# 00:01:45 |
y_morin |
quits : Quit: Night! |
# 07:33:05 |
imMute |
quits : Ping timeout: 260 seconds |
# 07:41:50 |
imMute |
joins #crosstool-ng |
# 08:49:28 |
smartin |
joins #crosstool-ng |
# 12:33:41 |
ccole |
joins #crosstool-ng |
# 17:38:48 |
y_morin |
joins #crosstool-ng |
# 18:08:59 |
smartin |
quits |
# 18:52:30 |
smartin |
joins #crosstool-ng |
# 19:00:51 |
bhundven |
joins #crosstool-ng |
# 19:05:26 |
smartin |
quits : Ping timeout: 245 seconds |
# 19:09:10 |
smartin |
joins #crosstool-ng |
# 19:46:23 |
daggs1 |
joins #crosstool-ng |
# 19:56:53 |
hank |
morning guys |
# 19:57:10 |
hank |
finally got a workable toolchain for my amrv4 target |
# 19:57:14 |
y_morin |
hank: good evening from here! ;-) |
# 19:57:24 |
y_morin |
hank: Good! :-) |
# 19:57:25 |
hank |
hehe good evening then. |
# 19:57:43 |
hank |
still having one issue. udivdi3 has bx instructions compiled in |
# 19:58:11 |
hank |
this is causing problems as it causes exceptions in userland every so often |
# 19:58:55 |
hank |
i've had a look into it and its a helper funciton from gcc but it looks like there isn't an arm specific implementation so I don't understand why all the other gcc helpers don't have the same problem |
# 19:59:30 |
hank |
any one had experience with compilation of gcc division functions |
# 20:00:49 |
y_morin |
hank: these are functions that do the maths in sowftware rather than using asm instructions. |
# 20:01:02 |
y_morin |
hank: I always had issues with those functions... |
# 20:01:23 |
hank |
ok. any places to start looking |
# 20:01:49 |
y_morin |
hank: just a few secs... |
# 20:01:56 |
hank |
i had a look through gcc/config/arm/... and there doesn't seem to be an implementation of divdi.. ok sure |
# 20:03:20 |
y_morin |
hank: I think it's the one from gcc/libgcc2.c that is used. |
# 20:04:02 |
y_morin |
__udivdi3 is just a call to __udivmoddi4 which is the actual code. |
# 20:05:09 |
hank |
ah ok. but looking at udivmoddi4, its just c |
# 20:05:20 |
hank |
or does arm have an implementation for it |
# 20:05:25 |
y_morin |
hank: I have an odea, let me find that code hunk... |
# 20:06:14 |
y_morin |
*idea |
# 20:06:58 |
hank |
ok thanks |
# 20:07:22 |
y_morin |
hank: look at: gcc/config/arm/linux-eabi.h@9 -> it defaults to arm10tdmi that has thumb. |
# 20:07:51 |
y_morin |
hank: crosstool-NG has a patch to unbreak it for armv4, but the patch has not be forward-ported to newer gcc-s: |
# 20:07:56 |
y_morin |
hank: http://crosstool-ng.org/hg/crosstool-ng/file/28ac8fbd9e6a/patches/gcc/4.4.6/210-arm-unbreak-armv4t.patch |
# 20:08:25 |
hank |
ah ok |
# 20:08:27 |
bhundven |
i thought that sounded too familiar |
# 20:08:34 |
y_morin |
(do not look at the cset log, it's not the corresponding one. Just lok at the patch) |
# 20:08:41 |
hank |
ok sure |
# 20:09:05 |
y_morin |
hank: probably that patch still applies to newer gcc-s, but I never needed anything |
# 20:09:37 |
hank |
alright thanks alot for the help.I will look into this and report back |
# 20:10:00 |
y_morin |
Hm, it's line 47, not 9. |
# 20:10:23 |
y_morin |
hank: still, arm9tdmi sounds like thumb-aware cpu. |
# 20:12:02 |
hank |
im using a arm922 core (no t at the end) in a cns2132 |
# 20:12:16 |
hank |
similar to the fa526 |
# 20:12:31 |
y_morin |
hank: sorry, I'm not that familiar with the ARM architecture! ;-) |
# 20:12:52 |
y_morin |
(although it's my preferred arch, I'm not expoert in ARM). |
# 20:14:50 |
kos_tom |
y_morin: hello. |
# 20:14:57 |
kos_tom |
y_morin: I faced a small download problem today with 1.13.4 |
# 20:15:03 |
kos_tom |
I don't know if it's still maintained or not. |
# 20:15:03 |
hank |
its not very well know it is wwaayyy old |
# 20:15:44 |
y_morin |
kos_tom: 1.13 is the version currently used by buildroot, right? |
# 20:15:57 |
y_morin |
(hello!) |
# 20:16:00 |
kos_tom |
y_morin: as far as I remember, yes. |
# 20:16:09 |
kos_tom |
basically, the problem is that the strace URL has changed |
# 20:16:10 |
y_morin |
kos_tom: then it is maintained! ;-) |
# 20:16:15 |
y_morin |
kos_tom: Arg... :-( |
# 20:16:22 |
kos_tom |
it lacks a /strace/${CT_STRACE_VERSION} at the end |
# 20:16:37 |
y_morin |
kos_tom: OK, thanks, I'll look. |
# 20:16:44 |
kos_tom |
ie, should be: http://blabla.sourceforge.net/strace/strace/${CT_STRACE_VERSION} |
# 20:16:49 |
kos_tom |
(i.e two times /strace/) |
# 20:16:50 |
y_morin |
kos_tom: BTW, any comments about the kconfig frontends package? ;-] |
# 20:17:00 |
kos_tom |
y_morin: no, sorry, I haven't had the time to look at it. |
# 20:17:01 |
kos_tom |
*but* |
# 20:17:12 |
kos_tom |
I gave the address of the project to Karim Yaghmour, who works on Cyborgstack |
# 20:17:22 |
kos_tom |
a remixed Android stack, for which he wants to use kconfig |
# 20:17:35 |
y_morin |
kos_tom: OK, nice! :-) |
# 20:17:43 |
kos_tom |
so when he asked me what was the best way to get kconfig running for a project, I said, "hey, look at Yann's stuff!" |
# 20:17:52 |
y_morin |
\o/ |
# 20:17:57 |
kos_tom |
and I gave the URL of it. So you might hear from Karim some day. |
# 20:18:30 |
kos_tom |
is fighting with the crappy LED API in U-Boot |
# 20:28:11 |
y_morin |
kos_tom: SF.net says: We are unable to display the page you requested. |
# 20:28:15 |
y_morin |
Dang!... :-( |
# 20:29:07 |
y_morin |
And they only serve .xz files for latest strace... Too bad... :-( |
# 20:30:22 |
kos_tom |
the 4.5.19 is still available |
# 20:30:40 |
kos_tom |
http://sourceforge.net/projects/strace/files/strace/4.5.19/ |
# 20:30:56 |
y_morin |
yes, it is. But I tried with latest, as usual... |
# 20:31:01 |
y_morin |
4.5.20 is, too. |
# 20:31:23 |
bhundven |
didn't I add 4.6.0? |
# 20:31:37 |
bhundven |
or just having problems with older versions? |
# 20:31:51 |
y_morin |
bhundven: hey! Yes, but the DL link changed, and we're fixing it. |
# 20:31:57 |
bhundven |
ah |
# 20:32:11 |
y_morin |
bhundven: and at the same time, I noticed 4.6 was available only as .xz, not .bz2. |
# 20:32:13 |
bhundven |
I hate it when they do that |
# 20:32:19 |
y_morin |
bhundven: Hey! |
# 20:32:23 |
y_morin |
:-) |
# 20:32:25 |
bhundven |
howdy :D |
# 20:33:10 |
bhundven |
is patiently waiting for gcc-4.7 and linux-3.3.0 |
# 20:33:25 |
y_morin |
bhundven: and I see that you did well to protect 4.6 behind has_xz :-) |
# 20:33:34 |
bhundven |
yes |
# 20:33:39 |
y_morin |
bhundven: let's form a group, then! :-) |
# 20:34:21 |
kos_tom |
what about ltrace ? |
# 20:34:28 |
y_morin |
although I have to admit that I am not really impatient to see 4.7. Means a lot of work to add this to ct-ng, yet once more. |
# 20:34:31 |
bhundven |
eh |
# 20:34:31 |
kos_tom |
no-one seems to be worried of the fact that it simply doesn't work on ARM |
# 20:34:35 |
bhundven |
ltrace... |
# 20:34:42 |
y_morin |
ltrace... |
# 20:35:51 |
y_morin |
I'm not even sure it once worked at all, except by chance... |
# 20:35:58 |
y_morin |
hides... |
# 20:37:04 |
bhundven |
I'm throwing multilib back at you, y_morin. |
# 20:37:22 |
bhundven |
I couldn't find a clean way to put the opposite arch in |
# 20:37:24 |
y_morin |
bhundven: Oh so? How do you dare? :-) |
# 20:37:39 |
daggs1 |
bhundven, hello, regarding the static gdbserver, where should your fix be applied? |
# 20:38:07 |
y_morin |
Hey, I'll handle that in a while... I have some hot stuff here to resolve first, but I have this issue on my TODO |
# 20:38:15 |
bhundven |
k |
# 20:38:26 |
y_morin |
bhundven: no problem. |
# 20:39:06 |
y_morin |
just got a brand new NAS to replace the old server: 8W idle, vs. 85W idle. Woot! :-) |
# 20:39:45 |
daggs1 |
y_morin, congrads :) |
# 20:40:11 |
y_morin |
daggs1: yet, I have to entirely review my install/backup/restore procedure, now... :-/ |
# 20:40:46 |
y_morin |
praises libvirt + virt-manager! :-) |
# 20:42:36 |
daggs1 |
y_morin, I feel your pain but you need to remember that you pay for new gadgets somehow... |
# 20:43:13 |
y_morin |
daggs1: well, first you pay with money. Then you pay with time. Then you win. Or so he said! ;-) |
# 20:43:50 |
daggs1 |
who said? |
# 20:44:31 |
y_morin |
daggs1: it's a rephrasing of Gandy's sentence: "First they ignore you, then they fight you, then you win" (or something like that) |
# 20:45:09 |
daggs1 |
where is Gandy now? |
# 20:46:17 |
y_morin |
daggs1: https://en.wikipedia.org/wiki/Gandhi |
# 20:46:28 |
y_morin |
And I mispelled his name. Shame on me... :-/ |
# 20:47:00 |
daggs1 |
y_morin, you are no fun, your expected output was dead... |
# 20:47:17 |
daggs1 |
*was suppose to be |
# 20:53:36 |
daggs1 |
ok, I'm going to bed, night all, bhundven, I know you see this, I'd appreciate it if you could point on where to patch your fix at. |
# 20:54:08 |
daggs1 |
quits : Quit: Leaving |
# 20:55:21 |
y_morin |
kos_tom: http://crosstool-ng.org/hg/crosstool-ng/rev/ce1a9c264de5 |
# 20:57:50 |
kos_tom |
right, sounds good! |
# 20:58:18 |
y_morin |
kos_tom: Might be interesting to add that patch to the ct-ng package in buildroot. |
# 20:58:45 |
kos_tom |
I don't think so, since strace is never being built by ct-ng in that case |
# 20:58:53 |
kos_tom |
so it's not really mandatory to have this patch, me thinks |
# 20:58:57 |
y_morin |
kos_tom: right. |
# 20:59:16 |
y_morin |
kos_tom: you had the issue when using ct-ng stand-alone, then? |
# 20:59:20 |
kos_tom |
yes |
# 20:59:27 |
y_morin |
Ah, OK. |
# 21:01:41 |
bhundven |
is now known as: bhundven|nap |
# 21:03:49 |
kos_tom |
y_morin: in fact, I almost never use ct-ng through the buildroot backend. |
# 21:04:00 |
kos_tom |
I really prefer using ct-ng stand-alone, and then use the external toolchain mechanism |
# 21:04:30 |
y_morin |
kos_tom: that's the way it makes sense to me, too: build the tools once and for all, then hack. |
# 21:04:41 |
y_morin |
Especially for big tools like a toolchain. |
# 21:05:03 |
kos_tom |
exactly. |
# 21:05:18 |
kos_tom |
would be me, I would remove all toolchain backends from Buildroot except the external one :) |
# 21:05:31 |
y_morin |
kos_tom: +1 |
# 21:06:01 |
y_morin |
but for new-comers, it's much simpler to work with an all-in-one build infra. |
# 21:06:17 |
kos_tom |
yes, that's what I've been told. |
# 21:06:31 |
kos_tom |
but in fact, it puts newcomers in a situation where they rebuild everything everytime they do a mistake |
# 21:06:36 |
kos_tom |
and this "everything" includes the toolchain |
# 21:06:51 |
kos_tom |
so I am not sure in the end it is really serving well the newcomers |
# 21:07:45 |
y_morin |
Yes, that's the drawback. You can't have the easy-to-use feeling of an all-in-one, and the comfort and reactiveness of a separate infra. |
# 21:08:42 |
y_morin |
The idea, I think, is to be well aware that internal+ct-ng backends are for starters. Once they get a bit more accustomed to bulding cross-rootfs, then they can investigate how to do the job more efficiently. |
# 21:09:18 |
y_morin |
That's were a tutorial like "Beyond buildroot" would make sense (a bit like there is for LFS and Beyond LFS). |
# 22:00:39 |
bhundven|nap |
is now known as: bhundven |
# 23:16:27 |
smartin |
quits |