00:00.50 | SpanKY | khem: only when using a bum compiler |
00:01.24 | SpanKY | JockeHome: no, i think we only do that to force it into bss |
00:01.31 | SpanKY | otherwise gcc defaults to putting it into comm |
00:01.47 | dalias | you mean data |
00:01.50 | dalias | bss=comm |
00:03.13 | SpanKY | i mean bss |
00:06.10 | JockeHome | why leave __curbrk at a random start value? doesn't it make mose sense to have it zeroed? |
00:06.30 | SpanKY | well look at the c versions ... it isnt used until it gets set |
00:07.04 | SpanKY | dalias: i dont really know what i mean |
00:08.18 | JockeHome | SpanKY: x86 does void *__curbrk = 0; |
00:09.02 | SpanKY | right |
00:09.15 | SpanKY | but we do it only to keep gcc/binutils happy |
00:09.29 | SpanKY | you cant make aliases to it otherwise |
00:09.38 | SpanKY | cause it is comm data |
00:09.57 | SpanKY | we should really be marking it with like attribute((nocommon)) or whatever |
00:10.03 | JockeHome | OK, then so should ppc |
00:10.28 | SpanKY | yeah, i think if we use __attribute__ ((nocommon)) instead, it should be fine |
00:10.32 | JockeHome | __curbrk: |
00:10.32 | JockeHome | <PROTECTED> |
00:10.32 | JockeHome | <PROTECTED> |
00:10.33 | JockeHome | <PROTECTED> |
00:11.23 | JockeHome | void * __curbrk=0; |
00:11.36 | JockeHome | complete asm for this: |
00:11.51 | JockeHome | .file "curbrk.c" |
00:11.51 | JockeHome | <PROTECTED> |
00:11.51 | JockeHome | <PROTECTED> |
00:11.51 | JockeHome | <PROTECTED> |
00:11.52 | JockeHome | __curbrk: |
00:11.52 | JockeHome | <PROTECTED> |
00:11.54 | JockeHome | <PROTECTED> |
00:11.56 | JockeHome | <PROTECTED> |
00:11.58 | JockeHome | <PROTECTED> |
00:12.00 | JockeHome | <PROTECTED> |
00:12.04 | SpanKY | hacker |
00:12.25 | JockeHome | what? gcc -S ... |
00:12.30 | SpanKY | HACKER |
00:12.37 | JockeHome | you too :) |
00:19.02 | JockeHome | --- brk.S (revision 14457) |
00:19.02 | JockeHome | +++ brk.S (working copy) |
00:19.02 | JockeHome | @@ -24,7 +24,14 @@ |
00:19.02 | JockeHome | <PROTECTED> |
00:19.02 | JockeHome | <PROTECTED> |
00:19.02 | JockeHome | - .comm __curbrk,4,4 |
00:19.05 | JockeHome | + .globl __curbrk |
00:19.09 | JockeHome | + .section ".sbss","aw",@nobits |
00:19.11 | JockeHome | + .align 2 |
00:19.13 | JockeHome | +__curbrk: |
00:19.15 | JockeHome | + .zero 4 |
00:19.17 | JockeHome | + .size __curbrk, 4 |
00:19.21 | JockeHome | + .type __curbrk, @object |
00:19.23 | JockeHome | + |
00:19.25 | JockeHome | compiles, but untested |
00:28.04 | SpanKY | JockeHome: go to sleep yet ? :) |
00:28.36 | SpanKY | oh well go to sleep |
00:28.39 | SpanKY | i'll send an e-mail |
00:37.11 | *** join/#uclibc thraxisp (n=thraxisp@ottawa-hs-206-191-33-114.d-ip.magma.ca) |
00:47.00 | *** join/#uclibc blindvt (n=bf@M897P010.adsl.highway.telekom.at) |
01:08.00 | khem | how can I remove a TEXTREL reference from a shared lib |
01:08.36 | khem | my problem is that the shared lib has this TAG in dynamic section and it means something is not PIC in the library |
01:13.20 | solar | depends. some arches like arm had a bug where every created executable would be marked at containing a text relocation |
01:13.48 | solar | you can't usually just remove the program header markings. As the executable will still containing the problem |
01:18.32 | khem | hmm my arch is arm |
01:18.54 | khem | but its just one shared lib which has it and others do not |
01:19.04 | khem | for e.g. I do not see it in libc.so |
01:19.10 | khem | but I see it in procps.so |
01:29.22 | khem | aha I see function user_from_uid() is creating this for me |
01:41.36 | SpanKY | khem: install scanelf and run `scanelf -T` :p |
01:41.43 | SpanKY | it'll tell you the source of TEXTRELs |
01:51.02 | khem | SpanKY: scanelf is it part of elfutils |
01:54.14 | SpanKY | no it isnt |
01:54.18 | SpanKY | it's something solar and i write |
01:54.19 | SpanKY | :P |
01:54.59 | SpanKY | http://gentoo.osuosl.org/distfiles/pax-utils-0.1.10.tar.bz2 |
01:59.11 | khem | SpanKY: does it tell me about the symbol which caused the issue |
01:59.35 | SpanKY | yes |
02:00.29 | SpanKY | it's quite small and is C only |
02:02.30 | khem | libproc-3.2.5.so: (memory/fake?) [0x4190] in $d [0x4180] |
02:02.40 | khem | hmmm what do I understand from this |
02:02.49 | khem | its the output of scanelf -T |
02:07.50 | SpanKY | does the .so have debugging ? |
02:08.00 | SpanKY | what version of gcc/binutils ? |
02:08.42 | khem | 3.4.3 and 2.15.94 (patched equal to 2.16.1 almost) |
02:08.50 | khem | I think it does not have debug info |
02:08.56 | khem | hmmm let me see |
02:09.01 | SpanKY | is it stripped ? |
02:13.25 | khem | hmmm no |
02:13.41 | khem | proc/libproc-3.2.5.so: ELF 32-bit LSB shared object, ARM, version 1 (ARM), not stripped |
02:14.36 | SpanKY | strip it and then run scanelf on it |
02:14.54 | SpanKY | sometimes having the debugging symbols in there can give better output, sometimes it hits intermediate symbols like $d |
02:15.14 | SpanKY | i should update it to skip things like .L12 |
02:15.24 | khem | yes locals |
02:15.40 | khem | http://dev.gentoo.org/~kevquinn/textrel-check/ |
02:15.45 | khem | also has something similar |
02:16.18 | khem | SpanKY: dollar symbols are not stripped in ARM |
02:16.43 | khem | they are used by other consumers like ld.so according to EABI |
02:16.59 | khem | I should have said "may be used by" |
02:18.22 | SpanKY | khem: i integrated kevquinn's work into scanelf |
02:18.50 | SpanKY | khem: well was libproc built with -fPIC ? |
02:44.05 | *** join/#uclibc thraxisp (n=thraxisp@ottawa-hs-206-191-33-114.d-ip.magma.ca) |
03:15.47 | *** join/#uclibc jacques (n=username@nslu2-linux/jacques) |
04:04.08 | solar | recall spanky arm had every executable with textrels. |
04:04.17 | solar | should be fixed in svn by psm however |
04:04.30 | SpanKY | yes, but khem said he's using a fixed version of binutils |
04:04.41 | SpanKY | and unless khem is also using uClibc-0.9.27, he shouldnt be seeing this issue |
04:05.08 | solar | khem: does it only happen with the one executable? |
04:06.32 | *** join/#uclibc Newsome (n=sorenson@70.131.69.51) |
04:49.46 | *** join/#uclibc andersee (n=andersee@codepoet.org) |
05:01.17 | *** join/#uclibc FireEgl (i=Atlantic@Atlantica.US) |
05:04.28 | *** join/#uclibc angom_h (n=angom@red-corp-201.130.118.46.telnor.net) |
05:28.50 | *** join/#uclibc Jenna (n=cherryRe@209.8.233.101) |
05:29.07 | *** part/#uclibc Jenna (n=cherryRe@209.8.233.101) |
06:07.12 | *** join/#uclibc Jenna (n=cherryRe@209.8.233.101) |
06:07.35 | *** part/#uclibc Jenna (n=cherryRe@209.8.233.101) |
06:52.04 | *** join/#uclibc Jenna (n=cherryRe@209.8.233.101) |
06:52.25 | *** part/#uclibc Jenna (n=cherryRe@209.8.233.101) |
06:53.33 | SkyNetIsHere_com | ok this is driving me nuts i just tried building buildroot on my THIRD gentoo box and still errors on busybox... libbusybox... |
07:55.29 | blindvt` | SkyN, what error are you seeing specifically |
07:58.42 | blindvt` | SkyNetIsHere_com, and what gcc are you using? Paste the error to http://rafb.net/paste |
08:39.03 | *** join/#uclibc psaksa (n=pate@ip212-226-134-137.adsl.kpnqwest.fi) |
08:48.26 | *** join/#uclibc blindvt (n=bf@M777P027.adsl.highway.telekom.at) |
08:50.36 | *** join/#uclibc Newsome (n=sorenson@ppp-69-218-176-246.dsl.emhril.ameritech.net) |
09:23.29 | CIA-6 | 03psm * r14458 10uClibc/ (3 files in 3 dirs): Add UCLIBC_HAS_SSP_COMPAT option guarding gcc-3.x ssp support |
10:14.34 | *** join/#uclibc Jenna_zz (n=cherryRe@209.8.233.100) |
10:15.09 | *** part/#uclibc Jenna_zz (n=cherryRe@209.8.233.100) |
10:36.35 | *** join/#uclibc psm (n=mps@host-6.mikroweb.hu) |
10:49.32 | *** join/#uclibc tty56 (n=johannes@p54A783A4.dip0.t-ipconnect.de) |
12:41.14 | *** join/#uclibc tty56 (n=johannes@p54A79757.dip0.t-ipconnect.de) |
12:55.24 | *** join/#uclibc prpplague (n=billybob@72.22.146.238) |
13:40.40 | *** join/#uclibc JockeHome (n=JockeHom@84-217-93-199.tn.glocalnet.net) |
13:55.35 | *** join/#uclibc tty56_ (n=johannes@p54A7A21C.dip0.t-ipconnect.de) |
13:55.38 | *** join/#uclibc vodz (i=1000@dzo.simtreas.ru) |
13:56.42 | blindvt` | hi |
14:00.05 | vodz | hi all |
14:08.58 | blindvt` | is anybody else seeing errors for busybox' defconfig (see ML)? |
14:10.21 | vodz | after removed "extern" keyword? |
14:13.20 | blindvt` | yes. Not sure if it's only me, though. |
14:13.24 | *** join/#uclibc sjhill (n=sjhill@eth13.com-link.com) |
14:14.37 | vodz | blindvt`: but it strange. For example coreutils/libcoreutils/coreutils.h: warning: 'struct stat' declared inside parameter list |
14:14.55 | vodz | but this line not changed |
14:16.47 | vodz | and for cp.c changed exterm main_cp() to main_cp(). Its eq for me |
14:20.43 | vodz | very very strange... May be macrox borken now? |
14:21.00 | blindvt` | erm, no! It's only me. |
14:22.44 | blindvt` | never mind |
14:25.22 | vodz | blindvt`: if i have strange error, I see preprocessor out. May be help this for you. |
14:35.18 | *** join/#uclibc tty56 (n=johannes@p54A7A21C.dip0.t-ipconnect.de) |
15:08.37 | *** join/#uclibc blindvt_ (n=bf@M950P028.adsl.highway.telekom.at) |
15:17.53 | *** join/#uclibc landley (n=landley@pool-70-17-176-135.pitt.east.verizon.net) |
15:30.31 | vodz | http://lambda-the-ultimate.org/classic/message1532.html#1533 |
15:39.21 | vodz | http://www.microsoft.com/whdc/driver/kernel/KMcode.mspx |
15:39.27 | vodz | hmm, intersetd text |
15:49.05 | vodz | If something can be made one hundred ways pervert and think up hundred first. Ideological principle of Perl. |
15:51.26 | *** join/#uclibc landley (n=landley@pool-70-17-176-135.pitt.east.verizon.net) |
16:00.24 | vodz | C++ is a great language for garbage collection primarily because it creates less garbage. (c) Stroustrup |
16:33.49 | sjhill | that's a bunch of crap |
16:34.55 | landley | The statement by Stroustrup, or the C++ language? :) |
16:37.08 | blindvt` | landley, re redhat9 is ok, so all is alright. no, it's not alright. vda said that plain make-3.80 didn't work (unfortunately he apparently updated to the current make-3.81-rc1). I'll try to find the time to fetch and install that vanilla 3.80 |
16:40.36 | CIA-6 | 03psm * r14459 10uClibc/ldso/ldso/ldso.c: Remove additional endif pointed out by vapier |
16:41.32 | psm | SpanKY: what does filter-out libgcc-glibc.ver remove? |
16:42.04 | vodz | blindvt`: And all I was right. Complication of makefile rules in what good has not resulted. Really it is impossible to make that there were no these dependences? |
16:42.45 | landley | Ubuntu claims to have make 3.80, and it's building with that. (Even e2fsprogs.) |
16:42.52 | SpanKY | psm: there is a gcc script which adds glibc versioning symbols |
16:42.54 | landley | Dunno what strange patches they've applied, though. |
16:46.32 | vodz | In comparison with e2fsprogs, fdisk code is the ideal sample. Though for some reason landley name fdisk code as the worst. |
16:46.59 | landley | No, fdisk is second. I agree e2fsprogs is the worst. |
16:47.42 | *** join/#uclibc thraxisp (n=thraxisp@ottgate.precidia.com) |
16:51.13 | blindvt` | landley, yes. I tested it on ubuntu and it worked there for me even with the initial patch. |
16:51.26 | landley | It's slackware it's failing on? |
16:51.28 | landley | Which version? |
16:52.07 | blindvt` | landley, no idea what version of slack. He said he uses vanilla make-3.80 from ftp.gnu.org (just installing that locally) |
16:52.39 | vodz | my Slackware 9.1.0. But i have not fresh SVN copy |
16:52.46 | landley | I'll have to resurrect my Firmware build before release. Mount's almost together again... |
16:54.33 | vodz | Slackware 9.1.0 have GNU Make 3.80 |
16:55.20 | vodz | Slackware 10.0 too |
16:57.33 | landley | It's Patrick's style to use the current release version of software in as unmodified a form as he can get away with. |
16:57.40 | vodz | hmm, and Slackware 10.1 |
16:58.22 | landley | I looked at moving to Slackware instead of Ubuntu a year ago, but that's when Patrick was still sick, and Slack had fallen a bit behind. |
16:58.29 | landley | (He wasn't using a 2.6 kernel yet.) |
16:59.41 | vodz | I use Slackware only, started from 1.0 (with 1.0.0-pre kernels) |
16:59.51 | landley | It's a nice distro. |
17:00.57 | vodz | heh, in my table I found Slackware-3.0 with 1.2.X kernel |
17:01.38 | dalias | :) |
17:03.02 | blindvt` | vodz, heh. i used to use slackware-3 and -4 |
17:03.40 | landley | I used SLS once, but it didn't like the hardware I tried it on. |
17:04.09 | landley | Wandered through OS/2 for a few years. Eventually came back to Debian and they drove me to Red Hat. |
17:04.50 | vodz | have debian is not vanilla kernel? |
17:05.20 | blindvt` | vodz, yes. debian does patch the kernel |
17:05.37 | landley | So does Gentoo. |
17:05.46 | landley | (Greg Kroah-Hartman maintains gentoo's kernel.) |
17:05.50 | landley | Dunno who maintains Debian's. |
17:05.57 | vodz | add other hardware suppopt? |
17:06.24 | blindvt` | tends to work mostly well (except the current 2.6.15 from debian has completely foobared pcmcia support) |
17:07.52 | vodz | ah, pcmcia... My pcmcia is very old and do tnot require 2.6 k |
17:08.23 | landley | pcmcia support moved into the kernel, and they were dropping support for the old external pcmcia package. |
17:08.31 | landley | (There was some discussion about this on the kernel list, a few months ago.) |
17:09.03 | landley | The new hotplug infrastructure makes supporting pcmcia in the kernel make sense. |
17:09.14 | vodz | may be, but pcmcia also old a technology |
17:09.28 | blindvt` | completely foobared as in it crashes about as fast as any suse kernel i've seen ;) |
17:10.07 | landley | New laptops still use pcmcia for wireless cards, they just build it in inside the case. |
17:10.14 | landley | Use it as a bus interface, or some such. |
17:10.28 | landley | (Only now it's "cardbus".) |
17:10.30 | blindvt` | vodz, well, it's just 2.6.15 from debian that doesn't work with pcmcia. The majority of the rest of debian's 2.6 series worked well for me with it |
17:10.40 | vodz | my two notebooks have pcmcia, third - without |
17:11.25 | landley | Gotta run... |
17:23.51 | vodz | Tomorrow in Russia a female holiday and the day off. In honour of a fire on a manufactory in New York where was lost about 150 women. |
17:25.47 | *** join/#uclibc samrobb_ (n=samrobb_@65-117-135-105.dia.cust.qwest.net) |
18:20.26 | khem | solar: spanKY: I only see it in few of shared libs |
18:20.50 | khem | one ofthem is procps.so |
18:21.06 | khem | libstdc++.so is another one |
18:22.01 | *** join/#uclibc carlg (i=clrsrv@www.clearcore.com) |
18:41.26 | psm | khem: what do you see in the libs? |
18:41.40 | psm | have you found what happens w/ opt* ? |
18:43.01 | khem | psm: I am stuck with textrels appearigng in some of shared libs |
18:43.31 | khem | other libs dont have it |
18:44.25 | khem | I have not got to a point where I could find the real reason for opt* issue |
18:48.19 | psm | khem: textrels are know issue w/ uClibc due to old libtool/ltconfig used by configure |
18:49.30 | psm | I have added to gentoo bugs a description how to detect this and how to solve (gentoo build system provides a way for it) |
18:50.51 | khem | psm: SpanKY gave me a tool which is able to detect it |
18:51.10 | khem | psm: i could not get around to solve it yet |
18:51.22 | khem | do you remember the bug nu,ber |
19:15.12 | psm | khem: #74088 |
19:39.20 | SpanKY | procps doesnt fall into the libtool category |
19:39.27 | SpanKY | it doesnt use any autotools |
19:47.03 | psm | procps is completely wrong, it needs including sys/user.h (ps/common.h and minimal.c) and LIBPROC := proc/libproc-$(TARVERSION).so to Makefile |
20:05.56 | khem | psm: are you looking in proc/module.mk file |
20:08.53 | khem | thats the one which builds library |
20:09.12 | *** join/#uclibc woglinde (i=woglinde@e178112099.adsl.alicedsl.de) |
20:09.18 | psm | khem: I was speaking of gentoo's build, havent looked to buildroot |
20:26.14 | CIA-6 | 03pgf * r14460 10busybox/shell/cmdedit.c: (log message trimmed) |
20:26.14 | CIA-6 | vi-mode changes: eliminate warnings from out-of-range-for-type |
20:26.14 | CIA-6 | case labels that are emitted by gcc4. rearrange macros to |
21:27.08 | *** join/#uclibc woody___ (n=woody@bi01p1.co.us.ibm.com) |
22:10.33 | *** join/#uclibc z0mb (n=satan@port54.ds1-sby.adsl.cybercity.dk) |
22:10.50 | z0mb | hi could someone answer a linux newbie question for me plz? :-O |
22:11.06 | z0mb | how do i set path on a BB? setenv doesnt work |
22:14.04 | khem | export directive should do |
22:14.38 | z0mb | so to set /opt/bin in path |
22:14.42 | z0mb | export /opt/bin |
22:14.44 | z0mb | ? |
22:14.57 | z0mb | in ø |
22:14.58 | z0mb | ' |
22:15.05 | khem | export PATH=$PATH:/opt/bin |
22:15.16 | z0mb | ok thanx alot :-) |
22:17.03 | *** part/#uclibc z0mb (n=satan@port54.ds1-sby.adsl.cybercity.dk) |