# 00:42:46 |
ragedragon |
quits : Ping timeout: 240 seconds |
# 00:43:04 |
aneyman |
quits : Ping timeout: 245 seconds |
# 00:43:22 |
aneyman |
joins #crosstool-ng |
# 04:28:51 |
bhundven |
wbx: you around? |
# 04:29:47 |
wbx |
bhundven: yes. |
# 04:31:27 |
bhundven |
wbx, do you use dejagnu with gcc much? |
# 04:33:26 |
bhundven |
gdb creates a file: sim/arm/run, which gets installed by crosstool-ng as: ${CT_TARGET}-run (in my case: arm-unknown-linux-gnueabi-run) |
# 04:34:00 |
bhundven |
dejagnu uses that to run the execution tests (as opposed to the compile tests) |
# 04:34:08 |
bhundven |
but I need to pass options to the -run command |
# 04:34:20 |
bhundven |
specifically, the sysroot |
# 04:36:43 |
wbx |
hmm. it is a time ago i used gcc dejagnu to run testsuite for gcc |
# 04:36:53 |
wbx |
run is the gdb simulator |
# 04:38:09 |
bhundven |
righ |
# 04:38:11 |
bhundven |
*right |
# 04:38:46 |
bhundven |
https://pastebin.osuosl.org/35801/ |
# 04:39:10 |
bhundven |
line 4, I need to add --sysroot= to -run |
# 04:40:46 |
wbx |
i have used env DEJAGNU=$(ADK_TOPDIR)/adk/tests/master.exp to customize the runs |
# 04:41:15 |
wbx |
need a coffee first |
# 04:41:56 |
bhundven |
:) |
# 04:45:48 |
wbx |
the docs for dejagnu are rather minimal. i amcreating a master.exp to ssh into a qemu running on the same machine via tcp/ip |
# 04:46:05 |
bhundven |
I'm mucking with the simulator |
# 04:46:14 |
bhundven |
so I don't need a vm |
# 04:49:20 |
wbx |
bhundven: then maybe you should install a wrapper around run which executes run with the options. at a first look dejagnu have no variable for passing options to run |
# 04:50:34 |
wbx |
gdb,target_sim_options"-sparclite"Special options to pass to the simulator. |
# 04:51:27 |
bhundven |
hrm, I just ran the -run command manually, adding the sysroot, and it still fails |
# 04:51:29 |
bhundven |
so... |
# 04:51:36 |
bhundven |
I may have other issues. |
# 04:52:01 |
wbx |
are you using any exp files? |
# 04:53:22 |
bhundven |
in the end, its just running: |
# 04:53:54 |
bhundven |
make check-gcc RUNTESTFLAGS="--target=arm-sim" |
# 04:54:24 |
bhundven |
the compile tests pass |
# 04:54:32 |
bhundven |
it's just the execute tests that fail |
# 04:55:16 |
bhundven |
I'm sure I'm missing something key |
# 04:55:16 |
bhundven |
gdb is built with sim |
# 04:55:16 |
bhundven |
so the -run command exists |
# 04:55:28 |
bhundven |
I can run it manually and see it returns 48, FAIL |
# 04:56:34 |
wbx |
don't know how good the arm simulator is. what arm cpu you are targeting? |
# 04:56:49 |
wbx |
i have only used bfin/h8300 simulator with gdb |
# 04:57:09 |
wbx |
with an exp file you can run in verbose mode |
# 04:57:39 |
bhundven |
I can pass the verbose options in RUNTESTFLAGS |
# 04:57:56 |
bhundven |
and they don't show me anything useful |
# 04:58:19 |
wbx |
no more info? |
# 04:58:37 |
wbx |
run run manually with the created executable. |
# 04:58:37 |
bhundven |
tons of stuff, that doesn't show me why the execution failed |
# 04:58:46 |
wbx |
can you paste it? |
# 04:59:00 |
bhundven |
that's what I've said: I can run it manually and see it returns 48, FAIL |
# 04:59:46 |
wbx |
i mean the verbose stuff |
# 05:00:14 |
bhundven |
https://pastebin.osuosl.org/35806/ |
# 05:04:43 |
wbx |
does the arm simulator have any options? run --help / -h |
# 05:05:20 |
bhundven |
https://pastebin.osuosl.org/35811/ |
# 05:05:33 |
bhundven |
None of the debug or trace options do anything |
# 05:06:18 |
wbx |
run --architecture-info? |
# 05:06:21 |
bhundven |
maybe I'm missing something from my gdb build |
# 05:06:39 |
bhundven |
~/x-tools/arm-unknown-linux-gnueabi/bin/arm-unknown-linux-gnueabi-run --architecture-info |
# 05:06:39 |
bhundven |
Possible architectures: arm armv2 armv2a armv3 armv3m armv4 armv4t armv5 armv5t armv5te xscale ep9312 iwmmxt iwmmxt2 plugin |
# 05:06:39 |
bhundven |
Target byte order unspecified |
# 05:07:20 |
wbx |
i just use --enable-sim --enable-sim-stdio for gdb compile |
# 05:07:40 |
wbx |
are you targeting armv6/armv7? |
# 05:08:31 |
bhundven |
it's generic, so I'm not specifically targeting anything specific. |
# 05:08:50 |
bhundven |
I think by default it's armv5 |
# 05:09:44 |
bhundven |
⯠file 20010124-1.x0 |
# 05:09:44 |
bhundven |
20010124-1.x0: ELF 32-bit LSB executable, ARM, EABI5 version 1 (SYSV), dynamically linked, interpreter /lib/ld-linux.so.3, for GNU/Linux 4.3.0, not stripped |
# 05:10:58 |
bhundven |
I just used --enable-sim |
# 05:11:16 |
bhundven |
do you also need --enable-sim-stdio? |
# 05:12:10 |
wbx |
i need it to boot uClinux. not sure if it is required for the gcc testsuite |
# 05:13:19 |
bhundven |
http://www.ifp.illinois.edu/~nakazato/tips/xgcc.html |
# 05:13:25 |
bhundven |
that page also uses it |
# 05:13:28 |
bhundven |
hrm |
# 05:14:16 |
bhundven |
well, I'll enable it and do a build again. The test suite has to run at the end of the ct-ng build so it's in the same environment (paths and flags...) so it takes a a while :P |
# 05:15:35 |
wbx |
bhundven: you could also try a hello world statically linked and use run |
# 05:15:47 |
bhundven |
true |
# 05:15:53 |
wbx |
to see if your compiler works with the simulator |
# 05:16:06 |
wbx |
maybe big endian is default in run ;) |
# 05:16:33 |
bhundven |
I'll give that a shot, It's still ~40 minutes on my laptop to built far enough to run -run |
# 05:21:30 |
bhundven |
hrm |
# 05:21:35 |
bhundven |
cd |
# 05:23:23 |
bhundven |
gdb-7.10/sim/arm/configure |
# 05:25:51 |
wbx |
puh. firefox needs so long to cross-compile. and then if it fails ... |
# 05:26:36 |
bhundven |
my family does not have the bald trait. If I worked on firefox, I would be bald. |
# 05:27:17 |
bhundven |
gdb and toolchains are annoying, but not enough to pull hair out |
# 05:27:41 |
bhundven |
I get frustrated using firefox... no clue what it is like to dev for. |
# 05:28:09 |
bhundven |
although, I used to use gentoo :) |
# 05:28:20 |
bhundven |
(2005-ish) |
# 05:29:55 |
wbx |
there is rather no good alternative for small laptops and cross-compiling. you could use dillo. |
# 05:31:41 |
bhundven |
I currently use debian. The only thing I compile is my kernel, and crosstool-ng. |
# 05:32:01 |
bhundven |
:) |
# 05:32:35 |
bhundven |
(well, other little projects I mess with, but they are really not worth mentioning) |
# 05:32:41 |
wbx |
debian is not running on all of my systems ;) |
# 05:33:12 |
bhundven |
well, I wouldn't get debian on my cortex-m3 |
# 05:34:00 |
wbx |
yeah. but it could be useful on a lemote yeelong. but i think you can only use Debian/mips no mips64 available. |
# 05:34:02 |
bhundven |
I guess it *might* be possible on my galileo or rasp-pi, but I'd rather run buildroot or yocto |
# 05:35:15 |
wbx |
needs another coffee to wake up before work |
# 05:35:32 |
bhundven |
I used to have a bunch of sparc station III's, an sgi Challange S, a BeBox... |
# 05:35:47 |
bhundven |
misses the BeBox... |
# 05:39:50 |
wbx |
have you given the machines away? |
# 05:39:57 |
bhundven |
yea |
# 05:40:02 |
bhundven |
sold them all |
# 05:40:22 |
bhundven |
I only have my embedded and dev/eval boards left. |
# 05:41:16 |
wbx |
embedded stuff is simpler if you are moving around a lot |
# 05:41:55 |
bhundven |
yea, I've moved quite a bit. Maybe when I buy a house, I'll buy up some old hardware again. I'm sure I'll find another BeBox ;) |
# 05:43:45 |
bhundven |
oh |
# 05:43:55 |
wbx |
i have some more bigger devices in my collection since I moved to a rented house. more place in my hobby room :) |
# 05:43:58 |
bhundven |
I guess MAME now can emulate both BeBox 66 and 133 |
# 05:44:58 |
bhundven |
wbx, yea. I lived out of my suite case for a few months after startup-fallout |
# 05:45:32 |
bhundven |
hard to carry 5 sparc station III's around ;) |
# 06:10:49 |
wbx |
bhundven: i think arm gdb simulator is only for bare-metal newlib compilers. so if you use toolchain targeting linux you better try qemu :) |
# 06:11:11 |
wbx |
i cannot exec any hello world app |
# 06:22:56 |
bhundven |
hrm |
# 06:22:56 |
bhundven |
ok |
# 06:23:44 |
wbx |
all refs in the net showing that only newlib compilers are used |
# 06:26:58 |
bhundven |
thanks for your help wbx! |
# 06:28:22 |
wbx |
no prob. |
# 06:29:08 |
wbx |
found a bug meanwhile in OpenADK, make check was broken. so it always is good for sth to help ;) |
# 06:32:22 |
wbx |
bhundven: you could use qemu-arm instead, you don't need to start full blown qemu-system-arm with ssh daemon |
# 06:36:36 |
bhundven |
do you know of any examples for running make check-... with? |
# 06:38:11 |
wbx |
no, i use qemu-system-arm with ssh. may be i'll try later to figure it out. |
# 07:11:48 |
bhundven |
sounds like I'd have to create a chroot |
# 07:36:41 |
diorcety |
joins #crosstool-ng |
# 08:40:47 |
diorcety |
quits : Quit: Leaving. |
# 08:58:52 |
y_morin |
joins #crosstool-ng |
# 09:19:09 |
diorcety |
joins #crosstool-ng |
# 09:54:29 |
Net147 |
quits : Ping timeout: 246 seconds |
# 09:56:53 |
Net147 |
joins #crosstool-ng |
# 12:07:44 |
y_morin |
quits : Ping timeout: 256 seconds |
# 12:10:13 |
y_morin |
joins #crosstool-ng |
# 12:20:59 |
bhundven |
quits : Quit: Leaving |
# 12:29:09 |
bhundven |
joins #crosstool-ng |
# 12:33:33 |
bhundven |
http://www.goes.noaa.gov/browsw.html |
# 12:33:35 |
bhundven |
nice |
# 12:33:54 |
bhundven |
the jet stream looks like a firehose |
# 12:52:21 |
y_morin |
quits : Remote host closed the connection |
# 12:57:37 |
y_morin |
joins #crosstool-ng |
# 13:15:44 |
bhundven |
y_morin, didn't buildroot used to have a external-toolchain/crosstool-ng option? |
# 14:11:56 |
y_morin |
bhundven: Hey! Yes, there was. But we removed it. |
# 14:12:04 |
y_morin |
bhundven: The reasons were: |
# 14:12:21 |
y_morin |
1) it was hard to match Buildroot's options down to ct-ng options |
# 14:12:43 |
y_morin |
2) Buildroot had gained support for glibc (and later musl), which made ct-ng support a bit moot. |
# 14:13:34 |
bhundven |
3) Buildroot gained support of y_morin :D |
# 14:13:40 |
bhundven |
jk |
# 14:13:44 |
bhundven |
lol |
# 14:14:03 |
y_morin |
bhundven: Maybe part of it, yes, but I was not involved in the toolchain stuff in Buildroot. kos_tom was. |
# 14:14:14 |
bhundven |
:) |
# 14:14:41 |
bhundven |
luckily you can still use an external toolchain. I just found configuring ct-ng a bit clunky |
# 14:15:15 |
y_morin |
bhundven: Yes, external toolchain support is there to stay in BR. |
# 14:15:33 |
y_morin |
bhundven: Sorry, I still need to review your two series... |
# 14:15:47 |
bhundven |
I committed 282 (less gcc versions part 2 |
# 14:16:08 |
bhundven |
the custom locations rewrite is currently broken... my fault |
# 14:16:09 |
y_morin |
yep, saw that. |
# 14:16:33 |
y_morin |
(wsa a comment on the less-gcc-version series) |
# 14:31:23 |
bhundven |
ok, fixed #288. I was missing an endif in config/libc/glibc.in from the rebase |
# 14:31:47 |
y_morin |
bhundven: OK, I'll review it now (after I launch a test-build in BR). |
# 14:31:48 |
bhundven |
I still want to update the help entries |
# 14:32:00 |
bhundven |
but I'm just looking for functional review |
# 14:32:04 |
bhundven |
Cool! |
# 14:32:41 |
bhundven |
also, I need to fix the order of options so they are consistent |
# 14:55:55 |
y_morin |
bhundven: It seems an easy cleanup would be to get rid of the mpfr autoreconf case for version 1.x ad n2.{0,1,2,3} since they are no longer supported. |
# 15:00:40 |
bhundven |
I just pushed another update that makes the options in consistent locations and with update help entries. |
# 15:00:53 |
bhundven |
s/update/updated |
# 15:01:15 |
bhundven |
y_morin: ok! |
# 15:02:17 |
y_morin |
bhundven: I'm doing the mpfr cleanup right now. |
# 15:02:26 |
bhundven |
nice! |
# 15:08:43 |
y_morin |
bhundven: I'm wondering why I added autoreconf for 2.4.x ... |
# 15:08:47 |
y_morin |
git log... |
# 15:21:45 |
bhundven |
y_morin: Updated 288 with your review suggestions |
# 15:22:16 |
bhundven |
hrm, the interactive rebase didn't re-order a commit... |
# 15:23:50 |
bhundven |
ok... It reordered it, but github doesn't reorder it? |
# 15:24:03 |
bhundven |
github is strange sometimes. |
# 15:31:55 |
bhundven |
y_morin: ok, I'm good to merge this or did you have anything else? |
# 15:33:08 |
y_morin |
bhundven: I still need to review the last two commits... |
# 15:33:32 |
bhundven |
ok |
# 15:33:49 |
y_morin |
bhundven: Weird, the MR is still not updated with the new commits order... ;-/ |
# 15:33:57 |
bhundven |
yea |
# 15:34:12 |
bhundven |
it looks right in my 'git log' |
# 15:34:56 |
bhundven |
first the xtensa removal of custom root dir, then the removal of custom root dir. |
# 15:35:12 |
y_morin |
bhundven: Yeah, lemme git-fetch your tree, then... |
# 15:36:04 |
y_morin |
bhundven: Not here. |
# 15:36:36 |
y_morin |
bhundven: Fetching bhundeven Already up-to-date. |
# 15:36:44 |
y_morin |
bhundven: Did you forgot to push? ;-) |
# 15:36:51 |
bhundven |
o.O |
# 15:36:53 |
bhundven |
hang on |
# 15:37:29 |
bhundven |
~/build/crosstool-ng @custom_locations_rewrite |
# 15:37:29 |
bhundven |
⯠git push --force bhundven custom_locations_rewrite |
# 15:37:29 |
bhundven |
Everything up-to-date |
# 15:38:29 |
bhundven |
strange... |
# 15:39:06 |
y_morin |
bhundven: Well... |
# 15:39:08 |
y_morin |
hides... |
# 15:39:34 |
y_morin |
bhundven: I have a local cache of a lot of git trees. |
# 15:39:53 |
y_morin |
bhundven: And my working dirs are cloned off that cache, not upstream. |
# 15:40:07 |
y_morin |
bhundven: So, when the cache is not up-to-date, then my working trees are not. |
# 15:40:22 |
y_morin |
bhundven: Now that I have updated the local cache, I get your new branch! ;-) |
# 15:40:29 |
bhundven |
yea, i think we both work the same way |
# 15:42:04 |
bhundven |
cd ~/src |
# 15:42:27 |
bhundven |
git clone --mirror https://github.com/crosstool-ng/crosstool-ng |
# 15:42:37 |
bhundven |
cd crosstool-ng.git |
# 15:42:55 |
bhundven |
git remote add bhundven git@github.com:bhundven/crosstool-ng |
# 15:43:01 |
bhundven |
cd ~/build |
# 15:43:11 |
bhundven |
git clone file:///home/bhundven/src/crosstool-ng.git |
# 15:43:15 |
bhundven |
;) |
# 15:44:32 |
y_morin |
bhundven: Exactly, except for ~/cache/upstream instead of ~/src ;-) |
# 15:48:27 |
bhundven |
punches his .gitconfig |
# 15:49:01 |
bhundven |
sometimes when you do interactive rebase, it does whitespace fixups on warning. |
# 15:49:11 |
bhundven |
when it's really breaking whitespace, not fixing it. |
# 15:57:44 |
bhundven |
ger, I've made a few whitespace errors |
# 16:00:50 |
bhundven |
y_morin: ok, whitespace fixes for those commits, although I need to go back and audit whitespace for the project -_- |
# 16:21:53 |
y_morin |
bhundven: Hmm... Sorry, but I think there was a misundertanding on the indentation... |
# 16:22:14 |
y_morin |
bhundven: In scripts/build/binutils/binutils.sh |
# 16:22:21 |
y_morin |
Arg, bad copy-paste.. |
# 16:23:01 |
y_morin |
bhundven: In scripts/build/binutils/binutils.sh: the CT_GetCustom is now indented by 4 spaces, while it was 8 before (and your previous patch made that 7). |
# 16:23:34 |
y_morin |
bhundven: I'm not an indentation fanatic, but IMHO it is best to keep sane indentatio. |
# 16:26:20 |
bhundven |
Ok. My gitconfig is doing this. |
# 16:26:46 |
bhundven |
I just went through and fixed the indentation, and now it's not how I set it. |
# 16:26:48 |
y_morin |
bhundven: What setting in your gitconifg is changing the indentation level? |
# 16:27:05 |
y_morin |
(so that I never turn it on!) |
# 16:27:24 |
bhundven |
It's from my home directory backup from a few years back |
# 16:27:37 |
bhundven |
I removed the option, but I lost my previous backup |
# 16:28:29 |
bhundven |
first was: |
# 16:28:33 |
bhundven |
[apply] |
# 16:28:38 |
bhundven |
whitespace = fix |
# 16:28:47 |
bhundven |
f that option, it always messes stuff up |
# 16:29:04 |
bhundven |
second is: |
# 16:29:08 |
bhundven |
[core] |
# 16:29:20 |
bhundven |
whitespace = space-before-tab,indent-with-non-tab,trailing-space |
# 16:29:55 |
y_morin |
indent-with-non-tab <-- bad... :-( |
# 16:30:01 |
bhundven |
yea |
# 16:30:09 |
bhundven |
So they are gone from my config now |
# 16:30:12 |
bhundven |
both |
# 16:39:24 |
bhundven |
y_morin: 4th times a charm? :D |
# 16:40:20 |
bhundven |
fixed indent, and removed default "" from custom location and custom version for all components |
# 16:53:20 |
y_morin |
bhundven: I think it's good now. |
# 17:09:29 |
diorcety |
quits : Quit: Leaving. |
# 17:11:54 |
bhundven |
y_morin: I added one more change to that series |
# 17:12:09 |
bhundven |
to clean up whitespace issues. |
# 17:12:24 |
bhundven |
https://github.com/bhundven/crosstool-ng/commit/ba1694051ad284d69e3268a54ef6295fcf7381dc |
# 17:49:31 |
daleg_ |
quits : Quit: Leaving... |
# 17:52:34 |
bhundven |
pushed again, to fix a bug with glibc's config |
# 17:53:15 |
bhundven |
custom was depending on LIBC_GLIBC_2_21_or_later, which was removed in the less gcc versions part 2 |
# 17:53:40 |
bhundven |
so now glibc custom depends on LIBC_GLIBC_2_20_or_later |
# 17:59:02 |
diorcety |
joins #crosstool-ng |
# 19:08:39 |
bhundven |
ghezz, I keep finding little things. Just fixed linux headers too. |
# 19:31:07 |
bhundven |
y_morin: ok, I think the scripts, configs, and new whitespace patch are ready to be reviewed again when you have time. |
# 19:31:34 |
bhundven |
I just ran a few different builds with custom sources, everything works now. |
# 19:50:35 |
alan_o |
quits : Ping timeout: 260 seconds |
# 20:03:38 |
alan_o |
joins #crosstool-ng |
# 20:23:54 |
bhundven |
quits : Quit: Leaving |
# 21:30:27 |
ragedragon |
joins #crosstool-ng |
# 22:56:50 |
y_morin |
quits : Quit: Nighty Night! :-) |
# 23:56:40 |
diorcety |
quits : Quit: Leaving. |