irclog2html for #elinux on 20040103

03:40.21*** join/#elinux jnork (~jnork@pool-141-158-44-95.phil.east.verizon.net)
03:57.18jnorkGood evening, chouimat.
03:57.58chouimathi jnork ... just spent the evening watching tv ...
04:02.34jnorkI just bought a DVD -+R/RW drive.
04:02.41jnorkI should probably install the software for it...
04:04.55jnorkHmm, this sez it'll require a reboot. BBS.
04:09.26*** join/#elinux jnork (~jnork@pool-141-158-44-95.phil.east.verizon.net)
04:56.01*** join/#elinux jnork (~jnork@pool-141-158-44-95.phil.east.verizon.net)
05:45.28*** join/#elinux TimRiker (~timr@proxyle01.ext.ti.com)
05:45.28*** mode/#elinux [+o TimRiker] by ChanServ
05:45.58kergothheh
05:52.11anderseeTimRiker: evening
05:52.16anderseeTimRiker: got a sec?
05:52.44kergothandersee: how are things going with the last 1.00 busybox rel?
05:52.58anderseekergoth: very good
05:53.03kergothgood good
05:53.10anderseekergoth: -pre5 is quite solid thus far
05:53.21anderseekergoth: a few little things, but nothing major
05:53.30TimRikerandersee: sure. hit me.
05:53.58TimRikeranyone working on an embedded depmod ?
05:54.23anderseeTimRiker: some x86 asm..
05:54.27anderseeTimRiker: http://www.uclibc.org/cgi-bin/cvsweb/*checkout*/uClibc/libc/sysdeps/linux/i386/clone.S?rev=1.8
05:55.00anderseeoops
05:55.08TimRikeroops?
05:55.10anderseeTimRiker: http://www.uclibc.org/cgi-bin/cvsweb/*checkout*/uClibc/libc/sysdeps/linux/i386/clone.S?rev=1.9
05:55.18anderseeanyway
05:55.35anderseeif I wrap the syscall like so:
05:55.37anderseeextern int __clone(int (*fn)(void *), void *child_stack, int flags, void *arg);
05:55.37anderseeint clone(int (*fn)(void *), void *child_stack, int flags, void *arg)
05:55.37andersee{
05:55.37andersee<PROTECTED>
05:55.37andersee<PROTECTED>
05:55.39andersee<PROTECTED>
05:55.41andersee<PROTECTED>
05:55.43andersee<PROTECTED>
05:55.45andersee}
05:55.47anderseeeverything is good
05:56.11anderseefor some reason I can't understand, checks for !fn and !child_stack in the asm don't seem to work
05:56.48anderseei.e. clone(valid_fn, NULL, blah, blah) segfaults rather than failing with EINVAL
05:57.54anderseeTimRiker: any ideas?
05:59.05Russonce you go risc, you never go back
05:59.30TimRikersounds riscy
06:00.51anderseeIt appears to assume the arg1 is at sp+4, and sp+8
06:01.24anderseeloads those into ecx, and jumps to .Lclone_error if it loads a 0
06:01.24TimRikerandersee: are you PIC or not?
06:01.32anderseeTimRiker: yes
06:01.36anderseeTimRiker: I am testing PIC
06:01.48TimRikerdoes the non-PIC work?
06:02.08TimRikerhave you looked at the asm for the C version?
06:02.58anderseeTimRiker: what C version?
06:03.31anderseeThe little wrapper thing above?
06:04.32anderseeit pulls stuff off the stack, and does, i.e. test %ecx,%ecx
06:04.48TimRikerwhat's the offset on the stack there?
06:05.11anderseeint clone(int (*fn)(void *), void *child_stack, int flags, void *arg)
06:05.11andersee{
06:05.11andersee<PROTECTED>
06:05.11andersee<PROTECTED>
06:05.11andersee<PROTECTED>
06:05.11andersee<PROTECTED>
06:05.13andersee<PROTECTED>
06:05.15andersee<PROTECTED>
06:05.17andersee<PROTECTED>
06:05.19andersee<PROTECTED>
06:05.21andersee<PROTECTED>
06:05.23andersee<PROTECTED>
06:05.25andersee<PROTECTED>
06:05.27andersee<PROTECTED>
06:05.29andersee<PROTECTED>
06:05.47TimRikerhmm 0x18?
06:07.27TimRikerwhat's the call d ? (me's x86 asm is rusty)
06:08.18TimRikerthat one shows off set 0x8 and 0xc which is what I would have guessed.
06:08.26TimRiker(not 4 and 8)
06:09.57anderseeso perhaps we are checking the wrong args?
06:11.52TimRikerI'd guess so. can you break there on an app?
06:12.06TimRikerthen look at the stack?
06:13.02anderseeI suppose I could do an 'info registers' in gdb
06:13.12anderseethat should give me esp
06:13.17anderseeand I could look there
06:13.40anderseeunless there is a better way to view the stack contents...
06:14.51TimRikerhmm. there is a dump stack command someplace. don't recall the command.
06:15.09TimRikerdisplay locals might get you something funky. ;-)
06:16.27TimRikertry "info frame" and "info locals" for kicks.
06:16.55TimRikerohh.. I mean "info args"
06:18.52anderseeI think part of what makes clone nasty is it is frameless
06:18.53TimRikerwhat does "x/5x %ebp" do? give an error?
06:19.29anderseeparse error %ebp
06:20.57anderseeI can do the same for the value of ebp though
06:23.45TimRikerthe x works? what do you get?
06:26.10TimRikerhmm. not sure you are setting up things ok for the error return. what happens if you just ret at .Lclone_error instead of erroring?
06:26.29TimRikeryou might be checking the right value and dieing in the error handler.
06:31.48anderseecould be
06:32.03anderseethe x works if I give it the value of esp
06:32.19anderseeIt prints the correct value for fn
06:32.39anderseethen something, I'm not sure what
06:32.46anderseethen some NULLs
06:33.59anderseeI think you are right
06:34.12anderseeI think it is working, then choking in the error handler
06:34.41anderseethe stuff I though was broken is in fact working properly
06:40.36*** join/#elinux noverlak (noverlak@42-183-174.adsl.cust.tie.cl)
07:00.27TimRikerwhat's the new rtai/rtlinux like thing that claims to avoid the rtlinux patent?
07:00.57TimRikerie it runs the kernel and the rt os under a parent instead of the kernel under the rt "os"?
07:03.31TimRikerahh. adeos
07:03.33TimRiker~adeos
07:07.00anderseeTimRiker: yeah
07:07.26anderseeTimRiker: but mostly you should just look at rtai
07:07.41anderseeTimRiker: since the current version is built on top of it
07:08.01anderseehttp://www.aero.polimi.it/~rtai/
07:36.33TimRikerhttp://www.pricewatch.com/1/33/1355-1.htm <- memory recommendations?
08:51.07anderseeTimRiker: got it working
08:51.25anderseeTimRiker: it was indeed the error handling that was broken
09:28.24TimRikersweet
09:45.11*** join/#elinux TheMasterMind1 (~aman@pool-141-156-161-12.esr.east.verizon.net)
13:55.55*** join/#elinux pb_ (~pb@cpc2-cmbg4-3-0-cust239.cmbg.cable.ntl.com)
14:08.13*** join/#elinux George- (~Test@adsl-165-63.freeuk.com)
14:26.32*** join/#elinux GPSFan (~Ken@65.121.49.59)
15:27.59*** join/#elinux jnork (~jnork@pool-141-158-46-12.phil.east.verizon.net)
16:50.55*** join/#elinux jnork (~jnork@pool-141-158-46-12.phil.east.verizon.net)
17:10.43*** join/#elinux jnork (~jnork@pool-141-158-46-12.phil.east.verizon.net)
17:15.39chouimathi
17:16.49jnorklo
19:35.51*** join/#elinux mallum (~mallum@82-35-113-164.cable.ubr07.dals.blueyonder.co.uk)
19:45.06mallumchouimat: yo
21:05.41chouimatmallum: ?
21:06.15*** join/#elinux andersee (~andersee@codepoet.org)
21:06.15*** mode/#elinux [+o andersee] by ChanServ
21:07.42mallumchouimat: hi, no new 24's over xmas ?
21:08.05chouimatmallum: i don't know probably not
21:48.50*** join/#elinux sjhill (~sjhill@pool-151-201-232-6.pitt.east.verizon.net)
22:10.39*** join/#elinux jnork (~jnork@pool-141-158-46-12.phil.east.verizon.net)

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