# 06:08:41 |
leming |
quits : Ping timeout: 240 seconds |
# 06:10:41 |
leming |
joins #crosstool-ng |
# 06:35:06 |
mnt_real |
quits : Remote host closed the connection |
# 07:15:15 |
kos_tom |
joins #crosstool-ng |
# 07:19:49 |
y_morin |
joins #crosstool-ng |
# 11:11:42 |
infobot |
quits : Remote host closed the connection |
# 11:44:36 |
infobot |
joins #crosstool-ng |
# 15:34:20 |
ctngbot_ |
joins #crosstool-ng |
# 16:21:49 |
y_morin_ |
joins #crosstool-ng |
# 16:22:28 |
ctngbot_ |
quits : Quit: I'll be back. Hopefully... |
# 16:26:33 |
y_morin |
quits |
# 16:26:43 |
y_morin_ |
is now known as: y_morin |
# 16:31:51 |
kos_tom |
quits : Quit: Leaving |
# 16:35:44 |
y_morin |
quits : Quit: back in a while... |
# 16:40:29 |
y_morin |
joins #crosstool-ng |
# 17:05:22 |
y_morin |
is now known as: y_morin|away |
# 17:05:28 |
y_morin|away |
is now known as: y_morin |
# 17:15:25 |
y_morin |
is now known as: y_morin|away |
# 17:15:28 |
y_morin|away |
is now known as: y_morin |
# 17:20:30 |
kos_tom |
joins #crosstool-ng |
# 17:24:29 |
ctngbot_ |
joins #crosstool-ng |
# 17:25:25 |
y_morin |
is now known as: y_morin|away |
# 17:25:28 |
y_morin|away |
is now known as: y_morin |
# 17:35:25 |
y_morin |
is now known as: y_morin|away |
# 17:35:28 |
y_morin|away |
is now known as: y_morin |
# 17:39:05 |
bhundven |
hmm. glibc loop? |
# 17:39:38 |
bhundven |
quits : Changing host |
# 17:39:38 |
bhundven |
joins #crosstool-ng |
# 17:45:25 |
y_morin |
is now known as: y_morin|away |
# 17:45:28 |
y_morin|away |
is now known as: y_morin |
# 17:55:25 |
y_morin |
is now known as: y_morin|away |
# 17:55:28 |
y_morin|away |
is now known as: y_morin |
# 18:05:25 |
y_morin |
is now known as: y_morin|away |
# 18:05:28 |
y_morin|away |
is now known as: y_morin |
# 18:15:25 |
y_morin |
is now known as: y_morin|away |
# 18:15:28 |
y_morin|away |
is now known as: y_morin |
# 18:21:40 |
ctngbot_ |
quits : Quit: I'll be back. Hopefully... |
# 18:26:22 |
mnt_real |
joins #crosstool-ng |
# 18:32:31 |
leming |
y_morin, glad you were able to recreate the loop.. i was out yesterday for the 4th here |
# 18:32:53 |
y_morin |
leming: Ha, I see. Hope you had fun! |
# 18:33:01 |
y_morin |
Yes, ineed, the loop happens here too. |
# 18:33:17 |
leming |
in my limited troubleshooting i tried recreating the command used to do the libc start files |
# 18:33:23 |
y_morin |
I haven't had time to check why, yet. |
# 18:33:30 |
leming |
basically copying/pasting the command that gets printed out for what gets executed |
# 18:33:47 |
y_morin |
Yes, that's the way I was going to go... |
# 18:34:07 |
y_morin |
crosstool-NG logs the full command, so it's easy to manually reproduce. |
# 18:34:16 |
leming |
and at one point when i ran that, it went through the configure and it died with the checking for forced unwind support: no error |
# 18:34:50 |
leming |
which stopped the build in the exact spot that it loops at |
# 18:35:03 |
leming |
right after doing that nptl check |
# 18:35:17 |
leming |
but actually printed the error and didn't just loop back on itself |
# 18:35:51 |
leming |
but then i figured out the save steps and restarting the build, and when i restarted the libc start files, it went through without an error |
# 18:36:04 |
y_morin |
I am just thinking that we do set and export some variables, such as $(sed). Maybe that makes configure break... :-/ |
# 18:36:39 |
leming |
i've tried with and without config.guess/sub, just to see if that had any effect and it doesn't |
# 18:37:06 |
leming |
it's almost like something is being carried through in the environment going from the beginning that makes that step barf |
# 18:37:25 |
leming |
but when you restart at that step, whatever it is isn't there, so it is successful |
# 18:37:43 |
leming |
but then again, it's so far just ubuntu that's exhibiting this particular issue |
# 18:38:20 |
y_morin |
leming: try setting the same variables as found in ./paths.mk before running. |
# 18:38:26 |
y_morin |
And then see if it breaks... |
# 18:38:54 |
leming |
i also had past versions of ct-ng hanging around in other directories with old configs, and those looped in the same way, so i'm thinking it's something new in ubuntu that it's not liking |
# 18:39:22 |
leming |
older versions, older configs, older source tarballs, ones that compiled without issue many many times before |
# 18:40:00 |
leming |
i can try some things later tonight, just checking in on my lunch break |
# 18:40:21 |
y_morin |
leming: I'll fire my ubuntu VM here and see if I can find something... |
# 18:40:54 |
leming |
i know ubuntu has moved over to their new standardized linaro toolchain setup |
# 18:41:10 |
leming |
distributions like arch (where it worked without issue) don't use linaro |
# 18:41:40 |
leming |
that's the only thing that's really changed that i'm aware of |
# 18:41:57 |
y_morin |
leming: do you mean that the default toolchain in ubuntu is from Linaro ? |
# 18:42:11 |
y_morin |
I mean, the native toolchain? |
# 18:42:53 |
leming |
yeah |
# 18:43:02 |
leming |
native ubuntu toolchain is their linaro thing |
# 18:43:24 |
y_morin |
But Linaro toolchains are pretty solid, from my experience... Especially for ARM. |
# 18:43:29 |
leming |
it's a recent switch they've made with that project gaining more steam |
# 18:43:49 |
leming |
yeah, they're standardizing it across all architectures, which is nice |
# 18:43:55 |
y_morin |
And anyway, glibc shall not be impacted by the native toolchain when building the headers... |
# 18:44:11 |
leming |
but i'm wondering if they've added something that changed just enough to cause the loop |
# 18:44:21 |
y_morin |
Indeed, standardisation is good. At least, if it breaks on one, it breaks on all! :-] |
# 18:44:24 |
y_morin |
hides... |
# 18:44:27 |
leming |
hah |
# 18:45:04 |
y_morin |
BTW, I'll probably have a beagleboard up-n-running here in a short while. I'll try to see if I can run crosstool-NG on ARM... |
# 18:45:27 |
y_morin |
Also, that'll give me the oportunity to at last (try to) implement the cross-native stuff in crosstool-NG... |
# 18:45:49 |
y_morin |
is longing for thos Cortex-A15 quad CPUs in the coming... |
# 18:46:06 |
y_morin |
is hungry |
# 18:46:12 |
leming |
oh yeah |
# 18:46:14 |
y_morin |
goes have dinner... |
# 18:46:28 |
y_morin |
is now known as: y_morin|away |
# 18:46:35 |
leming |
the cross-native feature would be nice, get away from having to deal with buildroot or trying to jumpstart a native compiler on your own |
# 18:47:00 |
leming |
i've been able to work off of my progress on armv5 to get armv7 working native |
# 18:47:22 |
leming |
going hard-float for our distribution |
# 18:51:09 |
leming |
i'll bbl |
# 19:01:59 |
kos_tom |
leming, any specific problem that you had with buildroot ? |
# 20:03:03 |
bhundven |
hmm |
# 20:03:34 |
bhundven |
I wonder if the glibc-eglibc-common stuff broke glibc. |
# 20:03:52 |
bhundven |
might be interesting to bisect |
# 20:04:35 |
bhundven |
sync to commit before that change, and test it. if it works, mark it as good and keep moving up till head. |
# 21:20:27 |
y_morin|away |
is now known as: y_morin |
# 21:21:18 |
y_morin |
bhundven: good idea! |
# 21:44:20 |
kos_tom |
quits : Remote host closed the connection |
# 21:55:23 |
y_morin |
bhundven: did you know that glibc-2.14 no longer has RPC headers+libs ? |
# 22:04:06 |
y_morin |
It is now needed to use the ti-rpc library: http://sourceforge.net/projects/libtirpc/ |
# 22:05:50 |
y_morin |
bhundven: I was wondering if t was a good idea to treat it as a sytem lib and build it with the toolchain, or treat it as a non-system lib, and leave the user do it. |
# 22:06:37 |
y_morin |
IMHO it really is a non-system lib, but people got accustomed to glibc providing it, so maybe as a transition it would make sense. |
# 22:06:49 |
y_morin |
I'll post a longer message on the list... |
# 22:49:04 |
leming |
if you commit changes that you think might impact the glibc loop thing i'd be happy to run a test against it |
# 22:50:04 |
y_morin |
leming: no fix/workaround yet. |
# 22:50:42 |
y_morin |
leming: I tested against revision 2369 (with glibc-2.14 shoehorned in there), but the loop also happens. |
# 22:50:51 |
y_morin |
(2369) is before the glibc/eglibc merge. |
# 22:50:56 |
leming |
workaround is just to ctrl+c the build at that point (with saved steps) and restart |
# 22:51:10 |
leming |
that has worked every time for me so far |
# 22:51:19 |
y_morin |
That's not a proper workaround for me! ;-) |
# 22:51:24 |
leming |
;) |
# 22:51:52 |
y_morin |
leming: anyway, that could mean that crosstool-NG is setting a variable that glibc does not like... |
# 22:52:05 |
leming |
that's what i've been thinking |
# 22:52:16 |
leming |
i doublechecked that everything passed to configure is fine |
# 22:52:37 |
leming |
my best guess is an environment variable that is upsetting it |
# 22:53:15 |
y_morin |
leming: could you do a test for me, please? Run from scratch in debug mode with save steps enabled, hit-Ctrl-C on loop, re-run. And once it passes that step, I'll ask for two files... |
# 22:53:30 |
y_morin |
leming: indeed, I was speaking about env vars, not cmd-line vars. |
# 22:53:48 |
leming |
alright |
# 22:54:52 |
leming |
takes about 10-12 minutes for me to hit that point on my computer |
# 22:55:08 |
y_morin |
leming: OK, thanks! |
# 22:55:27 |
leming |
CT_DEBUG_CT=y |
# 22:55:27 |
leming |
CT_DEBUG_CT_SAVE_STEPS=y |
# 22:55:30 |
leming |
those two, right? |
# 22:55:34 |
y_morin |
Yes. |
# 22:55:45 |
leming |
alright, it's off |
# 22:56:28 |
leming |
want me to just re-run the libc start files, or go from that point to the end? |
# 22:57:25 |
y_morin |
Run from scratch, and do as you do to make it work! ;-) |
# 22:57:42 |
leming |
will do |
# 22:57:46 |
y_morin |
Until the end, I mean, yes. |
# 22:58:17 |
y_morin |
But sucesful start_files is enough for that part, indeed. sorry... |
# 22:58:36 |
leming |
which files will you be needing? |
# 22:58:47 |
y_morin |
The two files are: .build/(tuple)/state/libc_start_files/env.sh just after you hit Ctr-C |
# 22:59:12 |
y_morin |
And the second is: .build/(tuple)/state/cc_core_pass_2/env.sh after start_files are OK. |
# 22:59:36 |
y_morin |
And you can do the diff by yourself, as well, of course! ;-) |
# 23:00:03 |
leming |
alright.. so i'll need to finish off the build to get the second one if i'm remembering the steps right |
# 23:01:06 |
y_morin |
leming: just starting the cc_core_pass2 is enough (it's just after start files). |
# 23:01:20 |
leming |
alright |
# 23:01:39 |
y_morin |
Thanks! Doing it in the VM here is rather slow... :-/ |
# 23:01:49 |
leming |
i'll put up the logs for both stages as well in case those reveal anything of interest |
# 23:02:01 |
y_morin |
OK, thx |
# 23:24:47 |
leming |
http://archlinuxarm.org/builder/ct-ng/ |
# 23:25:00 |
leming |
the build log is rather massive, heh |
# 23:25:32 |
leming |
also put up the configuration i used |
# 23:25:55 |
y_morin |
leming:thx. I'll have a look shortly. |
# 23:26:20 |
y_morin |
But it's 01:25 AM here, I'm soon going to bed... |
# 23:26:39 |
leming |
no rush, sleep is definitely important |
# 23:26:52 |
y_morin |
:-) |
# 23:58:31 |
y_morin |
quits : Quit: Night all, see ya! ;-) |