=== doko_ [n=doko@dslb-088-073-066-174.pools.arcor-ip.net] has joined #ubuntu-toolchain | ||
=== Keybuk [n=scott@217.205.109.249] has joined #ubuntu-toolchain | ||
fabbione | doko_: ping? | 10:40 |
---|---|---|
doko_ | fabbione: pong | 10:41 |
fabbione | doko_: what gcc do you expect for feisty? | 10:41 |
fabbione | i have glibc-2.5 almost done | 10:41 |
doko_ | currently 4.1; 4.2 is not released, and probably will take longer. | 10:44 |
fabbione | doko_: so the same gcc we have now in edgy? | 10:44 |
doko_ | so we have to backport the relevant binutils changes | 10:44 |
fabbione | or are you adding extra patches and stuff like that? | 10:44 |
fabbione | we will need a very new binutils to get hppa back in the game | 10:44 |
fabbione | AFAIK Jeff is looking at it | 10:44 |
doko_ | fabbione: no, extra patches, long double 64->128 bit ABI change on powerpc and sparc | 10:44 |
fabbione | doko_: ok, can you please let me have a source to build before monday? | 10:44 |
doko_ | fabbione: unreleased? an alternative might be the HJ Lu binutils for feisty | 10:44 |
fabbione | or if you have it done.. now | 10:44 |
fabbione | doko_: yes unreleased.. | 10:44 |
fabbione | dunno.. i hate binutils from the bottom of my heart | 10:44 |
fabbione | it freaks me out... | 10:44 |
doko_ | it's not my toy package either | 10:44 |
fabbione | eheh | 10:45 |
fabbione | what i am done up till now is to merge all the packaging bits for glibc. i will need help from Jeff to review debian/patches | 10:45 |
doko_ | I'll prepare gcc either over the weekend or today | 10:45 |
fabbione | and that's basically done | 10:45 |
fabbione | but i want to simulate a bootstrap over all my machines | 10:46 |
fabbione | build gcc with old gcc | 10:46 |
fabbione | build gcc | 10:46 |
fabbione | build kernel (that i know is going to suck) | 10:46 |
fabbione | rebuild glibc and gcc and kernel again for fun and profit | 10:46 |
fabbione | that would be lovely.. thanks | 10:46 |
fabbione | if we can have at least part of the toolchain ready before we open feisty that will be great | 10:47 |
=== mlpug [n=user@a84-231-238-186.elisa-laajakaista.fi] has joined #ubuntu-toolchain | ||
=== Dvalin [i=hawkeye@195.0.160.130] has joined #ubuntu-toolchain | ||
=== rodarvus [n=rodarvus@ubuntu/member/rodarvus] has joined #ubuntu-toolchain | ||
=== jbailey [n=jbailey@modemcable139.249-203-24.mc.videotron.ca] has joined #ubuntu-toolchain | ||
jbailey | fabbione: Yeah, what you said. | 03:46 |
jbailey | =) | 03:46 |
fabbione | ehe | 03:46 |
fabbione | i was saying that my target now is to 2.5 out for opening | 03:47 |
fabbione | then we can cleanup as much as we want | 03:47 |
jbailey | Yup. | 03:47 |
fabbione | afaict the only thing missing now for an upload is debian/patches/ubuntu from 2.4 to 2.5 | 03:47 |
jbailey | I need to merge in ports stuff for hppa/ | 03:47 |
jbailey | And get the TLS patches in for binutils. | 03:47 |
fabbione | and decide if we want NPTL or LT for hppa... | 03:47 |
jbailey | The nice part is that I now feel comfortable enough with BFD to do that. | 03:47 |
jbailey | No LT. | 03:48 |
fabbione | ok | 03:48 |
fabbione | i can fix that easily | 03:48 |
jbailey | Our .orig.tar.gz shouldn't even have the tar.bz2 in there. | 03:48 |
fabbione | well it has for now.. | 03:48 |
jbailey | But just in general. | 03:48 |
fabbione | should we kill it? | 03:48 |
fabbione | yeah | 03:48 |
jbailey | We're not Debian, we don't drag corpses behind us chained to the bumper. | 03:48 |
fabbione | ahah | 03:49 |
fabbione | there are a few entries in debian/changelog marked as (requires testing) or (requires review) | 03:49 |
fabbione | that's because i am not 100% of the reason why that stuff is there | 03:49 |
jbailey | The only thing that I see in there off hand is that some of the changes done directly to control.in/{main,opt} might be best done in sysdeps/depflags.pl | 03:52 |
jbailey | I *hate* sysdeps/depflags.pl | 03:52 |
jbailey | That was about where I ran out of steam redoing the packages for Debian in 2003. | 03:52 |
jbailey | What I think it should be is a tab delimited table or something like that which sits in control.in | 03:53 |
jbailey | Maybe a few different files, conflicts, depends, etc. | 03:54 |
jbailey | ARCH\tPACKAGE\tVERSION | 03:54 |
fabbione | imho we could simplify the packaging a lot with a set of shell scripts | 04:00 |
fabbione | or start eliminating tons of .mk stuff | 04:01 |
jbailey | There's a surprising amount of this that *can't* be simplified once you start to look at what it does. | 04:01 |
jbailey | But certainly half of rules.d/debhelper.mk can go. | 04:01 |
jbailey | I thought quilt provided a rules snippet to include, so I'm not sure why we have our own. | 04:02 |
doko_ | the tarball approach can go again; it was for the binutils-source package, elmo did prefer to have patched sources in this patch, so we can revert that again. | 04:08 |
doko_ | but keeping the binutils-source package | 04:08 |
jbailey | doko_: I've wondered a few times if it's better to use the upstream binutils releases or HJ Lu's snapshots, since his snapshots tend to include extra optimisation instructions and such. | 04:09 |
jbailey | But are still sort of crack-of-the-moment. | 04:10 |
jbailey | With his snapshots, we'dhave the HPPA TLS stuff already. | 04:10 |
jbailey | Otherwise we'll have to backport them to get hppa working. | 04:10 |
doko_ | jbailey: yes, I was thinking about it as well. but then, how switch back to FSF binutils without having any regressions? | 04:13 |
jbailey | Dunno. I've never understood why the binutils release cycle is so slow. | 04:14 |
jbailey | The changes that tend to go in there seem to be consistantly just opcode adds, etc. | 04:15 |
doko_ | if we can track these changes in a separate branch? | 04:17 |
jbailey | I don't think we have the bodies on toolchain to do that sort of work. | 04:18 |
jbailey | Since that drags the QA responsibility for those things onto us. | 04:18 |
jbailey | I think it just wouldn't get done in the end. | 04:18 |
fabbione | jbailey: i am getting a feisty amd64 chroot on ronne. the debs look good | 04:21 |
fabbione | jbailey: at least debdiff did spot only 2 changes.. one i knew about but we don't care (it's ppc only) and the other one it's just an extra file that needs investigation | 04:22 |
jbailey | We don't care about ppc anymore? =/ | 04:34 |
fabbione | it's already fixed in bzr.. i didn't feel restarting the builds for a ppc thingy | 04:35 |
fabbione | but at least i know it is fixed properly | 04:35 |
fabbione | dh_install: libc6-dev missing files (debian/tmp-libc/usr/lib/nptl*), aborting | 04:39 |
fabbione | GRRRRRRRRRRRRRRR | 04:39 |
jbailey | Right, debhelper 5 broke that construct, sadly. | 05:06 |
jbailey | I pouted at Joey for it. | 05:06 |
jbailey | foo* doesn't resolve to empty set in a glob anymore. | 05:06 |
fabbione | i don't think we ship anything with that name anyway | 05:08 |
fabbione | i386/ia64 did fail | 05:09 |
fabbione | amd64 is rebuilding and waiting sparc/ppc | 05:09 |
fabbione | we might as well chop it away if it's not required | 05:10 |
fabbione | (feisty-libc)fabbione@ronne:/lib$ | 05:17 |
fabbione | well.. they seem to work on amd64 at least | 05:17 |
jbailey | whee =) | 05:25 |
fabbione | waiting for all the builds to complete.. then i will respin with that postinst fix and the nptl fix | 05:26 |
fabbione | i am sure locales are broken | 05:26 |
fabbione | because we need to port the patch for belocs and the other search path from 2.4 | 05:27 |
fabbione | but at least we know that we can run | 05:27 |
=== __keybuk [n=scott@217.205.109.249] has joined #ubuntu-toolchain | ||
Dvalin | fabbione: what gcc do you guys use to build kernel? | 06:56 |
fabbione | gcc-4.1 | 06:56 |
Dvalin | okay | 06:56 |
fabbione | Dvalin: doko is our gcc god here | 06:56 |
Dvalin | yes | 06:56 |
Dvalin | that's why I ask in here ;) | 06:56 |
Dvalin | I have this weird issue | 06:57 |
fabbione | yeah but you are building on *mandriva* | 06:57 |
fabbione | you have issues.. no matter : | 06:57 |
fabbione | :P | 06:57 |
Dvalin | with the atyfb driver, I tried recompiling the same old kernel where it was working | 06:57 |
Dvalin | with current toolchain, not what I compiled it at that time | 06:57 |
fabbione | oh you are suspecting a gcc breakage? | 06:58 |
Dvalin | otherwise everything being the same, and problem exist in the same kernel release which previously worked (even with the same config) | 06:58 |
Dvalin | yes | 06:58 |
fabbione | Dvalin: i suggest you try with -O0 and check the asm output | 06:58 |
fabbione | in theory it should be the same | 06:59 |
Dvalin | I'm thinking about trying to build with gcc 4.0.3 which I built last working kernel with.. | 06:59 |
Dvalin | fabbione: okay, good tip, for checking the asm output.. I think you overestimating me ;) | 06:59 |
fabbione | Dvalin: nah.. that was a trick doko handed to me | 06:59 |
Dvalin | checking the asm output? or -O0? | 07:00 |
fabbione | asm output | 07:00 |
fabbione | and -O0 ;) | 07:00 |
fabbione | both | 07:00 |
Dvalin | okay | 07:01 |
Dvalin | I'm not competent at asm stuff.. ;p | 07:01 |
Dvalin | but I did score 100% at a simple asm asisgnment at an exam last spring ;) | 07:02 |
Dvalin | harrharr | 07:02 |
Dvalin | I think that landed me the C in stead of E which otherwise would be the final grade :p | 07:02 |
infinity | You don't need to write asm fluently to be able to compare it. | 07:03 |
Dvalin | ah | 07:03 |
infinity | (Thank god, too, because my i386 asm is less than perfect) | 07:03 |
Dvalin | I see your point :) | 07:03 |
infinity | Useless architectures, like m68k, powerpc, hppa, and sparc. Sure, they're easy. :P | 07:04 |
infinity | i386, not so much. | 07:04 |
Dvalin | hehe | 07:05 |
Dvalin | "useless" | 07:05 |
Dvalin | quite a bombastic statement, no? ;) | 07:05 |
infinity | Yes, that belonged in quotes. I missed adding the irony. :) | 07:05 |
jbailey | infinity: hppa can be annoyingly hard to read. It appears to be able to unable to load 32 bits into a register in a single operation. | 07:28 |
fabbione | jbailey: ia64 glibc are good | 07:29 |
fabbione | if i could only get sparc to complete the testsuite without a kernel crash | 07:29 |
fabbione | jbailey: i am uploading all the debs here: http://people.ubuntu.com/~fabbione/glibc-2.5/ | 07:29 |
jbailey | Joy. | 07:29 |
fabbione | amd64 is good too | 07:29 |
jbailey | Does it crash in a consistant place? | 07:29 |
fabbione | i386 will test tomorrow | 07:29 |
fabbione | it's a kernel problem.. | 07:29 |
fabbione | it lockups on high load | 07:30 |
fabbione | i need to downgrade to edgy kernel | 07:30 |
=== fabbione was running 2.6.19-rc2 | ||
jbailey | Any idea if Ben is going to target 2.6.19 or 2.6.20 for fiesty? | 07:30 |
fabbione | .19 afaik | 07:31 |
fabbione | 9391 fabbione 21 0 5088 448 304 D 300 0.0 9:22.63 ld-linux.so.2 | 07:31 |
fabbione | now i have a process on 400 of CPU | 07:31 |
jbailey | fabbione: Any idea when we're targetting to open the next one? | 07:32 |
jbailey | I remember that edgy had some hitches, openning up. | 07:32 |
fabbione | the kernel is already in git | 07:32 |
fabbione | we need to get a few important things fixed | 07:32 |
fabbione | like that headers issue i was mentioning | 07:32 |
fabbione | otherwise we won't even be able to rebuild the kernel | 07:32 |
jbailey | Headers issue? | 07:32 |
fabbione | yes.. build .19 | 07:33 |
fabbione | install linux-libc-dev | 07:33 |
fabbione | and you are doomed: ) | 07:33 |
fabbione | you can't even rebuild the kernel with that stuff | 07:33 |
=== jbailey knows of ia64 klibc headers and ppc merging ppc and ppc incorrectly. | ||
fabbione | it's totally broken | 07:33 |
jbailey | Err. | 07:33 |
fabbione | no.. this is worst | 07:33 |
jbailey | Kernel builds with -nostdinc. | 07:33 |
jbailey | Nothing in /usr/include should matter in the slighjtest. | 07:33 |
fabbione | not the scripts/* | 07:33 |
fabbione | and you need those | 07:33 |
fabbione | like.. hmmmm modpost? | 07:34 |
jbailey | I don't think I know about those. | 07:34 |
fabbione | you use them all the time.. you just don't know that :) | 07:34 |
jbailey | Can you toss a build failure into a pastebin? | 07:34 |
fabbione | jbailey: sure.. in a minute | 07:34 |
jbailey | No worries. | 07:34 |
fabbione | no actually i can't.. i don't have the deb handy.. | 07:35 |
fabbione | but i can show it here by memory | 07:35 |
fabbione | #include <errno.h> | 07:35 |
fabbione | #include <sys/socket.h> | 07:35 |
fabbione | int main() { | 07:35 |
fabbione | return -EINVAL; | 07:35 |
fabbione | return -SOL_SOCKET; | 07:36 |
fabbione | } | 07:36 |
fabbione | you can't build that one | 07:36 |
jbailey | Right, but I'm curious about the actual failure. | 07:36 |
fabbione | it's an include order issue | 07:37 |
fabbione | either SOL_SOCKET or EINVAL is not defined | 07:37 |
fabbione | if you swap the include's at the top, you get the other error | 07:37 |
doko_ | gcc-4.1_4.1.1-17ubuntu1 is on ronne:~doko/gcc/4.1/ | 07:37 |
fabbione | doko_: cool.. there is a feisty-libc chroot (amd64) on ronne | 07:38 |
fabbione | doko_: | 07:38 |
doko_ | enabling long-double-128 on powerpc and sparc, configuring with --enable-secureplt on powerpc | 07:38 |
fabbione | you can build there if you like | 07:38 |
fabbione | it has glibc-2.5 installed | 07:38 |
fabbione | doko_: i will do ia64/ppc/sparc/i386 tomorrow | 07:38 |
fabbione | i am just too tired today to do it | 07:38 |
fabbione | actually.. i could do ia64 overnight | 07:38 |
fabbione | ii libc6.1 2.5-0ubuntu1 GNU C Library: Shared libraries | 07:39 |
jbailey | doko_: Do you happen to know what binutils FC6 is shipping with? I remember that they have the GNU HASH lookups for the PLT stuff as one of their features. | 07:41 |
jbailey | But I think that requires a new binutils snapshot. | 07:41 |
doko_ | http://cvs.fedora.redhat.com/viewcvs/devel/binutils/ | 07:42 |
doko_ | 2.17.50 | 07:42 |
doko_ | 2.17.50.0.6, the version number smells like beeing the HJ stuff ... | 07:43 |
fabbione | doko_: gcc-4.1 building on ia64 | 07:44 |
fabbione | doko_: do you want to ask B-D on feisty-libc chroot or should I? i can do the others locally | 07:44 |
doko_ | they were installed | 07:45 |
doko_ | currently building on feisty | 07:45 |
fabbione | hmm ok | 07:45 |
fabbione | good | 07:45 |
jbailey | Hmm, here's going to be an interesting question. | 08:15 |
jbailey | Some of the libc tests probably need corresponding kernel functions to pass. | 08:15 |
fabbione | jbailey: ? | 08:15 |
jbailey | What do the buildds run for a kernel> | 08:15 |
fabbione | oh yes.. i already noticed that | 08:15 |
fabbione | 2.6.17 basically everywhere | 08:16 |
fabbione | i saw a bunch of test failures | 08:16 |
fabbione | but i didn't track them down yet | 08:16 |
fabbione | and i don't think for bootstrapping we care too much | 08:16 |
jbailey | If you toss those into a pastebin somewhere, I might be able to offer you opinions on them. | 08:16 |
fabbione | jbailey: ronne:/home/fabbione/{i386,amd64}/ | 08:17 |
fabbione | ppc is on davis | 08:17 |
fabbione | i can upload sparc and ia64 later | 08:17 |
fabbione | davis is still building FYI | 08:18 |
fabbione | oh acutally i did build ia64 with .19 | 08:19 |
jbailey | make[3] : bug-atexit3-lib.cc: Command not found | 08:20 |
jbailey | make[3] : *** [/home/fabbione/i386/glibc-2.5/build-tree/i386-i686/dlfcn/bug-atexi | 08:20 |
jbailey | t3-lib.os] Error 127 | 08:20 |
jbailey | Eh, weird. | 08:20 |
jbailey | Oh, no G++ installed. | 08:20 |
fabbione | scp glibc-2.5.tar.bz2 chinstrap.ubuntu.com:glibc-2.5-ia64-log.tar.bz2 | 08:21 |
fabbione | jbailey: uh? i did use dpkg-buildpackage.. do we miss a B-D ? | 08:22 |
jbailey | checking for g++... g++ | 08:23 |
fabbione | hmmm | 08:23 |
fabbione | i readded the CXX patch from you... | 08:24 |
fabbione | could that be related? | 08:24 |
jbailey | 604-421-2460 604-726-2460 | 08:24 |
jbailey | ls | 08:24 |
jbailey | BAH | 08:24 |
fabbione | 0wn3d | 08:24 |
Dvalin | mm/filemap.c: In function ?__generic_file_aio_write_nolock?: | 08:29 |
Dvalin | mm/filemap.c:1830: sorry, unimplemented: inlining failed in call to ?generic_write_checks?: function body not available | 08:29 |
Dvalin | mm/filemap.c:2133: sorry, unimplemented: called from here | 08:30 |
Dvalin | fabbione: are you sure you actually can compile the kernel with -O0? | 08:30 |
fabbione | Dvalin: i am pretty sure you can | 08:30 |
Dvalin | hmm | 08:31 |
Dvalin | then I don't get why I get that error when compiling with -O0 | 08:31 |
Dvalin | will try again with -O2 to see if it reproduces | 08:31 |
fabbione | i usually compile only the bits i need with -O0 | 08:31 |
Dvalin | okay | 08:31 |
Dvalin | so in other words, you don't know for sure that the *whole* kernel can be compiled with -O0? | 08:32 |
fabbione | no but it should be able to | 08:32 |
Dvalin | soo | 08:32 |
Dvalin | the error I got, a bug in kernel? | 08:32 |
fabbione | or gcc | 08:34 |
fabbione | it really depends what you are experimenting with | 08:35 |
jbailey | fabbione: It's entirely possibly that the kernel relies on optimiser tricks for correct code. | 08:35 |
jbailey | You can't compile glibc at -O0 for instance. | 08:35 |
fabbione | you are working on both at the same times so that makes it impossible to track easily | 08:35 |
Dvalin | fabbione: I got the same error with the "old" compiler | 08:35 |
fabbione | jbailey: it's a possibility.. | 08:35 |
Dvalin | I reverted back to last known good compiler | 08:35 |
Dvalin | 4.0.3 | 08:35 |
Dvalin | which I'm using now | 08:36 |
jbailey | fabbione: You didn't use debuild to build this? | 08:40 |
fabbione | dpkg-buildpackage -rfakeroot -uc -us -b | 08:40 |
fabbione | as i always do | 08:40 |
jbailey | debuild actually logs the whole build (as well as cleans some pieces of the environment) | 08:41 |
jbailey | So just "debuild -uc -us -b" is the equivalent. | 08:41 |
fabbione | hmm ok | 08:41 |
=== mlpug [n=user@a84-231-238-186.elisa-laajakaista.fi] has joined #ubuntu-toolchain | ||
fabbione | we still shouldn't really on debuild.. it should just work... | 08:42 |
fabbione | if we need to sanitize the ENV that should happen in debian/rules | 08:43 |
jbailey | eh, what do you mean? | 08:43 |
fabbione | sbuild doesn't call debuild or dpkg-buildpackage | 08:43 |
jbailey | No - sometimes I set external environment varilables to change what's happening intentionally. | 08:43 |
fabbione | ok.. those are well defined things to trigger certain behavious | 08:43 |
jbailey | Sucks to be sbuild. If the buildds don't use what all the developpers use then I consider the buildd broken. | 08:43 |
fabbione | but i don't set anything in my env other than ccache | 08:43 |
jbailey | Right, I don't know that it's a problem or not. Just that debuild is handy for that. | 08:44 |
jbailey | Mostly what I'm wishing for is the .build file. | 08:44 |
jbailey | The logs that the glibc build produces on its own start logging too late. | 08:44 |
fabbione | jbailey: actually.. no.. i disagree.. according to policy you should be able to call ./debian/rules $target and it should work. All the other tools are "wrappers" for that | 08:44 |
jbailey | So? | 08:44 |
jbailey | the buildd should do what the developpers are doing for maximum reliability. | 08:44 |
jbailey | whether the devs change or whether the buildds change is irrelevant. | 08:45 |
fabbione | if debuild does extra cleaning, then it might fail on the buildd | 08:45 |
fabbione | well i don't use debuild and like you i did build 95% of the biggest packages around without issues... | 08:45 |
fabbione | if you understand what i mean | 08:45 |
jbailey | It's less likely, I think, since the buildds don't tend to have random things in their environment. | 08:45 |
jbailey | Whereas a user running gnome and a pile of crap might. | 08:45 |
fabbione | right, but in my buildd/chroots i don't set random crap either | 08:46 |
fabbione | i know that for a fact of me being sane :) | 08:46 |
jbailey | So you agree, nice. =) | 08:46 |
=== jbailey hids | ||
jbailey | +e | 08:46 |
fabbione | ahah whatever :) | 08:46 |
fabbione | what i consider scary is this: | 08:46 |
fabbione | i have a clean environment | 08:46 |
jbailey | either way, I want a log file of the build more than I care about the environment. | 08:46 |
jbailey | It would've told me how make was called. | 08:46 |
fabbione | jbailey: the sources are stilll on ronne | 08:47 |
fabbione | it doesn't take long to build there | 08:47 |
jbailey | Nope, should be about 20 minutes. | 08:47 |
fabbione | more or less yeah | 08:47 |
jbailey | Which I can do. But I was asking that you consider using debuild since it just always produces a log file. | 08:47 |
fabbione | i will try.. i need to get used to type it :) | 08:48 |
fabbione | oh my wife made dinner! | 08:48 |
fabbione | bbl | 08:48 |
jbailey | enjoy! =) | 08:48 |
fabbione | thanks | 08:48 |
jbailey | bah! no source package! =) | 08:48 |
=== jbailey also talks fabbione into dropping the -b =) | ||
jbailey | echo "CXX = " >> build-tree/i386-libc/configparms | 08:55 |
jbailey | that would be the bug. =) | 08:55 |
jbailey | Need s BUILD_CXX line in debian/rules | 08:57 |
jbailey | Although more I wonder how much we care about the cross-compilation case. | 08:57 |
jbailey | echo "CXX = g++-4.1 -fno-stack-protector" >> build-tree/i386-libc/configpa | 08:58 |
jbailey | Better. | 08:58 |
jbailey | doko_: Any idea if upstream gcc is going to integrate cleanly an option to enable the stack-protector by default? | 08:58 |
fabbione | jbailey: yes.. in ~fabbione/ | 08:59 |
jbailey | fabbione: All good. cp -a was my bitch. =) | 09:00 |
fabbione | jbailey: so are you going to commit the fix? | 09:00 |
jbailey | After I see a succesful test suite on i386, yes. =) | 09:00 |
fabbione | jbailey: also note that the patch pristine from 2.4.. so it have been buggy in edgy too? | 09:01 |
jbailey | I don't think these two tests existed in 2.4. | 09:02 |
fabbione | ok | 09:02 |
jbailey | So the only thing would've been the C++ abi test, we I never paid attention to anyway. | 09:02 |
fabbione | see.. as i said.. harmless errors in the logs | 09:02 |
fabbione | you should have trust me before :P | 09:02 |
jbailey | I didn't say I didn't trust you. =) I merely offered my opinion. | 09:03 |
jbailey | "wow, this is broken. Here's the fix." Easy opinions like that. =) | 09:03 |
fabbione | ahah | 09:04 |
fabbione | yeah yeah | 09:04 |
jbailey | fabbione: There've got to be *some* advantages of staring at this package for 6 years. =) | 09:04 |
fabbione | jbailey: like an extra discount for the super-pack offer from your shrink? | 09:05 |
jbailey | Doesn't help much after the surcharge. | 09:05 |
fabbione | 10 hours at the price of 8 | 09:05 |
jbailey | Something about my reality check bouncing... | 09:06 |
fabbione | and if you are a glibc maintianer 10 hours at the price of 6! | 09:06 |
fabbione | ehhe | 09:06 |
doko_ | jbailey: no idea ... | 09:07 |
fabbione | doko_: you want to look at debian bug #391858 | 09:10 |
fabbione | doko_: it's not a glibc bug, but gcc.. the same we fixed in ubuntu not too long ago | 09:10 |
fabbione | their fix is just wrong | 09:10 |
fabbione | and they don't take into account other arches like sparc | 09:10 |
fabbione | that does the same thing | 09:10 |
doko_ | Tags: experimental, fixed-in-experimental | 09:24 |
fabbione | yes but still in glibc | 09:25 |
fabbione | and you are the gcc maintainer so you might want to explain to them that it is wrong | 09:25 |
doko_ | fabbione: fixed-in-experimental, they did fix it already ;-p | 09:28 |
fabbione | doko_: yes i know.. whatever.. our gcc is fine and will cope.. | 09:29 |
fabbione | jbailey: the ppc build seems to hang on a test suite | 09:30 |
fabbione | i have a bunch of processes in Z on davis | 09:30 |
jbailey | Joy. | 09:31 |
fabbione | this is the second timei see this | 09:32 |
fabbione | in the same build | 09:32 |
fabbione | the first time did "unblock" itself | 09:33 |
jbailey | fabbione: Can you tell which test? | 09:37 |
fabbione | x/regex.h posix/wordexp.h posix/fnmatch.h posix/getopt.h posix/tar.h posix/sys/unistd.h posix/sched.h posix/re_comp.h posix/wait.h posix/cpio.h posix/spawn.h pwd/pwd.h resolv/resolv.h resolv/netdb.h resolv/arpa/nameser.h resolv/arpa/nameser_compat.h resource/sys/resource.h resource/sys/vlimit.h resource/sys/vtimes.h resource/ulimit.h rt/aio.h rt/mqueue.h setjmp/setjmp.h shadow/shadow.h signal/signal.h signal/sys/signal.h socket/sys/soc | 09:37 |
fabbione | ket.h socket/sys/un.h stdio-common/printf.h stdio-common/stdio_ext.h stdlib/stdlib.h stdlib/alloca.h stdlib/monetary.h stdlib/fmtmsg.h stdlib/ucontext.h sysdeps/generic/inttypes.h sysdeps/generic/stdint.h stdlib/errno.h stdlib/sys/errno.h string/string.h string/strings.h string/memory.h string/endian.h string/argz.h string/envz.h string/byteswap.h sunrpc/rpc/auth.h sunrpc/rpc/auth_des.h sunrpc/rpc/auth_unix.h sunrpc/rpc/clnt.h sunrpc/r | 09:37 |
fabbione | pc/des_crypt.h sunrpc/rpc/key_prot.h sunrpc/rpc/netdb.h sunrpc/rpc/pmap_clnt.h sunrpc/rpc/pmap_prot.h sunrpc/rpc/pmap_rmt.h sunrpc/rpc/rpc.h sunrpc/rpc/rpc_des.h sunrpc/rpc/rpc_msg.h sunrpc/rpc/svc.h sunrpc/rpc/svc_auth.h sunrpc/rpc/types.h sunrpc/rpc/xdr.h sunrpc/rpcsvc/bootparam.h sysvipc/sys/ipc.h sysvipc/sys/msg.h sysvipc/sys/sem.h sysvipc/sys/shm.h termios/termios.h termios/sys/termios.h termios/sys/ttychars.h time/time.h time/sys | 09:37 |
fabbione | /time.h time/sys/timeb.h wcsmbs/wchar.h wctype/wctype.h > /home/fabbione/glibc-2.5/build-tree/powerpc-libc/begin-end-check.out | 09:37 |
fabbione | this is all i can get | 09:37 |
fabbione | or you need to remind me how to scroll back in screen | 09:38 |
jbailey | C-a { | 09:40 |
jbailey | err | 09:40 |
jbailey | C-a [ | 09:40 |
jbailey | but I don't remember what begin-end-check.out is. | 09:41 |
jbailey | I can look in my i386 build logs. | 09:41 |
jbailey | Meh, they haven't started the test suite yet. | 09:41 |
jbailey | I'm fighting doko for CPU time on ronne. =) | 09:42 |
fabbione | jbailey: if we really need a running .17 to build glibc we need to talk to elmo for the buildd | 09:43 |
jbailey | Shouldn't need it on ppc. | 09:44 |
jbailey | Will certainly need it on parisc. | 09:44 |
fabbione | what about i386/amd64/sparc? | 09:45 |
fabbione | and ia64? | 09:45 |
jbailey | Mainstream arches should all be fine. | 09:45 |
fabbione | ok | 09:46 |
jbailey | Have to check with Dave on sparc to be certain. | 09:46 |
jbailey | But since their nptl port worked fine, I wouldn't expect that to change. | 09:46 |
fabbione | no.. | 09:46 |
jbailey | The rest of the arches all had corporate funding for their ports, so they got done early. | 09:46 |
fabbione | ok.. i killed the build on ppc.. waiting Znarl to kill some Zl processes leftover | 09:47 |
fabbione | and restart with debuild | 09:47 |
fabbione | jbailey: ppc will wait tomorrow | 09:50 |
fabbione | i need to get some sleep | 09:51 |
jbailey | fabbione: g'n! =) | 09:52 |
fabbione | night | 09:52 |
Dvalin | seems like I was right | 09:53 |
Dvalin | building kernel with old gcc | 09:53 |
Dvalin | made it work | 09:53 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!