00:58.43 | CIA-2 | 03kraj * r23784 10uClibc/ (libc/string/arm/strncmp.S test/string/tester.c): |
00:58.43 | CIA-2 | Use C implementation of strncmp. |
00:58.43 | CIA-2 | Add a new strncmp testcase. |
01:10.56 | *** join/#uclibc hiyuh (n=hiyuh@KD125054017176.ppp-bb.dion.ne.jp) |
01:13.25 | *** join/#uclibc thomasez (n=thomasez@castle.linpro.no) |
01:21.53 | *** join/#uclibc wrobbie (n=rob@203.117.215.163) |
02:28.22 | *** join/#uclibc tsukasa (n=tsukasa@unaffiliated/tsukasa) |
02:46.03 | *** join/#uclibc landley (n=landley@cpe-70-116-30-223.austin.res.rr.com) |
02:47.50 | landley | blindvt`: You wanted me to build uClibc oabi for arm with -O0 to see if that made malloc() stop breaking? |
02:47.55 | landley | Or was that a misunderstanding? |
03:04.56 | *** join/#uclibc wbx (i=wbx@81.169.141.189) [NETSPLIT VICTIM] |
03:04.56 | *** join/#uclibc blindvt` (n=blindvt`@chello213047219050.surfer.at) [NETSPLIT VICTIM] |
03:04.56 | *** join/#uclibc Kaloz (i=kaloz@openwrt/developer/Kaloz) |
03:04.56 | *** join/#uclibc visitors (n=pulstar@198-tar-2.acn.waw.pl) |
04:35.05 | landley | Morning blindvt` |
05:34.57 | *** join/#uclibc dinar_away (n=dinar@212.176.242.38) [NETSPLIT VICTIM] |
06:50.42 | *** join/#uclibc swishy (n=swishy@60-234-136-97.bitstream.orcon.net.nz) |
08:04.11 | *** join/#uclibc wilson (n=wilson@150.81.broadband9.iol.cz) |
08:09.12 | blindvt | landley, morning |
08:09.23 | blindvt | landley, yes, oabi with O0 |
08:10.22 | blindvt | tsuramoto, which one is bigger? |
08:14.47 | landley | Morning blindvt. |
08:14.51 | landley | I'll try building it... |
08:15.25 | wilson | hi there. is possible to ask anybody here what to do with some compilation problem related to uclibc, but to buildroot too ? |
08:19.36 | blindvt | wilson, you can try |
08:20.23 | blindvt | landley, please turn on asserts too |
08:21.49 | wilson | i'm trying to compile buildroot for i386 target, but several packages is still exit with error undefined reference to lseek64, open64, etc. all related to .....64. uncle google tell me to enable large file support on, but it's take no effect.... |
08:22.13 | landley | blindvt: which config symbol enables the asserts? |
08:22.34 | blindvt | DOASSERTS |
08:23.04 | blindvt | should be one of the last entries in menuconfig (in debugging options) |
08:24.21 | blindvt | landley, i.e. turn on DODEBUG and DOASSERTS there and you should get O0 and asserts |
08:25.24 | landley | ok |
08:32.03 | *** join/#uclibc pcgeil (n=steffen@p549E4918.dip.t-dialin.net) |
08:57.46 | landley | Ah, darn it. |
08:57.55 | landley | I have the reversion patch in my build. Have to do it again. :) |
08:58.27 | landley | Second verse, same as the first. |
08:58.59 | landley | Hulu is reminding me just how deeply, thoroughly and _hugely_ Galactica 1980 sucked rocks. |
08:59.05 | landley | Wow it was bad. |
08:59.08 | landley | Astoundingly... |
09:01.41 | blindvt | landley, don't forget to revert both revisions. Ideally you would just use current trunk. Khem just fixed that strncmp thing on arm, too |
09:06.10 | landley | I have both svn I'm reverting together in one patch, so I just mv it up one directory level so the build doesn't pick it up. |
09:06.33 | landley | If I rm -rf build it'll re-extract and re-patch the source automatically. |
09:21.01 | landley | Darn it, Icann's next board of directors meeting is in Mexico City in march. |
09:21.21 | landley | How am I supposed to get cheeseburger coupons to distribute to them when they're in _mexico_? |
09:21.29 | landley | I don't know any good mexican cheeseburger restaurants. |
09:21.42 | landley | Here in the US I could get gift certificates from Fuddruckers or something... |
09:23.16 | landley | Sigh, vint cerf retired so it's probably not worth going through with it. |
09:23.48 | landley | Vint Cerf would have understood the importance of making sure that Icann has cheeseburgers. |
09:24.09 | landley | If you claim to administer the internet, there are certain requirements the internet imposes upon you. |
09:24.44 | landley | But I've never heard of Peter Thrush... |
09:25.26 | landley | vlindvt: build completed, I ran it, no detectable difference. |
09:25.34 | landley | The asserts didn't trigger, behavior's exactly the same. |
09:25.51 | landley | <PROTECTED> |
09:26.29 | landley | Hmmm, hang on... |
09:27.37 | CIA-2 | 03vda * r23785 10busybox/archival/tar.c: tar: fix typo in comment |
09:30.24 | landley | And I run the build a _third_ time. Sigh... |
09:32.06 | landley | Duh. I figured out what I did wrong. :) |
09:32.23 | landley | (I added the debug symbols to the 0.9.29 config file, not the 0.9.30-rc2 one. Right.) |
09:36.46 | landley | Eventually even I have to run out of ways to screw this up. |
09:37.50 | blindvt | ;) |
10:02.41 | landley | Ok, that one definitely added debugging symbols. |
10:02.53 | landley | Had to increase the size of the ext2 filesystem image to get it to package without running out of space. |
10:04.33 | landley | blindvt: Ok, this time it built with debugging symbols, it built with asserts... |
10:04.38 | landley | And this made no difference in its behavior. |
10:04.55 | landley | Freeing init memory: 80K |
10:04.55 | landley | /tools/bin/bash: xmalloc: cannot allocate 240 bytes (0 bytes allocated) |
10:04.55 | landley | Kernel panic - not syncing: Attempted to kill init! |
10:05.09 | landley | Just like in the non-debug case. |
10:05.41 | blindvt | landley, ok, can you use something like ash and run bash in the debugger to pinpoint this allocation failure some more? |
10:06.13 | landley | blindvt: as I said yesterday, it's not bash. |
10:06.21 | landley | Bash just has its own built-in asserts that catch the problem early. |
10:06.26 | landley | With ash, I get to run exactly one command. |
10:06.33 | landley | The second command, whatever it is, hangs and never runs. |
10:07.08 | landley | And actually the first command will run if it's something small and built into busybox (like ls -l). |
10:07.40 | landley | This isn't a bash bug, this is a "bash catches it early and reliably, other things misbehave in more complicated ways" bug. |
10:08.18 | landley | I'm guessing the other things will happily use the zero length allocations, stomp the heap, and things go nuts on a later malloc() or free(). |
10:09.13 | blindvt | landley, which threads impl do you use again? |
10:09.50 | landley | old. |
10:10.06 | landley | Didn't I send my .config to the list? |
10:10.33 | landley | I note that bash isn't a threaded program. |
10:11.02 | landley | I can try switching off threads and seeing if it makes a difference... |
10:12.30 | blindvt | landley, please do, just to verify this |
10:14.56 | landley | ok. Build #4... |
10:28.25 | *** join/#uclibc dinar_away (n=dinar@212.176.242.38) [NETSPLIT VICTIM] |
10:30.19 | blindvt | this is such a PITA. *grumble* |
10:32.54 | landley | Interesting. |
10:33.04 | blindvt | landley, hm? |
10:33.05 | landley | When I disabled threads, sched.h went away. |
10:33.23 | landley | Meaning the busybox miscutils/chrt.c build broke. |
10:33.41 | landley | miscutils/chrt.c:9:19: error: sched.h: No such file or directory |
10:33.47 | blindvt | landley, you disabled RT |
10:34.05 | landley | I told uClibc it didn't have threads. |
10:34.49 | landley | Commented out these two lines from my miniconfig: |
10:34.51 | blindvt | sed -i -e "/UCLIBC_HAS_REALTIME/s/.*\(UCLIBC_HAS_REALTIME\).*/\1=y/" .config |
10:34.51 | landley | #UCLIBC_HAS_THREADS=y |
10:34.51 | landley | #LINUXTHREADS_OLD=y |
10:35.37 | landley | Is UCLIBC_HAS_REALTIME set as a side effect of selecting UCLIBC_HAS_THREADS? |
10:35.43 | blindvt | landley, threads force RT support on |
10:35.50 | landley | That would do it then. |
10:36.36 | landley | Ok, building again with that symbol switched back on. |
10:36.39 | landley | Build #5. |
10:52.31 | blindvt | landley, so.. any "malloc(1);malloc(1)" goes hayward. Is that correct? |
10:54.22 | landley | I haven't tried. I can run some hello world programs to say what they do, maybe print the %p they return... |
10:54.58 | landley | But that might be tomorrow, since I'm near staggering bedwards. |
10:55.08 | landley | Right now I'm waiting for the threadless build to finish to see what it does.. |
10:57.15 | blindvt | the quality of that patch is really very low, even when ignoring this current problem |
11:02.01 | landley | Ah, darn it. |
11:02.13 | landley | I have to build the native gcc without --disable-threads if libc hasn't got it. |
11:02.20 | landley | Ok, one more build... |
11:06.25 | blindvt | landley, such a small testcase to reproduce it would be excellent. Meanwhile i'm redoing his cra^Wpatch so i can actually see the 18 lines he effectively changed |
11:07.06 | landley | If you could post the redone patch to the list, I'd like to read it myself. |
11:07.30 | landley | I should be able to come up with a small test case tomorrow, but changing the library itself requires a complete rebuild because everything's built against it... |
11:11.51 | blindvt | landley, k |
11:16.34 | landley | It's _not_ being happy about this "no threads" thing. I think I'll yank the gcc build entirely for the moment... |
11:17.14 | *** join/#uclibc wilson_ (n=wilson@150.81.broadband9.iol.cz) |
11:17.26 | landley | The native gcc build, I mean. |
11:17.30 | landley | Not needed to reproduce this problem... |
11:25.47 | landley | Ok, built. |
11:26.10 | landley | And the exact same failure. |
11:26.18 | landley | So no, disabling threads did not change behavior. |
11:26.24 | landley | And on that note, I go to bed. |
11:26.26 | landley | g'night. |
11:29.38 | blindvt | mhm |
11:32.27 | wilson_ | fopen64 solved, i just download latest snapshot. got know what I was doing wrong before. |
11:57.11 | blindvt | wilson_, k |
12:14.32 | *** join/#uclibc wilson__ (n=wilson@150.81.broadband9.iol.cz) |
12:26.48 | *** join/#uclibc wrobbie (n=rob@203.117.215.163) |
12:36.18 | *** join/#uclibc tsukasa (n=tsukasa@unaffiliated/tsukasa) |
12:36.50 | *** join/#uclibc irotas (n=irotas@CPE00131042e14d-CM0018c0b37a44.cpe.net.cable.rogers.com) |
12:52.07 | *** join/#uclibc wilson__ (n=wilson@150.81.broadband9.iol.cz) |
13:24.32 | *** join/#uclibc ivob (n=ivob@unaffiliated/tiny) |
13:38.33 | *** join/#uclibc wilson__ (n=wilson@150.81.broadband9.iol.cz) |
15:32.00 | *** join/#uclibc thomasez (n=thomasez@castle.linpro.no) [NETSPLIT VICTIM] |
15:32.00 | *** join/#uclibc khem (n=khem@adsl-71-146-3-103.dsl.pltn13.sbcglobal.net) |
15:32.00 | *** join/#uclibc Redhatter (n=vk4msl@gentoo/developer/redhatter) |
15:50.43 | tsuramoto | blindvt: trunk version is ~25kb bigger than 0.9.29 ( du -hs /lib ) |
15:58.27 | *** join/#uclibc ivob (n=ivob@unaffiliated/tiny) |
16:45.05 | *** join/#uclibc thomasez (n=thomasez@castle.linpro.no) [NETSPLIT VICTIM] |
16:45.06 | *** join/#uclibc khem (n=khem@adsl-71-146-3-103.dsl.pltn13.sbcglobal.net) |
16:45.06 | *** join/#uclibc Redhatter (n=vk4msl@gentoo/developer/redhatter) |
18:15.41 | blindvt | tsuramoto, for the same config and toolchain? That sounds suspicious |
18:21.19 | tsuramoto | uh |
18:22.02 | tsuramoto | blindvt, I build a toolchain with trunk |
18:22.29 | tsuramoto | blindvt, build a rootfs with it (busybox uClibc linux e2fsprogs) |
18:22.50 | tsuramoto | build a linux kernel with that rootfs cpio |
18:22.56 | tsuramoto | and booted it with qemu |
18:23.12 | tsuramoto | mem usages was the ~same |
18:23.49 | tsuramoto | and when I did du -hs /lib the difference was ~25kb |
18:24.02 | tsuramoto | trunk bigger than 0.9.29 |
18:24.52 | tsuramoto | I did a make oldconfig on my 0.9.29 config |
18:25.21 | tsuramoto | and did mostly answer as suggested |
18:26.47 | tsuramoto | revision 23781 |
19:13.49 | *** join/#uclibc likewise (n=likewise@84-104-101-183.cable.quicknet.nl) |
19:22.09 | *** join/#uclibc tsukasa (n=tsukasa@unaffiliated/tsukasa) |
20:59.11 | *** join/#uclibc tsukasa (n=tsukasa@unaffiliated/tsukasa) |
21:15.58 | *** join/#uclibc SpanKY (n=UserBah@pool-72-72-126-167.bstnma.east.verizon.net) |
21:22.38 | *** join/#uclibc ivob (n=ivob@unaffiliated/tiny) |
22:53.21 | *** join/#uclibc TokyoDan (n=TokyoDan@q014060.dynamic.ppp.asahi-net.or.jp) |
23:23.02 | CIA-2 | 03vda * r23786 10busybox/ (editors/vi.c include/libbb.h libbb/Kbuild): (log message trimmed) |
23:23.02 | CIA-2 | vi: move key reading routine out of vi into llbbb |
23:23.02 | CIA-2 | function old new delta |
23:23.33 | CIA-2 | 03vda * r23787 10busybox/libbb/read_key.c: libbb: add forgotten file from previous commit :( |
23:27.30 | CIA-2 | 03vda * r23788 10busybox/ (4 files in 4 dirs): (log message trimmed) |
23:27.30 | CIA-2 | less: reuse former vi's key reading code. Improve SIGWINCH handling. |
23:27.30 | CIA-2 | function old new delta |
23:50.51 | TokyoDan | Hello, Is busybox the shell that is used in Nokia N810-Maemo-Diablo? |
23:59.42 | CIA-2 | 03vda * r23789 10busybox/libbb/read_key.c: |
23:59.42 | CIA-2 | read_key: add comments how Shift and Meta keys looks like. |
23:59.42 | CIA-2 | <PROTECTED> |