**** BEGIN LOGGING AT Sat Jan 01 02:59:58 2011
Jan 01 06:57:47 <CIA-77> swalker * r24861 /packages/net/miau/ (Makefile files/miau files/miau.init files/miaurc):
Jan 01 06:57:47 <CIA-77> [packages] miau: update to 0.6.6
Jan 01 06:57:47 <CIA-77> * drop the local outdated miaurc in favor of upstream's example
Jan 01 06:57:47 <CIA-77> * drop /etc/default/miau
Jan 01 06:57:47 <CIA-77> * add conffile
Jan 01 06:57:47 <CIA-77> * use better upstream url
Jan 01 06:57:47 <CIA-77> * remove trailing whitespace
Jan 01 07:03:02 <CIA-77> swalker * r24862 /packages/utils/gnupg/ (Makefile patches/001-mips_gcc4.4.patch):
Jan 01 07:03:02 <CIA-77> [packages] gnupg: update to 1.4.11
Jan 01 07:03:02 <CIA-77> * update patch to include the post release fixes
Jan 01 07:03:02 <CIA-77> * use PKG_INSTALL
Jan 01 08:04:43 <builder> build #58 of at91 is complete: Failure [failed compile_4]  Build details are at http://tksite.gotdns.org:8010/builders/at91/builds/58
Jan 01 08:21:05 <swalker> ^ missing 2.6.35 AT91_ADC symbol
Jan 01 08:35:48 <builder> build #55 of ubicom32 is complete: Failure [failed compile_4]  Build details are at http://tksite.gotdns.org:8010/builders/ubicom32/builds/55
Jan 01 09:03:20 <builder> build #58 of brcm47xx is complete: Failure [failed compile_4]  Build details are at http://tksite.gotdns.org:8010/builders/brcm47xx/builds/58
Jan 01 09:45:25 <CIA-77> swalker * r24863 /packages/net/git/ (Makefile patches/001-convert_builtin.patch): [packages] git: update to 1.7.3.4, refresh patches
Jan 01 10:26:12 <builder> build #49 of sibyte is complete: Exception [exception failed slave lost shell_13 compile_12]  Build details are at http://tksite.gotdns.org:8010/builders/sibyte/builds/49
Jan 01 10:26:13 <builder> build #59 of atheros is complete: Exception [exception failed slave lost shell_13 compile_12]  Build details are at http://tksite.gotdns.org:8010/builders/atheros/builds/59
Jan 01 10:26:17 <builder> build #55 of cobalt is complete: Exception [exception failed slave lost shell_13 compile_12]  Build details are at http://tksite.gotdns.org:8010/builders/cobalt/builds/55
Jan 01 14:08:51 <CIA-77> swalker * r24864 /packages/net/mercurial/ (Makefile patches/001-no_bzip2.patch): [packages] mercurial: update to 1.7.2, refresh patches
Jan 01 14:17:54 <builder> build #56 of ppc40x is complete: Failure [failed shell_9]  Build details are at http://tksite.gotdns.org:8010/builders/ppc40x/builds/56
Jan 01 14:25:00 <mazilo> Has anyone looked into all OpenWRT packages that require perl package have their modules disabled and/or with "--without-perl=yes" or "--with-perl=no" option?
Jan 01 14:27:32 <[florian]> mazilo: I do not think anyone is doing real work with perl and openwrt, so if you volunteer doing that, do not hesitate
Jan 01 14:27:41 <[florian]> happy new year by the way ;)
Jan 01 14:28:58 <mazilo> florian: Happy new year to you too. Unfortunately, I am no programmer. I wish I could contribute in this sense. So, fixing problem with perl is out of my kills. :(
Jan 01 14:29:39 <mazilo> florian: It looks like perl package is to blame.
Jan 01 14:31:09 <mazilo> florian: When installed on a running OpenWRT system, the perl package provides its proper path. However, it is not so when it comes to a cross compilation that requires OpenWRT perl residing on the staging_dir.
Jan 01 14:31:49 <mazilo> florian: I start to noticed this when I tried to fix the Freeswitch mod_perl.
Jan 01 14:31:55 <[florian]> actually I am not really surprised about that
Jan 01 14:32:06 <[florian]> cross-compiling perl is not a straight process
Jan 01 14:33:05 <mazilo> florian: Exactly.
Jan 01 14:35:45 <xMff> mazilo: LOCAL_CFLAGS= -w -DMULTIPLICITY $(shell $(PERL) -MExtUtils::Embed -e ccopts)    is this the only place where it calls perl?
Jan 01 14:35:51 <xMff> in FS
Jan 01 14:36:06 <mazilo> xMff: It looks like it.
Jan 01 14:36:52 <mazilo> xMff: If I changed that line to "LOCAL_CFLAGS= -w -DMULTIPLICITY -DEMBED_PERL -I$(STAGING_DIR)/usr/lib/perl5/5.10/CORE', mod_perl compiled OK. But, that doesn't seem to be a proper way to do it, does it?
Jan 01 14:37:20 <xMff> you could try something like:  PERL:='echo $(STAGING_DIR)/usr/lib/perl5/5.10/CORE'
Jan 01 14:37:55 <mazilo> xMff: When perl version gets update, this mod_perl compilation will fail again until the version 5.10 on the quick dirty hack gets changed too.
Jan 01 14:38:33 <xMff> is this LOCAL_CFLAGS thing in one of the sources that are downloaded later?
Jan 01 14:39:58 <mazilo> xMff: If you mean the src/mod/languages/mod_perl/Makefile will have the same LOCAL_CFLAGS on the git version, then the answer is yes.
Jan 01 14:45:06 <xMff> you could patch it
Jan 01 14:45:53 <xMff> LOCAL_CFLAGS= -w -DMULTIPLICITY -DEMBED_PERL -I$(firstword $(wildcard $(STAGING_DIR)/usr/lib/perl5/5.*/CORE))
Jan 01 14:46:20 <mazilo> xMff: OK. I will try and let you know.
Jan 01 14:56:22 <builder> build #51 of rb532 is complete: Success [build successful]  Build details are at http://tksite.gotdns.org:8010/builders/rb532/builds/51
Jan 01 15:22:07 <CIA-77> hauke * r24865 /trunk/tools/Makefile: tools: add some missing dependencies
Jan 01 15:54:23 <CIA-77> jow * r24866 /trunk/tools/e2fsprogs/ (Makefile patches/004-big_endian_compile_fix.patch): [tools] revert r24851, it breaks compilation on many hosts
Jan 01 16:07:03 <CIA-77> jow * r24867 /trunk/package/e2fsprogs/Makefile: [package] e2fsprogs: revert r24848 as well
Jan 01 16:13:27 <CIA-77> hauke * r24868 /trunk/package/crda/Makefile: crda: update to version 1.1.1 and update regulatory database to most recent official version.
Jan 01 16:55:47 <CIA-77> jow * r24869 /packages/net/ddns-scripts/ (3 files in 2 dirs): [packages] ddns-scripts: use /proc/uptime as time reference, this makes it immune against time warps (#8557)
Jan 01 17:34:31 <generatorglukoff> hello
Jan 01 17:34:41 <generatorglukoff> i'm using asus wl520gc(u) router
Jan 01 17:34:53 <generatorglukoff> and have cpu frequency problem
Jan 01 17:35:14 <generatorglukoff> what is the best way to fix it?
Jan 01 17:35:42 <xMff> generatorglukoff: 200 vs. 240mhz ?
Jan 01 17:36:23 <generatorglukoff> yeap
Jan 01 17:36:54 <xMff> ping Hauke
Jan 01 17:37:04 <Hauke> xMff: pong
Jan 01 17:37:44 <xMff> afaik the current brcm47xx code hardcodes 240MHz for the bcm5354 but some models run with only 200mhz
Jan 01 17:38:02 <xMff> this must be decided by examining the clkfreq nvram var
Jan 01 17:38:23 <xMff> let me find the appropriate spot
Jan 01 17:39:15 <generatorglukoff> time.c?
Jan 01 17:39:27 <xMff> yes, that could be the right one
Jan 01 17:39:27 <generatorglukoff> arch/mips/bcm47xx/time.c?
Jan 01 17:39:57 <generatorglukoff> or in ssb code
Jan 01 17:40:16 <Hauke> I think there is something in the ssb code
Jan 01 17:40:31 <generatorglukoff> ssb_cpu_clock()
Jan 01 17:40:44 <generatorglukoff> drivers/ssb/driver_mipscode.c
Jan 01 17:40:59 <generatorglukoff> anyway
Jan 01 17:41:16 <generatorglukoff> right now i have 240 in cpufreq
Jan 01 17:41:21 <generatorglukoff> *clkfreq
Jan 01 17:41:41 <xMff> and your cpu runs slower?
Jan 01 17:41:54 <generatorglukoff> seems so
Jan 01 17:42:42 <generatorglukoff> i'm not sure who set that value
Jan 01 17:43:07 <generatorglukoff> because i've used dd-wrt for some time
Jan 01 17:43:26 <mazilo> xMff: I have rebuilt FS v1.0.6 from scratch using your patch and it works ONLY if the line PERL=`which perl` is replaced with PERL="$(STAGING_DIR)/usr/bin/perl" or add PERL="$(STAGING_DIR)/usr/bin/perl" to the section on "Build/Compile" of the FS Makefile for OpenWRT (net/freeswitch/Makefile). Without this, the compilation won't produce mod_perl.so file.
Jan 01 17:43:29 <generatorglukoff> however dd-wrt have hack around that
Jan 01 17:44:56 <xMff> http://wl500g.googlecode.com/svn/trunk/kernel/002-bcm5354.patch
Jan 01 17:45:09 <xMff>  /* 5354 could run both on 200 & 240 Mhz -- use nvram setting */
Jan 01 17:45:17 <xMff> afaik the upstream code has nothing like that
Jan 01 17:45:21 <xMff> ssb
Jan 01 17:49:08 <generatorglukoff> btw, does cfe use value of clkfreq?
Jan 01 17:49:16 <xMff> I don't know
Jan 01 17:53:35 <KanjiMonster> generatorglukoff: yes
Jan 01 17:59:49 <generatorglukoff> changing clkfreq to 200 doesn't change anything
Jan 01 18:17:55 <generatorglukoff> maybe it's just a board quirk?
Jan 01 18:18:11 <generatorglukoff> something that should be detected from get_router() ?
Jan 01 19:10:03 <KanjiMonster> generatorglukoff: "5354 chipcommon pll setting can't be changed." according to broadcom source, so the 200 mhz/ 240 mhz might be pin strapping setting, so nothing that can be changed through software
Jan 01 19:11:11 <generatorglukoff> ok
Jan 01 19:11:29 <generatorglukoff> so, we should just detect that, right?
Jan 01 19:11:42 <generatorglukoff> i have
Jan 01 19:11:53 <generatorglukoff> boardnum=45
Jan 01 19:12:01 <generatorglukoff> boardtype=0x48E
Jan 01 19:12:07 <generatorglukoff> and boardrev=0x10
Jan 01 19:12:48 <generatorglukoff> KanjiMonster: could this values be used to add cpu timing quirk for wl520gc(u) board?
Jan 01 19:13:36 <KanjiMonster> generatorglukoff: without knowing if there are other 5354 boards have different values there is no way to tell
Jan 01 19:16:29 <generatorglukoff> KanjiMonster: maybe there is some list of this values?
Jan 01 19:19:32 <KanjiMonster> generatorglukoff: I do not know of any. Perhaps broadcom has one, but I doubt you'll get access to it ;)
Jan 01 19:23:47 <generatorglukoff> hmm
Jan 01 19:24:17 <generatorglukoff> KanjiMonster: i'm just thinking about the best way to fix that issue
Jan 01 19:24:57 <generatorglukoff> KanjiMonster: maybe dd-wrt way with separate Fix_WL520GUGC_clock=1
Jan 01 19:25:04 <generatorglukoff> in nvram
Jan 01 21:39:25 <xMff> ping cshore_
Jan 01 23:05:50 <generatorglukoff> Hauke: ping
Jan 01 23:15:01 <CIA-77> jow * r24870 /packages/libs/libiconv/ (30 files in 4 dirs): [packages] libiconv: extend the library to support more charset conversions
Jan 01 23:17:58 <CIA-77> jow * r24871 /packages/net/ettercap/Makefile: [packages] ettercap: use mini iconv (again)
Jan 01 23:18:57 <CIA-77> jow * r24872 /packages/net/lsm/Makefile: [packages] lsm: fix compilation with ccache (#8539)
Jan 01 23:19:02 <cshore_> pong xmff
Jan 01 23:19:28 <cshore_> pong xMff
Jan 01 23:20:26 <xMff> cshore_: I spotted some oddities in preinit, 50_choose_console
Jan 01 23:21:00 <xMff> cshore_: the if / elif statement is supposed to pick a device node based on available hotplug mechanisms
Jan 01 23:21:27 <cshore_> right
Jan 01 23:21:33 <xMff> the dd check afterwards will allways succeed if hotplug2 is included, therfore M0, M1 and M2 get overwritten
Jan 01 23:21:57 <xMff> the 30_something does mknod /dev/console as part of init_hotplug2
Jan 01 23:22:01 <cshore_> the dd will only succeed if there is actually a console
Jan 01 23:22:08 <xMff> so it will basically make the logic in 50 fail
Jan 01 23:22:25 <xMff> cshore_: seems to not be the case
Jan 01 23:22:47 <cshore_> hmm....it's supposed to be able to work without a console
Jan 01 23:23:10 <cshore_> the dd is the check that was there before I made the modularization
Jan 01 23:23:29 <xMff> right
Jan 01 23:23:36 <PaulFertser> cshore_: hi :)
Jan 01 23:24:00 <cshore_> PaulFerster: hi
Jan 01 23:24:58 <PaulFertser> cshore_: so, the issue is basically that when an x86 target has no serial port preinit aborts without giving any clue to why that happens.
Jan 01 23:25:37 <PaulFertser> cshore_: i determined it happens on first "echo" invocation after the preinit execs itself with input/output redirected to /dev/console
Jan 01 23:26:24 <PaulFertser> cshore_: the trick to read 0 bytes with dd from /dev/console (present in 50_choose_console) doesn't help, it considers /dev/console to be sane and functional.
Jan 01 23:28:21 <cshore_> PaulFerster: do you know a way that works?
Jan 01 23:28:37 <PaulFertser> Examing busybox's sources lead me nowhere, i couldn't see why it would abort, even when output is non-functional.
Jan 01 23:29:02 <PaulFertser> cshore_: unfortunately, no idea.
Jan 01 23:29:31 <cshore_> PaulFerster: the odd thing is that apparently it used to work
Jan 01 23:29:49 <CIA-77> jow * r24873 /packages/net/subversion/Makefile: [packages] subversion: get rid of libintl dependency
Jan 01 23:30:17 <PaulFertser> cshore_: it probably used to work for some other case. mgm seems to have introduced that trick.
Jan 01 23:30:38 <xMff> PaulFertser: did you check what happens if /dev/ptmx is used instead?
Jan 01 23:31:48 <PaulFertser> xMff: looks like it doesn't abort that way but no messages from preinit (like "press f", etc) are seen either.
Jan 01 23:31:58 <xMff> http://paste.openwrt.org/d6039aa56
Jan 01 23:33:03 <xMff> PaulFertser: what is the current menu.lst btw?
Jan 01 23:33:21 <xMff> PaulFertser: kernel cmdline in particular
Jan 01 23:34:14 <PaulFertser> xMff: root=/dev/sda2 rootfstype=ext2 rootwait console=tty0 console=ttyS0,38400n8 noinitrd reboot=bios
Jan 01 23:34:21 <xMff> hm
Jan 01 23:34:42 <xMff> shouldn't /dev/ptmx correlate to the first virtual terminal then somehow?
Jan 01 23:35:01 <PaulFertser> I boot the image in qemu, and it works that way, but fails the way i describe when i change ttyS0 to ttyS1 (that's how i test it).
Jan 01 23:35:03 <xMff> or is the second console= taking precedence
Jan 01 23:36:36 <PaulFertser> xMff: i guess ptmx terminals are not attached to the console by default in any way.
Jan 01 23:36:56 <xMff> hm true
Jan 01 23:37:29 <xMff> aha
Jan 01 23:37:46 <xMff> "The last device [given on the kernel cmdline] will be used when you open /dev/console"
Jan 01 23:38:26 <xMff> so ttyS0 always takes precedence, even if not present
Jan 01 23:38:59 <xMff> for generic x86 one would need to swap serial and vga terminals
Jan 01 23:39:15 <xMff> for embedded x86 this would be undesirable though
Jan 01 23:40:19 <xMff> if I see it correctly, the only reason we specify a console at all is to pass a baudrate
Jan 01 23:40:40 <xMff> unfortunately there seems to be no way to give a baudrate without mentioning a specific device
Jan 01 23:41:03 <xMff> the default behaviour is described as " If no console device is specified, the first device found capable of acting as a system console will be used. At this time, the system first looks for a VGA card and then for a serial port."
Jan 01 23:41:11 <xMff> which is basically what we want
Jan 01 23:41:54 <PaulFertser> xMff: i think it would be rather ok if the boot aborted or wasn't fully functional if only the user would be able to get the message asking to fix consoles manually.
Jan 01 23:42:38 <PaulFertser> But i see no way to detect this condition :/
Jan 01 23:43:15 <xMff> hm
Jan 01 23:44:53 <xMff> interssting
Jan 01 23:45:33 <xMff> when I do "echo foo > /dev/ttyS3" on my desktop I get an I/O error when running under bash and a no-op when running as "sh"
Jan 01 23:48:40 <xMff> I wonder why the preinit script aborts at all, I don't see a "set -e" anywhere
Jan 01 23:48:54 <xMff> so it should just continue running
Jan 01 23:49:14 <PaulFertser> I tried reading ash.c sources but that didn't help me to understand why it aborts.
Jan 01 23:49:16 <xMff> PaulFertser: can you try an explicit "set +e" ?
Jan 01 23:49:21 <PaulFertser> xMff: sure
Jan 01 23:50:27 <xMff> provided that we find a way to survive the writes to the not existing terminal we could wrap echo into a procedure which echoes into both /dev/console and /dev/tty0
Jan 01 23:50:58 <xMff> hm, stupid idea
Jan 01 23:52:05 <PaulFertser> xMff: i added "set +e" to the beginning of /etc/preinit, didn't help
Jan 01 23:52:33 <xMff> odd
Jan 01 23:54:45 <cshore_> PaulFerster have you seen the thread that talks about debugging preinit?
Jan 01 23:55:53 <xMff> PaulFertser: here's a completely different approach: http://paste.openwrt.org/d5f69fb57
Jan 01 23:56:15 <PaulFertser> cshore_: no
Jan 01 23:56:15 <xMff> it bqasically gives vga priority if its enabled
Jan 01 23:56:56 <xMff> meh stupid
Jan 01 23:57:40 <xMff> this way: http://paste.openwrt.org/d59ebc6d6
Jan 01 23:59:48 <PaulFertser> cshore_: found and skimmed it.
Jan 02 00:00:45 <cshore_> PaulFerster: glad you found it...I didn't - didn't where it was and missed it when I was looking
Jan 02 00:00:57 <cshore_> *didn't remember
Jan 02 00:01:11 <PaulFertser> cshore_: google is my friend :)
Jan 02 00:02:11 <cshore_> :)
Jan 02 00:02:58 <PaulFertser> xMff: of course it would be nicer to have a unified x86 image working. I'm not sure if it worth spending your time on this question anymore, this proved to be harder than i first thought :/
Jan 02 00:05:44 <PaulFertser> Probably it would be enough to change "Attempting to kill init" to "Preinit died. See http://wiki.openwrt.org/... for the list of possible reasons".
Jan 02 00:07:43 <PaulFertser> And this message would be seen on all working consoles because it'd come from the kernel.
Jan 02 00:12:23 <CIA-77> jow * r24874 /packages/libs/libshout/Makefile: [packages] libshout: fix linking
Jan 02 00:12:55 <Hauke> generatorglukoff: pong
Jan 02 00:14:50 <generatorglukoff> Hauke: i've send mail into list
Jan 02 00:15:00 <generatorglukoff> it's about early serial console
Jan 02 00:15:20 <generatorglukoff> i can't compile kernel with serial driver as module
Jan 02 00:15:30 <generatorglukoff> or switch it off completely
Jan 02 00:15:48 <generatorglukoff> i've posted update to your patch
Jan 02 00:15:49 <Hauke> generatorglukoff: ok thank you I will add your patch tomorrow
Jan 02 00:16:01 <generatorglukoff> thanks
Jan 02 00:21:09 <PaulFertser> While we're at it, can someone tell me anything about my yaddns patch? Should i fix it somehow or just wait or remind about it in some time or forget about it?
Jan 02 00:22:34 <xMff> will look into it later
Jan 02 00:24:15 <PaulFertser> Ok, thanks. ttyl, gotta sleep now. Good night :)
Jan 02 00:25:45 <CIA-77> jow * r24875 /packages/libs/libtheora/patches/001-no_docs_tests.patch: [packages] libtheora: avoid libSDL build dependency by removing call to AM_PATH_SDL in configure.ac
Jan 02 00:35:08 <cshore_> xMff: what's the status on rc4/10.03.1?
Jan 02 00:35:24 <xMff> I have no idea
Jan 02 00:35:30 <xMff> our release team is mia
Jan 02 00:35:49 <cshore_> ic
Jan 02 00:37:33 <cshore_> How long do you think we should wait before really getting worried?
Jan 02 00:38:40 <xMff> well
Jan 02 00:39:48 <cshore_> ah, i wasn't aware of that.
Jan 02 00:57:20 <CIA-77> jow * r24876 /packages/net/imspector/patches/: [packages] imspector: remove iconv workarounds
Jan 02 01:17:58 <CIA-77> jow * r24877 /trunk/tools/missing-macros/ (4 files in 3 dirs): [tools] missing-macros: add as-compiler-flag.m4 and as-version.m4
Jan 02 01:26:58 <CIA-77> jow * r24878 /trunk/tools/missing-macros/ (Makefile src/README src/m4/as-unaligned-access.m4): [tools] missing-macros: add as-unaligned-access.m4
**** ENDING LOGGING AT Sun Jan 02 02:59:58 2011