**** BEGIN LOGGING AT Fri Jun 15 03:00:11 2018 Jun 15 07:21:40 Omegamoon: interesting, maybe because of the USB-C adapter? Jun 15 07:25:19 btw I did the OTA update today and it worked.. Jun 15 07:52:01 greguu: I am still puzzled why this usb/uart cable doesn't work to be honest Jun 15 07:52:53 I don't have that much usb-c cables, so I'm reluctant to cut one to pieces to bypass the usb-c adapter Jun 15 07:53:47 as far as I know, the usb-c adapter is a pretty dumb 1:1 connection Jun 15 07:54:12 so that should work... i think ;) Jun 15 08:01:21 greguu: switching channels to #gemini-pda :) Jun 15 08:09:15 Omegamoon: NotKit = TheKit Jun 15 08:09:22 greguu, gm Jun 15 08:10:00 so problem arises when building armv5e. armv5t or armv5 are ok Jun 15 08:10:00 ant_work: good evening Jun 15 08:10:05 definitely gcc bug Jun 15 08:10:45 hm, interesting. why did you build for arm5e ? Jun 15 08:10:55 tried as well with 'xscale' feature but gcc-runtime fails Jun 15 08:11:04 my void box is on gcc 7.3 Jun 15 08:11:23 that was my Q. Do you build with 'e' = dsp feature? Jun 15 08:12:53 I tried: Jun 15 08:13:09 DEFAULTTUNE = "armv5" -> TUNE_FEATURES = "arm armv5" Jun 15 08:13:34 DEFAULTTUNE = "armv5e" -> TUNE_FEATURES = "arm armv5 dsp" Jun 15 08:13:57 DEFAULTTUNE = "armv5t" -> TUNE_FEATURES = "arm armv5 thumb" Jun 15 08:14:19 previously it was Jun 15 08:14:27 DEFAULTTUNE = "armv5te" -> TUNE_FEATURES = "arm armv5 thumb dsp" Jun 15 08:14:41 these 2 with 'e'/dsp get miscompiled Jun 15 08:15:26 I build "-march=armv5te -mtune=xscale -msoft-float -mfloat-abi=soft -Os -pipe -fomit-frame-pointer -funit-at-a-time" Jun 15 08:16:19 must be a OE config specific issue maybe Jun 15 08:16:31 ok, I tried xscale as well, then I get with our OE toolchain Jun 15 08:16:46 DEFAULTTUNE = "xscale" -> TUNE_FEATURES = "arm armv5 thumb dsp xscale" Jun 15 08:16:57 but CC warning about -mcpu=xscale switch conflicts with -march=armv5te switch Jun 15 08:16:58 and without *thumb* Jun 15 08:17:22 its -mtune Jun 15 08:17:27 not -march Jun 15 08:17:29 so maybe the trick is like you did -mtune=xscale Jun 15 08:17:42 yes Jun 15 08:18:04 I'll repeat with gcc7.3, all above tests done with gcc8.1 Jun 15 08:18:33 ok, and drop the "thumb" Jun 15 08:19:17 well, is not the culprit here Jun 15 08:19:51 it only has to do with gcc being over enthusiastic about combining store/loads Jun 15 08:20:51 I think is a regression after 7.2 Jun 15 08:25:44 build5/voidz-packages/masterdir-armv5tel-musl/bin/gcc -v Jun 15 08:25:52 gcc version 7.3.0 (GCC) Jun 15 08:26:09 not sure Jun 15 08:26:26 what version did you try so far ? Jun 15 08:26:41 7.3 and 8.1 Jun 15 08:28:00 hm, all my void builds since May have been with 7.3 I think Jun 15 08:28:07 and work fine Jun 15 08:28:18 never tried 8.x Jun 15 08:28:54 http://cgit.openembedded.org/openembedded-core/tree/meta/recipes-devtools/gcc/gcc-7.3.inc?h=master-next Jun 15 08:29:00 with these patches Jun 15 08:29:27 do you use a vanilla 7.3.0? Jun 15 08:32:02 no Jun 15 08:32:06 void patches Jun 15 08:32:07 https://github.com/greguu/voidz-packages/tree/voidz-packages-v01/srcpkgs/gcc/patches Jun 15 08:32:45 https://github.com/greguu/voidz-packages/blob/voidz-packages-v01/srcpkgs/gcc/template Jun 15 08:34:05 the patches are not special Jun 15 08:34:13 so almost 7.3 vanilla Jun 15 08:35:22 wow, so much patches in OE gcc Jun 15 08:35:35 bound to break things... Jun 15 08:38:45 heh, I keep my fingers away from toolchain Jun 15 08:39:57 try vanilla Jun 15 08:45:35 the problem we have is, ARMv5 and ARMv5E are deprecated Jun 15 08:45:49 gcc9 will get rid of armv4 and maybe of these Jun 15 08:45:55 https://gcc.gnu.org/gcc-7/changes.html Jun 15 08:46:13 Support for the ARMv5 and ARMv5E architectures has been deprecated (which have no known implementations) and will be removed in a future GCC release. Note that ARMv5T, ARMv5TE and ARMv5TEJ architectures remain supported. The values armv5 and armv5e of -march are thus deprecated. Jun 15 08:46:53 so they want us to use thumb Jun 15 09:03:46 I see now, Thumb as a way to support that legacy instructions... instead of supporting them proper Jun 15 09:04:04 well, this should not affect Cxx00 Jun 15 09:04:19 but any before ? Jun 15 09:04:33 Cxx0 should be fine too Jun 15 09:11:28 yes Jun 15 09:11:49 but on this patch we lose xscale/iwmmx afais Jun 15 09:11:59 *on this path Jun 15 09:17:21 no, xscale and iwmmxt shoudld be fine Jun 15 09:59:04 bbl Jun 15 20:06:05 greguu, applie this fresh patch but no luck.. Jun 15 20:06:06 https://patchwork.ozlabs.org/patch/924885/ Jun 15 20:07:51 I am now back to 7.2 to test Jun 15 23:21:30 greguu, heh, that one is ok **** ENDING LOGGING AT Sat Jun 16 03:00:04 2018