irclog2html for blob on 2002.07.09

00:20:00BZFlagRuss: where's that reflash app you wrote?
00:20:45Russ|werkwhich one?
00:21:29BZFlagany. all. ;-)
00:21:43BZFlagthe one that runs from user space and forces a reboot afterwards.
00:22:58BZFlagwatches lotr while hacking. ;-)
00:24:27BZFlagdon't you hate it when you get stabbed by a mordor blade?
00:39:43BZFlagRuss|werk: you around? you remember the app I speak of? you had one that bb init runs?
00:39:58Russ|werkBZFlag: thats jflash
00:40:04Russ|werkoh, that
00:40:04Russ|werkright
00:40:21Russ|werkwhich part? the /linuxex script?
00:40:45Russ|werkI was thinking jtag, I have an app that flashes from kernel space over jtag... (mtd map)
00:41:31BZFlagah. yeah I meant the bb init one. might be /linuxex
00:41:50BZFlagdoes it just copy bb + libc etc to a ramfs first?
00:42:19BZFlagI just want to dd a new rootfs and then reboot. just looking for an easy way to reboot after the dd.
00:42:23Russ|werkit copies a very simple staticly linked flash program
00:42:41BZFlagthat sounds great. ;-)
00:42:47Russ|werkthe flash program opens /out (copied from /dev/mtd4)
00:43:23Russ|werkthen opens mmaps /image.bin (image file)
00:43:34Russ|werkuses ioctl's to erase the flash
00:43:52Russ|werkthen starts doing write()'s to flash
00:44:07BZFlagthen how do you reboot?
00:44:18Russ|werkcloses everything, sync(), sleep(3), reboot(LINUX_REBOOT_CMD_RESTART);
00:45:15BZFlagis there a reason for the sleep(3)? or just to be safe?
00:46:43Russ|werkjust to be safe
00:46:57Russ|werkit doesn't hurt anything
00:46:58BZFlagsource? (if you don't mind)
00:47:11Russ|werksure
00:48:10Russ|werkto get any debug output, you'll have to redirect stdout/err to ttySA0
00:48:18BZFlaggack. fw inhibited.
00:48:38BZFlagemail? Tim@Rikers.org
00:48:48Russ|werkhttp://68.2.221.180:8000/flash.c
00:50:18Russ|werkcompiled statically, its 18k
00:51:22BZFlaguclibc?
00:51:49Russ|werkya
00:51:59BZFlagsweetness. thanx!
00:51:59Russ|werkarm-uclibc-gcc -Os -Wall -static flash.c -o flash
00:52:10Russ|werkand then strip it
00:52:15BZFlagyep
00:52:46Russ|werker, I'm actually forgetting to strip it
00:52:47BZFlagso it chroots and then calls itself?
00:53:15BZFlag        if (!strcmp(basename(argv[0]), "chroot")) chroot_main(argc, argv);
00:53:15BZFlag        else flash();
00:53:22Russ|werknm, I am
00:53:32Russ|werkno, you can ln -s flash chroot
00:53:46Russ|werkpivot_root . old
00:53:46Russ|werkexec ./chroot . ./flash
00:54:10Russ|werkso that way you don't have to worry about getting bb chroot to find bb and uclibc
00:55:02Russ|werk(LD_LIBRARY_PATH=./old/lib exec ./old/bin/busybox chroot . ./flash) or something
00:55:05BZFlaghmm.. if you do a pivot_root, why do you need to do a chroot?
00:55:25Russ|werkso then you can umount /old
00:55:41BZFlagjust an exec would get you that, no?
00:55:52Russ|werknope
00:55:59BZFlagwhy not?
00:56:25Russ|werkbecause lib is now in /old/lib
00:56:30BZFlagI do that on the tux /linuxrc. exec /linuxrc, which then execs sbin/init. I can unmount after the second exec.
00:56:38Russ|werkso bb won't know where /lib is unless you ln -s stuff
00:57:11Russ|werkreally, for me, since I'm on a cramfs, its not critical that I umount it
00:57:16BZFlagbut if you exec /foo (linked to flash) which just execs /flash then you would be ok.
00:57:35BZFlagno libs would be open.
00:57:57Russ|werkthere isn't only one way to do it
00:58:14Russ|werkchroot is really simple anyway
00:58:21BZFlagok, just asking. ;-)
00:58:31Russ|werkchroot(argv[1]), chdir("/"), execvp(argv[2], argv + 2);
00:59:48BZFlagI'm just thinking that an execvp would be enough.
01:00:29Russ|werkprobably, I think I read it in the kernel piviot_root doc
01:01:00BZFlagI think the bb pivot_root does the equiv of a chroot.
01:01:35Russ|werkya, I got that from Documentation/initrd.txt
01:02:12Russ|werk"- chroot to the new root afterwards"
01:06:24BZFlaghmm. perhaps I'm doing it wrong. /me goes off to experiment.
01:07:01Russ|werkdon't forget, after /linuxex start running, you need to > /dev/ttySA0 (or maybe /dev/console) to see anything
01:13:19BZFlagright. thanx.
01:13:50BZFlagmight just work on vercel, as there is a console/kybd
02:06:57Sammyuh , lot of typhoon ....
02:07:37Russ|werkpretty wet, eh?
02:08:10SammyI hate that ...
02:08:46SammyRuss|werk: do you check mtdblock.c again ?
02:09:53Russ|werkwhat do you mean?
02:10:13Sammyabout the mount -a hang problem ...
02:10:43Russ|werkstill getting it?
02:11:24SammyI put it on my puppy board , yes , but tux I change to 2.4.18 No
02:11:45Sammyhow can I fix that prob ?
02:12:10Russ|werkdoes grep invalidate_device drivers/mtd/mtdblock.c return anything?
02:12:42Russ|werkthis is probably the problem:
02:12:43Russ|werkhttp://lists.infradead.org/pipermail/linux-mtd-cvs/2002-May/002271.html
02:13:03Russ|werkhttp://lists.arm.linux.org.uk/pipermail/linux-arm-kernel/2002-March/007940.html
02:13:47Sammydrivers/mtd/mtdblock.c
02:14:00Sammyinvalidate_device(inode->i_rdev, 1);
02:14:31Sammycheck...
02:18:30Russ|werkjust take that line out
03:07:05Sammyok , fix it ...thanx Russ
07:50:24mmattenhi
08:28:44Sammyis comming ...
08:30:22Sammytyphoon
08:30:56Sammydouble blob site
08:31:16Sammyany different ?
08:31:50erikmyes, we moved to oftc because all other relevant channels moved
08:32:30Sammyerikm: and bot ?
08:32:54erikmlxrbot will move when I have time to visit the uni
08:34:28Sammywill here be shotdown after finish move ?
08:35:36Sammyer shutdown
08:35:55Sammyerikm: you must be more busy this day ...
08:40:25erikmI think I'll remove the channel
11:02:46harshahello everyone
11:03:19harshadoes the blob-2.-.5 support the Sleep Wakeup completly
11:03:49harshaor is still a lot of tweaking needed to get it up
11:04:06erikmharsha: should work
11:04:27harshano because I have been trying it and no progress
11:04:47erikmtry the latest CVS version
11:05:10harshathe other thing is DRAM and Peripheral HOLD as given in the datasheets of SA1110 says something diferrent about them
11:05:17harshalet me try the CVS
12:52:46harshaerik, how is the assbet supposed to wake, if the serial port hasnt been configured
12:53:20harshaif the blob branches to PSPR without initializing the serial ports
12:57:54erikm... which is exactly how it should be done
12:58:28harshabut how will the kernel come up then, if the serial ports are not initialized
12:58:37harshaI am a little confused, so please dont mind
12:58:48erikmthe kernel will reinitialise all hardware
12:59:29erikmthe idea was to make the bootloader part as simple as possible and let the kernel do the dirty work, simply because the kernel already knows in what state the hardware should be
12:59:41erikms/idea/design objective/
13:01:04harshaok
13:01:20harshathank you
14:01:34Speedy2Blob
14:10:47prpplaguemorning all
14:10:54prpplagueseems we have some new faces
14:13:59Speedy2Heh
14:15:15erikmprpplague: note that seletz, mmatten and me moved to oftc.net
14:15:31Speedy2Hey Erik
14:15:42erikmwaves
14:16:57Speedy2Is LART going to move to the SA-1110?
14:17:03Speedy2Can you guys even get SA-1100s?
14:21:34prpplagueerikm: you got a #blob on oftc?
14:34:43erikmSpeedy2: lart is not moving to sa1110. there are already enough sa1110 designs out there
15:47:44FareGakuk!
15:48:04Farecan anyone here help with my linux porting problem?
15:49:18erikmfood
15:49:59Fareindeed, food will help me
15:50:12FareI mean, without food, I'll have much more troubles trying to port linux

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