**** BEGIN LOGGING AT Mon Oct 23 02:59:57 2006 Oct 23 03:31:40 [g2]: pong Oct 23 07:25:43 good morning Oct 23 07:28:46 'morning Oct 23 07:30:55 I gained roughly 30% to 40% speed increase reducing the number of extended access cycles to flash. Oct 23 07:31:21 speed increase as in flash bandwidth performance Oct 23 07:32:16 after boot? Oct 23 07:37:06 in the bootloader, but it is used throughout Linux, so also JFFS2 would benefit from it. Or loading the kernel from flash. Oct 23 07:43:41 hmm Oct 23 07:43:58 wouldn't speed up loading the kernel on the nslu2, but it could help later on Oct 23 07:47:41 NAiL: why not on the nslu2? Oct 23 07:49:46 because RedBoot on the nslu2 (afaik) just reads 1MB of flash before doing anything Oct 23 07:50:07 we can't speed up that part Oct 23 08:28:23 morning sluggers Oct 23 08:42:08 morning barnseen2o Oct 23 08:46:24 Anyone knows who is doing the libata IDE driver for the IXP4xx? Oct 23 08:49:16 03rwhitby * r511 10kernel/trunk/patches/ (2.6.18/defconfig 2.6.19/defconfig): Removed -slugos from LOCALVERSION in defconfig. We don't do that. Oct 23 08:49:56 03rwhitby * r512 10kernel/trunk/Makefile: Updated top-level Makefile to build 2.6.19 Oct 23 08:59:44 03marcusb * r59 10debian/ixp400-accesslib/trunk/ixp400-accesslib/debian/changelog: Add ~unreleased tag to version. Oct 23 09:00:26 [g2]: ping Oct 23 09:00:33 03marcusb * r60 10debian/ixp400-accesslib/trunk/ixp400-accesslib/debian/ (README.Debian control.modules.in rules): Generate a module symvers file, needed when building the ixp400-eth module. Oct 23 09:04:01 03marcusb * r61 10debian/ixp400-eth/trunk/ixp400-eth/ (10 files in 3 dirs): New upstream version. Oct 23 09:08:17 likewise: that is dwery Oct 23 09:08:33 ~seen dwery Oct 23 09:08:39 dwery was last seen on IRC in channel #nslu2-linux, 3d 17h 46m 39s ago, saying: 'yep'. Oct 23 09:08:49 tnx Oct 23 09:10:41 any other loft users here, besides [g2]? Oct 23 09:11:43 [g2] sold more than one then? ;-) Oct 23 09:34:32 rwhitby: the libata patch is to make the ixp4xx work with any/most IDE controllers, right? As I understand, the Gateworks Avila IDE driver is for IDE directly-connected to the IXP4xx expansion bus. Oct 23 09:34:51 likewise: I dunno. Oct 23 09:35:16 [g2] is supposed to keep our kernel repo up to date for the Loft Oct 23 09:35:17 rwhitby: me neither, at least not for sure yet. Oct 23 09:35:32 I have no way of knowing what the current status is, apart from asking [g2] Oct 23 09:36:03 rwhitby: the avila-ide.c supposedly is not upstream, not for the gateworks board, nor for the loft. I'll ask [g2] when he is up and running. Oct 23 11:40:04 03rwhitby * r4262 10optware/trunk/ (6 files in 3 dirs): coreutils, adduser: Added update-alternatives support for the su binary. Oct 23 11:49:22 03rwhitby * r4263 10optware/trunk/ (4 files in 3 dirs): coreutils, adduser: Fixed update-alternatives support for the su binary. Oct 23 11:51:04 that should allow clamav to install and use adduser if you have coreutils installed. Oct 23 11:51:12 03rwhitby * r4264 10optware/trunk/ (make/clamav.mk sources/clamav/postinst): clamav: Ensure /opt/var/log exists Oct 23 12:02:30 03rwhitby * r4265 10optware/trunk/make/clamav.mk: clamav: added missing bzip2 dependency Oct 23 12:03:02 <[g2]> likewise pong Oct 23 12:06:50 03rwhitby * r4266 10optware/trunk/ (3 files in 2 dirs): clamav: moved log file into a subdirectory so the permissions can be set correctly Oct 23 12:12:14 03rwhitby * r4267 10optware/trunk/ (make/clamav.mk sources/clamav/postinst): clamav: fixed the postinst Oct 23 12:16:45 03rwhitby * r4268 10optware/trunk/ (3 files in 2 dirs): clamav: Move the log file back up to where it was, and fix the real problem (permissions on /opt/var/log) Oct 23 12:17:49 03rwhitby * r4269 10optware/trunk/make/clamav.mk: clamav: Update to latest upstream version. Oct 23 13:01:13 03rwhitby * r4270 10optware/trunk/make/fsg3-bootstrap.mk: fsg3-bootstrap: Updated postinst to chmod created directories correctly Oct 23 13:02:33 03rwhitby * r4271 10optware/trunk/ (3 files in 3 dirs): clamav: Updated postinst to chmod created directories correctly Oct 23 13:02:39 [g2]: ping 2 Oct 23 13:13:36 03rwhitby * r4272 10optware/trunk/Makefile: Makefile: enabled libstdc++ for fsg3 Oct 23 13:31:02 03rwhitby * r4273 10optware/trunk/make/fsg3-bootstrap.mk: fsg3-bootstrap: Enabled inclusion of libgcc_s.so Oct 23 13:36:04 that allows "man" to run :-) Oct 23 13:40:30 03rwhitby * r4274 10optware/trunk/Makefile: fsg3: enabled PERL_PACKAGES - let's see how many build Oct 23 13:47:38 <[g2]> likewise pong Oct 23 13:59:22 <[g2]> likewise pong Oct 23 14:10:37 03rwhitby * r4275 10optware/trunk/ (Makefile make/optware-devel.mk): optware-devel: Added - renamed from unslung-devel Oct 23 14:23:08 [g2]: I am interested to see what you did with the CF interface on the loft. Oct 23 14:23:45 <[g2]> likewise yeah, and ? Oct 23 14:24:04 [g2]: it is directly connected to the ext bus? Oct 23 14:24:10 <[g2]> yup Oct 23 14:24:14 [g2]: this is the avila design Oct 23 14:24:42 Is the avila-ide.c driver available from you? Oct 23 14:25:21 <[g2]> likewise my understanding is that gateworks significantly contributed to the Intel ap note on using the EXP bus to talk to the CF Oct 23 14:25:58 <[g2]> I originally just the avila-ide.c driver forward ported to 2.6 in BE mode Oct 23 14:26:20 <[g2]> however, it's totally busted in le mode due to the IXP bus handling Oct 23 14:26:45 [g2]: we happen to run BE and we are considering adding CF. Oct 23 14:27:02 [g2]: What's the loft list price? :-) Oct 23 14:27:28 <[g2]> likewise you wanna buy one for the BE driver ? Oct 23 14:27:52 well, maybe to prototype the CF, maybe the driver standalone. Oct 23 14:28:12 <[g2]> likewise you'd love my dev kit setup Oct 23 14:28:36 [g2]: i gotta go in a few minutes, but could you forward info to leonw@mailcan.com? Oct 23 14:28:46 <[g2]> sure Oct 23 14:28:52 <[g2]> we'll exchange e-mails Oct 23 14:29:04 <[g2]> likewise you're in the UK right ? Oct 23 14:29:16 in short, we want to add CF support for BE IXP4xx. Any help is welcome, maybe we can arrnage something to help shorten dev time. we're .nl based Oct 23 14:29:26 This is for www.axon.nl Oct 23 14:30:14 <[g2]> likewise ok the BE stuff is done and working. I can give you the numbers on the performance Oct 23 14:30:36 <[g2]> it really depends on what you are trying to accomplish Oct 23 14:30:48 <[g2]> (as do most engineering efforts) :) Oct 23 14:31:02 [g2]: slow massive (2GB) storage is main task Oct 23 14:31:23 [g2]: gotta run now, hope to spend more time here later this week. Oct 23 14:31:24 <[g2]> slow (cheap :) Oct 23 14:31:29 <[g2]> nod Oct 23 14:31:29 :-) yup Oct 23 14:31:32 CU Oct 23 14:31:34 <[g2]> cheers Oct 23 14:31:35 tnx Oct 23 14:31:39 <[g2]> np Oct 23 14:32:01 [g2]: you .uk? Oct 23 14:32:21 <[g2]> barnseen2o nah, US (East coast) Oct 23 14:37:11 03bzhou * r4276 10optware/trunk/ (Makefile make/py-pastewebkit.mk): added and promoted py-pastewebkit Oct 23 16:02:58 Hello everyone Oct 23 16:04:28 03g2-tbillman * r513 10kernel/trunk/patches/2.6.18/ (2 files): Improve RedBoot FIS directory handling. Reduce verbosity. Update Loft defconfig. Oct 23 16:04:41 [g2]: ping (again :-) Oct 23 16:07:39 <[g2]> likewise pong again Oct 23 16:08:45 <[g2]> likewise out for food bbl Oct 23 16:08:53 [g2]: me too Oct 23 18:57:05 wow... Oct 23 18:57:13 wrong channel ;) Oct 23 19:28:00 03rwhitby * r4277 10optware/trunk/Makefile: Makefile: Promoted crosstool-native and optware-devel for fsg3 Oct 23 19:32:49 dwery: ping Oct 23 19:33:57 likewise: pong Oct 23 19:34:20 dwery: You are working on libata support for the ixp4xx? Oct 23 19:34:30 yes Oct 23 19:34:45 CF support on Loft to be exact Oct 23 19:35:02 dwery: Is that for PCI IDE controller chip, or CF on the expansion bus of the ixp4xx? Oct 23 19:35:09 s/chip/chips Oct 23 19:35:24 CF. PCI ide is alreay supported by libata Oct 23 19:36:03 dwery: Do you know of avila-ide.c? Oct 23 19:36:16 yes. Oct 23 19:36:23 old ide style iirc Oct 23 19:36:58 dwery: OK, so you are working on new-style libata support for the Avila design CF interface (as on the loft)? Oct 23 19:37:02 yes Oct 23 19:37:28 dwery: we are interested in that as well, is this open-source, can we sponsor you? :-) Oct 23 19:37:38 of course! :) Oct 23 19:37:48 Yes to both? Oct 23 19:37:56 yes. Oct 23 19:38:00 <[g2]> :) Oct 23 19:38:11 I love doing open source Oct 23 19:38:16 [g2]: :-) Oct 23 19:38:18 and get paid as well :D Oct 23 19:40:24 [g2] and dwery: Maybe I can arrange to sponsor you both in open-source fashion (i.e. not much :-), [g2] for a bit of hardware design (and a old skool driver maybe?), dwery for the open-source libata driver. Oct 23 19:41:19 eheheh :) Oct 23 19:41:28 I'd be happy. Oct 23 19:41:40 I'm on #nslu2 Oct 23 19:50:44 <[g2]> likewise are you using anything for JTAG debugging on your boards ? Oct 23 20:41:20 NAiL: my crosstool-native build for fsg3 segfaults when running gcc Oct 23 20:44:20 anything more specific? Oct 23 20:46:08 03rwhitby * r4278 10optware/trunk/ (3 files in 2 dirs): crosstool-native: Updated fsg3 config Oct 23 20:46:41 NAiL: I've now copied the different softfloat options over to try again Oct 23 20:47:06 ok Oct 23 20:48:12 (copied from the cross-built crosstool) Oct 23 21:17:21 rwhitby: does perl & perl packages run on fsg3? Oct 23 21:20:57 eno: yes. well perl at least executes perl -v Oct 23 21:22:34 cool Oct 23 21:23:13 I didn't need to do anything, it just worked. Oct 23 21:24:18 gah, my home fuse box just popped Oct 23 21:25:11 barnseenio: s/your pc/some slug/ Oct 23 21:25:55 i have no buffer im not sure what your sedding! Oct 24 02:01:10 ping g2 Oct 24 02:03:12 oops, I mean ping [g2] Oct 24 02:04:23 hey dumfrac Oct 24 02:05:04 hey rwhitby Oct 24 02:05:24 I have a proposed solution for testing the endianess of the FIS table Oct 24 02:06:17 couldn't we do something like if swab32(buf[i].flash_base) == (offset + 256*i) in redboot.c ? Oct 24 02:07:06 dumfrac: I haven't looked at the swapping issue close enough to comment. Best to reply to the email thread (cause Marc is not on IRC) Oct 24 02:07:44 will do - I was hoping that [g2] would be around to discuss it first (just in case I was off my rocker) Oct 24 02:08:48 my ultra slow (but faster than the slug) build machine is currently building a new kernel to test [g2]'s new patch on the slug **** ENDING LOGGING AT Tue Oct 24 02:59:57 2006