**** BEGIN LOGGING AT Sat Oct 21 02:59:57 2006
Oct 21 09:59:52 <likewise> ~seen [g2]
Oct 21 10:00:20 <jbot> [g2] <n=g2@nslu2-linux/g2> was last seen on IRC in channel #nslu2-linux, 7h 53m 59s ago, saying: 'dumfrac nice to meet you, I've seen you e-mails'.
Oct 21 11:30:08 <[g2]> dumfrac morning
Oct 21 11:34:10 <blaster8> ~seen dumfrac
Oct 21 11:34:23 <jbot> dumfrac is currently on #nslu2-linux (10h 51m 32s). Has said a total of 3 messages. Is idling for 10h 30m 8s, last said: '[g2]: i guess I mean that I'll test 02-mtdpart-redboot-byteswap-partition-truncate.patch :-)'.
Oct 21 11:40:01 <barnseenio> ~morning
Oct 21 11:40:07 <jbot> Good morning!
Oct 21 11:40:35 <blaster8> morning
Oct 21 11:40:48 <barnseenio> i have to go trim hedges & lawns today
Oct 21 11:40:57 <barnseenio> the joys of home ownership
Oct 21 11:42:08 <likewise> good morning /afternoon fooks
Oct 21 11:42:10 <[g2]> barnseenio s/have/get/  :)
Oct 21 11:42:20 <[g2]> hey likewise 
Oct 21 11:42:22 <barnseenio> lol
Oct 21 11:42:37 <barnseenio> my wifes taken my son swimming, i got around 2 hours to do jobs
Oct 21 11:42:41 <[g2]> barnseenio doesn't that _sound_ so much better
Oct 21 11:42:47 <likewise> Anyone done some bootloader programming? I find that the default boot loaders have the flash chip select / access timing  way too conservative.
Oct 21 11:42:48 <barnseenio> of which ive already wasted 3o mins doing email/irc
Oct 21 11:43:11 <[g2]> likewise that's on purpose
Oct 21 11:43:35 <CIA-8> 03blaster8 * r509 10kernel/trunk/patches/ (4 files in 2 dirs): Add Redboot Partition Truncate to 2.6.19 and clean byteswap patch
Oct 21 11:43:57 <barnseenio> hang tough sluggers 
Oct 21 11:45:21 <likewise> [g2]: so that all flashes work?
Oct 21 11:45:50 <[g2]> likewise yeah, 1) have hw work, 2) make it faster
Oct 21 11:45:58 <likewise> [g2]: but does any platform modify their timings in Linux afterwards?
Oct 21 11:46:22 <likewise> [g2]: I am sure every platform can get a two to threefold access time reduction.
Oct 21 11:46:30 <likewise> ixp4xx platform, that is
Oct 21 11:46:33 <[g2]> likewise once you start doing a custom bootloard, well the box is yours :)
Oct 21 11:46:51 <[g2]> s/bootloard/bootloader/
Oct 21 11:46:51 <jbot> [g2] meant: likewise once you start doing a custom bootloader, well the box is yours :)
Oct 21 11:47:11 <likewise> [g2]: I know, just wondered why there is no existing stuff to be found that optimizes this for the existing ixp4xx platforms.
Oct 21 11:47:35 <[g2]> likewise it depends on what you are trying to optimize
Oct 21 11:48:02 <likewise> [g2]: throw away redundant clock cycles in the expansion bus accesses.
Oct 21 11:48:44 <[g2]> I'd be all up for testing that with you
Oct 21 11:49:00 <likewise> [g2]: what's your boot loader?
Oct 21 11:49:09 <blaster8> likewise: are you running BE?
Oct 21 11:49:17 <[g2]> but I really get nearly a 30x by booting to USB
Oct 21 11:49:30 <likewise> [g2]: I copied a 8 MB FPGA bitstream from Flash to expansion bus, and wondered why it took 3 seconds...
Oct 21 11:49:36 <[g2]> likewise Redboot 2.02 or 03
Oct 21 11:49:42 <[g2]> custom
Oct 21 11:49:57 <likewise> oh yes, already told me that. I should move to .03 first...
Oct 21 11:50:20 <[g2]> I don't think there's any big improvements there
Oct 21 11:50:48 * [g2] _loves_ the -m disk hda1:/boot/zimage stuff
Oct 21 11:51:45 <[g2]> likewise have you tested the processing time from loading the FPGA from RAM ?
Oct 21 11:51:55 <[g2]> like preload the image into RAM then boot
Oct 21 11:52:50 <[g2]> you should be able itemize the 3 seconds into Expansion I/O and other
Oct 21 11:53:53 <NAiL> hmm
Oct 21 11:54:19 <NAiL> so it would be possible to speed up the flash on eg. the nslu2?
Oct 21 11:54:28 <NAiL> (after redboot, of course)
Oct 21 11:54:54 <blaster8> possibly
Oct 21 11:55:02 <[g2]> NAiL I think he's talking about both during and after
Oct 21 11:55:03 <NAiL> That'd be way cool
Oct 21 11:55:22 <blaster8> [g2]: we aren't modifying the stock bootloader, remember :)
Oct 21 11:55:30 <NAiL> [g2]: yeah, but the nslu2 won't (in most cases) be touching redboot
Oct 21 11:55:34 <[g2]> NAiL sure anybody with JTAG can play around with it
Oct 21 11:55:54 <blaster8> [g2]: no warnings on 2.6.19-rc2 build with your redboot patch
Oct 21 11:56:13 <[g2]> blaster8 and then that's something you'll need to leave on the table
Oct 21 11:56:14 <NAiL> [g2]: I think the number of JTAG'ed NSLU2's are <10
Oct 21 11:56:58 <[g2]> well if (when) you go too fast with the timing adjustments, you'll need jtag to recover
Oct 21 11:57:12 <blaster8> not so good
Oct 21 11:57:22 <NAiL> anyway, combined with the coherency change/npe DMA stuff, the slug just might be a bit faster :)
Oct 21 11:57:28 <[g2]> and don't forget there are parts variance and substitutions across the product line
Oct 21 11:57:41 <[g2]> remember this box has been in production for nearly 2.5 years
Oct 21 11:57:43 <blaster8> so, basically, best to be conservative, just in case
Oct 21 11:57:48 <blaster8> done
Oct 21 11:58:04 <CIA-8> 03blaster8 * r510 10kernel/trunk/patches/2.6.19/02-mtdpart-redboot-byteswap-partition-truncate.patch: Refresh Redboot Partition Truncate patch
Oct 21 11:58:12 <[g2]> blaster8 it really depends on what you are trying to do and what it's worth to you
Oct 21 11:58:21 <blaster8> exactly
Oct 21 11:58:23 <[g2]> remember the slugs are only $100 new
Oct 21 11:58:37 <[g2]> it's not like you are toasting a car or your house
Oct 21 11:59:03 <[g2]> the NPU dev system I worked on was $45K
Oct 21 11:59:15 <blaster8> that's ridiculous
Oct 21 11:59:21 <blaster8> how much is an ixdp425?
Oct 21 11:59:52 <[g2]> a couple years ago they were 3K but that's a reference design
Oct 21 12:00:03 <[g2]> it's different
Oct 21 12:00:34 <[g2]> I didn't get any ixdp425s (I had one on loan for about 6 weeks though)
Oct 21 12:00:59 <[g2]> the NPU dev system was really bada**
Oct 21 12:01:25 <[g2]> that was back in '01-'02
Oct 21 12:01:51 <[g2]> now you can to the same stuff on a $300 PC
Oct 21 12:02:11 <[g2]> well nearly the same stuff
Oct 21 12:09:02 <[g2]> dumfrac I think you are correct about the FIS directory size adjustment for the NSLU2
Oct 21 12:09:58 <likewise> what is the flash part on a slug?
Oct 21 12:10:09 <[g2]> 8MB Intel
Oct 21 12:10:19 <likewise> Damned they are so cheap I lost mine in the paper work :-)
Oct 21 12:10:20 <likewise> s/in/between
Oct 21 12:10:25 <likewise> Strata?
Oct 21 12:10:30 <[g2]> yeah
Oct 21 12:11:09 <likewise> things like jffs2 on flash will benefit from the speedup
Oct 21 12:11:22 <[g2]> and small partitions
Oct 21 12:11:44 <[g2]> likewise have you played with jffs2 option that optimizes for boot ?
Oct 21 12:12:01 <[g2]> or maybe I'm thinking squashfs 3.0
Oct 21 12:12:02 <likewise> [g2]: summarize option? not yet, no
Oct 21 12:12:25 <likewise> [g2]: Yes, did most testing on squashfs(-lzma) 3.0 (3.1r2 release)
Oct 21 12:13:02 <[g2]> either jffs2 or squahsfs has a specific option for optimizing boots
Oct 21 12:13:25 <[g2]> I remember seeing the option and thinking I should try it out, but haven't been back to test it
Oct 21 12:14:11 <[g2]> I'd guess hat lzma saves space for sure and possibly time
Oct 21 12:14:56 <[g2]> wonderful chatting guys....  bb in a few hours
Oct 21 12:15:17 <[g2]> blaster8 what's the magic chant in the Makefile to run the .19-rc2 ?
Oct 21 12:15:25 <NAiL> §IT
Oct 21 12:15:28 <NAiL> uh
Oct 21 12:15:43 <blaster8> change PATCH thingy from 2.6.18 to 2.6.19
Oct 21 12:15:45 <blaster8> make clobber && make
Oct 21 12:16:04 <[g2]> blaster8 cool, I try that later today
Oct 21 12:16:10 <blaster8> remember, 2.6.19 should have loft patches which can be sent upstream :)
Oct 21 12:16:31 <blaster8> i.e. separating out the Avila startup code and adding a libata CF driver
Oct 21 12:16:38 <[g2]> the ATA patch is gonna change a lot
Oct 21 12:17:02 <blaster8> fine by me, as long as it's the CF support patch in 2.6.19
Oct 21 12:17:09 <blaster8> 2.6.19 experimental, 2.6.18 stable
Oct 21 12:27:39 <likewise> Where is IDE and CF support and usability currently at with ixp4xx platforms?
Oct 21 12:28:32 <blaster8> mainline support non-existent
Oct 21 12:28:52 <blaster8> hoping to fix this for 2.6.20, probably (but possibly 2.6.19 :) )
Oct 21 12:29:11 <blaster8> are you running BE?
Oct 21 12:29:24 <likewise> blaster8: there are no IXP4xx platforms with IDE or CF in existance?
Oct 21 12:29:40 <blaster8> mainline kernel support non-existant
Oct 21 12:29:55 <blaster8> Avila make ixp4xx with CF attached to expansion bus
Oct 21 12:30:21 <blaster8> no widespread ixp4xx with ide attached to expansion bus - most have pci devices to provide ide
Oct 21 12:30:34 <likewise> [g2]: considering and testing several options for FPGA (pre-linux, post-linux, 1-1 from flash, compressed/uncompressed, etc)
Oct 21 12:31:22 <blaster8> loading from CF card would be quite neat
Oct 21 15:17:55 <AdamBaker> Hi all - I'm still trying to get an RT73 based USB WiFI adaptor working on a slug. I've now switched to DebianSlug so I've got an LE machine which the older driver version needs but it also needs PREEMPT disabled and it is on by default so I'd need to change that at least and maybe fix other stuff. The other option would be to use the newer, better written (serialmonkey.com) driver but that needs kernel >=2.6.17 which would mean switching to a
Oct 21 15:44:40 <NAiL> VoodooZ: ping?
Oct 21 15:44:55 <NAiL> Did you know anything about WinTV PVR USB2? 
Oct 21 15:46:03 * NAiL is considering one for his Thecus N2100
Oct 21 15:46:27 <NAiL> I want to use my n2100 as a mythtv backend :)
Oct 21 15:49:22 <VoodooZ_Home> pong
Oct 21 15:49:37 <VoodooZ_Home> Nope. Ne3ver played w/ one
Oct 21 15:49:45 <VoodooZ_Home> sounds nice
Oct 21 16:41:53 <CieD> NAiL, are you in the UK?
Oct 21 16:44:18 <CieD> we were talking about a freeview version earlier in #hantslug 
Oct 21 16:44:22 <blaster8> CieD: No, but I am
Oct 21 16:47:17 <CieD> thanks.. he was talking about WinTV dongles
Oct 21 16:47:24 <CieD> that's why I asked him
Oct 21 16:52:32 <NAiL> CieD: nope, Norway
Oct 21 16:53:13 <CieD> Ok.. our freeview dongles would be useless then :)
Oct 21 16:54:29 <blaster8> dumfrac: ping
Oct 21 16:55:23 <VoodooZ> hehehe. I had 2 xchat open! no wonder my nick was used!
Oct 21 16:56:28 <NAiL> CieD: What's frieeview?
Oct 21 16:57:04 <CieD> It's digital TV, over terrestrial frequencies rather than satellite
Oct 21 16:57:48 <CieD> so here, it'd be ideal to plug into the slug and record
Oct 21 17:00:04 <blaster8> except that on-the-fly video recompression is not terribly realistic
Oct 21 17:00:21 <blaster8> so you'd be recording MPEG-2, which is quite heavy space-wise
Oct 21 17:00:59 <CieD> I've not looked that far into it, I just know others that use them
Oct 21 17:04:20 <AdamBaker> Freeview MPEG-2 is on average 2GB/hr so not hugely compressed but quite usable with a big disk
Oct 21 17:32:16 <NAiL> DVB-T?
Oct 21 17:32:40 * NAiL leaves for the cinema
Oct 21 20:12:04 <dumfrac> blaster8: I have posted the output of version to the nslu2-developers mailing list - useful command :-)
Oct 21 20:12:33 <blaster8> thanks dumfrac - it hasn't come through yet, but I'll give it another few minutes
Oct 21 20:12:58 <blaster8> I'm keen to find out exactly which utility builds the FIS directory entry
Oct 21 20:13:33 <dumfrac> blaster8: i think that slugimage creates the FIS directory - I could be wrong, but it seemed so from my brief look at the code this morning
Oct 21 20:13:52 <blaster8> I think it should, but I'm no perl expert
Oct 21 20:14:43 <dumfrac> me neither :-)
Oct 21 20:16:16 <dumfrac> hmm... should that be "me either"
Oct 21 20:16:52 <blaster8> who knows?
Oct 21 20:16:54 <blaster8> ;)
Oct 21 20:16:57 <dumfrac> :-)
Oct 21 20:17:03 <blaster8> ok, well something odd is going on
Oct 21 20:17:26 <dumfrac> i'm going to take a look at the apex code to see if it could be some assumption made there
Oct 21 20:18:42 <blaster8> no, it's slugimage, I'm sure
Oct 21 20:18:51 <blaster8> well, pretty sure
Oct 21 20:19:19 <dumfrac> you may be right - it is more likely because I really only made a simple minded hack to try it out
Oct 21 20:22:03 <blaster8> A hexdump of the FIS partition area of the NSLU2 would be good too :)
Oct 21 20:22:07 <blaster8> you can do this with APEX
Oct 21 20:23:05 <dumfrac> ok - i'll do that now...
Oct 21 20:23:22 <blaster8> thanks, http://pastebin.ca
Oct 21 20:23:53 <blaster8> Hopefully seeing the hexdump should make it obvious whether APEX is misinterpreting the flash or whether Slugimage is writing out the wrong thing
Oct 21 20:26:50 <dumfrac> it should be 'dump -4 nor:0xb0007e50' right ?
Oct 21 20:27:10 <blaster8> hopefully something like that
Oct 21 20:29:18 <dumfrac> i'm confused - i'm not seeing anything that makes sense - just checking that i have the right address
Oct 21 20:30:47 <blaster8> dump nor:0x7e0000+4k
Oct 21 20:33:49 <dumfrac> http://pastebin.ca/214093
Oct 21 20:35:25 <blaster8> so why does APEX think that FIS directory is at nor:0xb0007e50?
Oct 21 20:35:35 <blaster8> time to get the developer I think
Oct 21 20:36:06 <blaster8> I'm not entirely clear these skip regions are a good idea
Oct 21 20:36:41 <dumfrac> i sort of know what the skip regions are for, but I don't really know how they work
Oct 21 20:36:58 <blaster8> http://trac.nslu2-linux.org/slugimage/changeset/72
Oct 21 20:38:26 <blaster8> we need rwhitby too, and he won't be back till Monday
Oct 21 20:38:44 <dumfrac> right - it's the mechanics of the skip regions that I'm not familar with :-)
Oct 21 20:38:55 <blaster8> can you send a reply with the pastebin contents to the mailing list
Oct 21 20:39:11 <dumfrac> ok
Oct 21 20:39:16 <blaster8> we'll just have to wait this one out, I think
Oct 21 20:39:58 <dumfrac> i'll keep on looking to see if I can figure it out - i have leant a lot about the slug in a short period of time so learning apex should be too hard (ha ha) :-)
Oct 21 20:40:16 <blaster8> it's good code, well commented
Oct 21 20:40:21 <blaster8> thanks for doing this testing, btw
Oct 21 20:40:23 <dumfrac> s/leant/learnt
Oct 21 20:40:38 <blaster8> you're bringing up issues faster than we can solve them :)
Oct 21 20:40:40 <blaster8> which is a good thing
Oct 21 20:40:46 <dumfrac> it is a pleasure - i'm just glad that i have found a way to contribute to the project
Oct 21 20:40:53 <dumfrac> :-)
Oct 21 20:41:09 <dumfrac> hopefully one day i may know enough to solve some
Oct 21 20:42:02 <dumfrac> how long is a patebin post kept ?
Oct 21 20:42:15 <dumfrac> s/patebin/pastebin
Oct 21 20:44:29 <blaster8> forever
Oct 21 20:44:38 <blaster8> unless you specify otherwise
Oct 21 20:45:14 <dumfrac> ok - i'll just post the pastebin link to the mailing list
Oct 21 22:40:23 <[g2]> dumfrac hey thx for testing the new patch on the nslu2
Oct 21 23:55:38 <dumfrac> [g2]: you re welcome. I'm am now about test the daily build of the Debian installer
Oct 22 01:07:18 <[g2]> dumfrac you're running etch right ?
Oct 22 01:15:39 <dumfrac> [g2]: on the slug I'm running etch with the kernel from sid (2.6.18)
Oct 22 01:15:58 <dumfrac> and patched (obviously) :-)
Oct 22 01:16:38 <[g2]> dumfrac so are you using a patched debian sid kernel ?
Oct 22 01:16:54 <dumfrac> yup
Oct 22 01:17:08 <[g2]> as opposed to a nslu2-linux kernel
Oct 22 01:17:55 <dumfrac> ah yes, the ixp4xx kernel. The nslu2 specific kernel has been removed from sid.
Oct 22 01:18:16 <dumfrac> yes, as oppsoed to the nslu2-kernel
Oct 22 01:18:32 <dumfrac> s/oppsoed/opposed
Oct 22 01:18:58 <dumfrac> I have applied the networking patches and the reboot patch
Oct 22 01:20:21 <dumfrac> the patches are taken directly from the nslu2 subversion respository
Oct 22 01:21:41 * [g2] is dist-upgrade a box to etch now
Oct 22 01:22:44 <dumfrac> [g2]: not sure what you mean
Oct 22 01:23:04 <[g2]> I'm dist-upgrading a box from sarge to etch
Oct 22 01:25:25 <dumfrac> I started out with etch. I used beta 2 and then beta 3 of the debian installer. The unofficial beta 2 installer that tbm vreated had the old Intel ether driver. To use the beta three, you need still need to use a USB to ethernet adapter.
Oct 22 01:26:13 <[g2]> dumfrac I've got my own hardare (it's has serial ports and CF that looks like hda
Oct 22 01:26:27 <dumfrac> I guess dist-upgrade is going to give you a 2.6.17 kernel
Oct 22 01:26:36 <dumfrac> the Loft system ?
Oct 22 01:26:45 <[g2]> yeah, I don't have a kernel installed
Oct 22 01:27:13 <dumfrac> ah, right. What is the ethernet hardware
Oct 22 01:27:14 <[g2]> I build my own which are mostly the nslu2-linux repo with tweaks
Oct 22 01:27:27 <[g2]> dual ethers (POE)
Oct 22 01:27:29 <dumfrac> hence the defconfig_loft file
Oct 22 01:27:56 <[g2]> yeah.  This test box has 128MB, 16MB flash, lots of fun stuff
Oct 22 01:28:19 <dumfrac> Do you build the Loft systems ? And, if so, for what purpose ?
Oct 22 01:28:53 <dumfrac> brb - got to check on my chicken curry
Oct 22 01:29:00 <[g2]> yum
Oct 22 01:35:40 <dumfrac> back - I have another 10 minutes or so before I have to tend to again :-)
Oct 22 01:37:09 <[g2]> heh
Oct 22 01:37:35 <[g2]> I've probably got another 75 packages before the upgrade is finished
Oct 22 01:38:18 <[g2]> I'm up the the "p"s
Oct 22 01:39:30 <dumfrac> :-) I imagine that your hardware is a touch faster than the slug - the slug lives up to it's name when installing a new distribution. I know, because I've been trying out the latest build of the Debian installer.
Oct 22 01:40:30 <[g2]> yeah. the usb 2.0 disks can be pretty quick
Oct 22 01:41:01 <[g2]> right now I'm bw limited to the Microdrive
Oct 22 01:41:47 <[g2]> If I was upgrading the external drive the I/O would be way faster
Oct 22 01:42:08 <dumfrac> I tried to find some information about the Loft hardware but all I can find is http://www.giantshoulderinc.com/. Is this you ?
Oct 22 01:42:57 <[g2]> yup
Oct 22 01:43:39 <[g2]> dumfrac it's really just a custom Gateworks GW2348
Oct 22 01:43:40 <dumfrac> do you sell these ?
Oct 22 01:44:01 <[g2]> yeah
Oct 22 01:44:30 <[g2]> http://www.giantshoulderinc.com/hw-4533
Oct 22 01:45:09 <dumfrac> cool, looks like a nice system - cheick curry calls again - time to make the sauce
Oct 22 01:45:22 <dumfrac> s/cheick/chicken
Oct 22 01:45:33 <[g2]> http://www.giantshoulderinc.com/ab3/case.jpg
Oct 22 01:45:42 <[g2]> yum yum
Oct 22 01:47:34 * [g2] reboots etch box
Oct 22 02:00:56 <dumfrac> [g2]: did it work ?
Oct 22 02:01:03 <[g2]> dumfrac sure
Oct 22 02:01:29 <dumfrac> cool
Oct 22 02:01:43 <[g2]> nod. thx.  
Oct 22 02:02:01 <[g2]> barnseen2o ping
Oct 22 02:04:02 <[g2]> anybody know anything about rtorrent ?
Oct 22 02:04:34 <[g2]> dumfrac how's the curry ?
Oct 22 02:05:10 <dumfrac> it is now in the oven baking - will be ready in about half an hour
Oct 22 02:06:23 <[g2]> dumfrac got a good recipe ?
Oct 22 02:08:15 <dumfrac> it is one I got from my mother years ago, when I first came to the US - I can type it up and send it to you, if you'd like ?
Oct 22 02:10:23 <[g2]> dumfrac that'd be great if you don't mind
Oct 22 02:10:37 * [g2] has never made curry
Oct 22 02:11:10 <[g2]> I've eaten it plenty at Indian/Thai resturants
Oct 22 02:11:42 <dumfrac> no problem - now I've got help clean the mess I made in the kitchen - that's the problem with cooking :-) I'll send the recipe later.
Oct 22 02:11:43 <[g2]> There a good Roti Prata nearby
Oct 22 02:12:15 <[g2]> sure anytime. It's not like I'm gonna rush out at midnight to cookup a 3AM curry :)
Oct 22 02:12:20 <[g2]> but THX
Oct 22 02:12:31 <dumfrac> :-) talk to you later
Oct 22 02:12:51 <[g2]> dumfrac I'm guessing you are PST/MNT time
Oct 22 02:13:08 <[g2]> GMT-7
Oct 22 02:20:53 <joshin> GMT-7 is the best!
**** ENDING LOGGING AT Sun Oct 22 02:59:56 2006