[12:13] <dilinger> BenC: libata supports smart ioctls?  as of when?
[12:14] <dilinger> also, has anyone done any reverse engineering of win32 drivers?  what do people recommend for disassembly (and if something exists, decompilation to something resembling C)
[12:28] <BenC> dilinger: libata passthrough patch went in during last month of breezy
[12:33] <infinity> BenC : I'll fiddle with initramfs and the above-mentioned libata modprobe argument, since I'm one of the people who whined about his disappearing CDROM.
[12:34] <infinity> (The resource conflict with ide-generic thing is a red herring, BTW, 2.6.12 gave the same output.  It's the 5 or 6 lines BEFORE that that went missing that are important)
[12:35] <infinity> Of course, given our focus as a desktop OS, it may be in our best interest to just hack our kernels to always turn that flag on in the driver, and not mess with it in initramfs.
[12:36] <infinity> (We can revisit that the day someone complains that ATAPI is always on)
[12:56] <dilinger> BenC: oh
[12:57] <dilinger> BenC: you mean went into breezy or upstream?
[12:57] <dilinger> but i haven't tested the amd64 breezy machine i just installed
[01:05] <BenC> infinity: that's what I'm thinking too
[01:06] <BenC> dilinger: breezy, it's in 2.6.15
[01:11] <infinity> BenC : Well, I'll do some quick reboot testing and verify that this really does bring my drive back from the dead, and if so, we can decide which direction to go with it.
[01:11] <infinity> I assume it'll work, though.
[01:11] <BenC> infinity: worked for a guy on kernel-team list
[01:12] <BenC> I'm trying to catch jgarzik to see how stable the sata atapi code is right now
[01:12] <infinity> Oh, you've already pinged him with the fix?  I only saw his message with dmesg output, etc... Nothing else.
[01:13] <infinity> The only danger seems to be in having libata drive really old PATA ATAPI devices over an SATA->PATA bridge, which seems a pretty unlikely combination to me.
[01:23] <BenC> yeah, I redid my email client, and got moderated to the list
[01:23] <BenC> but he got my email and replied
[01:26] <infinity> Ahh, cool.
[01:29] <BenC> jgarzik says sata_promise and sata_sx4 are known to not work with atapi
[01:29] <BenC> the drivers that is
[01:30] <infinity> Well, that sort of leaves us at an impasse.
[01:30] <BenC> I'm going to blacklist those two drivers and set atapi_enabled=1 by default
[01:30] <infinity> Those two drivers are kinda popular. :/
[01:30] <infinity> (not as popular as piix or via but still)
[01:49] <mjg59> infinity: I've found various drivers that probably ought to be in l-r-m
[01:50] <infinity> You mean stuff that should be removed from the kernel, or stuff we don't yet ship?
[01:50] <infinity> (speaking of the latter, I need to make ltmodem build again)
[01:50] <mjg59> infinity: Stuff we don't yet ship
[01:51] <mjg59> Modem drivers (hahaha)
[01:51] <infinity> Mmkay.
[01:51] <infinity> More ltmodem-like crap?
[01:51] <mjg59> Intel ones for a start
[01:51] <infinity> Why can't we just finance a worldwide effort to get broadband to everyone's house?
[01:51] <mjg59> There's one for the 536 and one for the 537
[01:52] <mjg59> The 537 one looks like it needs light hacking to get it to work on recent 2.6, the 536 one looks ok
[01:52] <mjg59> But they're both 2.6 compatible
[01:52] <infinity> You do the light hacking required, and I'll happily include them.
[01:52] <mjg59> I'm also looking at the old (GPLed) Conexant drivers
[01:52] <mjg59> They involve rather more pain
[01:52] <infinity> I want, bsically, upstream tarball, + your patches in a diff.
[01:53] <mjg59> Oh, when I say GPLed, I mean "GPL + closed shit"
[01:53] <infinity> The conexant stuff looked scary.
[01:53] <mjg59> The conexant stuff scares the shit out of me
[01:53] <mjg59> I'm trying to forward port the last release of it to 2.6
[01:53] <mjg59> Sorry, the last free release
[01:53] <mjg59> It still looks scary, but, well
[01:54] <infinity> We don't even attempt to support modems/ppp in d-i as a network method, right?
[01:54] <mjg59> Nope
[01:54] <infinity> So these don't need to hit a udeb, just the main LRM deb...
[01:54] <infinity> Good.
[01:56] <infinity> Well, if you make the intel stuff work and I make the ltmodem stuff work, we've suddenly got coverage of a crapload of softmodems.  The conexant stuff would be a nice bonus, though.
[01:57] <infinity> Personal anecdotal statistical analysis, and some really bad math would lead me to eblieve that 8 out of 5 people have an LT-based softmodem somewhere in their house.
[01:58] <mjg59> Yeah
[01:58] <mjg59> I've got at least one
[01:58] <mjg59> Most of them are slmodem-compatible now, though
[02:01] <infinity> Of course, I haven't actually used a modem in Linux since, like, 1995.
[02:02] <infinity> Perhaps this makes me an ideal candidate to test this stuff and see if it's foolproof.
[02:03] <mjg59> Modern hardware is almost all AC97. Of that, pretty much everything uses a smartlink-compatible codec except for ones which use Conexants
[02:03] <mjg59> Oh, and Apples, which use some horrid USB conexant thing
[02:03] <infinity> Well, enough people have complained about ltmodem going away that I assume someone still needs it.
[02:04] <infinity> Anyhow, a fork that builds on moden kernels has been pointed out to me, so I'll re-enable it and close the bugs, and not care much about the particulars. :)
[02:06] <infinity> Whatever you want in LRM, though, just get me upstreamish source (with an upstream URL of sorts), plus your patches to make it function, and I'll integrate it.
[02:06] <infinity> And make sure we can distribute it, of course.
[02:20] <mjg59> Oh, sure
[02:42] <infinity> BenC : Kay, just tested with "break=premount && modprobe libata atapi_enabled=1 && exit", and the resulting boot had my CD/DVD drive too.  Yay.
[02:44] <infinity> Of course, I still don't have a /dev/dvd, but that's probably something I can blame on Keybuk.
[02:52] <infinity> Hrm.
[02:53] <infinity> WEEKEND OFFTOPIC QUESTION:
[02:53] <infinity> Does anyone know how to force my DVD drive (or software player) to pretend to be in a specific region?
[02:54] <infinity> (Rather than doing auto-region, region-guessing, which breaks on Star Wars DVDs which try multiple regions and bomb out if more than 1 is found to be "valid")
[02:59] <mjg59> infinity: Your player should have an option to force the region
[02:59] <mjg59> Then it can just lie to the DVD
[02:59] <infinity> Don't see one in mplayer's manpage, and xine's doesn't appear to actually do anything.  Yay.
[02:59] <mjg59> Hm. Getting the hcfusbmodem driver working ought to be practical
[02:59] <mjg59> Which ought to mean working Apple modems. Yay.
[03:00] <infinity> Oo, more stuff for powerpc LRM users?... Fun.
[03:00] <mjg59> The hacking in most of it is fairly easy, but it'll need a bit of reconstructive work in the serial layer
[03:00] <mjg59> And then, with luck, that'll be applicable to the hsf code as well
[03:00] <mjg59> Only real problem is that we'll be running with older Conexant code
[03:01] <mjg59> Oh, and all these drivers want to spew huge gobs of stuff across /
[03:01] <mjg59> Need to look into minimising that
[03:08] <infinity> Ahh, well, hard-setting the region on the drive helped at any rate.
[03:35] <BenC> infinity: cool
[09:26] <infinity> BenC : Around?
[11:55] <AcidPils1> good morning
[11:56] <AcidPils1> is there a reason for not having the acx module in the 2.6.15-6-k7 kernel?
[01:36] <CataEnry> hi *
[01:36] <AcidPils1> hi
[02:07] <CataEnry> bye
[03:12] <siretart> AcidPils1: I'd say, file a bug in bugzilla about this
[03:35] <CataEnry> hi all
[04:54] <CataEnry> bye
[05:42] <mjg59> BenC: If we're switching on the PATA libata stuff, should we switch on ata_piix's PATA support?
[05:43] <BenC> I'll check it, it may be on from the merge I did with libata-dev.git
[05:44] <BenC> infinity: pong
[05:47] <mjg59> BenC: It's a single #define in ata_piix.c, IIRC
[05:47] <BenC> yeah, I see it
[05:47] <AcidPils1> ah, BenC siretart told me to ask you about the missing acx module in the 1.6.15 kernel
[05:47] <AcidPils1> bug or feature?
[05:47] <BenC> I'm not really enabling PATA except where there are PATA drivers
[05:47] <AcidPils1> 1.6.15? 2.6.15 i mean ;)
[05:47] <BenC> acid: what is the acx module?
[05:48] <AcidPils1> drivers for acx100/acx111 wlan cards
[05:49] <AcidPils1> in 2.6.12 they were included, but in 2.6.15 they are not
[05:50] <BenC> acidpils: uh, it is enabled
[05:51] <BenC> check the config, CONFIG_ACX100=m
[05:52] <BenC> mjg59: let me talk to jgarzik about how safe PATA is in that driver (atiix and one other, are the only ones that need the define to enable PATA)
[05:52] <BenC> everything else is just on
[05:54] <mjg59> BenC: If we could get that and the SATA hotswap patches, then I can whip up support for swapping bay devices on laptops
[05:54] <mjg59> It's a small amount of acpi-integration code
[05:55] <BenC> what exactly is PATA anyway (how's it different from SATA)?
[05:55] <mjg59> Parallel ATA
[05:55] <mjg59> Or legacy stuff
[05:55] <mjg59> The ones with 40 or 80-wire cables
[05:56] <BenC> ah
[05:58] <Yagisan> the "though cables must be shorter then ~80cm or it won't work" version
[05:59] <AcidPils1> BenC: i must have been blind or something like that 
[05:59] <AcidPils1> *reboot and try*
[06:00] <siretart> oh nice.. my r50e is Oopsing on power down.. grml..
[06:02] <BenC> 80cm? Who runs > 80cm cables to their drives?
[06:02] <BenC> or did you mean 8cm? :)
[06:02] <mjg59> 18"
[06:02] <mjg59> So 50cm
[06:02] <mjg59> On x86 boards, it's quite common
[06:02] <BenC> ok, 18" makes more sense
[06:03] <Yagisan> BenC: I have a full tower case here, and due to design error, the PATA cable and drives are on opposite sides of the case
[06:05] <Yagisan> BenC: funnily enough there are a lot of pcs with pata cables up to 1m long in them (and I wonder how they still work)
[06:11] <Yagisan> any chance of the swap prefetch patch from http://ck.kolivas.org/patches/swap-prefetch/mm-swap_prefetch-18.patch ?
[06:12] <BenC> I'll let that stuff simmer in -mm first
[06:13] <Yagisan> OK, couldn't hurt asking. Thanks BenC
[06:15] <BenC> np
[06:15] <Yagisan> night all
[06:19] <CataEnry> hi :)
[06:50] <mjg59> BenC: Current git doesn't build
[06:50] <mjg59> drivers/scsi/libata-scsi.c: At top level:
[06:50] <mjg59> drivers/scsi/libata-scsi.c:2194: error: syntax error before return
[06:50] <BenC> yeah, just pushed fixes for that
[06:50] <BenC> one typo for me, and another for jgarzik, both fixed
[06:56] <mjg59> Ok, pulling
[06:57] <AcidPils1> BenC: am i stupid? http://paste.ubuntulinux.nl/5378
[07:07] <siretart> BenC: I checked the build logs, and AcidPils seems to have a point somewhere. Despite of being enabled in the kernel config, it seems that the kernel module acx_pci.ko is not built any longer.
[07:08] <siretart> so either it has moved to another kernel module, or it vanished somehow..
[07:12] <mjg59> drivers/scsi/sata_sil.c:339: error: ATA_FLAG_NOINTR undeclared (first use in this function)
[07:12] <mjg59> drivers/scsi/sata_sil.c:339: error: (Each undeclared identifier is reported only once
[07:13] <mjg59> drivers/scsi/sata_sil.c:339: error: for each function it appears in.)
[07:13] <mjg59> Hmm.
[07:21] <mjg59> BenC: ZD1211 seems broken, too
[07:22] <mjg59>   DEVLIST drivers/usb/net/zd1211/zddevlist.h
[07:22] <mjg59> make[6] : *** [drivers/usb/net/zd1211/zddevlist.h]  Error 1
[07:22] <mjg59> Or is that me needing a different awk?
[08:06] <BenC> gawk
[08:10] <BenC> siretart: you're correct, the acx100/Makefile is broken
[08:13] <siretart> ah