irclog2html for uclibc on 2002.10.14

00:32:03psypeteHmm
00:32:41psypetewould it be easy or difficult to embed udhcp in one of my projects? i.e., can it just "drop in" like in busybox with a wrapper or two?
01:16:54psypeteis away: food food
06:24:40aaronlandersee: grrrr
06:24:47aaronl/home/aaronl/src/busybox/busybox/archival/dpkg.c: In function `unpack_package':
06:24:47aaronl/home/aaronl/src/busybox/busybox/archival/dpkg.c:1336: warning: implicit declaration of function `deb_extract'
06:24:51aaronl/home/aaronl/src/busybox/busybox/archival/dpkg.c:1336: `extract_quiet' undeclared (first use in this function)
06:29:52anderseelooks like bug1 broke it...
06:46:51aaronlhehe
06:46:59aaronlandersee: this is cool
06:47:16aaronlandersee: i just made a graph of -finline-limit value vs binary size
06:47:35aaronlandersee: there is a range where it dips below the -finline-limit=0 binary size
06:50:45psypetehullo?
06:51:03psypeteis back (gone 05:34:07)
06:53:24anderseeaaronl: that sounds very odd
06:53:33aaronlnot to me
06:53:47aaronlinlining small functions will always be a win
06:53:52aaronlbut -Os won't inline any functions by itself
06:54:58anderseeInlining small functions (~40 bytes) that would otherwise be in a shared library is certainly a win
06:56:00aaronli found that between -finline-limit=35 and -finline-limit=75 the size of busybox is decreased
06:56:00anderseeInlining very small functions also eliminates ELF header entries.
06:56:08aaronli'd ratehr bitch and moan to the gcc developers though
06:56:18aaronlto make gcc automatically inline any functions with -Os that are wins
06:56:19psypeteandersee: i have a question
06:56:34anderseeSounds like the right thing...
06:56:55anderseepsypete: the answer is maybe
06:57:04anderseepsypete: :)
06:57:43psypeteandersee: do you know the udhcp code well?
06:58:38anderseeNot especially so.  Russ Dill is the author.
06:59:02anderseepsypete: Russ is on at least #tuxscreen at the moment
06:59:34psypeteandersee: thanks
07:01:16psypeteandersee: how easily does udhcp incorporate into busybox? would it be hard to drop into one of my own projects?
07:01:40anderseepsypete: it is already in the devel version of busybox...
07:03:43psypeteandersee: i know... i was wondering how hard it is to drop into another program, as a non-main type function?
07:05:21anderseeI'm not sure I follow you.
07:05:29psypetefrom what i read
07:05:29anderseeWell, Russ is on looks like...
07:05:33psypeteooo
08:15:09psypeteis away: .
16:55:34kergothokay guys, question. I've got /dev -> /var/dev, setup identical to the devices in the tuxscreen root. problem is, umount /dev/hda1 fails. fstab references /dev/hda1, but i think umount is reacting badly to the fact that the real path to the device is /var/dev/hda1.
16:55:38kergothideas? :)
21:50:16coobhi, anyone alive?
21:50:20kergothno
21:50:29coobcool, a channel of zombies :D
21:51:10kergothmumbles about needing to eat human flesh
21:51:15coobI've installed a uClibc toolchain into /toolchain, installation went fine, however I'm having problems compiling busybox
21:51:18coobusing...
21:51:20coobmake CROSS=/toolchain/bin/arm-uclibc- PREFIX=/toolchain/newroot/
21:52:04coobi get... bah closed the aterm lemme recompile shouldn't take a sec.
21:52:41sjhillhey andersee
21:52:55coobok.
21:53:00coobthis odd, it works now
21:53:02coobirc magic
21:53:04coob:)
21:59:22anderseesjhill: hey
21:59:44anderseecoob: whats the problem
21:59:57sjhillandersee: i'm taking time now to fix that 'crt0.s' problem and then i'll have a look at the dload issue
22:00:01anderseecoob: oh.  it works.
22:00:11anderseesjhill: wohoo!  cool, thanks
22:00:24anderseesjhill: I think I fixed mips already
22:00:27anderseelooking
22:00:59anderseesjhill: yup, I fixed crt0.S already
22:01:18sjhillandersee: in that native toolchain?
22:01:25anderseesjhill: So nothing to worry about there.  Just the wierd dlopen problem
22:01:43sjhillandersee: well, how did you fix the crt0.s thing? i need that
22:02:12anderseesjhill: It should work just fine with the updated wrapper or with a native toolchain.  Just needs to call __uClibc_start_main().  See libc/sysdeps/linux/mips/crt0.S in CVS.
22:02:45anderseerevision 1.6
22:02:45anderseedate: 2002/10/01 17:25:57;  author: andersen;  state: Exp;  lines: +12 -14
22:02:45anderseePatch from Marshall M. Midden <m4@brecis.com> to fixup crt0 for
22:02:45anderseemips where I had made a mess
22:03:25coobtrying to put together a newer verison of revol
22:03:36cooblast revision was like may i think
22:03:50coobjust trying to figure out what the autor did heh
22:08:04coobhrm
22:08:16coobwonder if openss{l,h} wil work.
22:09:53anderseecoob: I've heard reports of openssh problems, which I plan to look into today.
22:10:46coobah ok.
22:10:50coobI just need the client.
22:10:55coobit's for a psion heh
22:11:06coobmight be overkill.
22:11:14coobwonder if there's a 'lite' ssh client.
22:12:16BZFlaghttp://www.lysator.liu.se/~nisse/lsh/
22:12:29coob:D thanks
22:12:44kergothexplodes
22:13:24BZFlagducks to avoid the flying kergoth pieces
22:23:44coobdoes busybox come with al components compiled in by default?
22:25:51coobdiscovers applet_source_list
22:48:39coobah Config.h even better heh
22:51:22anderseecoob: Config.h is your friend.  And no, not everything is compiled in by default
22:52:10coobyeah figured thanks
22:52:16coobjeez bustbox is sweet :D
22:52:22coobbusybox rather
22:52:26coobdamn tiny keyboard
23:05:50coobum, to cross compile a kernel will standard gcc suffice or do i need to tell the Makefile to use my cross compiler
23:10:17BZFlagset ARCH in the Makefile or on the command line.
23:10:37BZFlagand you need a xcompile toolchain of course.
23:11:19coobARCH := arm
23:11:29coobyeah i have a toolchain in /toolchain
23:11:38coobhow do i point the kernel to it.
23:11:50coobCROSS_COMPILE   = arm-linux-
23:11:51coobaha
23:12:05BZFlagyep. and be sure it's on the path.
23:12:25coobchange that to arm-uclibc-
23:12:58coob:D thanks

Generated by irclog2html.pl by Jeff Waugh - find it at freshmeat.net! Modified by Tim Riker to work with infobot logs, split per channel and by date, etc.