ibotlog2html for #crosstool-ng

<< Previous 2016-03-03 Next >>

# 00:20:22 mingwandroid joins #crosstool-ng
# 00:25:13 bhundven https://github.com/crosstool-ng/crosstool-ng/issues/361
# 00:25:20 bhundven *sigh*
# 01:36:53 blueness quits : Quit: blueness
# 02:15:02 bhundven_ joins #crosstool-ng
# 02:23:48 bhundven quits : Ping timeout: 248 seconds
# 02:23:52 systmkor quits : Ping timeout: 248 seconds
# 02:23:56 systmkor_ joins #crosstool-ng
# 02:37:31 blueness joins #crosstool-ng
# 02:38:14 blueness quits : Client Quit
# 02:58:12 mingwandroid quits : Ping timeout: 260 seconds
# 03:44:24 ctngbot joins #crosstool-ng
# 03:45:05 aleek1 joins #crosstool-ng
# 03:45:05 s1341 joins #crosstool-ng
# 03:45:05 systmkor_ joins #crosstool-ng
# 03:45:05 bhundven_ joins #crosstool-ng
# 03:45:05 Triskelios joins #crosstool-ng
# 03:45:05 coffeeholic_ joins #crosstool-ng
# 03:45:05 philenotfound joins #crosstool-ng
# 03:45:05 JakeSays joins #crosstool-ng
# 03:45:05 enunes joins #crosstool-ng
# 03:45:05 xenoxaos joins #crosstool-ng
# 03:45:05 Net147 joins #crosstool-ng
# 03:45:05 feepbot joins #crosstool-ng
# 03:45:05 stefanct joins #crosstool-ng
# 03:45:05 imMute joins #crosstool-ng
# 03:45:05 aquarat joins #crosstool-ng
# 03:45:05 ius joins #crosstool-ng
# 03:45:05 ovf joins #crosstool-ng
# 03:45:05 blino joins #crosstool-ng
# 03:45:05 roolebo joins #crosstool-ng
# 03:45:05 trigg_ joins #crosstool-ng
# 03:45:05 trog-lo-bit joins #crosstool-ng
# 03:45:05 blueness_ joins #crosstool-ng
# 03:45:07 muep joins #crosstool-ng
# 03:45:07 devster31 joins #crosstool-ng
# 03:45:07 KaibutsuX joins #crosstool-ng
# 03:45:07 smartin joins #crosstool-ng
# 03:45:07 wbx joins #crosstool-ng
# 03:45:07 ChanServ joins #crosstool-ng
# 03:45:07 kos_tom joins #crosstool-ng
# 03:45:07 giminy joins #crosstool-ng
# 04:24:21 blueness joins #crosstool-ng
# 14:05:21 ctngbot joins #crosstool-ng
# 14:05:55 bhundven ctngbot ...
# 14:09:53 enunes bhundven: I have already generated a aarch64 toolchain before with ct-ng and it built a fine Buildroot image which at least boots aarch64 kernel
# 14:10:10 enunes bhundven: in qemu
# 14:12:31 enunes bhundven: IIRC I just took a working arm-unknown-linux-gnueabi and changed 32-bit to 64-bit and it worked...
# 14:18:12 bhundven huh
# 14:21:06 bhundven yup, maybe I failed twice
# 14:21:39 bhundven Apocalypse:armv8-rpi3-linux-gnueabihf bhundven$ ct-ng saveconfig MKDIR samples
# 14:21:42 bhundven Invalid configuration `aarch64v8-rpi3-linux-gnueabi': machine `aarch64v8-rpi3' not recognized
# 14:21:45 bhundven that's a no go
# 14:22:11 bhundven well, that didn't paste well to irc
# 14:22:23 bhundven Apocalypse:armv8-rpi3-linux-gnueabihf bhundven$ ct-ng saveconfig
# 14:22:33 bhundven MKDIR samples
# 14:22:40 bhundven Invalid configuration `aarch64v8-rpi3-linux-gnueabi': machine `aarch64v8-rpi3' not recognized
# 14:24:30 enunes bhundven: well maybe there's some fix to support these specific cpu options, I just found the issue 361 title a bit heavy, support for aarch64 is already there
# 14:24:52 bhundven enunes: yes: 06:21 <@bhundven> yup, maybe I failed twice
# 14:28:41 enunes well at the time I noticed the lack of a very basic sample for aarch64 as we have it for arm-unknown-linux-gnueabi , I guess I can submit one
# 14:28:45 enunes maybe also add to travis?
# 14:28:52 bhundven would be good
# 14:29:02 bhundven ct-ng saveconfig, ftw
# 14:29:04 bhundven :)
# 14:30:06 enunes ok then, will do that soon
# 14:31:01 bhundven ah, dang, that error I pasted earlier, more me being dumb
# 14:31:21 bhundven (suffix to the arch name)
# 14:31:26 bhundven XP
# 14:33:15 bhundven not my best last two days, focus/attention wise
# 15:19:33 bhundven enunes: https://github.com/crosstool-ng/crosstool-ng/pull/362
# 15:19:35 bhundven :)
# 15:25:06 bhundven just updated it to add CT_LOG_EXTRA=y
# 16:01:48 enunes bhundven: I'm a little confused now, the other rpi3 config was for aarch32 then?
# 16:02:14 enunes yeah apparently so
# 16:31:27 diorcety quits : Ping timeout: 268 seconds
# 16:41:43 bhundven enunes: yup
# 16:42:47 bhundven well, in config.sub there isn't aarch32
# 16:42:56 bhundven arm* is
# 16:43:02 bhundven so arm* is 32
# 16:43:08 bhundven aarch64 is 64-bit
# 16:45:46 bhundven big endian as well, where they would be armeb and aarch64_be.
# 16:47:43 bhundven enunes: interestingly, since RPI3 only has 1G of memory, and really all the RPI developers wanted was more speed, they opted to ship with a 32-bit OS
# 16:48:11 bhundven hence, flipping to 64-bit has no real advantage over 32-bit in that case.
# 16:48:54 bhundven which is why there is no real push to have 64-bit, besides someone that just wants to run their RPI3 in 64-bit. (which, hey, I totally get!)
# 16:49:48 bhundven But now for the fun part!
# 16:50:01 bhundven waiting patiently...
# 16:50:12 bhundven for somone to ask why...
# 16:50:37 bhundven the binaries created with the aarch64 toolchain don't work on their RPI3... >:)
# 16:51:20 bhundven or why u-boot won't boot their new 64-bit kernel
# 16:51:39 bhundven becuase I think you need to rebuild u-boot for 64-bit
# 16:51:49 enunes bhundven: well the first rpi3 advertisements enforce the 64-bit part, which seems to be odd given that they don't take advantage of it
# 16:52:32 bhundven compatibility
# 16:52:38 enunes bhundven: as for the u-boot part, isn't the limitation that gpu firmware? powerpc64 kernel at least runs fine from a 32-bit u-boot
# 16:53:25 bhundven I'm throwing sticks in the dark... It's a guess.
# 16:53:52 bhundven I'm trying to anticipate the support questions that will come in, once I merge the aarch64 sample.
# 16:53:54 JakeSays is now known as: ForEachSays
# 16:54:04 enunes well I totally agree with the part that eventually someone is going to complain that their aarch64 binaries don't work :)
# 16:55:32 bhundven should check the rpi forums to see if they have released 64bit firmware yet
# 16:56:36 ForEachSays is now known as: ForEachSaysNo
# 16:57:07 enunes I don't own any raspberry pi due to this boot firmware from gpu stuff
# 16:57:56 bhundven I have a v1 b+
# 16:58:09 ForEachSaysNo is now known as: ForEachSays-fore
# 16:58:16 ForEachSays-fore is now known as: JakeSays
# 16:58:41 bhundven I'm slightly sore that the hardware developers did not work with upstream u-boot to get their work in upstream.
# 16:59:06 bhundven I mean, it's one of the most popular maker development boards, why not?
# 16:59:32 bhundven I know there are people working on it
# 16:59:47 bhundven but I think that should have been one of those "up front" kind of things.
# 17:01:01 bhundven intel has their chips in the linux kernel and boards working in u-boot before their released.
# 17:01:18 bhundven s/their/they're/g
# 17:01:33 bhundven maybe not /g
# 17:01:35 bhundven :P
# 17:02:06 bhundven not sure how you'd only match and replace the second instance
# 17:02:23 enunes haha
# 17:02:32 enunes well maybe adding the previous word in the pattern
# 17:04:12 bhundven stackoverflow found a few ideas
# 17:04:49 bhundven shows how easily distracted he can get
# 17:05:29 diorcety joins #crosstool-ng
# 17:06:07 bhundven well, this sample builds and on the surface looks good
# 17:06:31 bhundven I don't have a board to test with, and it's early enough that it may take some work before it's bootable/usable
# 17:07:07 bhundven thinking about just merging the PR for the aarch64 sample
# 17:08:29 bhundven Guessing Mike (UK) is either asleep or on the way...
# 17:15:32 enunes I can build test the aarch64 sample, but I don't have a rpi3 to boot it
# 17:15:52 bhundven well, it's cortex-a53
# 17:16:02 bhundven I'll add a generic aarch64 sample
# 17:21:40 y_morin joins #crosstool-ng
# 17:34:44 bhundven enunes: https://github.com/crosstool-ng/crosstool-ng/pull/364
# 17:34:52 bhundven https://travis-ci.org/crosstool-ng/crosstool-ng/builds/113460984
# 17:34:55 bhundven specifically
# 17:35:05 bhundven https://travis-ci.org/crosstool-ng/crosstool-ng/jobs/113460989
# 17:35:17 bhundven but it will take a bit before that build starts
# 17:35:43 bhundven #364 and #366 are ahead of it.
# 17:36:02 bhundven but that doesn't stop you from downloading the PR and giving it a shot ;)
# 17:37:42 enunes bhundven: yeah I'll do that. I'm on a ubuntu machine right now, which isn't my unsupported Arch machine :)
# 17:38:01 bhundven heh
# 17:38:37 bhundven well, now that the gcc check for static is in (although still needs to be changed), the default static settings won't affect you
# 17:39:11 bhundven the change that needs to be made was a slight oversite on my part
# 17:39:36 bhundven while the default system cc (gcc/clang) may or may not support static linking
# 17:39:53 bhundven it's possible to provide a different "build compiler" by setting the build tuple
# 17:40:13 bhundven and it's possible that toolchain may support static linking
# 17:40:37 bhundven so checking in the ./configure script might be too early
# 17:41:27 bhundven there is a static check in crosstool-NG.sh.in
# 17:42:13 bhundven but it needs to be setup to toggle CT_CONFIGURE_has_static_link=y if available and unset it if not.
# 17:43:15 bhundven technically speaking, it could be fine to have both a global check (in ./configure) and a local (in crosstool-NG.sh.in), as the later will/can overwrite the former.
# 17:46:38 enunes bhundven: I guess I understand it, it takes the settings from the host cc which may not be the same cc to build the cross toolchain
# 17:46:50 enunes does it mean that with the current implementation, it doesn't break with lack of static libs anymore?
# 17:46:56 enunes for cc == build cc
# 17:48:39 bhundven well, if the build cc does not support static linking, then any option that would enable any static linking would be unavailable.
# 18:00:41 bhundven gonna bail out for a few hours.
# 18:01:55 enunes bhundven: ok, I'll leave the report here for the tests with aarch64-unknown-linux-gnueabi
# 18:05:03 bhundven ok, thanks
# 18:05:06 bhundven !
# 18:15:10 enunes bhundven: aarch64-unknown-linux-gnueabi built successfully and, bonus test, it builds my aarch64 buildroot config for qemu and boots fine too
# 21:25:30 diorcety quits : Quit: Leaving.
# 22:34:45 y_morin quits : Quit: Nighty Night!

Generated by ibotlog2html by Yann E. MORIN