**** BEGIN LOGGING AT Mon Mar 14 23:59:56 2005
Mar 15 01:28:12 * perlguru byebye
Mar 15 03:15:23 <rwhitby> Wouldn't you know it, as soon as you buy a 1.1 hub for testing, 1.1 devices start working on a 2.0 hub with Openslug with EHCI as a mdule.
Mar 15 03:15:26 <rwhitby> module.
Mar 15 03:16:00 <rwhitby> Now to put EHCI back into the kernel and retest that (since that's what was originally failing).
Mar 15 03:27:22 <perlguru> self-repairing-code ;)
Mar 15 03:37:41 <rwhitby> or transient faults :-(
Mar 15 03:48:35 * dyoung-zzzz groans
Mar 15 03:50:30 <rwhitby> hey dyoung-zzzz
Mar 15 03:50:39 * perlguru byebye
Mar 15 03:51:28 * perlguru Leaving
Mar 15 03:51:33 <dyoung-zzzz> I was going to implement .altroot tonight.
Mar 15 03:51:42 <dyoung-zzzz> but I got distracted.
Mar 15 03:52:32 <rwhitby> so I've tested 1.1 pl2303, 1.1 ftdi_sio, 1.1 pegasus, 1.1 and 2.0 flash disks, and a 1.1 bluetooth dongle, all on (a) directly connected, (b), 1.1 hub, (c) 2.0 hub, and all with ehci as a module, but enabled (i.e. it is loaded after the pivot)
Mar 15 03:53:30 <rwhitby> the 1.1 bluetooth on 2.0 hub gave a couple of errors (hci_usb_isoc_rx_submit_urb error -38) but still worked fine (rfcomm uses bulk, not isoc anyway)
Mar 15 03:54:33 <rwhitby> so I'm going to retest all on unslung while openslug is building, and then retest again on openslug with ehci built-in (although I suspect I will have to do some kernel trimming to achieve that)
Mar 15 04:02:31 <dyoung-zzzz> Good Plan.
Mar 15 04:04:43 <rwhitby> what's the line in /etc/fstab need to be to mount /proc/bus/usb ...
Mar 15 04:04:45 <rwhitby> ?
Mar 15 04:04:53 <rwhitby> (in openslug)
Mar 15 04:05:26 <rwhitby> and how do you mount it manually
Mar 15 04:08:17 <rwhitby> root@nslu2:~# mount -t usbdevfs usbdevfs /proc/bus/usb
Mar 15 04:08:17 <rwhitby> mount: Mounting usbdevfs on /proc/bus/usb failed: No such device
Mar 15 04:09:00 <rwhitby> should usbdevfs be in /proc/filesystems?
Mar 15 04:09:06 <rwhitby> (it's not on my openslug)
Mar 15 04:55:08 <rwhitby> root@nslu2:~# microcom -D/dev/ttyUSB1
Mar 15 04:55:08 <rwhitby> PL-2303 ttyUSB1: pl2303_open - failed submitting interrupt urb, error -28
Mar 15 04:55:30 <rwhitby> ok, I got a failure on a PL2303 attached to a usb2.0 hub on openslug running ehci
Mar 15 05:01:50 <rwhitby> let's try again with a 1.1 hub ...
Mar 15 05:06:09 <rwhitby-away> error is repeatable after a reboot (with the same 2.0 hub)
Mar 15 05:06:33 <rwhitby-away> let's take everything else off the hub first ..
Mar 15 06:02:48 * perlguru Leaving
Mar 15 07:44:55 <rwhitby> OK, the pl2303 problems seems to occur only when you have two pl2303 cables plugged into the same 2.0 hub, as well as a bluetooth dongle.  Take any one of those three ingredients away, and everything works fine.  Don't have the bluetooth dongle plugged in at boot, and plug it in later, and everything works fine.  Use a 1.1 hub, and everything works fine.
Mar 15 07:46:31 <rwhitby> The error occurs on the open of the second pl2303 device.  Hence I think it might be a pl2303 error rather than a ehci error.  But you never know.
Mar 15 08:17:22 <CIA-1> 03jp30 * 10unslung/make/vim.mk: generate a control file for vim, and stage ncurses
Mar 15 08:23:37 <CIA-1> 03jp30 * 10unslung/Makefile: promote vim
Mar 15 08:32:51 <CIA-1> 03jp30 * 10unslung/make/ (template.mk vim.mk): correct use of <foo> in template.mk; and fix resulting error in vim
Mar 15 11:55:28 <kergoth> rwhitby-away: fyi, i added uclibc 0.9.19 to oe.  took about 2 minutes to add it, and about 4 to fix the slight differences in its buildsystem.  testing builds atm
Mar 15 13:15:32 <dyoung-zzzz> rwhitby, is your Prolific device a PL2303 or a 2303X ?  The 2303x has a know driver bug.
Mar 15 13:39:08 <rwhitby-away> kergoth: thx for adding uclibc .19 to OE - did you push the changeset?
Mar 15 13:39:37 <rwhitby-away> dyoung-zzzz: the pl2303 devices in question are the ones in the RS usb serial cables
Mar 15 13:40:19 <dyoung-zzzz> those are reguular ones.  Hmm
Mar 15 13:40:26 <kergoth> rwhitby-away: not yet, still testing builds with it
Mar 15 13:40:57 <rwhitby-away> kergoth: cool, thx.  It will be cool to have all the OE packages for the Asus WL500g too :-)
Mar 15 13:41:08 <kergoth> rwhitby-away: that platform is mipsel?
Mar 15 13:41:11 <rwhitby-away> yep
Mar 15 13:41:13 <kergoth> k
Mar 15 13:41:40 <rwhitby-away> think of Oleg's wl500g firmware as the Unslung version of OpenWRT ....
Mar 15 13:42:23 <rwhitby-away> [g2]-away: so I got the OpenSlug kernel down to 955676 !
Mar 15 13:42:52 <dyoung-zzzz> Geez, what you yankn from it?
Mar 15 13:43:05 <dyoung-zzzz> I imagine DEBUG takes a lot of space...
Mar 15 13:43:06 <rwhitby-away> ! CONFIG_KALLSYMS=n
Mar 15 13:43:06 <rwhitby-away> ! CONFIG_USB_EHCI_HCD=y
Mar 15 13:44:02 <rwhitby-away> I could get it within 1k by removing other stuff (like /proc/pci, ...), but that's the big one.
Mar 15 13:44:16 <[g2]-away> rwhitby-away, Excellent!
Mar 15 13:44:31 <kergoth>  /proc/pci really isnt necessary anymore anyway.  lspci is our friend
Mar 15 13:44:34 <kergoth> heh
Mar 15 13:44:48 <rwhitby-away> the question is: do we really want KALLSYMS.  I expect it is very useful for those who can actually do something with oopses.
Mar 15 13:45:36 <[g2]-away> I'd lose the symbols for now
Mar 15 13:45:55 <rwhitby-away> yeah, the apex build can put it back in
Mar 15 13:46:03 <[g2]-away> I think it's more important that it be easy for VoodooZ_log and perlguru to be able to build on AMD
Mar 15 13:46:35 <[g2]-away> and truth is we'll have the symbol map for the kernel
Mar 15 13:46:45 <[g2]-away> or the ppl that build locally have the map
Mar 15 13:47:03 <kergoth> yep, can always ksymoops em
Mar 15 13:47:04 <rwhitby-away> yeah - this will solve the 64 bit problem too
Mar 15 13:47:23 <rwhitby-away> or at least mask the 64 bit problem :-)
Mar 15 13:48:05 <[g2]-away> We'll it makes total sense to me that building on a 64-bit may have some default alignments different
Mar 15 13:48:40 <kergoth> hm?
Mar 15 13:48:48 <kergoth> the target arch should be completely independent of build arch.
Mar 15 13:48:48 <[g2]-away> I think there may be an issue there, but this is why I think we should have a tool that verifies the actual ELF text and other sections
Mar 15 13:48:56 <kergoth> if there's leakage, its broken
Mar 15 13:50:32 <[g2]-away> I don't think it's leakage, I think its different alignment and sizing 
Mar 15 13:52:42 <kergoth> what are you talking about?
Mar 15 13:52:59 <kergoth> you're targeting arm.  if your build machine's type sizes or alignment affect target binaries, its BROKEN
Mar 15 13:53:02 <kergoth> period
Mar 15 14:02:04 <rwhitby-away> kergoth: I agree.
Mar 15 14:02:26 <kergoth> it isnt too much to expect, to expect that the target binaries will not be affected by the build environment
Mar 15 14:02:42 <kergoth> if they are, your builds are no longer portable, consistent, or reproducable.  you ahve to require that everyone build it from the same environment you do
Mar 15 14:02:46 <kergoth> which is just completely unreasonable
Mar 15 14:02:59 <kergoth> heh
Mar 15 14:06:49 <CIA-1> 03rwhitby 07 * r1.3078 10openembedded/ (4 files in 4 dirs): OpenSlug 1.7 - removed KALLSYMS, and EHCI is now compiled back into the kernel.
Mar 15 14:17:57 <wipmac> hi . is there an easy way to control bandwidth of the nslu2?
Mar 15 14:18:20 <rwhitby-away> kergoth: I will be adding the asusoe distro for the wl500g machine soon after you push the uclibc :-)
Mar 15 14:19:01 <kergoth> currently swiping some patches for uclibc 0.9.19 from elsewhere
Mar 15 14:31:40 <rwhitby-away> kergoth: I'm going to do a bk send to oe@handhelds.org of a patch for ipkg to make it use sysconfdir for /etc/ipkg - can you pick it up from there and review it?  It's a pretty basic packages, so I don't want to screw it up.
Mar 15 14:32:02 <kergoth> use bk -send -d
Mar 15 14:32:07 <kergoth> prepends with a unified diff
Mar 15 14:32:12 <kergoth> so we can reply to the mail with commetns
Mar 15 14:32:20 <rwhitby-away> ok, will do.
Mar 15 14:35:39 <rwhitby-away> kergoth: sent.  thanks in advance for your review.
Mar 15 14:35:42 <kergoth> np
Mar 15 14:36:22 <rwhitby-away> back later
Mar 15 14:39:25 <chris650> I'm curious if anyone has gotten mod_dav to work with apache on an NSLU .... does anyone know?
Mar 15 16:06:48 <rwhitby-web> I think I just found my next router: http://wl500g.info/showpost.php?p=11960&postcount=22
Mar 15 16:07:49 <rwhitby-web> mipsel, 200MHz, 32 (yes 32) MB RAM.  Only 4MB flash, but Oleg's custom firmware can load rootfs from attached storage.  2 external USB 2.0 ports, plus headers for another two inside.  Header for two UARTS on the board!
Mar 15 16:08:11 <rwhitby-web> http://www.newegg.com/app/ViewProductDesc.asp?description=39-112-005&depa=0
Mar 15 16:09:58 <rwhitby-web> And the ability to burn the bootloader without JTAG: http://wl500g.info/showpost.php?p=11987&postcount=4
Mar 15 16:10:07 <rwhitby-web> (it uses pmon)
Mar 15 17:09:08 <MrSLU2> Anyone around?
Mar 15 17:11:37 <[g2]> yes
**** ENDING LOGGING AT Tue Mar 15 23:59:56 2005