: error: gcc must provide the header |
# 19:59:12 |
bhundven |
gives memleak a bot snack |
# 19:59:51 |
mingwandroid |
mmm, cogs. |
# 20:28:32 |
memleak |
you mean cigs? |
# 20:28:45 |
memleak |
real men smoke cigars |
# 20:29:20 |
memleak |
hey how up tight is this channel btw? not a lot of people talk so i dont really know the rules around here |
# 20:29:56 |
memleak |
some channels be all like "stop joking around" and others be like "it's been 20 minutes since you last made a joke" |
# 20:36:19 |
mingwandroid |
couldn't really say, friendly but somewhat serious would be my best assessment. |
# 20:36:37 |
memleak |
ok sounds great! |
# 20:37:23 |
mingwandroid |
there's comprehensive logs in the topic though so you can get a feel for it there. |
# 20:40:08 |
memleak |
ah |
# 20:40:38 |
memleak |
so basically anything i say can be used against me in the court of law. |
# 20:41:27 |
memleak |
except crosstool-ng doesn't have a legal thingy because money donations directly to the project won't work |
# 21:30:08 |
oskude |
hmm, where can i change/add the mirrors ct-ng uses to download sources? |
# 21:57:11 |
bhundven |
oskude: path and misc options > Use a mirror (CT_USE_MIRROR) |
# 21:57:57 |
bhundven |
besides that, some of the mirrors are in the build scripts themselves. |
# 21:58:12 |
oskude |
bhundven: roger, thanks |
# 21:58:17 |
bhundven |
np |
# 21:59:37 |
bhundven |
y_morin: I get the same problem with cpuid.h without multilib enabled |
# 22:19:18 |
bhundven |
something up with my system or something. |
# 22:28:26 |
y_morin |
bhundven: Yes, ibot! :-) |
# 22:28:49 |
y_morin |
memleak: What is it about kick/banning people? |
# 22:29:41 |
y_morin |
bhundven: Sorry, did not have a look at the multilib stuff yet. I'm just back from FOSDEM + two Buildroot Developpers Days. I'm pretty exhausted... |
# 22:29:52 |
bhundven |
no problem |
# 22:29:59 |
bhundven |
it's kind of a mess |
# 22:30:35 |
bhundven |
on top of that, I have a new laptop and going through the pains of getting the environment right |
# 22:30:42 |
y_morin |
codyps: If your bash is not in a standard place, you can tell where to find it at ./configure time: ./configure --with-bash=/path/to/bash |
# 22:30:59 |
y_morin |
codyps: Otherwise, as mingwandroid said. |
# 22:32:05 |
bhundven |
y_morin: it was because of a problem I had. I randomly spoke about toolchains being a pain, because I spent 30 minutes hunting down a problem that was dash being the default shell |
# 22:32:13 |
bhundven |
but it was kind of not related to ct-ng |
# 22:32:32 |
bhundven |
it's why off-topic is a pain in irc sometimes. |
# 22:32:36 |
bhundven |
:/ |
# 22:32:58 |
bhundven |
how was fosdem? |
# 22:33:10 |
y_morin |
memleak: Rules for the chan are about the same as in all channels: 1) we are all real people, please act as civili as posible; 2) this is a volunteer effort; 3) this is about crosstool-NG, off-topic is OK, but if it drifts for tool ong, move to another chan. |
# 22:33:30 |
y_morin |
s/tool ong/too long/ |
# 22:34:17 |
y_morin |
bhundven: FOSDEM: great as usual! :-) I was not very enthusiast about the program this year, but it's a personal feeling. But the conferences are not all: you get to meet the people, too! :-) |
# 22:35:14 |
y_morin |
bhundven: Yes, dash as default shell is a pain... |
# 22:35:17 |
mingwandroid |
y_morin: I'd hold off looking at multilib for a bit (of course comments are welcome) |
# 22:35:41 |
y_morin |
mingwandroid: You mean your v3 is not yet final? |
# 22:35:46 |
mingwandroid |
y_morin: but I'm going to wrap my head around mercurial a bit and make a better patchset taking with help from bhundven |
# 22:35:52 |
bhundven |
mingwandroid: after messing with multilib all night, I think there is a better way. |
# 22:35:57 |
mingwandroid |
y_morin: we'll move the bit from codyps into per-arch bits .. |
# 22:36:03 |
mingwandroid |
bhundven: do tell! |
# 22:36:31 |
bhundven |
well, I don't know yet, but I think we need to setup for multilib earlier |
# 22:36:41 |
mingwandroid |
bhundven: I think my patch cleans out some stuff which is good (it clarifies the multi-os vs multi-lib distinction) |
# 22:36:41 |
y_morin |
mingwandroid: I know multilib is a pain. Some archs have some configure options others do not, and it's complex to know what supports what. |
# 22:36:50 |
bhundven |
y_morin: yup |
# 22:37:04 |
bhundven |
the problem is, the paths are not setup properly |
# 22:37:15 |
y_morin |
mingwandroid: what do you mean "multi-os" ? Did you mean "multi-arch" ? |
# 22:37:18 |
bhundven |
I know that's why codyps added the multi_dir to the function |
# 22:37:28 |
mingwandroid |
well, the old stuff uses -print-multi-lib but that's internal to GCC whereas -print-multi-os relates to the sysroot layout |
# 22:37:51 |
mingwandroid |
then there's a load of messing about to account for this, which I was able to remove entirely. |
# 22:38:19 |
y_morin |
mingwandroid: but -print-multi-lib should give you pairs of {options,sub-dir} where to find/install the non-default libs |
# 22:38:22 |
mingwandroid |
so 32/ subdir is an implementation detail of GCC, where it puts it's libraries only. |
# 22:38:37 |
mingwandroid |
y_morin: print-multi-os is the option for that. |
# 22:39:21 |
y_morin |
mingwandroid: gcc -print-multi-os |
# 22:39:21 |
mingwandroid |
gcc -m32 --print-multi-os |
# 22:39:22 |
y_morin |
mingwandroid: gcc: error: unrecognized command line option â-print-multi-osâ |
# 22:39:28 |
mingwandroid |
sorry -- |
# 22:39:31 |
y_morin |
mingwandroid: while -print-multi-lib is OK. |
# 22:39:42 |
y_morin |
Da! |
# 22:40:17 |
mingwandroid |
AFAIUI, you take the compiler flags from -print-multi-lib and throw them through --print-multi-os to get the sysroot subfolder layoyut |
# 22:40:18 |
y_morin |
mingwandroid: OK. But -print-multi-lib returns three configurations, while --print-multi-os only returns one: ../lib |
# 22:40:31 |
y_morin |
mingwandroid: Ah! :-) |
# 22:40:33 |
mingwandroid |
yup, so take the 2nd part of each of the three m32 etc etc. |
# 22:40:35 |
mingwandroid |
yup.. |
# 22:40:40 |
bhundven |
ah |
# 22:40:58 |
y_morin |
hates it when mesages croses... |
# 22:41:03 |
y_morin |
*srosses |
# 22:41:05 |
bhundven |
notes the moment of enlightenment |
# 22:41:09 |
y_morin |
Doh... *croses |
# 22:41:11 |
mingwandroid |
that realization allowed a lot of cleanup. |
# 22:41:26 |
y_morin |
mingwandroid: Yeah! I like the idea! |
# 22:41:42 |
mingwandroid |
they don't make it easy though, also, we must track when that was added to GCC so we can do something fudgy for earlier versions |
# 22:41:42 |
y_morin |
Tss... Bad keybiard, bad... |
# 22:42:05 |
mingwandroid |
bhundven: what was your new thinking? |
# 22:42:05 |
bhundven |
mingwandroid: or just make multilib depend on newer gcc |
# 22:42:11 |
y_morin |
mingwandroid: Too old gcc: no support for multi-lib. Sounds a deal to me if we can simplify the code drastically. |
# 22:42:17 |
y_morin |
bhundven: +1 |
# 22:42:18 |
mingwandroid |
bhundven: indeed, personally, I only care for 4.7 and above. |
# 22:42:24 |
bhundven |
yea |
# 22:43:42 |
bhundven |
well, a lot of paths are setup earlier, and with codyps's patches (that I've refactored into scripts/build/arch) it modifies the tuple to handle the different variants. |
# 22:43:52 |
mingwandroid |
I had a bit of stuff in there for when it doesn't work: https://www.sourceware.org/ml/crossgcc/2014-02/msg00017/0003-_e_glibc__Fix_various_multilib_build_bugs.patch |
# 22:43:53 |
bhundven |
which happens much later |
# 22:44:31 |
oskude |
quits : Quit: Leaving |
# 22:44:31 |
mingwandroid |
bhundven: yes, I'd love to be able to get rid of all the path making that ctng does, and linking lib to lib32 and lib64 in 4 different locations on the filesystem .. |
# 22:44:35 |
bhundven |
it's good stuff, I just think that the paths need to be setup earlier |
# 22:45:00 |
bhundven |
so gcc and others see the paths too |
# 22:45:18 |
codyps |
also on multi-lib: there is a bug with --enable-multilib, the recommended course of action is to just not pass --disable-multilib to enable a multilib gcc (multilib "has been the default for a long time") |
# 22:45:28 |
mingwandroid |
bhundven: ok, for multilib I removed it and it seemed to be fine. |
# 22:45:35 |
bhundven |
codyps: that is in my mq |
# 22:45:36 |
y_morin |
mingwandroid: did you did it on-purpose to have all the lines cleanly aligned? 8-O |
# 22:45:36 |
bhundven |
:) |
# 22:45:56 |
bhundven |
I'll post what I have soon |
# 22:46:05 |
codyps |
great! |
# 22:46:09 |
bhundven |
it's still in your wip state |
# 22:46:14 |
bhundven |
but refactored a bit |
# 22:46:30 |
mingwandroid |
ok sorry. I'll pastebin my latest patch that applies to diorcety btw, I messed up quite badly due to being crap with hg and in a rush! |
# 22:46:51 |
y_morin |
mingwandroid: Agreed about the lib32/lib64/lib symlinks. They are a mess. |
# 22:47:05 |
y_morin |
mingwandroid: I'd welcome such a cleanup! |
# 22:47:15 |
mingwandroid |
I'm not convinced anything needs them you know, certainly multilib doesn't. |
# 22:47:21 |
y_morin |
really has to find time to migrate to git... |
# 22:47:36 |
bhundven |
y_morin: awh, I actually love mq |
# 22:47:46 |
bhundven |
haha |
# 22:47:51 |
y_morin |
mingwandroid: Well, bakc in the day, I did need to add them, or something would break (but can't remember what, t'was a long time ago) |
# 22:48:02 |
bhundven |
git rocks though |
# 22:48:19 |
mingwandroid |
y_morin: ok, talkikng to bhundven the big problem with ctng is that the scope is vast and it's hard to know you didn't break things |
# 22:48:25 |
y_morin |
bhundven: When you learn to use "git rebase -i", you'll never come back, I promise! ;-) |
# 22:48:31 |
mingwandroid |
indeed. |
# 22:48:39 |
mingwandroid |
http://pastebin.kde.org/pvpiuwd4h |
# 22:48:56 |
bhundven |
mingwandroid: yes, I broke so much stuff with static toolchain |
# 22:49:41 |
y_morin |
mingwandroid: the 'die' function does not exist. We have CT_Abort instead |
# 22:50:18 |
mingwandroid |
ok, bhundven is refactoring that bit |
# 22:50:29 |
mingwandroid |
look at line 183 |
# 22:50:41 |
mingwandroid |
that's the block I was able to get rid of with --print-multi-os |
# 22:50:50 |
y_morin |
mingwandroid, bhundven: I think you also need to test all the cases, since you could have more than one flag, no? |
# 22:50:56 |
bhundven |
y_morin: yea, I'm gonna post a new patch queue here soon, thats a bit more focused |
# 22:51:11 |
bhundven |
y_morin: yup. |
# 22:51:18 |
mingwandroid |
good riddence. |
# 22:51:28 |
bhundven |
needs new hardware |
# 22:51:32 |
y_morin |
mingwandroid, bhundven: Since we use a decent bash, you can use: case "sdfsd" in test1) list1 ;;& test2) list2;;& esa |
# 22:51:49 |
mingwandroid |
also check line 375, that's where I stopped doing all the ln -sf for lib32 etc etc, when multilib |
# 22:51:57 |
y_morin |
mingwandroid, bhundven: the ';;&' means: continue testing. |
# 22:52:24 |
mingwandroid |
y_morin: define "decent bash" >= 3.0? |
# 22:52:35 |
mingwandroid |
y_morin: as OSX is still on 3.2 or thereaboutes |
# 22:52:39 |
mingwandroid |
thereabouts. |
# 22:53:24 |
y_morin |
mingwandroid: Yes, bash 3.2 should be OK. |
# 22:53:40 |
mingwandroid |
y_morin: you are right there could be multiple flags, support neon etc.. |
# 22:53:50 |
y_morin |
Just check the man page (if OS-X stil has a man program, that is! :-] ) |
# 22:54:14 |
mingwandroid |
y_morin: I was tempted to port Pacman to it since we got that on Windows and Linux. |
# 22:54:48 |
bhundven |
mingwandroid: it would be good to build a table of multilib combinations, and draw a line around the ones we want to initially target. |
# 22:54:54 |
y_morin |
mingwandroid: Oh no... ;;& is a bash4ism... |
# 22:55:01 |
bhundven |
isn't there also multiarch? |
# 22:55:14 |
y_morin |
bhundven: No, please, no. NOT multiarch, please... |
# 22:55:18 |
bhundven |
lol |
# 22:55:19 |
mingwandroid |
debian came up with that. |
# 22:55:26 |
bhundven |
looked down the rabbit hole |
# 22:55:28 |
y_morin |
bhundven: multilib: OK. multiarch: not OK. |
# 22:55:34 |
mingwandroid |
they decided things weren't complicated enough |
# 22:55:50 |
mingwandroid |
https://wiki.debian.org/Multiarch/Tuples |
# 22:55:56 |
y_morin |
mingwandroid: And in the end, they managed to get their way to upstream that beast... :-( Sight... |
# 22:56:00 |
y_morin |
Sigh... |
# 22:56:15 |
bhundven |
ok, hrm, just built my trusty e500v2 toolchain just fine |
# 22:56:29 |
bhundven |
no cpuid.h header issue there |
# 22:56:44 |
bhundven |
e5500/e6500 is cursed |
# 22:56:46 |
bhundven |
lol |
# 22:56:49 |
y_morin |
mingwandroid: Fact is, when I asked a Debian toolchain guy (Wookey), he was not even capable of telling me what multiarch was even trying to _solve_ in the first place... |
# 22:57:01 |
bhundven |
y_morin: lol |
# 22:57:05 |
mingwandroid |
I'd rather have fat binaries that that nonsense. |
# 22:57:18 |
mingwandroid |
than that |
# 22:57:42 |
bhundven |
op, I spoke too soon. I thought it was on the final gcc. just died in the same spot |
# 22:58:03 |
y_morin |
mingwandroid: In that page, in the paragraph titiled "The Problem", they just explain their solution. No trace of the problem at all... :-( |
# 22:59:09 |
mingwandroid |
it strikes me that they said "the standard names aren't standard enough, lets make a new set of standard names" |
# 22:59:17 |
mingwandroid |
I think xkcd have something for that ;-) |
# 23:00:14 |
y_morin |
mingwandroid: exactly what I was gonna say. Go XKCD! :-) |
# 23:00:34 |
y_morin |
mingwandroid: http://xkcd.com/927/ |
# 23:00:49 |
mingwandroid |
y_morin: yeah, that's the one. |
# 23:01:23 |
bhundven |
hehe |
# 23:01:34 |
y_morin |
mingwandroid: Actually, there *is* one new tuple that makes sense (somewhat): the x86_64-linux-gnux32 (for x32 ABI). |
# 23:02:26 |
mingwandroid |
y_morin: just as we were all getting used to conflating arch with abi.. |
# 23:03:10 |
mingwandroid |
bhundven: y_morin: so is there a github for mercurial or anything like that? or can we just move ctng there and be happy? ;-) |
# 23:03:24 |
y_morin |
mingwandroid: bitbucket.org |
# 23:03:31 |
bhundven |
bitbucket.org supports both |
# 23:03:47 |
bhundven |
and supports free private hosting |
# 23:03:48 |
mingwandroid |
in an interoperable way? (hoping..) |
# 23:03:58 |
y_morin |
mingwandroid: why do you want to migrate the repository off the current host? |
# 23:04:23 |
mingwandroid |
y_morin: I just don't want to learn mercuiral, and would prefer it on github as I'm used to that is all. |
# 23:04:33 |
mingwandroid |
y_morin: no offense meant .. |
# 23:04:58 |
y_morin |
mingwandroid: Problem: currently using Mercurial. github has no Hg support. |
# 23:05:06 |
y_morin |
does not like githuib much, either... |
# 23:05:40 |
y_morin |
mingwandroid: but when we migrate to git, it will still be hosted on crosstool-ng.org anyway. |
# 23:06:05 |
y_morin |
mingwandroid: And I won;t do the github fake pull-request stuff, they are just a pain... |
# 23:06:22 |
bhundven |
so, I'm not finding cpuid.h under .build// |
# 23:06:33 |
mingwandroid |
y_morin: really? I quite like that stuff. |
# 23:06:43 |
bhundven |
after the first gcc build, during headers & start files with glibc |
# 23:06:54 |
y_morin |
mingwandroid: They don;t leave a trail in the mailing list archives, so they are a pain. |
# 23:07:05 |
mingwandroid |
y_morin: have you used Gerrit much? |
# 23:07:33 |
y_morin |
mingwandroid: the list should be the all-complete location to find what happened on the devl process (tentative patches, etc...) |
# 23:07:47 |
y_morin |
mingwandroid: No, I am really fine and happy with git, thank you! ;-) |
# 23:09:23 |
mingwandroid |
y_morin: Gerrit is a just a collaborative frontend-y thing for git to discuss patches, but yes, it also doesn't leave a nice mailing list trail. |
# 23:10:08 |
mingwandroid |
y_morin: I'll wrap my head around Mercurial for now. |
# 23:10:43 |
mingwandroid |
bhundven: in terms of the multilib stuff, how do we proceed? You're going to put up a patch? |
# 23:10:46 |
y_morin |
mingwandroid: Also, I don't like that it is a web-stuff... |
# 23:11:05 |
bhundven |
mingwandroid: well, I need to get past my #firstworldproblem of even building any toolchain |
# 23:11:15 |
bhundven |
>.< |
# 23:11:32 |
y_morin |
OK, guys, I've largely past the time I should have gone to bed. I'll let you make plans for this multib stuff. See ya! :-) |
# 23:11:33 |
mingwandroid |
bhundven: right. I've got a bad problem on Windows with our fancy pants MSYS2 thing .. gnumake segfaulting randomly. |
# 23:11:40 |
bhundven |
forgot that I had my mq applied, so I just poped that and trying again |
# 23:11:43 |
bhundven |
l8r y_morin |
# 23:11:45 |
y_morin |
quits : Quit: Nighty Night! |
# 23:12:38 |
mingwandroid |
bhundven: is mq basically like local branches then? |
# 23:12:47 |
bhundven |
it's a local patch queue |
# 23:12:49 |
mingwandroid |
you have a number of patch queues? |
# 23:13:03 |
mingwandroid |
so I could import codys and yours and also have my own? |
# 23:13:11 |
bhundven |
well, I think you can only have one patch queue, I think. |
# 23:13:27 |
bhundven |
redundant, on purpose |
# 23:13:49 |
mingwandroid |
I guess you can have multiple complete repos if you want to fake it, crosstool-ng.bhundven etc. |
# 23:14:05 |
bhundven |
but I've only used it to queue up and prepare patches for patchbomb (hg email) |
# 23:14:31 |
mingwandroid |
there's an ordering/indexing thing though? so you can drop some in and out that way I guess? |
# 23:17:31 |
bhundven |
kinda, yea |
# 23:18:05 |
bhundven |
I should make some scripts to do mq management stuff |
# 23:18:15 |
bhundven |
but, seeming that things are going the git route... |
# 23:18:35 |
mingwandroid |
yeah, but I think that's been talked about for a long time now :-( |
# 23:18:47 |
bhundven |
yea, it's been beat to death here |
# 23:19:02 |
mingwandroid |
I'm figuring that I'll be learning Mercurial just for crosstool-ng to be honest! |
# 23:19:29 |
bhundven |
I use it a lot |
# 23:19:30 |
smartin_ |
quits : Quit: leaving |
# 23:19:40 |
bhundven |
I used to have to work with perforce |
# 23:19:45 |
mingwandroid |
due to the things you hack on or through choice? |
# 23:20:09 |
mingwandroid |
People in my industry have a lot of love for perforce generally, but I've never tried it. |
# 23:20:42 |
mingwandroid |
I don't like central server based solutions, and couldn't imagine not using git really! |
# 23:20:55 |
bhundven |
and user sandboxing (Centralized) was a pain, so I'd check out code like normal, and then make a directory a mercurial repo for me to manage until my changes were acceptable, and I'd queue them up in mq so that I could just push a bunch of changes at once. |
# 23:21:43 |
mingwandroid |
right. I guess "git init" would be similar. |
# 23:21:49 |
bhundven |
yea |
# 23:21:56 |
bhundven |
you could use git for the same thing |
# 23:22:08 |
bhundven |
hg is more light weight |
# 23:22:21 |
mingwandroid |
yay! I successfully built Linux targeting compilers on Windows. |
# 23:22:23 |
bhundven |
git is really complex underneath |
# 23:22:51 |
doc2 |
quits : Remote host closed the connection |
# 23:24:20 |
mingwandroid |
building binutils libraries for target, compiling Linux software on Windows seems a bit filthy to me :-) |
# 23:24:35 |
bhundven |
mingwandroid: well, anyways, until I can get my laptop to stop sending my toolchain builds off the rails, I'm stuck |
# 23:25:05 |
mingwandroid |
bhundven: can you fire up a fresh VM? |
# 23:25:19 |
mingwandroid |
bhundven: that's my usual solution when things get too hairy. |
# 23:25:29 |
bhundven |
hehe |
# 23:25:33 |
bhundven |
yea, my computers suck |
# 23:26:58 |
bhundven |
the mac needs to be reinstalled, upgraded to 10.9 and things have been weird to say the least. new hp laptop is an i3 quad core 2.4ghz. 4g ram. |
# 23:27:01 |
mingwandroid |
bhundven: I do most stuff on my laptop too; the screen flashes every few seconds from when I dropped it and badly replaced it, the power cable falls out 5 times an hour and the battery lasts about 2 minutes. |
# 23:27:43 |
mingwandroid |
bhundven: but it creaks along like a trouper running Linux and OSX VMs. |
# 23:27:55 |
mingwandroid |
trooper even. |
# 23:28:32 |
mingwandroid |
bhundven: I'd like a new mac pro, but I'd hate myself if I spent that much on an Apple thing. |
# 23:28:41 |
bhundven |
heh |
# 23:29:11 |
mingwandroid |
10.9 wasn't Apple's best moment I don't think. |
# 23:29:21 |
bhundven |
hopefully I get paid soon, I'd like to get this ibm rd340, go stick it in a rack somewhere, and not worry about building on my laptops anymore. |
# 23:29:35 |
mingwandroid |
you'll need ventilation for that I guess? |
# 23:29:52 |
mingwandroid |
finished!!! yay! |
# 23:30:15 |
bhundven |
I'm in seattle, so I can walk down the street and run into a wall that is probably a building with a data center in it. |
# 23:30:50 |
bhundven |
probably get something cheap at the westin for a 2u one-off |
# 23:31:14 |
bhundven |
ok, now my e500v2 build got past the cpuid.h issue |
# 23:31:33 |
bhundven |
but died in build-libc-final |
# 23:31:42 |
bhundven |
a bug |
# 23:31:43 |
bhundven |
fun |
# 23:31:57 |
bhundven |
"Please submit a full bug report," |
# 23:32:46 |
mingwandroid |
ICE? |
# 23:33:05 |
bhundven |
ira |
# 23:33:28 |
bhundven |
oh |
# 23:33:35 |
bhundven |
unrecognizable insn |
# 23:33:40 |
mingwandroid |
ooh. |
# 23:34:09 |
bhundven |
maybe I need to step back a version or two and try again |
# 23:34:23 |
bhundven |
4.7.2? glibc 2.17? |
# 23:34:25 |
bhundven |
:D |
# 23:34:42 |
bhundven |
currently: 4.8.2, glibc-2.18 |
# 23:35:57 |
bhundven |
just trying to work on multilib, start pulling on the thread that is the sweater on crosstool-ng |
# 23:37:54 |
mingwandroid |
I've been doing 4.8.2 glibc-2.18 on intel (multilib) and arm (arm6) so far .. |
# 23:38:07 |
mingwandroid |
and also eglibc-2.18 and glibc-2.15/2.16. |
# 23:38:26 |
bhundven |
yea, I'm about to do an arm toolchain, maybe it will squeal about something in my setup |
# 23:38:27 |
mingwandroid |
I can switch to trying to get PPC going if it'd help? |
# 23:50:07 |
mingwandroid |
heh, I missed this earlier: |
# 23:50:14 |
mingwandroid |
"y_morin: mingwandroid: did you did it on-purpose to have all the lines cleanly aligned? 8-O" |
# 23:51:14 |
mingwandroid |
damn right I did. I worked and reworked multilib so much I had many gos at the commit msg ;-) |
# 23:55:17 |
Botanic |
mingwandroid, hey I been tryin your binaries and I get ld: library not found for -lcrt1.10.6.o |
# 23:55:33 |
Botanic |
i assume that i have the wrong sdk or something |
# 23:55:38 |
mingwandroid |
Botanic: you've got a broken SDK yes |
# 23:55:50 |
mingwandroid |
or you are not passing -sysroot to ld. |
# 23:55:50 |
Botanic |
which did you use? |
# 23:56:28 |
mingwandroid |
one I got from a Mac with "tar -cjhf MacOSX10.6.sdk.tar.bz2 MacOSX10.6.sdk" |
# 23:56:48 |
mingwandroid |
the h is essential to follow symlinks, problem is Mac SDK has absolute symlinks in there... |
# 23:56:52 |
mingwandroid |
that's dumb. |
# 23:57:13 |
mingwandroid |
well, I can't speak for all Mac SDKs, but the ones I've known. |
# 23:57:17 |
Botanic |
i have the dmg i assume ill need to extract that on a mac firts then |
# 23:57:31 |
mingwandroid |
well, there's tools around but I've not used them in years. |
# 23:57:36 |
mingwandroid |
dmg2img |
# 23:57:47 |
mingwandroid |
hfs+ fs kernel drivers etc. |
# 23:57:47 |
mingwandroid |
cpio |
# 23:57:59 |
mingwandroid |
all the fun of the fair! |
# 23:58:57 |
Botanic |
humm my ld flags were -Wl,-headerpad_max_install_names -Wl,-search_paths_first |
# 23:59:02 |
Botanic |
ill try -sysroot |