**** BEGIN LOGGING AT Sat Jun 04 02:59:56 2011
Jun 04 21:18:36 <ant__> Jay7: hello
Jun 04 21:19:14 <ant__> some interesting findings:
Jun 04 21:19:24 <ant__> kexecboot-klibc (static):  Alignment trap: kexecboot (429) PC=0x00014438 Instr=0xe5933000 Address=0x401a7c6f FSR 0x013
Jun 04 21:19:40 <ant__> and
Jun 04 21:19:47 <ant__> kexecboot: ELF 32-bit LSB executable, ARM, version 1 (SYSV), dynamically linked (uses shared libs), for GNU/Linux 2.6.16, stripped
Jun 04 21:19:57 <ant__> Alignment trap: kexecboot (410) PC=0x0001031c Instr=0xe5931000 Address=0x000288d6 FSR 0x013
Jun 04 21:20:20 <ant__> both not running as init (see pid)
Jun 04 21:20:54 <Jay7> ant__: better to ask some experienced arm devel
Jun 04 21:21:03 <Jay7> I'm not sure I know any..
Jun 04 21:21:14 <ant__> if the eglibc-shared does the same then it's suspicious
Jun 04 21:21:51 <ant__> well, pb or khem
Jun 04 21:21:58 <Jay7> looks like compiler problem
Jun 04 21:22:00 <ant__> w worked for arm
Jun 04 21:22:12 <Jay7> do we using thumb?
Jun 04 21:22:33 <ant__> yes, angstrom and minimal do
Jun 04 21:22:37 <ant__> but klcc..
Jun 04 21:24:16 <ant__> fwiw klibc is not compiled in thumb mode
Jun 04 21:24:33 <Jay7> try to disable it and rebuild
Jun 04 21:25:37 <ant__> I'm pretty sure we are in arm mode
Jun 04 21:26:11 <ant__> export CC=${TARGET_PREFIX}klcc
Jun 04 21:26:11 <ant__> # reset inherited OE flags to avoid e.g. ggdb3 and keep size small
Jun 04 21:26:11 <ant__> export CFLAGS=""
Jun 04 21:27:06 <ant__> arm-angstrom-linux-gnueabi-klcc -DHAVE_CONFIG_H -I.    -Wall  -MT kexecboot.o -MD -MP -MF .deps/kexecboot.Tpo -c -o kexecboot.o kexecboot.c
Jun 04 21:27:40 <ant__> btw there is still the xpm warning
Jun 04 21:27:49 <Jay7> I know
Jun 04 21:27:53 <Jay7> about unused vars
Jun 04 21:28:29 <ant__> have you seen the alignment trap examples? pointing to RGBA...
Jun 04 21:28:33 <ant__> :p
Jun 04 21:28:51 <ant__> must be very common error
Jun 04 21:31:05 <ant__> I'll try following this guide
Jun 04 21:31:07 <ant__> http://www.rt-embedded.com/blog/archives/resolving-alignment-traps/
Jun 04 21:31:32 <ant__> but don't know whether shared would be easier to debug
Jun 04 21:31:47 <ant__> seems is kexecboot bug, not *libc, though
Jun 04 21:32:18 <ant__> I hope that instruction points to real k. code
Jun 04 21:32:40 <Jay7> hm..
Jun 04 21:32:47 <Jay7> may be, may be..
Jun 04 21:33:10 <Jay7> I'm using void pointers to store icon reference in menu
Jun 04 21:39:41 <ant__> well, how can it be the .dbg package is so small?
Jun 04 21:40:00 <ant__> *kexecboot         │  89416│
Jun 04 21:40:26 <ant__> debug:
Jun 04 21:40:33 <ant__> kexecboot                                      │  35731│
Jun 04 21:43:07 <Jay7> ant__: btw, do we still have 115200 console speed?
Jun 04 21:43:58 <ant__> yes I think
Jun 04 21:44:18 <Jay7> ok
Jun 04 21:53:06 <ant__> ok, same instruction
Jun 04 21:53:06 <ant__> Alignment trap: kexecboot (385) PC=0x0000e17c Instr=0xe5931000 Address=0x401e19bf FSR 0x013
Jun 04 22:05:51 <ant__> ok
Jun 04 22:05:53 <ant__>     e17c:	e5931000 	ldr	r1, [r3]
Jun 04 22:07:07 <ant__> grep reveals 3 calls
Jun 04 22:07:15 <ant__>     b280:	e5931000 	ldr	r1, [r3]
Jun 04 22:07:16 <ant__>     e17c:	e5931000 	ldr	r1, [r3]
Jun 04 22:07:16 <ant__>     e420:	e5931000 	ldr	r1, [r3]
Jun 04 22:07:28 <ant__> seems one fails
Jun 04 22:07:46 <ant__> ..now I have to see how to spit out C code from this...
Jun 04 22:07:58 <ant__> (is compiled whithout -g)
Jun 04 22:08:09 <ant__> mission impossible?
Jun 04 22:14:08 <ant__> 0000e124 T process_ctx_menu
Jun 04 22:14:08 <ant__> 0000e354 T process_ctx_textview
Jun 04 22:14:17 <ant__> I can only guess
Jun 04 22:14:37 <ant__> anyway now I'll recompile with -g
Jun 04 22:25:51 <ant__> Jay7: hmmm
Jun 04 22:26:15 <ant__> I think I have found
Jun 04 22:27:47 <ant__> Jay7: http://paste.debian.net/118855/
Jun 04 22:28:01 <ant__> hex2rgb
Jun 04 22:28:28 <ant__> hm..?
Jun 04 22:28:52 <Jay7> what line it was?
Jun 04 22:29:18 <ant__> full debug is too big:/
Jun 04 22:29:27 <ant__> see gmail ;)
Jun 04 22:29:33 <Jay7> e17c ?
Jun 04 22:29:43 <Jay7>     e17c:	e1800206 	orr	r0, r0, r6, lsl #4
Jun 04 22:29:51 <ant__> yes, but different address (see .dbg package)
Jun 04 22:30:02 <ant__>   e17c: e5931000  ldr r1, [r3]
Jun 04 22:30:27 <ant__> wait, same address different instruction?
Jun 04 22:30:42 <Jay7> as I see from your paste
Jun 04 22:33:10 <ant__> see the .dbg/kexecboot symbols  http://paste.debian.net/118856/
Jun 04 22:33:38 <Jay7> ELF



Jun 04 22:33:43 <Jay7> thats all
Jun 04 22:33:53 <ant__> ok, I'm using mc editor...
Jun 04 22:38:07 <ant__> /home/andrea/tmp/aa/kexecboot: ELF 32-bit LSB executable, ARM, version 1 (SYSV), statically linked, not stripped
Jun 04 22:38:07 <ant__> 000080b4 t $a
Jun 04 22:38:07 <ant__> 000080c0 t $a
Jun 04 22:45:09 <ant__> see http://paste.debian.net/118857/
Jun 04 22:54:20 <ant__> bah, seems indeed e17c, don't know yet why...
Jun 04 22:57:13 <ant__> if it's here
Jun 04 22:57:14 <ant__> <------>case 12 + 1:<-->/* #32329999CCCC */
Jun 04 22:57:14 <ant__> <------><------>/* so for now only take two digits */
Jun 04 22:58:51 <ant__> heh
Jun 04 22:59:26 <ant__> fatality or not it seems around RGB like they deprecate here  http://www.aleph1.co.uk/oldsite/armlinux/book/afaq.html
Jun 04 22:59:46 <ant__> well, enough indices I hope :)
Jun 04 23:06:08 <Jay7> yeah..
Jun 04 23:07:07 <ant__> it's indeed about padding/align
Jun 04 23:07:17 <ant__> maybe something goes wrong internally
Jun 04 23:07:28 <ant__> in gcc
Jun 04 23:08:05 <ant__> or do you see evident gugs?
Jun 04 23:08:13 <ant__> *bugs?
Jun 04 23:10:54 <Jay7> there is possibility to alignment problem in code
Jun 04 23:11:10 <Jay7> easy way is disable icons
Jun 04 23:11:31 <Jay7> but I'm unsure that code is clean enough to deal with --disable-icons
Jun 04 23:11:57 <Jay7> I have no tried it even iirc
Jun 04 23:16:51 <Jay7> time to sleep
Jun 04 23:27:04 <ant__> good night
Jun 04 23:27:19 <ant__> indeed in text mode no-crash
Jun 04 23:41:23 <ant__> hm.. static char * back_xpm[] = {
Jun 04 23:41:23 <ant__> "32 32 91 1",
Jun 04 23:54:10 <ant__> after all above, using old back.xpm makes alignment disappear...
Jun 04 23:56:46 <ant__> well, not on second time :/
Jun 04 23:58:47 <ant__> too late ;)
Jun 04 23:58:49 <ant__> gn
**** ENDING LOGGING AT Sun Jun 05 02:59:56 2011