IRC log for #uclibc on 20100126

00:06.42*** join/#uclibc animeloe[net] (n=animeloe@unaffiliated/animeloe)
00:24.38*** join/#uclibc malloc_ (n=malloc@2001:470:1f11:80b:216:3eff:fe70:1be0)
00:24.38*** join/#uclibc djdeath_ (n=djdeath@potipota.net)
00:24.38*** join/#uclibc gnomon (n=gnomon@CPE0022158a8221-CM000f9f776f96.cpe.net.cable.rogers.com) [NETSPLIT VICTIM]
00:25.08*** join/#uclibc vapier (i=UserBah@wh0rd.org)
01:36.46*** join/#uclibc animeloe (n=animeloe@unaffiliated/animeloe)
02:05.43*** join/#uclibc DuratarskeyK (n=wedge@77-234-14-61.pppoe.yaroslavl.ru)
02:06.55*** join/#uclibc animeloe (n=animeloe@unaffiliated/animeloe)
02:20.48*** join/#uclibc dileX (n=sd@p5B2ED1A0.dip.t-dialin.net)
04:41.31*** join/#uclibc animeloe (n=animeloe@unaffiliated/animeloe)
04:49.07*** join/#uclibc animeloe (n=animeloe@unaffiliated/animeloe)
05:16.17*** join/#uclibc dileX (n=sd@vpn-eu2.unidsl.de)
06:15.28*** join/#uclibc dileX (n=sd@vpn-eu1.unidsl.de)
06:37.18*** join/#uclibc dileX_ (n=sd@vpn-eu1.unidsl.de)
06:39.18*** join/#uclibc y_morin (n=ymorin@ARennes-252-1-56-112.w83-195.abo.wanadoo.fr)
06:54.24*** join/#uclibc YeYo (n=YeYo@ns2.kolporter.com.pl)
06:57.47*** join/#uclibc ncopa (n=ncopa@245.39.189.109.customer.cdi.no)
07:04.36CIA-5903vda.linux 07master * r1aaf1cb096e2 10busybox/examples/var_service/getty_tty1/ru_koi8r.keymap: whitespace fix
07:18.09CIA-5903vda.linux 07master * rda879ec2e967 10busybox/modutils/modprobe.c: modprobe: protect against possible SEGV
07:20.00CIA-5903vda.linux 07master * re52b735231bd 10busybox/Makefile: Bump version to 1.16.0
07:20.48CIA-5903vda.linux 07refs/tags/1_16_0 * re52b735231bd 10busybox/Makefile: Bump version to 1.16.0
07:21.06CIA-5903vda.linux 071_16_stable * re52b735231bd 10busybox/Makefile: Bump version to 1.16.0
07:26.41*** join/#uclibc hw_ (n=hw@p578b3905.dip0.t-ipconnect.de)
07:31.09*** join/#uclibc hw_ (n=hw@p578b3905.dip0.t-ipconnect.de)
07:32.00Jacmetdjdeath_: posts to mailing list are also fine - Normally I handle things quicklier, but things have been kind of crazy here
07:32.21Jacmetdjdeath_: and I've been ill
07:35.11CIA-5903vda.linux 07master * re83877942982 10busybox/Makefile: start 1.17.x development
07:40.29*** join/#uclibc dileX (n=sd@p5B2ED1A0.dip.t-dialin.net)
07:41.35*** join/#uclibc hw_ (n=hw@p578b3905.dip0.t-ipconnect.de)
08:54.42CIA-5903jacmet 07master * r99f74cbec5d5 10buildroot/package/busybox/Config.in: busybox: bump version
09:01.41CIA-5903rep.dot.nop 07master * r7295c13f4be2 10uClibc-website/.gitignore: add some ignores
09:01.43CIA-5903rep.dot.nop 07master * rcc31f2cdaa48 10uClibc-website/ (4 files in 2 dirs): update copyright year and bribe address
09:17.46djdeath_Jacmet: ok, good recovery
09:18.17djdeath_Jacmet: indeed, I have seen a lot of patches on the mailing list
10:06.21CIA-5903vda.linux 07master * rffc4bced958e 10busybox/networking/libiproute/ (ipaddress.c iproute.c): libiproute: eliminate unused fields in struct filter_t's; style fixes
10:17.09dileXhttp://busybox.net/downloads/fixes-1.15.3/ seems to be a dead-link
11:15.13*** join/#uclibc _Lucretia__ (n=munkee@5ac2a2d0.bb.sky.com)
11:20.42mnemocdileX: that happens when there are no fixes for that version yet
11:31.19dileXOK. when 1st patch arrives the dir is created?
11:59.51*** join/#uclibc p0ker_ (n=p0ker@93.124.192.66)
12:12.10*** join/#uclibc gustavoz (n=gustavoz@host5.201-253-139.telecom.net.ar)
12:18.37CIA-5903julien.boibessot 07master * r48b628b8613d 10buildroot/ (CHANGES package/sdl/sdl.mk): sdl: needs tslib to be compiled first when selected.
12:36.27*** part/#uclibc YeYo (n=YeYo@ns2.kolporter.com.pl)
12:47.09ncopaanyone having issues with libxml2-2.7.6 and uclibc?
12:47.38ncopaxsltproc breaks when reading from stdin
12:48.12ncopaseems like it believes it hits EOF too early
12:54.54CIA-5903rep.dot.nop 07master * r5682ae1fcf3f 10uClibc/libc/ (signal/Makefile.in stdlib/Makefile.in): convert to foo-y kbuild style
13:02.44*** join/#uclibc animeloe[net] (n=animeloe@unaffiliated/animeloe)
14:05.25*** join/#uclibc animeloe[net] (n=animeloe@unaffiliated/animeloe)
14:14.03*** join/#uclibc tchan (n=tchan@lunar-linux/developer/tchan)
14:49.58*** join/#uclibc sjhill (n=sjhill@home.bethel-hill.org)
14:55.10*** join/#uclibc tsukasa` (n=tsukasa@unaffiliated/tsukasa)
15:01.56*** join/#uclibc animeloe[net] (n=animeloe@unaffiliated/animeloe)
15:04.44*** join/#uclibc animeloe (n=animeloe@unaffiliated/animeloe)
15:15.02*** join/#uclibc matteo (n=matteo@openwrt/developer/matteo)
15:35.06*** join/#uclibc adb (n=adb@144.85.45.157)
15:45.24*** join/#uclibc austinf (n=austinf@74.92.231.110)
15:59.33*** join/#uclibc wberrier (n=wberrier@12.191.193.40)
16:21.32*** join/#uclibc mnt_real (n=sinan@bas1-montreal43-1177754708.dsl.bell.ca)
16:29.00busybot[New Bug] https://bugs.uclibc.org/961 min, P5, buildroot, gustavo@zacarias.com.ar->unassigned@buildroot.uclibc.org, NEW, pending, Bump dnsmasq to 2.52
16:30.01busybot[New Bug] https://bugs.uclibc.org/967 min, P5, buildroot, gustavo@zacarias.com.ar->unassigned@buildroot.uclibc.org, NEW, pending, Bump netstat-nat to 1.4.10
16:33.01busybot[New Bug] https://bugs.uclibc.org/973 min, P5, buildroot, gustavo@zacarias.com.ar->unassigned@buildroot.uclibc.org, NEW, pending, Bump iw to 0.9.19
16:33.43gustavozJacmet: hi, around?
16:47.37*** join/#uclibc psypete (n=realname@li62-9.members.linode.com)
16:49.15psypetei have a buildroot issue... on slack 12, using i686 defconfig of buildroot-2009.11, both binutils 2.19 and 2.19.1 fail with "bucomm.c:154: error: 'TARGET' undeclared (first use in this function)"
16:50.13psypeteif anyone has seen this error before, do they know what this refers to? it seems like building binutils by itself outside of buildroot completes successfully, so i'm assuming maybe a patch or some compile option specified by buildroot may be causing it
17:01.15CIA-5903rep.dot.nop 07master * r29db573ed42b 10uClibc/Makefile.help: document UCLIBC_EXTRA_CFLAGS
17:28.53psypeteit appears as if the output/toolbox/binutils-2.19-build/ directory that configure and make run from are specifying all the -I and other gcc flags to point to the output/toolbox/binutils-2.19/ directory instead of the -build path. yet binutils-2.19-build/binutils/config.h is what defines TARGET... so the right configuration includes are not being found/used
17:45.19*** join/#uclibc lewellyn (n=lewellyn@greenviolet/lewellyn)
18:12.53blindvtaustinf, ping
18:13.10austinfyo
18:14.01blindvtaustinf, git grep builtin_define gcc/config/sparc; # doesn't enlighten me as to which CPP token denoted v7/v8
18:14.10blindvtaustinf, there's v9 though.
18:15.05blindvtaustinf, can you please either show me the respective tokens or follow-up to yann's mail to remove the sparc block and fixup the config tokens CONFIG_SPARC_V* ? TIA
18:16.39austinfI don't know the tokens off the top of my head, I'll have to look
18:18.06blindvtaustinf, (you most likely know, so just for your convenience..) echo '' | gcc-4.5-unknown-sparc -xc -E -dD -c - -o -
18:19.52blindvtfor j in 7 8 9 9b whatever;do ... -mcpu=v$i;done
18:20.09austinfyeah I don't see anything relevant either on my gcc
18:20.19blindvtcool stuff
18:20.27austinfyeah
18:21.23blindvtfucking broken pile of slowaris compat crap^C
18:21.30austinfyeah I hate it
18:22.09austinfit's caused me nothing but trouble
18:25.05Jacmetgustavoz: I am now
18:26.13gustavozJacmet: hi. i'll be working on the menu reorg in the next days, question is, do we wanna move the packages into subdirectories to order them up by category?
18:26.49gustavozat first i'd say "no" and leave it for later
18:29.25Jacmetgustavoz: I think the first step can just be to keep all files as they are and move libs to a seperate submenu in kconfig
18:29.44gustavozJacmet: exactly, also move the "dev" apps to a new submenu too
18:29.52Jacmetgustavoz: yes
18:30.22gustavozJacmet: i'll be touching descriptions also, mostly adding some useful ones since some packages lack the basic "what's this for"
18:30.37Jacmetgustavoz: ok, good
18:31.43gustavozJacmet: at some point we'll want to solve the category in dirs (or not) issue since some packages are that way now (database comes to mind)
18:33.02Jacmetgustavoz: yeah, the database one is imho a bit silly as it only has 2 entries, but multimedia is fairly big
18:33.29Jacmetgustavoz: but that's imho a secondary (internal) issue - users normally won't look under package/
18:34.27gustavozJacmet: the only useful thing about moving stuff is easier Config.in handling which can be accomplished by splitting it too
18:34.28austinfblindvt, yeah I don't see any easy way to pull those config options without losing the ability to pick the right version of certain .S files
18:34.58austinfblindvt, thanks to gcc
18:35.03Jacmetgustavoz: true
18:37.59gustavozJacmet: also there's mtd-utils-1.3.0 at last, though it requires some touching to get it rolling (ubifs at least)
18:38.50Jacmetgustavoz: I know - Actually there's 1.3.1, but I haven't gotten them to put a tarball on the ftp yet
18:39.09Jacmetgustavoz: 1.3.0 had a fairly serious bug in ubifs
18:39.52gustavozJacmet: i didn't get it to build cleanly in br, mostly a headers thingy, not too interested in ubi atm either
18:40.40Jacmetgustavoz: ubi is still on my todo list to look closer into, but it isn't a high priority atm
18:40.56gustavozJacmet: same thing here. did you get to play with the 9g45?
18:41.26Jacmetgustavoz: yes, a bit
18:42.05Jacmetgustavoz: I did a quick performance test comparing it to 9263 and a ppc platform we're using (mpc8347)
18:42.53Jacmetgustavoz: cpu intensive stuff (mp3 decoding) is ~2x of 9263 (400MHz vs 200MHz), but memory bandwidth (memcpy) is quite bad, worse than 9263
18:43.19gustavozJacmet: odd, it should perform better with 2x ddr2 channels
18:43.20Jacmetgustavoz: but it's an engineering sample board / sw, so perhaps they just need to tweak the DDR controller params
18:43.48gustavozJacmet: though ddr2 has more latency than the SDR the older ones use
18:44.22gustavozJacmet: did you test the ehci controller? as in, does it get even close to 480?
18:45.29Jacmetgustavoz: I got ~19MB/s for 1MB memcpy on the 9g45, ~22MB/s for the 9263, and ~40MB/s for the ppc
18:45.57Jacmetgustavoz: sorry, not yet - the first design we'll use it in won't have usb
18:48.56blindvtaustinf, please file a bug
18:49.17blindvtaustinf, (against gcc.gnu.org/bugzilla/ of course, not us)
18:49.51austinfwill do
18:51.12*** join/#uclibc fsvend (n=frank@77.233.16.62.customer.cdi.no)
18:59.17blindvtaustinf, TIA
18:59.41austinfso for now, can we not remove those config options :) ?
18:59.46austinffor sparc at least
19:02.16blindvtaustinf, if (TARGET_FPU) set_optab_libfunc (sqrt_optab, TFmode, "_Q_sqrt");
19:02.18gustavozJacmet: which benchmark did you use?
19:02.38gustavozJacmet: i'm getting generally better numbers with stream (http://www.greyhound-data.com/gunnar/glibc/index.htm?page=downloads)
19:02.48blindvtaustinf, you'll want to cross-compile perl and rely on auto-tools for configury ;)
19:03.05austinfoh my
19:03.35Jacmetgustavoz: I just decompressed a mp3 with madplay and used a small test program timing memcpy() calls
19:04.03gustavozJacmet: k give stream.c a shot since it measures with varying sizes/methods
19:04.11blindvttries his sardistic grin ;)
19:04.44Jacmetgustavoz: ok, I'll do a new test friday/early next week
19:05.07blindvtaustinf, see http://repo.or.cz/w/buildroot.git/blob/HEAD:/target/Config.in.arch#l364
19:05.44blindvtaustinf, and the sparc block below. I can imagine your pain :P
19:06.06austinfI only really care about v7,v8,v9,v9b for uClibc
19:06.34austinfand actually v8 is the only one I _really_ care about
19:06.37austinfand test
19:07.09blindvt<PROTECTED>
19:07.10blindvt429         string
19:07.10blindvt430         default V7      if BR2_sparc_v7 || BR2_sparc_cypress || BR2_sparc_sparclite || BR2_sparc_f930 || BR2_sparc_f934 || BR2_sparc_sparclite86x || BR2_sparc_sparclet || BR2_sparc_tsc701
19:07.10blindvt431         default V8      if BR2_sparc_v8 || BR2_sparc_supersparc || BR2_sparc_hypersparc
19:07.10blindvt432         default V9      if BR2_sparc_v9 || BR2_sparc_ultrasparc || BR2_sparc_ultrasparc3 || BR2_sparc_niagara || BR2_sparc64_v9 || BR2_sparc64_ultrasparc || BR2_sparc64_ultrasparc3 || BR2_sparc64_niagara
19:07.12blindvt433         default V9      if BR2_sparc_v9a || BR2_sparc64_v9a
19:07.14blindvt434         default V9B     if BR2_sparc_v9b || BR2_sparc64_v9b
19:07.37austinfyep
19:07.54blindvtthe marketing names are just that. But it's really a mess
19:08.14austinfthere might be like three users of sparc uClibc, and they all use v8 to my knowledge
19:08.47gustavozJacmet: did atmel hit final silicon for the 9g45? they are past due date AFAIK
19:09.10blindvtdoes v8 have vis?
19:09.26Jacmetgustavoz: I actually don't know - Atmel seems to have had some issues lately - strike at their plant in .fr
19:09.48gustavozblindvt: nope, VIS is a v9 thingy
19:09.50blindvtaustinf, or is vis just !!fpu
19:10.20austinfcorrect
19:10.47blindvtgustavoz, right (sorry for asking too fast):
19:10.50blindvt<PROTECTED>
19:10.50blindvt<PROTECTED>
19:10.50blindvt<PROTECTED>
19:10.50blindvt<PROTECTED>
19:10.50blindvt<PROTECTED>
19:10.51blindvt<PROTECTED>
19:10.52blindvt<PROTECTED>
19:10.55blindvt<PROTECTED>
19:11.53blindvthm.. so we basically need to see how to distinguish v9 from v9b. Ideas?
19:11.58gustavozblindvt: these days you should probably care about >=v8
19:14.56gustavozblindvt: v9 alone is baseline V9 (64 bit, starting with ultrasparc1), v9a is v9+vis, v9b is ultrasparc3+
19:15.20gustavozblindvt: v9b being V9 and VIS 2.0
19:15.26austinfyep
19:15.57austinf__sparc__ && __arch64__ will let you know you are at least at v9
19:16.58gustavozless than v8 is probably a waste of time, v8 would be gaisler sparc processors
19:18.52gustavozlast time i checked 2.6 kernels ran really bad or not at all for v8 hardware except for gaisler IP with their own patches
19:19.19austinfcorrect, I have an extensive patch adding a whole leon subarch to 2.6.30
19:19.26blindvti think that we don't care (much) about FPU in this respect, so for us v9+fpu==v9b(-vis, since we it isn't explicitly impled ATM) so we effectively can rely on defined __sparc_v9__ && __WORDSIZE
19:19.47blindvts/we it/it/
19:19.58blindvt~botcake
19:19.59ibotMmmmmm.  Cake.... droool
19:20.47blindvtaustinf, unless you add fenv support that cares about vis, we should get away with the available info. Thanks for flying with us
19:21.14austinfok
19:21.17blindvtenough. let's get real stuff done now
19:23.19austinfblindvt, I'm about 50% through syncing nptl from glibc btw
19:24.24blindvtaustinf, oh, and i'd add one prominent #if __WORDSIZE != 64 && defined __arch64__ || __WORDSIZE == 64 && !defined __arch64__\n#error you are toast\n#endif somewhere to sparc, but maybe that's just me ;)
19:24.47austinfwe'll get PI mutexes, private futexes, robust futexes
19:24.52blindvtaustinf, cool! fore!
19:25.42blindvtaustinf, i'm reimplementing /etc/* parsers as we speak (modulo arguing about freaking sparc stuff ;)
19:26.00*** join/#uclibc dougmencken (n=Douglas@93.123.156.139)
19:27.01busybot[New Bug] https://bugs.uclibc.org/979 nor, P5, Busybox, joao.fs.cardoso@gmail.com->unassigned@busybox.net, NEW, pending, "delgroup" doesnt delete a user from a secondary group
19:36.12*** join/#uclibc trem (n=trem@AToulouse-256-1-64-94.w86-205.abo.wanadoo.fr)
20:08.23*** join/#uclibc Dj-Death (n=djdeath@potipota.net)
20:53.14*** join/#uclibc y_morin (n=ymorin@ARennes-252-1-56-112.w83-195.abo.wanadoo.fr)
21:30.47*** join/#uclibc Christos_N (n=Christos@ppp-94-66-51-223.home.otenet.gr)
21:47.01busybot[New Bug] https://bugs.uclibc.org/985 enh, P5, buildroot, daniel@bristot.eti.br->unassigned@buildroot.uclibc.org, NEW, pending, Bump usb_modeswitch package to 1.1.0
21:49.20tremnite all, sweet dreams
22:34.34*** join/#uclibc animeloe[net] (n=animeloe@unaffiliated/animeloe)
22:42.13y_morinaustinf: ping?
22:42.33austinfi'm here
22:43.09y_morinDid you discuss the sparc stuff here, or on another channel?
22:43.24y_morinSo I can have a look at the archives...
22:43.57austinfit was here
22:44.08y_morinFact is that sparc is not something I know much about
22:44.14y_morin(here: ok)
22:44.22austinfdon't worry too much about it
22:45.08austinfsummary: basically CONFIG_SPARC_v7 we don't care about at all, and will probably drop altogether
22:45.43y_morinwhich means I'll have to drop my IPX :-)
22:46.04austinfCONFIG_SPARC_V8 has embedded users (myself + plus gaisler's customers)...and we can detect that with __sparc__ && WORDSIZE == 32
22:46.21austinfwe can't distinguish v7, v8 via gcc's builtin defines
22:46.52y_morinOK, so the deal is drop the now ageing v7 in favour of v8 ?
22:47.04austinfyes
22:47.25austinfalso, we can't distinguish between v9 and v9b, v9a
22:47.48austinfso we'll have to just select the v9 assembly versions
22:48.07austinfbut I don't know of any users of v9 and uClibc
22:48.12austinfanyways
22:48.44*** join/#uclibc CIA-40 (n=CIA@208.69.182.149)
22:50.20y_morinMy gcc does not know about v9a and v9b, am looking at gcc-4.4...
22:50.38y_morinanyway, v9a and v9b seem to be 64-bit, right?
22:50.48austinfyes
22:56.36*** join/#uclibc ibot (i=ibot@rikers.org)
22:56.36*** topic/#uclibc is discussion of uClibc and Busybox | uClibc 0.9.30.2 was released 13 Jan 2010 | busybox 1.15.3 was released 12 Dec 2009 (unstable busybox 1.16.git) | buildroot 2009.11 was released 1 December 2009 | For general setup issues try #elinux or #edevreminds me that
22:56.52y_morinBTW, is there a who's who on irc, smthg like nick_name -> real_name?
22:57.11austinfblindvt = Bernhard
22:57.18austinfsolar = Mike F I think
22:57.34austinfkhem = Khem Raj
22:57.40austinfaustinf = austinf
22:57.43austinf:)
22:58.14austinf~botcake
22:58.15ibotMmmmmm.  Cake.... droool
23:00.56y_morin:-)
23:10.55*** part/#uclibc wberrier (n=wberrier@12.191.193.40)
23:13.54khemaustinf: Mike Frysinger is vapier or spanKY
23:13.59austinfoh
23:14.03austinfwho si solar?
23:14.10austinfs/si/is/
23:20.11khemaustinf: solar is Ned Ludd
23:20.20austinfah
23:21.37khemPeter Korsgaard is Jacmet
23:22.03khemaustinf: how is nptl pull coming along ?
23:22.07austinfthis seems like something we should teach ibot
23:22.39austinfkhem, coming. I will need help with arm and mips most likely. I'll mail when I get there
23:22.52khemsure ok
23:23.28khemand Carmelo is not on IRC
23:33.24austinfibot, who is austinf?
23:33.25ibotyou are probably Austin Foxley
23:33.30austinfdone
23:33.52austinfibot, who is blindvt?
23:33.53ibotit has been said that blindvt is Bernhard Fischer
23:34.57*** join/#uclibc animeloe[net] (n=animeloe@unaffiliated/animeloe)
23:37.13austinfibot, who is y_morin?
23:37.15ibotI think you lost me on that one, austinf
23:45.56*** join/#uclibc tchan (n=tchan@lunar-linux/developer/tchan)

Generated by irclog2html.pl Modified by Tim Riker to work with infobot.