**** BEGIN LOGGING AT Thu Feb 22 02:59:58 2007 Feb 22 03:45:39 hey angstrom gurus, question about making a change to the pcmcia subsystem from a user perspective Feb 22 03:45:53 there's a possible workaround to a wifi card issue i'm having only in recent kernels Feb 22 03:45:59 http://mail.iocaine.com/pipermail/hostap/2003-January/001060.html has the possible fix Feb 22 03:46:18 it says to try the PCIC_OPTS="irq_mode=1" option for pcmcia Feb 22 03:46:28 what would be equivillent for angstrom? Feb 22 03:47:01 jnc: What does PCIC_OPTS pertain to? Feb 22 03:47:49 on a redhat box it is a setting from rc.pcmcia Feb 22 03:47:55 and /etc/default/pcmcia on debian Feb 22 03:48:00 I'm not exactly sure. Feb 22 03:48:26 at this point I have to say my dmesg output is very similar to problems people are having with recent openembedded based distro builds Feb 22 03:49:10 http://www.oesf.org/forums/lofiversion/index.php/t18141.html Feb 22 03:49:11 jnc: Change your modprobe settings to use that as a parameter for the PCIC module Feb 22 03:49:27 This is what debian does with PCIC_OPTS: Feb 22 03:49:28 modprobe -q $PCIC $PCIC_OPTS >/dev/null 2>&1 Feb 22 03:49:32 pcic module? Feb 22 03:49:44 not seeing one grepping insensitive case for lsmod output Feb 22 03:49:59 PCIC's the controller module, iirc Feb 22 03:50:52 It may be i82365, yenta_socket etc Feb 22 03:51:08 hmm for spitz Feb 22 03:51:52 /lib/modules/`uname -r`/kernel/drivers/pcmcia/ should have the relevant driver Feb 22 03:52:15 good call Feb 22 03:52:30 except um Feb 22 03:52:37 there's no pcmcia directory? Feb 22 03:52:44 you've got a spitz-class device too right Feb 22 03:53:11 I haven't looked on my spitz yet Feb 22 03:53:17 * jnc pokes openembedded "Is this thing compiled in" Feb 22 03:53:20 I'm just going on the debian pcmcia stuff Feb 22 03:54:04 It may be compiled in though, CF dependance and all Feb 22 03:54:27 groking the kernel then Feb 22 03:54:32 In that case.. your best bet may be to create a kernel patch for linux-rp-currentversion Feb 22 03:54:37 this is a genuine bug though Feb 22 03:55:00 I thought I was the only body to have this trouble, there are other reports now I'm seeing in google searches Feb 22 03:55:23 jnc: In that case.. patch + mailing list RFC if it fixes a common issue? Feb 22 03:55:35 yeah Feb 22 03:56:02 surgery tomorrow though. puts me out of use for my left arm for a month Feb 22 03:56:03 oh well Feb 22 03:56:05 :P Feb 22 03:56:11 I can borrow a real laptop computer maybe Feb 22 03:56:33 I tried for a few minutes to get by with one hand on the spitz, just to see Feb 22 03:56:38 it is unpossible. Feb 22 03:57:25 Yeah, you really need both hands.. I can type with 4 fingers on it, but that's 2 from each Feb 22 03:58:13 the spitz config appears to have pcmcia related code compiled in Feb 22 03:59:04 it's not clear to me what's going on though Feb 22 03:59:06 jnc: Does defconfig-spitz have any way to insert the irq_mode=1 option? Feb 22 03:59:18 not that I can tell Feb 22 03:59:26 there's something CIS related, whatever it means Feb 22 03:59:39 CONFIG_PCMCIA_LOAD_CIS=y Feb 22 03:59:57 the hard disk is on pcmcia though Feb 22 04:00:05 as you said, a dependence. Feb 22 04:00:29 CIS = Card Info database, basically Feb 22 04:00:33 oh Feb 22 04:00:44 The one that's relevant looks to be CONFIG_PCMCIA_PXA2XX Feb 22 04:01:30 that's not in teh jnc@macco:~/build_angstrom/build/temp/work/spitz-angstrom-linux-gnueabi/linux-rp-2.6.20-r2/linux-2.6.20/drivers/pcmcia$ grep irq_mode pxa2xx_* Feb 22 04:01:35 no result Feb 22 04:01:51 Yeah Feb 22 04:01:55 only pd6279 appears to have it Feb 22 04:02:42 I don't like bugging RP anymore about things, since I know so little Feb 22 04:02:55 this may be a sort of thing he will know about? Feb 22 04:03:16 Yeah, I'd imagine he would.. AFAIK he's the kernel guru for pxa2xx Feb 22 04:03:56 jnc: Do any of the pages w/ irq_mode suggest why it works, rather than just the fix? Feb 22 04:04:11 no indication of how or why Feb 22 04:04:20 just that it needed to be done with some cards, not with others Feb 22 04:04:34 and only on recent RP kernels if I'm reading it correctly Feb 22 04:04:45 The only reference to irq_mode I can find in pcmcia suggests that it enables PCI mode irq handling over ISA, which I don't think applies to pxa2xx, but I could be seriously wrong Feb 22 04:05:31 hm Feb 22 04:05:39 I'm up for trying things if I am able Feb 22 04:06:34 I'm not really sure where to go from there, TBH.. I'll have a dig around though, and see if there's anything obvious Feb 22 04:08:31 it bombs immediately when I try to use wpa_supplicant for more than a minute Feb 22 04:09:05 if I'm acting quick, sometimes it will even get far enough to authenticate with the AP and get an ipaddress after an 'ifup wlan0' Feb 22 04:09:25 few seconds later, crap fills up dmesg and the system comes to a crawl Feb 22 04:10:33 thanks for suggestions Feb 22 04:11:19 jnc: Hmm, I can see three major things that irq_mode affects in one of the drivers - specifies which interrupt the controller should flag (0x03, PCI INTA#), in card setup as to how interrupts are signalled from the controller, and registering the interrupt w/ the kernel, all of which are only done in PCI mode Feb 22 04:13:21 Beyond that.. I can't see any similarities in the pxa2xx_* codebase; interrupt handling seems to be done differently for SOC Feb 22 04:14:57 As always, standard disclaimer.. IANAKerneldev, so your best bet would be having a chat to one of the others Feb 22 04:15:57 aye Feb 22 04:16:07 i'm trying all the other possibilities first Feb 22 04:16:19 i.e. changing out card firmware Feb 22 04:16:45 which wifi card have you paired up with your spitz? Feb 22 04:21:23 DCF-660W Feb 22 04:23:08 same as I have Feb 22 04:23:15 yours works fine? Feb 22 04:31:03 Yeah Feb 22 04:31:09 Well, as well as the card on the other end :) Feb 22 04:31:18 I've got an orinoco silver in my host machine which is a bit funny Feb 22 04:34:17 hm Feb 22 04:34:22 it appears to be working now? Feb 22 04:34:32 I don't know what causes the problems I am seeing Feb 22 04:34:45 jnc: What issues do you get? Feb 22 04:34:59 If it's WPA-only, I can't really test that here Feb 22 04:35:24 connection drops, card's light flashes, system freezes or slows down quite a bit, weird stuff fills the screen, dmesg fills with crap Feb 22 04:35:27 yeah Feb 22 04:35:29 it's WPA-only Feb 22 04:35:35 you don't use WPA hmm Feb 22 04:36:56 Nah, mine's completely insecure.. I only allow ssh in on that interface Feb 22 04:37:03 I don't think firefox, minimo, or gpe web browser work that well Feb 22 04:37:22 gpe web browser is the only one I am able to get a matchbox wm frame on, the others sit doing nothing Feb 22 10:44:15 koen: can you build tosa ansgtrom images? Feb 22 10:45:28 tosa has a buildable kernel again? Feb 22 10:45:57 <_law_> koen, there is a udev build problem ERROR: QA issue: libvolume_id.pc failed sanity test (reference to /usr/include or workdir) Feb 22 10:46:13 _law_: right, see http://openembedded.org Feb 22 10:46:19 koen: hm. have to check it then Feb 22 10:47:59 * koen bitbake angstrom-bootstrap-image for tosa Feb 22 10:49:01 _law_: and read the openembedded-devel mailinglist Feb 22 10:51:07 <_law_> koen, set PKG_CONFIG_PATH to ${STAGING_LIBDIR}/pkgconfig in my enviorment? Feb 22 10:51:20 <_law_> koen, or where should i set it? Feb 22 15:03:18 we won'weren't expecting to hear anything until next month/msg marcus but then, they didn't say anything about the Feb 22 15:03:36 hmm, wrong window. Feb 22 15:03:40 lots of lag. Feb 22 17:13:18 hi all Feb 22 17:13:24 hey den-ros Feb 22 17:13:32 :) Feb 22 17:50:23 arg. Feb 22 17:50:39 why does osso_initialize fail without verbose errors messages? Feb 22 18:27:18 what package supplies /etc/ipkg/*feed* Feb 22 18:27:55 none Feb 22 18:28:11 how do they get there? Feb 22 18:28:17 do_rootfs() Feb 22 18:28:28 ack Feb 22 18:28:30 yes, that is a bad way to do that Feb 22 18:28:41 no wonder I couldnt find them :-) Feb 22 18:28:50 :) Feb 22 18:30:40 koen: cheers anyway I found the ones I needed in emergency :-) Feb 22 22:08:00 hi **** ENDING LOGGING AT Fri Feb 23 02:59:57 2007