[12:48] <mjg59> ispiked: No, hangs on suspend
[01:44] <zul> lheylo
[03:19] <BenC> fabbione: sparc64 will be a go with -4.4
[03:20] <BenC> atleast it will build
[03:20] <BenC> getting ready to boot the e3k with it
[03:29] <zul> cool
[05:39] <fabbione> BenC: cool
[08:13] <Earthpig> anyone about?
[09:51] <Earthpig> ANy kernel dev types awake?
[10:00] <fabbione> hi
[10:03] <Earthpig> Is that a hi, I'm awake? :)
[11:04] <CataEnry> hi all
[12:06] <CataEnry> bye all
[01:20] <jbailey> Anyone know off hand if Ben has a new ppc64 kernel to test with rtc fixes?
[01:29] <CataEnry> hi all
[02:33] <zul> heylo
[02:36] <jbailey> DoubleChuck!
[02:42] <zul> hehe
[02:44] <Earthpig> Anyone know the best way to get a driver added to the kernel images?
[02:44] <Yagisan> submit bug + patch ?
[02:45] <Earthpig> Is that preferable to packaging it as a -module thing?
[02:46] <Earthpig> Basically, options are: 1) patch straight into the kernel source; 2) package as a separate source .deb that builds a -module.deb which make-kpkg can use.
[02:47] <Earthpig> There are additional utilities associated with the driver, so I can see some merit in a single source .deb that spits out multiple, but my understanding is that users still have to build the kernel module themselves, even with the -module.deb option.
[02:49] <Yagisan> well, if you do bug + patch, and the kernel devs are happy with it, then the users won't need to build it themselves
[02:49] <jbailey> Earthpig: At this point we don't have a good way of making sure that modules are automatically rebuilt when kernel 
[02:49] <jbailey> ABIs change.
[02:49] <jbailey> When that happens, packaging separate will work.
[02:49] <Earthpig> True. The code looks like shit though ;) WOuldn't suprise me though if I was told to run it through indent first.
[02:49] <jbailey> But in most cases the truly correct answer is to get it in upstream.
[02:50] <Earthpig> Well, the driver is in -mm but not yet vanilla.
[02:50] <Earthpig> (arcmsr, for reference)
[02:52] <Earthpig> When you say upstream, do you mean vanilla or debian? 
[02:52] <Yagisan> vanilla
[02:52] <Earthpig> Crap. :) If it's in -mm but not yet vanilla, I'm guessing I can't push any harder.,
[02:52] <Yagisan> Ubuntu doesn't use Debian kernels TTBOMK
[02:53] <Earthpig> I've noticed the patch list is different.
[02:56] <Earthpig> Hrm. I only just discovered that protected methods are also package-wide.
[02:56] <Earthpig> oops wrong channel
[03:02] <jbailey> Earthpig: If it's in -mm, you might be able to ask Ben to sync it in.
[03:06] <Earthpig> does ben have an address?
[03:06] <Earthpig> (not registered... can't privmsg)
[03:10] <jbailey> Private message is almost always the wrong solution anyway.  There's no reason for the discussion to be private.
[03:10] <Earthpig> agreed.
[03:11] <jbailey> If BenC's not around atm, I'd suggest emailing the ubuntu-kernel list, or filing a bug in bugzilla with the patchsets from -mm that you want.
[03:11] <jbailey> That way a discussion can happen around the idea.
[03:11] <BenC> what's up?
[03:12] <jbailey> (Or we could remember that Ben doesn't nick highlight on his name.. *g*)
[03:12] <BenC> yes, Ben gives me too many false positives :)
[03:12] <jbailey> Yeah, I still do it for Jeff.  It helps me catch people talking about me. =)
[03:12] <BenC> lol
[03:13] <jbailey> I want per-channel nick highlights.
[03:13] <BenC> Earthpig: what driver?
[03:17] <Earthpig> ben: arcmsr
[03:18] <BenC> sata driver?
[03:18] <fabbione> hey BenC
[03:18] <Earthpig> RAID controller.
[03:18] <BenC> fabbione: hey, sparc64-smp fails to boot, but I'm tracking it down
[03:18] <Earthpig> PCIe or PCI-X host-side, SATA-II on the disk side.
[03:18] <fabbione> BenC: ah
[03:18] <fabbione> BenC: what about UP?
[03:18] <BenC> didn't try UP
[03:18] <fabbione> ok
[03:19] <BenC> the failure is in alloc_percpu() (kmalloc_node fails), so I suspect UP works
[03:20] <Earthpig> -mm has been tracking the driver from Areca; 2.6.15-rc1 contains the current driver; broken-out/areca-raid-linux-scsi-driver.patch should apply cleanly against any 2.6.x tree.
[03:20] <BenC> Earthpig: send me an email with the brokenout patch from -mm and I'll look at it
[03:21] <BenC> 2.6.15-rc1 contains the driver?
[03:21] <BenC> then we're all set, since dapper has 2.6.15-rc1
[03:21] <Earthpig> Sorry. 2.6.15-rc1-mm2
[03:21] <BenC> ah
[03:21] <Earthpig> I meant the -mm patch against it :)
[03:21] <BenC> ok, can you get the driver patch broken out and email me that?
[03:21] <Earthpig> http://kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.15-rc1/2.6.15-rc1-mm2/broken-out/areca-raid-linux-scsi-driver.patch
[03:21] <Earthpig> That's the broken out patch.
[03:21] <BenC> ok, thanks
[03:21] <Earthpig> You might want to read the comments at the top.
[03:23] <Earthpig> I can compile it out of tree easily enough, but it is inconvenient. :) Especially since / and /boot are on it.
[03:24] <Earthpig> Since it's a module and self-contained, the risk that it'll b0rk other stuff should be quite low. Dunno how strict your criteria are for inclusion.
[03:26] <BenC> wow, this code is commented to the hilt
[03:26] <Earthpig> good or bad? :D
[03:26] <BenC> like 20 lines per #define in the header :)
[03:27] <BenC> depends on your opinion of proper commenting I guess
[03:27] <Earthpig> Well, good that there are comments. Bad if they all say it's shit. :)
[03:27] <BenC> Earthpig: any crashes with it?
[03:28] <Earthpig> Not sure yet. Will know by the end of the week; awaiting a new motherboard.
[03:28] <Earthpig> It doesn't crash on load :D
[03:28] <BenC> I'll mark it for inclusion...are the tools required to use it?
[03:28] <Earthpig> Nope; it has BIOS configuration.
[03:28] <BenC> ok
[03:28] <Earthpig> Tools are handy though; supports online expansion and raid level migration.
[03:29] <Earthpig> (and stripe-size migration)
[03:29] <BenC> you have to agree to be my bug-bitch if there are any problems reported on it though :)
[03:29] <Earthpig> If it means it normally shouldJustWork[tm] , I'm fine with that.
[03:30] <BenC> it should, just nice to have an extra tester for external-drivers like this
[03:30] <Earthpig> Yeah, no problem.
[03:31] <BenC> wow, nice that the driver is maintained by areca directly
[03:31] <BenC> should be able to get support from Erich easily enough
[03:32] <Earthpig> Yeah. They're a pretty friendly bunch.
[03:32] <Earthpig> Hell, they did a press release when it make it into -mm.
[03:32] <BenC> hehe
[03:32] <BenC> maybe they'll do one when we put it into ubuntu? :)
[03:33] <Earthpig> If you let them know, that wouldn't surprise me.
[03:33] <Earthpig> Support have also been good; they took my mobo issue up with Asus directly.
[03:38] <jbailey> Earthpig: Do you have contact information for them? =)
[03:39] <jbailey> I can take care of that arrangement. =)
[03:39] <Earthpig> For Areca?
[03:39] <Earthpig> Asus?
[03:40] <jbailey> Whichever company did the press relesae.  Was that Areca?
[03:40] <Earthpig> Yeah.
[03:40] <jbailey> I can't see it ebing asus.
[03:40] <jbailey> But we have a hardware certification program, and I'd love to introduce them to it.
[03:41] <Earthpig> http://www.areca.com.tw/news/html/n_press%20release.htm is the existing press releases.
[03:42] <Earthpig> March/April are the ones I'm thinking of.
[03:42] <BenC> erich@areca.com.tw is the guy supporting the driver
[03:43] <Earthpig> kevin34@areca.com.tw is the support guy i've been dealing with.
[03:51] <dilinger> i keep hearing really good things about areca
[03:51] <dilinger> too bad their stuff is so much more expensive than everything else
[03:51] <Earthpig> It's cheaper that 3ware.
[03:51] <dilinger> nah
[03:52] <dilinger> 8port sata controllers are like $200 more than 3ware
[03:52] <dilinger> sorry, $100 more.  i bought a refurb'd controller, which is where the other $100 came from ;)
[03:53] <Earthpig> Native SATA-II?
[03:53] <Earthpig> 3ware for a long time were using PATA internally and bridge-chips.
[03:54] <Earthpig> I can also only see PCI or PCI-X. PCI gets saturated easily once you go to any 2-disk setup. And PCI-X adds more than $100 to the cost of the motherboard.
[03:55] <dilinger> i doubt it; the 9500s-8 is what i ended up getting
[03:55] <Earthpig> Do you get better than 90MB/sec from it?
[03:56] <dilinger> it hasn't arrived yet
[03:56] <Earthpig> Heh ok. :)
[03:56] <Earthpig> At least your drivers are in already. ;)
[04:04] <Earthpig> benc: want my email address for later?
[04:04] <BenC> email me, or I'll lose it
[04:04] <BenC> bcollins@ubuntu.com
[04:05] <BenC> my inbox is the only static data store I have
[04:05] <Earthpig> ok will do
[04:05] <BenC> thanks
[04:05] <BenC> if I could turn emails into stickies, I'd be happy
[04:06] <chmj> neccesity is the mother of invention 
[04:06] <Earthpig> With BeOS, you could. :)
[04:07] <jbailey> BenC: You need on of those SLP label printers. =)
[04:09] <BenC> hehe
[04:09] <BenC> anyone know if the sky2 driver is stable now?
[04:09] <BenC> if I can avoid sk98lin for dapper I will
[04:26] <zul> bah...how can you not have a bootdisk for a server?
[04:26] <fabbione> BenC: .15 on ppc seems pretty stable
[04:27] <fabbione> have been running it for a while now
[04:27] <fabbione> except suspend/resume that choacks on some large penises..
[04:27] <fabbione> but that's a known upstream problem
[04:27] <fabbione> benh is working on it already
[04:40] <BenC> yeah, it's been running for over a week on my G4
[04:41] <BenC> glad it's not an isolated incident :)
[04:44] <fabbione> ehhehe
[04:44] <fabbione> unfortunatly the new model is barely supported
[04:44] <fabbione> i need to bitch Olof 
[04:51] <mjg59> BenC: Uhm.
[04:51] <mjg59> BenC: Why did we disable apic by default on uniprocessor machines?
[04:51] <mjg59> That just breaks a different set of machines
[04:51] <mjg59> There's hardware that needs apic support now
[04:51] <BenC> mjg59: it was a patch I pulled from fedora, but I hadn't decided on whether to leave it on or not
[04:52] <mjg59> Yeah. I think it's the wrong solution.
[04:52] <BenC> easy enough to disable it, it's a config option
[04:53] <BenC> will be disabled in my next push
[04:54] <mjg59> BenC: For instance, the bug you pulled the patch from is about hardware that requires the apic to be on...
[04:54] <mjg59> Ok, cool
[04:55] <BenC> I didn't get it from a bug, just went through all the fedora patches, and that one looked atleast interesting to have in
[04:55] <mjg59> Ah, ok
[04:55] <mjg59> The bugzilla entry referenced, then
[05:00] <mjg59> BenC: Any chance we can get the driver from http://rtl8180-sa2400.sourceforge.net/ ?
[05:01] <mjg59> Moderately common 802.11 chipset
[05:01] <BenC> sure, I'll look into it
[05:01] <mjg59> (I'm just going through the list of wireless drivers)
[05:04] <infinity> BenC : Say, any idea why my laptop claims to not have a CD drive when I boot with 2.6.15?
[05:05] <BenC> infinity: my G4 does the same thing
[05:05] <BenC> does it show in dmesg, and just not work in the desktop?
[05:06] <BenC> mines actually there, just gnome doesn't use it
[05:06] <mjg59> BenC: Also, there seems to be a driver for the softmac prism54 chipsets appearing
[05:06] <infinity> BenC : No, mine doesn't even show in dmesg.  Just plain ain't there.
[05:06] <BenC> mjg59: yeah, I have a softmac patch, which is also needed for bcm430x
[05:06] <mjg59> Ok
[05:06] <mjg59> The stuff from http://jbnote.free.fr/prism54usb/ ?
[05:06] <BenC> infinity: ide?
[05:06] <infinity> Yep.
[05:06] <mjg59> (Most. Misleading. Name. Ever)
[05:06] <BenC> infinty: drives show up?
[05:07] <BenC> fucking {arch}
[05:07] <infinity> Erm, SATA, not IDE.
[05:08] <zul> mjg59: send me an email and ill do it tonight
[05:08] <infinity> And yes, hard drive shows up, CD not.
[05:08] <infinity> The CD drive ALMOST shows up.
[05:08] <BenC> infinity: "almost"?
[05:08] <infinity> [17179573.012000]  ata2: dev 0 cfg 49:0f00 82:0000 83:0000 84:0000 85:0000 86:0000 87:0000 88:0407
[05:08] <infinity> [17179573.012000]  ata2: dev 0 ATAPI, max UDMA/33
[05:08] <infinity> But nothing after that about the drive (like vendor ID, etc)
[05:09] <BenC> tried modprobe sr_mod?
[05:10] <infinity> No dice.
[05:11] <infinity> Oh well, I have other stuff to do, I'll have to whine to you about it later.
[05:12] <mjg59> I should really try to get hold of one of the softmac cards for testing
[05:12] <Earthpig> benc: any idea when the next push is going to be?
[05:14] <mjg59> BenC: Also, last time I checked we didn't seem to have the zd1211 firmware
[05:14] <mjg59> Not sure what we want to do about that, or what license it's under
[05:15] <BenC> is it 100% useless without the firmware?
[05:15] <mjg59> BenC: Yeah
[05:15] <mjg59> By "we" I mean Ubuntu - it's in Debian
[05:15] <BenC> interesting that we even have the driver in the first place then
[05:15] <mjg59> As a package, rather than in the kernel
[05:16] <BenC> well, if it's similar to the current firmware distribution terms we can just include it like we do now
[05:17] <BenC> debian package doesn't even have a license or anything in the .orig.tar.gz
[05:18] <mjg59> It's in the upstream driver release, as a bunch of hex in a GPLed header file
[05:18] <mjg59> BenC: Ok, we don't seem to have prism54 support for softmac PCI devices
[05:18] <infinity> My favourite.
[05:18] <mjg59> But there's a driver that ought to (at least partially) work now
[05:18] <BenC> GPL'd hex, lovely
[05:19] <BenC> totally contradicts it's own license
[05:19] <mjg59> infinity: Oh, I still need to do that vga16fb patch, don't I?
[05:19] <mjg59> BenC: Well, they're the copyright holders, so.
[05:19] <mjg59> It ought to be distributable
[05:19] <infinity> mjg59 : I have something on my hard drive here.
[05:20] <mjg59> infinity: Oh, cool. Just changing the ysize and timings?
[05:20] <infinity> mjg59 : I stole the 640x400 timings from the X vga drive.  If you happen to know that those aren't "correct", let me know.
[05:20] <infinity> s/drive/driver/
[05:20] <infinity> I need to do some testing here before I push the patch to Ben.
[05:20] <BenC> mjg59: source code == "prefered form of the work for making modifications"
[05:20] <infinity> And I also need to make the same changes to syslinux.
[05:21] <mjg59> infinity: As far as I know, they're good
[05:21] <infinity> BenC : It's hazy if you're the copyright holder.  Much less hazy if you are obfuscating someone else's code.
[05:22] <infinity> mjg59 : If you have an urge to dig through the asm in syslinux and find the right spot to make the change, that'd be cool.
[05:22] <infinity> mjg59 : Though, I can find it easily enough.
[05:22] <mjg59> infinity: Hahahahaahahahahaha (dies)
[05:23] <infinity> mjg59 : What I'd really dig is for someone (*stare*) to make usplash a proper blocking daemon, so we can stop this silly "usplash & ; sleep 1" business.
[05:23] <mjg59> Oh, yeah. What's the sleep for, again?
[05:23] <mjg59> Should really just add a daemon() call
[05:23] <infinity> mjg59 : Because usplash takes a while before it's ready to accept the first write commant.
[05:23] <infinity> command, too.
[05:23] <mjg59> Oh, yes
[05:24] <BenC> ok, zd1211 will be in -4.4 too
[05:24] <infinity> mjg59 : If it blocked until it was ready, then backgrounded itself, all would be well.
[05:24] <BenC> zd1211-firmware that is
[05:24] <mjg59> Just add daemon() before it starts the idle loop
[05:24] <mjg59> BenC: Rock
[05:25] <infinity> mjg59 : Oh.  And a curiosity.  Did yuo have a good reason for using tty8, or did it just seem the thing to do?
[05:25] <mjg59> infinity: I can't remember
[05:25] <mjg59> I'm not even sure if I did
[05:25] <mjg59> That might have been someone else's diff
[05:25] <mjg59> I thought I just grabbed the current console
[05:25] <infinity> mjg59 : It cauases a certain amount of scary if, say, your disk check fails, you get dropped to tty8, and you can't type.
[05:26] <infinity> mjg59 : If that's someone else's fault, and you have no idea why, I may unfault it back to using the current VC, cause it seems a bit... Wrong.
[05:26] <mjg59> Yeah. 
[05:26] <mjg59> infinity: Ok, syslinux just does                 mov ax,0012h            ; Set mode = 640x480 VGA 16 colors int 10h
[05:26] <mjg59> (nngh whitespace breakage)
[05:26] <infinity> Yeah, but in which file? :)
[05:26] <mjg59> graphics.inc
[05:27] <infinity> I grepped for 10h, found a few too many, and put it on my "look closer in a few days" list.
[05:27] <mjg59> In the "vgasetmode" function
[05:28] <infinity> Rock.  Lifesaver, you are.
[05:28] <mjg59> infinity: There doesn't seem to be an int 10 mode for 640x400
[05:28] <infinity> I'll fix both this week then, and if we get 2.6.15 as the default kernel on flight-2, we can have end-to-end testing of the new mode.
[05:28] <mjg59> 640x350 is 0x10
[05:28] <infinity> mjg59 : I'm pretty sure I found one when I was googling before.
[05:28] <infinity> Maybe not.
[05:29] <mjg59> Oh, I want to do something really nasty to the framebuffer
[05:29] <mjg59> I want a sysfs parameter that, when set, discards all framebuffer reads or writes silently
[05:29] <mjg59> So that way we can avoid anything touching the framebuffer until we've POSTed and so on
[05:30] <infinity> Perverse.  I like it.
[05:30] <mjg59> It's trivial
[05:33] <infinity> You'd do that at the fbcon level, I assume, not touch each fb driver?
[05:33] <mjg59> Yeah
[05:33] <infinity> Well, put me on the "yes please" bandwagon for that.
[05:34] <infinity> That may fix my "breezy doesn't resume from hibernate" issue on my laptop.
[06:45] <Mithrandir> mjg59: you're evil.  But in a good way.
[06:46] <mjg59> Mithrandir: ?
[06:46] <Mithrandir> mjg59: the discard writes hack.
[06:47] <mjg59> Ah
[07:06] <BenC> zul: ping
[07:21] <zul> BenC: yo
[07:22] <zul> i mean pong
[07:22] <BenC> zul: the i8042 patch you sent me, is there any reason that it should not be submitted upstream?
[07:22] <zul> no there shouldnt be
[07:26] <BenC> ok, thanks
[07:27] <zul> nopr
[07:27] <zul> heh...my kotd almost works
[07:34] <zul> http://zulinux.homelinux.net/kotd <-- daily kernel builds automated (its using an old git though)
[07:42] <BenC> are they apt'able?
[07:43] <zul> not yet..
[07:43] <zul> its on my personal box though
[07:44] <zul> and its only x86
[07:54] <zul> blah
[08:03] <zul> BenC: its a start atleast
[08:04] <BenC> yeah, it'll be needed soon
[08:06] <jbailey> BenC: I could probably fire off a KOTD for ppc/ppc64 here if you're not setup for it.
[08:06] <BenC> not setup for any of the automated stuff yet
[08:06] <BenC> zul: how long do your builds take?
[08:07] <zul> couple of hours
[08:07] <jbailey> Using ccache?
[08:07] <zul> not yet..
[08:08] <zul> it was just a test run hasnt been optimized yet
[08:08] <jbailey> I have my package builder hooked to always use ccache.
[08:08] <zul> jbailey: i wrote a script that does it for me
[08:08] <jbailey> alias db='linux32 debuild -e PATH=~/bin:/usr/local/bin:/usr/bin:/bin:/usr/bin/X11:/usr/games'
[08:09] <zul> is there something on the wiki about ccache?
[08:09] <jbailey> No ide.a
[08:09] <jbailey> But apt-get install ccache
[08:09] <jbailey> mkdir ~/bin
[08:09] <jbailey> cd ~/bin
[08:09] <jbailey> ln -s /usr/bin/ccache gcc
[08:10] <jbailey> (repeat for gcc cpp g++ gcc-3.4 cpp-3.4 g++-3.4, etc...)
[08:10] <jbailey> Then anytime those are called, ccache gets used.
[08:10] <Mithrandir> echo 'export PATH=/usr/lib/ccache:$PATH' >> ~/.${SHELL}rc ; exec $SHELL
[08:10] <jbailey> Ooo.
[08:10] <jbailey> I didn't know about /usr/lib/ccache
[08:10] <Mithrandir> it's very convenient.
[08:11] <jbailey> Yup
[08:11] <jbailey> alias db='linux32 debuild -e PATH=/usr/lib/ccache:/usr/local/bin:/usr/bin:/bin:/usr/bin/X11:/usr/games'
[08:11] <jbailey> Then run 'db' when you want a ccache enabled build.
[08:12] <zul> cool...
[08:12] <jbailey> Note that debuild cleans the path if you're not otherwise careful.
[08:12] <jbailey> The linux32 applies to me since I'm on ppc64.
[08:12] <jbailey> You may want to take it out for your arch.
[08:16] <Earthpig> ccache doesn't go via alternatives?
[08:17] <Earthpig> oops... you're inserting ccache on the path somewhere else so that it calls alternatives.
[08:19] <jbailey> Earthpig: No, because you need to be able to trivially *not* use gcc for validation.
[08:30] <jbailey> BenC: Thinking of which, do you have a ppc64 image for me to test?  Otherwise I will start my afternoon's worth of work.
[09:17] <BenC> jbailey: no images yet
[09:19] <BenC> jbailey: CCACHE_DISABLE=y debuild, btw
[10:01] <jbailey> BenC: Ah, handy.
[10:17] <makx> infinity: ping