00:00.22 | landley | It seems it's needed to build gcc 4.0 on x86. :( |
00:01.10 | psm | landley: the addition influences too much, that cant be done as an option (it makes all future compiles link to libc.so and ld.so as well (the latter only if needed) |
00:01.33 | landley | I can't comment since I have no idea what your patch does. |
00:01.51 | landley | Alas, I stopped following the uclibc mailing list over 700 messages ago, when it turned into the buildroot list and stopped having anything to do with uclibc. |
00:02.10 | psm | begin reading again ;) |
00:02.11 | landley | Since I don't use buildroot, slogging through just stopped being worth it... |
00:02.48 | psm | send *buildroot* > /dev/null |
00:04.50 | landley | Would there be anything left? |
00:05.15 | landley | P.S: ashes, thanks for pointing me to the patch. The build continues to chug along... |
00:05.35 | andersee | landley: evening |
00:06.01 | landley | Hiya andersee! |
00:06.15 | landley | I got my Firmware Linux distro unblocked a few days ago, and am almost ready to make a release. |
00:06.28 | landley | Except that gcc 4.0.2 and uclibc don't like each other at all. :) |
00:06.41 | landley | Been fighting with that for... three days now? |
00:07.06 | andersee | landley: and you choose to go with the bleading edge, because....? |
00:07.09 | andersee | :-) |
00:07.26 | landley | Because I've got to upgrade at some point, might as well be now. |
00:07.43 | landley | (I could have jumped to 3.4, and it remains an option.) |
00:08.00 | landley | But considering that last time I worked out how to get the sucker to function from first principles... |
00:08.07 | landley | (And sticking &%&*#& printfs in the source code...) |
00:08.21 | landley | Oh: what stops gcc 4.0.2 from compiling with uClibc. 2 things: |
00:08.33 | landley | One I posted to the buildroot mailing list about |
00:09.07 | landley | That's the dl_iterate_phdr missing thing. |
00:09.30 | landley | (You ripped it out right before gcc started depending on it, apparently. Although only for something called "libmudflap", which nobody seems to know what it does.) |
00:09.33 | landley | Not even google. |
00:09.41 | landley | Which I found impressive, actually. |
00:10.00 | landley | The other is the libc_stack_end thing, which gcc patches (apparently wrongly, but it seems to work anyway) here: |
00:10.16 | landley | http://www.linuxfromscratch.org/patches/hlfs/svn/uClibc-0.9.28-libc_stack_end-1.patch |
00:10.54 | psypete | ...mudflap? |
00:11.02 | landley | I'm not making that up. |
00:11.37 | landley | http://www.tummy.com/Products/krud/packages/libmudflap.html |
00:11.52 | landley | It's built as part of gcc, but nobody anywhere will say what it's _for_. |
00:12.31 | landley | andersee: I'm trying to get it working. With a chainsaw if necessary. |
00:12.41 | andersee | heh |
00:13.06 | landley | Great. Bash 2.05b won't build with gcc 4.02 |
00:13.09 | landley | This seems to be unrelated. |
00:13.35 | landley | ake[1]: Entering directory `/home/landley/newbuild/firmware-build/tools/sources/bash-2.05b/lib/malloc' |
00:13.35 | landley | gcc -I. -I../.. -I../.. -I../../include -I../../lib -DHAVE_CONFIG_H -DSHELL -g -O2 -DRCHECK -Dbotch=programming_error -c malloc.c |
00:13.35 | landley | malloc.c:239: warning: integer constant is too large for 'unsigned long' type |
00:13.35 | landley | malloc.c: In function 'internal_free': |
00:13.35 | landley | malloc.c:904: error: label at end of compound statement |
00:13.37 | landley | make[1]: *** [malloc.o] Error 1 |
00:13.40 | landley | That's just painful. |
00:13.51 | landley | (No I am _not_ upgrading to bash 3, I'm going to FIX BUSYBOX'S SHELL. But not today.) |
00:14.06 | landley | Yes, but that would be _sane_. |
00:14.14 | andersee | heh |
00:15.20 | mjn3 | landley: i think libmudflap is some type of invalid pointer protection. as i recall, i did get it building when with gcc 4.0 in buildroot |
00:15.44 | landley | Is there any way to disable it? |
00:15.58 | landley | (So it doesn't try to build?) |
00:16.04 | mjn3 | thought you had to actually enable it with a config option |
00:16.10 | landley | It's far and away the most persnickety of the gcc build bits. |
00:16.17 | mjn3 | perhaps that's changed since the initial release |
00:16.19 | landley | Nope. It's building anyway in 4.0.2 |
00:17.12 | landley | Unless you're cross compiling, libmudflap tries to build with the gcc that just compiled but has not yet been installed. |
00:17.45 | landley | When tweaking things to get libraries from a nonstandard location and use nonstandard libraries anyway, this involves duct tape to get the sequencing right. |
00:18.12 | psypete | $landley_vocabulary++ |
00:18.13 | mjn3 | all uclibc toolchain builds are done as cross compilations in buildroot |
00:18.27 | landley | _why_ does bash have its own built-in implementation of malloc? |
00:18.46 | psypete | portability thingie? |
00:18.50 | landley | mjn3: I noticed that at least some code for --target=i386-linux-uclibc got merged. |
00:19.12 | landley | I intend to play with that later to see if I can simplify anything, but for right now I'm trying to get it to work with a soldering iron and tweezers, playing bug whack-a-mole... |
00:19.21 | mjn3 | vapier pushed some binutils changes and perhaps some gcc config stuff |
00:19.28 | landley | Cool. |
00:19.36 | landley | I look forward to using it. |
00:19.39 | landley | After I get a release out. :) |
00:20.08 | landley | I'm not kidding, bash has its own malloc implementation. |
00:20.21 | landley | It calls sbrk, and does coalescing and everything. |
00:20.47 | landley | And people wonder why emacs turned into an operating system if they can do this to a poor defenseless command shell... |
00:21.59 | landley | WANT TO HURT FSF! |
00:22.15 | landley | I have to wander away from the keyboard for a bit to calm down. |
00:22.30 | psypete | could you grab me some coffee while you're up pls? |
00:26.28 | landley | Just tea around here. |
00:26.38 | landley | I don't do coffee. |
00:26.45 | landley | It's really strong tea, though. |
00:26.58 | landley | And I've got some diet coke. |
00:27.04 | landley | Can't email that though. (Carbonation.) |
00:30.10 | landley | Right. gcc 4.0 objects if a label is the last thing in a function. |
00:31.30 | landley | Adding any function after the label (including do {;} while(0);) fixes it. |
00:31.36 | landley | That's not an error, that's a warning at best. |
00:32.01 | landley | I'd harange them about it, but I'm keeping my distance. |
00:32.06 | landley | I really don't want to get any of that on me. |
00:33.57 | landley | The leaves of what? |
00:34.07 | psypete | red tea leaves |
00:34.26 | landley | I just use lipton bags, which I'm told contain the sweepings from the factory floor of real tea factories. |
00:34.50 | psypete | yep, pretty much dirt sand and some tea extract |
00:35.09 | psypete | btw why not have a return() at the end of your functions? |
00:35.13 | landley | I find it's all in the preparation, although that's clearly because I don't have real tea readily available. |
00:35.21 | landley | http://www.bbc.co.uk/dna/h2g2/A61345 |
00:35.25 | landley | A great man. |
00:35.38 | landley | it's not my function, it's the fsf's function. (And it's returning void, that's why.) |
00:35.51 | landley | And it's got #ifdefs after the label. |
00:35.59 | landley | Four or five of them, with various functionality blocks. |
00:36.08 | psypete | yeah, if the water's not right you're gonna shock the tea and it'll be screwed anyway |
00:36.14 | landley | It's just that in this case, they're all disabled. But return; could potentially change behavior by cutting off the unreachable code. |
00:36.31 | landley | Oh I shock the tea anyway, I just make sure it's _hot_ enough. |
00:36.47 | landley | Most tea isn't made with really really really hot water. |
00:37.12 | landley | It's made with just sort of hot water, and yes there is a difference. |
00:37.55 | landley | (I actually throw the tea bags into the water while it's still boiling, on the theory "it's lipton". Turn off burner, throw in tea bags, put lid back on, go do something else. |
00:38.05 | landley | I've sometimes forgotten it's there for upwards of 24 hours. That's _good_ tea. :) |
00:38.18 | landley | (Diluted about 1/3 with milk, anyway. Wouldn't try drinking it straight...) |
00:38.38 | psypete | heh, yeah i would hope... i did that with green tea once. was NOT a happy camper. |
00:39.02 | landley | I sometimes use one bag of green tea to 8 bags normal tea (in a gallon of water), but green tea is a flavoring, not a tea. |
00:39.17 | landley | (And oolong tea has just plain gone rancid. I don't know what the point of it is.) |
00:40.39 | psypete | green tea will keep me up at night if i steep enough of the stuff |
00:41.03 | psypete | plus i only put about 2lbs (or 40 metric liters) of sugar per cup |
00:41.53 | landley | One cup of sugar per gallon works for me. I could do metric measurements but that's not what they sell here. |
00:42.08 | landley | So I have to occasionally figure out how to convert gilberts into furlongs and so on... |
00:45.50 | landley | hotplug? |
00:46.12 | psypete | yep, he linked me some patch for a dynamic hotplug busybox applet |
00:46.23 | landley | udev? |
00:46.32 | psypete | don't think so, this is for a primarily 2.4 system |
00:46.41 | landley | Ah. Much less interesting then. |
00:46.55 | landley | One of my near-term todo items is getting udev integrated into my firmware system. |
00:46.59 | psypete | hey, whatever makes my USB disks auto-unmount |
00:47.07 | landley | But that's after I release... |
00:47.22 | psypete | whats the application & medium size? |
00:47.38 | landley | I wonder how big/complicated udev is? (I know it contains a complete copy of klibc, which is dumb. It did back around .40, anyway.) |
00:49.18 | landley | The general design seemed nice and simple, though... |
00:49.29 | psypete | it's supposed to me more "intuitivey" |
00:49.36 | landley | What is? |
00:49.39 | psypete | udev |
00:49.53 | psypete | when i looked at it before, i kept going "ohh... ahhh..." |
00:49.59 | landley | If you mean it doesn't have 18,000 dev entries for every possible device... |
00:50.22 | psypete | well that and the layout of /sys/ made a lot more useful information accessible |
00:50.24 | landley | And it doesn't rename /dev/hda to /dev/fruitbasket/ampitheatre/bus1/slave/toshiba... |
00:50.37 | landley | I forgot lots of numbers in that path, by the way. |
00:50.42 | landley | And random punctuation... |
00:50.49 | psypete | yeah i was gonna say, add some $RAND here and there |
00:51.06 | landley | I'm entirely in favor of udev. |
00:51.19 | landley | Greg Kroah-Hartmann is a brilliant programmer. |
00:51.29 | landley | He should be kept away from release management via restrainign order, but other than that... |
00:51.48 | landley | And he handed it off recently. I forget to who. |
00:52.30 | landley | I think busybox is probably going to grow a "2.6" directory, with the new pivot_root replacement and udev and such. |
00:52.47 | landley | Although if 2.8 ever ships that'd look a bit silly, but I think we're safe for a while there. :) |
00:53.21 | psypete | heh, if 3.0 doesn't ship first \o/ |
00:54.44 | landley | Hey, I was arguing 2.6 should have been 3.0. |
00:54.51 | landley | In retrospect, that looks right, doesnt' it? |
01:03.09 | psypete | i'm never in favor of verion number jumps "for the heck of it", or even when there's been a major addition |
01:03.52 | psypete | if the software has been rewritten from scratch or otherwise is virtually a different product, yes a new major version number is welcomed |
01:04.44 | psypete | but we could go to 2.16 for all i care, if the codebase is the same one as the last, with just things like VM or tcp/ip stack changing |
01:05.29 | psypete | then again i'm not really a kernel hacker so i'm not one to talk. and i'm not Linus so i don't have much say in the matter. |
01:06.10 | landley | Where'd the bug list go? |
01:06.49 | landley | Version numbers should indicate major new functionality. We went to 1.0 and 2.0. |
01:07.05 | landley | The point of 2.0 is that 1.0 wouldn't really substitute for it. |
01:07.39 | psypete | well the same thing could be said for minor revisions in some cases |
01:07.42 | landley | Is anybody else getting the bug list showing them no bugs (0-0)? |
01:08.35 | landley | What's this changed(hrs) thing? |
01:08.49 | psypete | the bugs are still there... |
01:08.59 | psypete | theres just 0 unassigned and 0 resolved |
01:09.48 | landley | How do I get it to show me the list of all open busybox bugs? |
01:09.53 | landley | Did somebody go through and clean up the entire list? |
01:10.19 | psypete | /view_all_bug_page.php |
01:10.30 | landley | I'm on it. |
01:10.37 | psypete | so'm i... tons of bugs |
01:10.43 | landley | Viewing Issues (0 - 0 / 0) |
01:10.47 | psypete | in both the busybox and uclibc project pages |
01:10.53 | psypete | Viewing Issues (1 - 50 / 191) |
01:11.07 | psypete | http://bugs.uclibc.org/view_all_bug_page.php - make sure your filters are clear too |
01:11.39 | psypete | might also make a difference to select the project from the right drop-down box |
01:12.18 | landley | It's pointing at "busybox". |
01:12.23 | landley | It was working last time I went there. |
01:12.35 | landley | I just clicked "use filter" (without changing anything) and now it's showing me 77 bugs. |
01:13.09 | psypete | well you can start by using a different browser and fix it later |
01:13.55 | landley | A) it worked with this browser before, many times. B) clicking "use filter" brought up a list of bugs. |
01:14.16 | landley | Dunno if it's the full list, but it's a list... |
01:14.24 | psypete | C) just incase it wasn't all the bugs (??) you can use a different browser -temporarily- which won't have cookies or whatever saved with weird settings |
01:15.26 | landley | I don't have another browser installed on this laptop. I can clear the cookies for uclibc.org easily enough, but I didn't _change_ the cookies. |
01:16.17 | psypete | in Lisp |
01:17.59 | *** join/#uclibc ChanServ (ChanServ@services.) |
01:18.00 | *** mode/#uclibc [+o ChanServ] by irc.freenode.net |
01:20.04 | *** join/#uclibc ashes_ (n=ashes@modemcable080.175-201-24.mc.videotron.ca) |
01:42.31 | landley | That's "lithp". |
01:55.32 | landley | andersee: so apparently there are two issues with combining gcc 4.0.2 and uClibc. |
01:55.53 | landley | dl_iterate_phdr and __libc_stack_end. |
02:02.02 | *** join/#uclibc ashes (n=ashes@modemcable080.175-201-24.mc.videotron.ca) |
02:16.12 | *** join/#uclibc tchan (n=tchan@lunar-linux/developer/tchan) |
02:19.05 | landley | The /tools build made it to the end, with uclibc 0.9.28 and gcc 4.0.2. |
02:19.12 | landley | Now I need to fix busybox to understand --exclude... |
02:31.53 | landley | Can we perhaps buy vodz some "learn to speak english" tapes? |
02:31.56 | landley | A telecourse, perhaps? |
02:32.12 | landley | The last argument (val) can undefined from applet_opts. |
02:32.12 | landley | <PROTECTED> |
02:32.13 | landley | <PROTECTED> |
02:32.13 | landley | <PROTECTED> |
02:32.13 | landley | <PROTECTED> |
02:32.28 | landley | I want to add --exclude. |
02:32.45 | landley | The logic is there in --exclude from, except I want to append these arguments to the linked list directly, not get them out of a file. |
02:32.57 | landley | in --exclude-from, I mean. |
02:33.29 | landley | I've never had a clear idea how to use getopt_ulargs(). |
02:33.43 | landley | I want to add a long option, which doesn't have a short equivalent, and can be invoked more than once. |
02:33.47 | landley | I don't even know if this is possible. |
02:34.06 | landley | I just know that uClibc-0.9.28 can't use busybox tar during its install phase without it. |
02:34.33 | *** join/#uclibc tchan (n=tchan@lunar-linux/developer/tchan) |
03:06.25 | CIA-10 | 03landley * r11865 10busybox/archival/tar.c: (log message trimmed) |
03:06.25 | CIA-10 | Add --exclude option (to make uClibc-0.9.28 headers install using busybox tar). |
03:06.26 | CIA-10 | I have no idea how to apply bb_getopt_complementally to a --longopt that |
03:06.37 | landley | Now to find out if that does, in point of fact, actually work. |
03:28.09 | landley | Wow, gcc 4.0 may actually have an improvement over 3.3! |
03:28.22 | landley | The stupid hardwired library paths in collect2.c don't seem to be there anymore... |
04:39.40 | *** join/#uclibc _cow (n=cow@M871P001.adsl.highway.telekom.at) |
04:54.56 | *** join/#uclibc EvilRick (n=Ev1lRick@196-28-86-129.wdsl.co.za) |
06:26.48 | *** join/#uclibc andersee (n=andersee@codepoet.org) |
06:32.40 | CIA-10 | 03vodz * r11866 10busybox/archival/tar.c: llist must initialize, add loses ifdef, small indent correction |
08:28.27 | *** join/#uclibc Wes (n=pwn3r@d142-179-76-233.bchsia.telus.net) |
09:38.55 | *** join/#uclibc JockeHome (n=JockeHom@84-217-1-108.tn.glocalnet.net) |
10:24.03 | CIA-10 | 03vodz * r11867 10busybox/ (11 files in 7 dirs): common BUFSIZ BSS buffer, small reduce code, data and bss |
12:39.59 | *** join/#uclibc _cow (n=cow@M905P010.adsl.highway.telekom.at) |
12:50.06 | *** join/#uclibc JockeHome (n=JockeHom@84-217-1-108.tn.glocalnet.net) |
13:45.35 | CIA-10 | 03vodz * r11868 10busybox/ (coreutils/date.c coreutils/realpath.c procps/pidof.c): RESERVE_CONFIG_BUFFER --> bb_common_bufsiz1 |
13:49.23 | CIA-10 | 03vodz * r11869 10busybox/sysklogd/syslogd.c: complex RESERVE_CONFIG_BUFFER --> bb_common_bufsiz1 |
13:56.19 | CIA-10 | 03vodz * r11870 10busybox/util-linux/mkfs_minix.c: data --> bss |
14:10.38 | CIA-10 | 03vodz * r11871 10busybox/miscutils/hdparm.c: data --> code |
14:13.11 | CIA-10 | 03aldot * r11872 10busybox/libbb/loop.c: |
14:13.12 | CIA-10 | - an empty middle term in ?: violates ISO C |
14:13.12 | CIA-10 | - use shorter boilerplate and use C89 style comments |
14:24.03 | CIA-10 | 03vodz * r11873 10busybox/sysklogd/syslogd.c: 20 bytes data to bss |
14:41.10 | CIA-10 | 03vodz * r11874 10busybox/e2fsprogs/blkid/ (probe.c probe.h): more const, all data to text |
14:50.03 | *** join/#uclibc tchan (n=tchan@lunar-linux/developer/tchan) |
16:21.28 | *** join/#uclibc ashes_ (n=ashes@modemcable080.175-201-24.mc.videotron.ca) |
16:44.58 | *** join/#uclibc ashes (n=ashes@modemcable080.175-201-24.mc.videotron.ca) |
16:51.19 | enerv | cow, ping |
16:51.44 | enerv | cow, my english is not good, but... let's go. |
16:51.52 | enerv | This project aims to provide a i386-uclibc port of debian. The idea is to make possible to do a full fresh installation in this arch. |
17:05.20 | enerv | they i'am, compiling GCC without libc6... but i not, obtaining compile accurately... i getting error catching to header of libc6 signal.c. |
17:05.35 | enerv | ../../gcc-3.4.3/gcc/config/i386/linux.h:217:20: signal.h: No such file or directory |
17:05.36 | enerv | ../../gcc-3.4.3/gcc/config/i386/linux.h:218:26: sys/ucontext.h: No such file or directory |
17:06.29 | enerv | my ./configure configure --target=i386-pc-linux-uclibc --prefix=$(DESTDIR)/usr/i386-pc-linux-uclibc/ -prefix=/static --disable-nls --disable-shared -with-as=$LFS/static/bin/as --with-ld=$LFS/static/bin/ld -v |
17:08.38 | enerv | i try also, ./configure configure --target=i386-pc-linux-uclibc --prefix=$(DESTDIR)/usr/i386-pc-linux-uclibc/ --with-headers |
17:09.34 | enerv | resuming, i need compile gcc static |
17:53.19 | *** join/#uclibc tchan (n=tchan@lunar-linux/developer/tchan) |
19:14.06 | *** join/#uclibc jamie (n=jamie@softmodem.org) |
19:42.08 | *** join/#uclibc ccjoe (i=clrsrv@www.clearcore.com) |
19:42.09 | *** join/#uclibc carlg (i=clrsrv@www.clearcore.com) |
20:02.12 | *** join/#uclibc psypete (n=psypete@adsl-144-98-74.mia.bellsouth.net) |
20:35.53 | *** join/#uclibc ashes_ (n=ashes@modemcable080.175-201-24.mc.videotron.ca) |
20:40.04 | *** join/#uclibc _cow (n=cow@M885P011.adsl.highway.telekom.at) |
20:56.32 | CIA-10 | 03aldot * r11875 10busybox/libbb/loop.c: |
20:56.33 | CIA-10 | - don't update copyright years. He might release it, but apparently will have |
20:56.33 | CIA-10 | <PROTECTED> |