ibotlog2html for #crosstool-ng

<< Previous 2013-04-14 Next >>

# 00:40:17 Doug__ joins #crosstool-ng
# 00:40:26 Doug__ quits : Client Quit
# 00:50:26 doc2 quits : Ping timeout: 245 seconds
# 04:22:11 alan_o quits : Quit: Leaving
# 05:03:44 sh4rm4 quits : Remote host closed the connection
# 05:33:26 sfan5|OFF is now known as: sfan5
# 06:15:37 sfan5 is now known as: sfan5|OFF
# 06:23:25 akio joins #crosstool-ng
# 06:33:30 sh4rm4 joins #crosstool-ng
# 06:56:12 smartin_ joins #crosstool-ng
# 08:11:18 doc2 joins #crosstool-ng
# 08:24:32 y_morin joins #crosstool-ng
# 09:12:55 sh4rm4 quits : Quit: sh4rm4
# 09:15:32 sh4rm4 joins #crosstool-ng
# 09:18:26 codyps quits : Read error: Operation timed out
# 09:23:13 codyps joins #crosstool-ng
# 09:31:58 codyps quits : Read error: Connection reset by peer
# 09:33:13 Net147 joins #crosstool-ng
# 09:33:38 codyps joins #crosstool-ng
# 12:20:22 sfan5|OFF is now known as: sfan5
# 14:37:45 Net147 quits : Quit: HydraIRC -> http://www.hydrairc.com <- In tests, 0x09 out of 0x0A l33t h4x0rz prefer it :)
# 14:42:42 doc2 quits : Quit: Verlassend
# 16:52:38 diorcety joins #crosstool-ng
# 18:24:06 sfan5 is now known as: sfan5|OFF
# 18:27:08 sfan5|OFF is now known as: sfan5
# 19:43:08 Doug__ joins #crosstool-ng
# 19:44:52 Doug__ y_morin:
# 19:44:53 Doug__ Yann,
# 19:44:55 Doug__ We chatted Friday about the failure to accept kernel.org's certificate due to lack of recognition of the issuer.
# 19:44:57 Doug__ I wonder about the following: Lines 52-55 of linux.sh make reference to--
# 19:44:59 Doug__ korg_base="http://ftp.kernel.org/pub/linux/kernel/${rel_dir}"
# 19:45:01 Doug__ CT_GetFile "linux-${CT_KERNEL_VERSION}" \
# 19:45:02 Doug__ "${korg_base}" \
# 19:45:04 Doug__ "${korg_base}/longterm/v${k_ver}" \
# 19:45:06 Doug__ "${korg_base}/longterm"
# 19:45:08 Doug__ There is no "longterm" subdirectory under /v3.x/ on the kernel.com site, as there is under /v2.6/.
# 19:45:10 Doug__ Could this have led to a spurious message about the certificate?
# 19:45:12 Doug__ I can't begin to follow the scripting, so I can't personally diagnose.
# 20:04:03 alan_o joins #crosstool-ng
# 20:04:18 b_rare joins #crosstool-ng
# 20:07:56 codyps quits : Ping timeout: 245 seconds
# 20:40:35 smartin_ quits : Quit: leaving
# 20:51:52 y_morin Doug__: Hello! No that's not the reason for the certificate issue. I typed the command manually here, and it failed straight the first time.
# 20:52:08 y_morin Doug__: The problem is really Cygwin-related.
# 20:52:32 muep is it intentional that gmp gets built for host but then ppl is built using the system-provided gmp headers?
# 20:53:37 y_morin Doug__: You can however work this issue around by manually downloading the linux kernel (eg. with Firefox), put it in (say) ~/downloads, and point ct-ng to use ~/downloads as a local dir with tarballs
# 20:53:48 y_morin muep: No, this should not be the case.
# 20:54:19 y_morin muep: How did you notice that the host headers are used instead of our own version?
# 20:55:08 muep I think I saw that in a log, but it looks I lost the part of the log, so a minute
# 20:55:12 y_morin Doug__: The option is in "Paths and misc options" ---> "Local tarballs directory"
# 20:56:14 muep y_morin: near the end of build.log, the error message has this: /usr/include/gmpxx.h:3269:21: error: previous definition of 'class std::numeric_limits<__gmp_expr<__mpz_struct [1], __mpz_struct [1]>
# 20:56:52 y_morin muep: what host distro?
# 20:57:10 muep y_morin: Fedora's rawhide branch
# 20:58:08 y_morin muep: Ah. Fedora. IIRC, that's not the first report, but I can't remember anyone suggestign a solution so far.
# 20:58:18 muep ct-ng does work for me in Fedora 18, but it has a bit older gmp
# 20:58:42 muep so maybe it would not run into this http://www.cs.unipr.it/pipermail/ppl-devel/2012-December/018574.html
# 20:59:25 muep but rawhide has gmp 5.1.1
# 20:59:26 y_morin muep: the companion libs (gmp, mpfr, ppl, cloog) should not be used from the host, only the versions we build ourselves.
# 20:59:36 hrubi quits : Ping timeout: 245 seconds
# 20:59:59 y_morin muep: so it does not (should not) matter what host version you have installed, or even that you have any installed or not.
# 21:00:43 muep true, but I suspect this problem might exist in many other systems besides this one
# 21:01:17 muep unless other operating systems ship their gmp libraries in a significantly different place
# 21:02:21 y_morin muep: indeed, but since I'm not impacted on my system (Debian Squeeze), and no one provided a fix so far, I can't see how to fix this issue.
# 21:04:04 muep maybe I'll just have to wait until gmp 5.1 becomes more common, unless I happen to find a fix myself
# 21:05:31 y_morin muep: I might be able to install, and test with, a rawhide in a VM, but it's 23:00 here, so maybe not today...
# 21:05:33 muep where do the companion libs get installed during the build?
# 21:06:01 muep gmp is not yet anywhere under the prefix dir
# 21:07:02 y_morin muep: muep the complibs are not installed where the toolchain is installed.
# 21:07:29 y_morin muep: they are installed in .build/TARGET/builtools (where TARGET is the tuple of the toolchain)
# 21:08:13 muep I am not sure if I read the compiler command lines correctly, but I do not immediately see any -I options pointing there
# 21:09:00 Doug__ yann: Sorry to interrupt, but it's a one-liner.....just ftp ..../kernel/v3.x/linux-3.ZZZ.tar.xz, where ZZZ is the version on my raspi? I understand ow to set menuconfig.
# 21:09:10 y_morin muep: Oh, then that's a hint. It smells like libtool borks the command line... :-/
# 21:09:29 muep just in case you wish to check, end of build log is here: http://paste.fedoraproject.org/7463/73724136/
# 21:10:23 y_morin Doug__: No, ZZZ is for the headers used to build the toolchain. Of course, it should match the running kernel.
# 21:10:50 y_morin Doug__: However, what is important is, wrt to headers 3.X.y and 3.X.z are equivalent.
# 21:11:30 Doug__ OK, so it's 3.7.3 You last comment baffles me.
# 21:11:43 y_morin Doug__: So, if you have 3.2.27 running on the RPi, then you can use 3.2.43 with no problem.
# 21:11:46 codyps joins #crosstool-ng
# 21:12:07 y_morin Doug__: OK, so you can use the 3.7.10 headers with no problem.
# 21:12:42 y_morin (3.7.10 is what is currently in the ct-ng repos, the latest release may have another version)
# 21:12:55 y_morin muep: OK, I'll have a look
# 21:12:58 Doug__ Got it. Thanks. I'm out of your hair for now. Bye.
# 21:13:31 Doug__ quits : Quit: ChatZilla 0.9.90 [Firefox 21.0/20130408165307]
# 21:14:58 muep y_morin: oh, and this is with the latest crosstool-ng code from the mercurial repository
# 21:15:43 muep IIRC it (or some other issue) did happen with released crosstool-ng versions, too, which is why I got the development version
# 21:16:04 sfan5 is now known as: sfan5|OFF
# 21:16:15 y_morin muep: From your build.log extract, look at line 363: the GMP prefix is properly passed. But then on line 802 (the first error) we can't see any -I indeed.
# 21:16:37 y_morin muep: yep, there's been no change in this respect since the last release.
# 21:17:19 y_morin muep: line 801, in fact.
# 21:17:41 hrubi joins #crosstool-ng
# 21:21:11 muep y_morin: it is starting to be a time for me to sleep, too
# 21:21:49 muep thanks for helping me understand the issue, even if there is no obvious immediate solution
# 21:22:52 y_morin muep: cheers!
# 21:23:04 y_morin muep: I'll still continue to investigate.
# 21:23:36 muep thank you for looking at it, see you
# 21:23:39 y_morin muep can you send the config.log at the rootof the PPL build dir?
# 21:23:51 muep yes
# 21:24:11 y_morin muep: it's in .build/TARGET/build/build-ppl-VERSION
# 21:25:31 muep y_morin: http://paste.fedoraproject.org/7464/59746841
# 21:25:42 y_morin muep: Thanks!
# 21:29:34 muep thank you :-)
# 22:16:16 y_morin quits : Quit: Nighty Night!

Generated by ibotlog2html by Yann E. MORIN