**** BEGIN LOGGING AT Sat Jun 11 23:59:56 2005 Jun 12 00:57:38 mickeyl: ping Jun 12 01:06:19 morning Jun 12 01:18:37 Morning Jun 12 01:19:09 hi CoreDump|home, uv1 Jun 12 01:19:27 evening CoreDump|home uv1 bluelightning Jun 12 01:19:37 hi kared Jun 12 01:36:42 RP: It seems that the initialization AtiCore code is no longer valid under 2.6 Jun 12 01:37:01 RP: It looks like the framebuffer has changed it's physical address. Jun 12 01:46:04 hi guys Jun 12 01:47:59 Spyro: Hi. Jun 12 01:48:17 Problem with S5500. Does not automount mmc card. Image is from yesterday. Any ideas ? Jun 12 01:48:36 uv1: yeah Jun 12 01:48:46 let me guess, a manual mount works? Jun 12 01:48:53 CoreDump|home: yes Jun 12 01:48:59 see bug #47 :) Jun 12 01:49:08 CoreDump|home: ok. Jun 12 01:49:43 busybox is fubared due to a certain patch. remove the patch, rebuild BB and all is good Jun 12 01:53:04 yay. I tricked oe into building a kernel but not packaging it. Jun 12 01:55:29 is familiar distro really using latest CVS gcc ? Jun 12 01:55:33 is that wise? Jun 12 01:57:43 I guess not, as my bootstrap-image build just failed on gcc Jun 12 01:58:01 The following requested languages could not be built: f77 Jun 12 01:58:01 Recognised languages are: c,ada,c++,f95,java,objc,obj-c++,treelang Jun 12 01:58:08 jacques: I forced an earlier version Jun 12 01:58:25 is it possible to suppress bitbakes building of all locales? Jun 12 01:58:35 Spyro, which version do you suggest? Jun 12 01:59:02 jacques: see the familiar 0.8.2 conf file - it has some commented out PREFERRED_VERSION lines Jun 12 01:59:21 I transplanted those into the familiar 0.8.3 file and am using that, so far, sucessfully Jun 12 01:59:48 Spyro, great advice - thanks Jun 12 02:00:01 np. Jun 12 02:03:31 I guess I shouldn't have been using familiar dist anyway - should be using fmiliar-0.8.3 Jun 12 02:05:31 :) Jun 12 02:13:30 morning all Jun 12 02:14:12 hi RP Jun 12 02:14:24 Sirfred: Under 2.6, we need to alter atilib to use /dev/mem Jun 12 02:15:03 RP: Good morning. Jun 12 02:15:11 RP: So, there's no chance to use it in it's current state. Jun 12 02:16:22 RP: ? Jun 12 02:16:43 Sirfred: Correct, it needs a simple patch Jun 12 02:17:03 I'm just looking for it - in fact its in oe if I remember correctly Jun 12 02:17:14 RP: What assumptions is libaticore making to fail in this way? Jun 12 02:17:34 RP: Visually, it seems that it's only the framebuffer offset what's bad. Jun 12 02:17:42 Sirfred: It assumes the static mapping of the ati registers that the 2.4 kernel makes Jun 12 02:17:56 That static mapping is world writeable Jun 12 02:18:07 Under 2.6, you can't easily make such a mapping anymore Jun 12 02:18:15 and I doubt we should be making it anyway Jun 12 02:19:23 RP: If I understand you, that means that there's no way to go before completing libaticore disassembling and translating? Jun 12 02:19:44 Sirfred: No, we can patch the assembler Jun 12 02:20:17 RP: But all reference to registers need to be patched? Jun 12 02:20:26 RP: To add some offset or something so? Jun 12 02:20:33 RP: I mean some base offset. Jun 12 02:20:34 Sirfred: Have a look at openembedded/packages/sharp-binary-only/sharp-aticore-oss-1.0.1/aticore-2.6.patch Jun 12 02:20:47 RP: There I go Jun 12 02:21:50 RP: I see you only needed to patch MapBar function. Jun 12 02:21:58 Sirfred: Correct Jun 12 02:22:23 RP: So, if I use the sharp-aticore-oss package instead of sharp-aticore that I'm using now... Jun 12 02:22:24 Sirfred: This appears to work fine under 2.6 Jun 12 02:22:32 Sirfred: It will work :) Jun 12 02:22:37 RP: Great. Jun 12 02:22:45 RP: I cannot wait to test it Jun 12 02:23:43 RP: I saw that sharp-aticore-oss package yesterday, but I didn't know what it was. Jun 12 02:23:53 is this a benign or blocker error: cvs checkout: warning: cannot open /cvs/CVSROOT/val-tags read/write: Permission denied Jun 12 02:24:01 Sirfred: Its a libati that has been split open a bit... Jun 12 02:24:08 RP: :) Jun 12 02:24:09 trying to get handhelds.org 2.6 kernel Jun 12 02:24:10 Sirfred: Its the one I've been working from Jun 12 02:24:21 RP: Should be sharp-aticore-abitoss Jun 12 02:24:54 ah, OK the checkout started now so I guess the error was benign Jun 12 02:24:54 Sirfred: Its being worked on, as you know ;-) Jun 12 02:25:00 RP: I know. Jun 12 02:29:02 RP: I've uninstalled sharp-aticore and installed sharp-aticore-oss Jun 12 02:29:12 RP: But now, I have not libaticore.so, only libaticore.so.1.0.1 Jun 12 02:29:33 RP: Shouldn't sharp-aticore-oss make symbolic links to libaticore.so.1 and libaticore.so ? Jun 12 02:31:23 RP: Well, I think that I was using this libaticore-oss before, it was automatically installed while I requested sharp-aticore. Jun 12 02:31:51 RP: Now, I see that the "MapBar: %08x" logs were produced by your new library version Jun 12 02:32:13 RP: Anyway, I still have that framebuffer offset problem. Jun 12 02:32:28 RP: Did you test my code examples under this library? Jun 12 02:32:36 RP: I mean , the ones to bitblt, drawline,... Jun 12 02:33:36 I've only ever used the atitest program which has always appeared to work Jun 12 02:34:01 Does your code make assumptions of the framebuffer offset? Jun 12 02:36:51 RP: Yes, the infamous: Jun 12 02:36:57 VRAMoffset = (scr_w == 320)?0:0xf000000; Jun 12 02:37:03 ret = AtiCore_SetFrontBuffer(VRAMoffset, 0, 0); Jun 12 02:37:23 RP: But I suppose that that is the framebuffer address as seen by the w100, isn't it? Jun 12 02:37:32 That's what I thought... Jun 12 02:37:42 RP: Could you send me a working example? Jun 12 02:37:55 RP: Mine are not drawing fine on the framebuffer. Jun 12 02:38:06 Does atitest work? Jun 12 02:38:13 RP: Wait... Jun 12 02:38:41 RP: That's shipped into sharp-aticore package, isn't it? Jun 12 02:38:54 Its shipped in both Jun 12 02:39:08 sharp-aticore-oss should also have it Jun 12 02:40:15 Package sharp-aticore-oss (1.0.1-r1) is installed on root and has the following files: Jun 12 02:40:16 /usr/lib/libaticore.so.1.0.1 Jun 12 02:40:26 RP: It isn't Jun 12 02:40:45 Sirfred: Did you build your own image? Jun 12 02:40:58 RP: No. Jun 12 02:41:01 RP: It's the official package. Jun 12 02:41:08 RP: Well, I've installed the sharp-aticore package. Jun 12 02:41:22 RP: I've checked that libaticore.so.1 is a symbolic link to libaticore.so.1.0.1 Jun 12 02:41:26 RP: atitest is not working. Jun 12 02:41:43 RP: It has the same problem, the framebuffer seems to be assumed in a wrong address. Jun 12 02:42:24 Sirfred: Thinking about it, atitest comes as a seperate package Jun 12 02:43:02 RP: But it's also in sharp-aticore Jun 12 02:43:35 RP: And I don't see any other candidate on sharp-binary-only directory. Jun 12 02:44:10 Sirfred: Try the two ipks at http://www.rpsys.net/openzaurus/temp Jun 12 02:44:55 Sirfred: They were generated from when I last worked on atilib. I'm hoping they work... Jun 12 02:45:38 RP: You mean: libaticore1_1.0.1-r0_armv5te.ipk and sharp-aticore-oss_1.0.1-r0_armv5te.ipk ? Jun 12 02:46:28 Sirfred: yes Jun 12 02:46:35 RP: Let's try Jun 12 02:46:39 hmm, openswan is in bootstrap-image ? that's new Jun 12 02:50:52 RP: Your sharp-aticore-oss depends on sharp-aticore. Jun 12 02:51:02 RP: And there are conflicting files among the two packages. Jun 12 02:51:10 Package sharp-aticore wants to install file /usr/lib/libaticore.so.1 Jun 12 02:51:12 But that file is already provided by package sharp-aticore-oss Jun 12 02:51:14 RP: And so on Jun 12 02:51:21 Does anyone know, why a CL860 (2.6) is slower, than a SL5500 (2.4) ? Jun 12 02:51:36 Sirfred: remove sharp-aticore and then force my packaged to install Jun 12 02:51:56 THey're obviously going to conflict as they provide the same thing Jun 12 02:52:15 RP: Yes, but sharp-aticore is automatically downloaded by ipkg when I request your package to install. Jun 12 02:52:19 RP: How can I avoid that? Jun 12 02:53:01 Sirfred: I usually put them on a CF card and don't tell ipk anything about urls... Jun 12 02:53:26 RP: Well, it's a way. Jun 12 02:53:40 uv1: The SL5500 has a more streamlined shape so reaches a greater terminal velocity as it falls through the air... Jun 12 02:53:59 RP: Havent' you consider the option of erasing that dependency? Jun 12 02:54:01 RP: :) Jun 12 02:54:12 uv1: You might want to define "slower" ;-) Jun 12 02:54:39 Sirfred: These are my development packages. They're broken in many ways.... Jun 12 02:55:27 RP: So, I should need to comment out my ipkg sources Jun 12 02:55:36 Sirfred: No Jun 12 02:55:52 RP: Could you be so kind to explain me the way to proceed? Jun 12 02:55:57 Sirfred: Just download them onto the device and ipkg instal them Jun 12 02:56:00 manually Jun 12 02:56:06 RP: That's what i made Jun 12 02:56:24 RP: I said: ipkg install ./sharp-aticore-oss_1.0.1-r0_armv5te.ipk Jun 12 02:56:37 RP: And the clever ipkg automatically dowloaded its dependencies. Jun 12 02:56:53 Mine never does that... Jun 12 02:57:00 Maybe becuase I've never configured it perhaps... Jun 12 02:57:01 RP: Mine seems to ve clever. Jun 12 02:57:19 RP: Neither I do. Its untouched after the fresh install. Jun 12 02:57:41 RP: Is your zaurus connected to the network? Jun 12 02:57:49 RP: I mean, internet connected? Jun 12 02:58:24 Sirfred: Not when I've been testing atilib... Jun 12 02:58:32 RP: That could be the reason. Jun 12 02:58:41 RP: I'm going to temporaly disable the default route Jun 12 02:59:50 No way. Jun 12 02:59:51 Failed to download sharp-aticore. Perhaps you need to run 'ipkg update'? Jun 12 03:00:04 Too clever, I'm starting to hate it. Jun 12 03:00:54 Sirfred: My current image is too corrupt to try and test on. ipkg won't run :-( Jun 12 03:00:55 RP: Well, I'm going to unpack your ipk and put the files manually on my SD. Jun 12 03:01:10 RP: I think it would be the fastest way. Jun 12 03:01:10 Sirfred: That might be best Jun 12 03:01:54 ipkg: "Package (null) has no Architecture defined" Jun 12 03:02:24 Is this a known problem with ipkg I can easily fix? Jun 12 03:02:41 RP: Well, this atitest seems to work fine. Jun 12 03:03:59 RP: I WANT that code. ;-) Jun 12 03:04:06 RP: Could you please... ? Jun 12 03:04:15 Sirfred: I suspect you were using sharp-aticore instead of sharp-aticore-oss Jun 12 03:04:33 RP: No, because the MapBar trace was showing. Jun 12 03:04:48 RP: Now, more traces are showing, so I'm using the same library as your atitest. Jun 12 03:04:57 RP: But I'm missing something in my code Jun 12 03:05:03 Is this http://www.rpsys.net/openzaurus/temp/atitest.c your atitest ? Jun 12 03:06:16 It seems that it's my fault, because my test program are showing equally bad under this new libaticore and the packaged by sharp-aticore-oss. Jun 12 03:06:57 Yes, that's my atitest.c Jun 12 03:07:31 atilib-rp.tgz is the source to the ipks you installed Jun 12 03:07:54 Keep in mind this is my hacked working copy Jun 12 03:08:24 Sirfred: If you do find a different between what we need to do under 2.6 and 2.4, I'd like to know about it Jun 12 03:08:55 RP: Of course. Jun 12 03:09:05 RP: I've compiled this atitest of yours, and works fine, also Jun 12 03:09:20 RP: Now, I'm going to revert to the "official" sharp-aticore-oss Jun 12 03:29:04 I'm really confused now. Jun 12 03:29:23 RP: After installing your sharp-aticore-oss, ipkg is finding that package, I don't know how. Jun 12 03:29:50 someone here have build succesfully gpdf ? Jun 12 03:29:52 RP: ipkg info sharp-aticore-oss dump information about two packages, the official and yours. But, how is it able to find information for yours, when I have purged and erased the ipkg? Jun 12 03:31:06 RP: In /usr/lib/ipkg/lists there's only one sharp-aticore-oss. Jun 12 03:32:23 Going to restart the zaurus Jun 12 03:35:05 jacques: I checked in a new patch for the busybox environ thing. Would you like to give that a try? Jun 12 03:36:37 Twiun: good morning Jun 12 03:36:39 03pb 07 * r1.3555 10openembedded/packages/busybox/ (busybox_1.00.bb busybox-1.00/ifupdown-spurious-environ.patch): second attempt at fixing the busybox environ thing Jun 12 03:36:53 pb_, great, I'll try it in maybe 20 minutes when my familiar bootstrap build finishes :-D Jun 12 03:37:21 CIA-4: better late than never Jun 12 03:37:24 jacques: cool, thanks Jun 12 03:37:32 pb_: Hey! Jun 12 03:37:38 pb_: long time no see :) Jun 12 03:37:42 pb_: back in the uk? Jun 12 03:37:46 gpe-image take FOREVER ;-) Jun 12 03:45:57 Twiun: yep, been back for a while Jun 12 03:47:09 pb_: I lose track of what's happening in this chan now I don't have any hw to play with Jun 12 03:49:28 yeah, that's too bad Jun 12 03:51:23 03pb 07 * r1.3556 10openembedded/packages/xserver/xserver-xorg_6.8.99.10.bb: new xorg snapshot Jun 12 03:54:14 is it possible to suppress locale support in bitbake? I only really want en_gb Jun 12 03:54:54 don't I wish Jun 12 03:56:25 I assume you mean suppress building of all locales and packaging them Jun 12 03:56:39 yeah Jun 12 03:56:45 which adds probably 30-40 minutes to the average build :-\ Jun 12 03:56:53 yeah Jun 12 03:57:10 I've asked this before, apparently it's harder than it sounds... or something Jun 12 03:57:21 really? it should only be a few minutes, certainly not 30 or 40. Jun 12 03:57:38 pb_: theres a LOT of it Jun 12 03:57:44 hundreds of packages Jun 12 03:57:50 literally Jun 12 03:57:53 pb_: either way its a big waste of flash :) Jun 12 03:57:57 right, but each of those packages is tiny Jun 12 03:58:00 Spyro: eh? Jun 12 03:58:07 are you running bitbake on flash? Jun 12 03:58:11 but packaging them is not so fast Jun 12 03:58:20 Spyro, you don't ahve to *install* them all heh :-) Jun 12 03:58:23 installing the resultant rootfs on flash... Jun 12 03:58:48 jacques: doesnt it install everything by defualt ? Jun 12 03:58:51 I don;t think they would all fit - last I checked glibg had 16MB of locale data alone Jun 12 03:58:53 Spyro: They're not all installed into the rootfs... Jun 12 03:58:59 glibc Jun 12 03:59:22 oh, thats not so bad then Jun 12 04:00:34 do IP/licensing issues prevent 4-bit SD open source drivers? Jun 12 04:00:53 no. Jun 12 04:00:59 jacques: tmio_mmc is 4 bit Jun 12 04:01:04 there are open source 4-bit drivers for at least some of the ipaqs. Jun 12 04:01:24 mreimer posted to the 2200 list that he got SD working on 2200, and was getting 1.81MB/s hdparm read Jun 12 04:01:27 which is 1bit speeds Jun 12 04:01:46 and very close to the speeds I get with my MMC cards in a cardreader Jun 12 04:01:56 jacques: he posted an update this morning Jun 12 04:02:04 I just haven't seen any numbers like the windows ones Jun 12 04:02:21 pb_, after the one with the 1.81MB/s post ? Jun 12 04:02:25 yeah Jun 12 04:02:33 RP: I've found the difference. Jun 12 04:02:35 ah cool I guess gmane hadn't gotten it yet Jun 12 04:02:45 RP: It's strange, in my opinion. Jun 12 04:02:52 jacques: he claims to be getting 124.75 MB/sec now. I'm not quite sure I believe that number, but it certainly sounds like an improvement. Jun 12 04:03:05 that's buffered read speed Jun 12 04:03:17 that's hdparm -T need hdparm -t Jun 12 04:03:27 oh, right, yeah. sorry, I was reading the mail wrong. Jun 12 04:03:28 that will approach memory read speed Jun 12 04:03:42 that was indeed the one with the 1.81MB transfer rate. Jun 12 04:03:48 ah :-\ Jun 12 04:03:53 RP: The problem seems to be related with Aticore_SetupGraphicWindow Jun 12 04:04:13 my fastest SD cards get just over 6MB/s in a usb2.0 reader Jun 12 04:04:20 (linux hdparm -t) Jun 12 04:04:35 there are lots of cards out now which can to 10MB/s + Jun 12 04:05:22 RP: I had changed win.Width to scr_w and win.Height to scr_h, that's the problem. I don't understand why, but those dimensions should be, doesn't matter what the screen orientation is, 480 and 640 respectively. Jun 12 04:06:04 damn, waitasecond.... Jun 12 04:06:27 I just remembered - my OLD OLD sandisk 256MB card only gets about 1.75MB/s..... Jun 12 04:06:37 maybe he just needs a faster card :-) :-) Jun 12 04:07:36 heh, maybe Jun 12 04:07:48 hmm, he says 512MB tho which _should_ be faster - my old nasty one is a 256MB Jun 12 04:07:56 and my fastest are panasonic 512MB ones Jun 12 04:08:08 I guess we should get him to try his card in another machine. Jun 12 04:08:24 yeah in a usb2.0 cardreader under linux hdparm -t would be ideal Jun 12 04:08:33 I should get my finger out and test it on my h5500, too. Jun 12 04:09:00 need to find the 5500 first, though, which could be a challenge. I haven't seen it for weeks. Jun 12 04:09:07 lol Jun 12 04:13:19 What is the mmcqd for ? Jun 12 04:13:25 What is the mmcqd process for ? Jun 12 04:13:35 Sirfred: Under 2.4, the framebuffer was rotated, under 2.6 it isn't Jun 12 04:13:59 RP: Anyway, it's a little inconsistent. Jun 12 04:14:14 Sirfred: Not really, we're just removing sharp hacks Jun 12 04:14:39 RP: Now, I'm using: win.Size.Width = 640, and win.Width = 480 Jun 12 04:15:24 Sirfred: Which is correct - the screen is that size Jun 12 04:15:54 RP: But why win.Width needs to be 480 ? Jun 12 04:16:24 Sirfred: I misread that. That is confusing Jun 12 04:16:43 It could be confusing panel width with actual width Jun 12 04:16:46 RP: Using win.Width == 480 as you're using in atitest makes it work. Jun 12 04:16:53 RP: Could be. Jun 12 04:17:00 RP: Well, now it's working fine under 2.6 Jun 12 04:17:03 We need to see where SetupGraphicWindow uses those values Jun 12 04:17:16 RP: Do you know what is the process mmcqd for ? Jun 12 04:17:33 RP: I suspect it has something to do with sd card io Jun 12 04:18:00 Sirfred: It is mmc related Jun 12 04:18:41 RP: I have my test programs on the sd card. The test program stopped, mmcqd was at 30% CPU, my test program freezed and 0% Cpu Jun 12 04:18:56 RP: After a while, it continues. Jun 12 04:19:24 Sirfred: It probably hit some timeout problem Jun 12 04:22:42 RP: Finally, I have opie running with the accelerated driver. Jun 12 04:23:11 Sirfred: excellent :D Jun 12 04:23:45 RP: Well, a lot of work remaining, but I like the better scroll performance. Jun 12 04:23:59 mickeyl: ping Jun 12 04:24:24 Sirfred: The scroll performance would be a big bonus Jun 12 04:24:39 RP: Anyway, it seems to work different that on 2.4 Jun 12 04:24:50 RP: On 2.4 it seemed... synchronized with the redrawing. Jun 12 04:26:35 Sirfred: Is this not a result of preempt under 2.6? Jun 12 04:26:47 RP: Don't know. Jun 12 04:28:20 RP: There're things that perhaps are working slower than in software mode. Jun 12 04:28:39 run xperf (j/k) Jun 12 04:28:46 or is it x11perf Jun 12 04:29:17 RP: I think that when the number of cliprects is high, perhaps software mode could be better, depending on the operation. Jun 12 04:31:52 NOTE: preferred version "3.4.4 of gcc-cross-initial not available Jun 12 04:32:27 which is weird because it is indeed available Jun 12 04:33:23 but it built gcc-cross-initial-3.4.4+csl-arm-20050603-r0 so I guess that's close enough :-) Jun 12 04:35:00 afternoon Jun 12 04:35:01 maybe I should tell it to build gcc-cross-3.4.4+csl-arm-20050603 tho... Jun 12 04:35:06 Sirfred: That's probably true... Jun 12 04:35:39 RP: I've prepared a .bb and a patch to build this libqte. Jun 12 04:35:58 jacques: that is a bit odd Jun 12 04:36:21 RP: I've modified qte_2.3.10.bb and added a patch to that existing package. Perhaps I should create a new one ? Jun 12 04:36:53 pb_, I'll be flashing the image with the new bb/ifupdown soon Jun 12 04:37:04 jacques: very good Jun 12 04:37:44 Sirfred: You should probably talk to mickeyl about this Jun 12 04:37:54 Curious, pb_ messages show the ip addr on opie-irc. Jun 12 04:38:00 Sirfred: We can probably apply the patch conditionally Jun 12 04:38:06 It seems that it doesn't like ipv6. Jun 12 04:38:13 RP: How is that? Jun 12 04:39:29 Sirfred: Something like SRC_URI_append_c7x0 Jun 12 04:39:40 That would just make it apply for the c7x0 machines Jun 12 04:39:41 RP: Well, that's what I made Jun 12 04:40:28 RP: Anyway, the patch will not show if you don't specify QWS_DISPLAY=W100:0 Jun 12 04:40:41 Sirfred: That's fine then Jun 12 04:40:42 RP: So, more changes are needed. Jun 12 04:40:57 RP: Well, I'll try to talk with mickeyl Jun 12 04:41:03 RP: It seems to be missing now. Jun 12 04:41:07 s/It/He/ Jun 12 04:43:53 Sirfred: He's the best person to talk to about this Jun 12 04:44:30 RP: First bug: Screen rotation doesn't work. Jun 12 04:44:54 03pb 07 * r1.3557 10openembedded/packages/libxvmc/libxvmc_20050612.bb: new XvMC snapshot Jun 12 04:45:12 Sirfred: That will probably be because we do the rotation in hardware Jun 12 04:45:26 (I think) Jun 12 04:49:15 good afternoon all Jun 12 04:49:30 hi koen Jun 12 04:49:36 hey woglinde Jun 12 04:52:25 reenoo_: do you have any preference where classpath-native should stage it's classpath? Jun 12 04:57:24 koen: no, nothing in particular. Jun 12 05:21:53 If I alter my local.conf, my entire toolchain, etc. (and glibc :( ) appears to be rebuilt now. Is this normal behaviour? Jun 12 05:22:51 only if someone bumped the PR for gcc and glibc Jun 12 05:23:29 hmm, I don;t think that's happened since last night (when I built an opie-image) Jun 12 05:23:32 do you have a TMPDIR defined? Jun 12 05:23:55 no Jun 12 05:24:04 that would be a problem Jun 12 05:24:12 why is that? Jun 12 05:24:27 OE use $PWD/tmp as a tmpdir Jun 12 05:24:40 yep Jun 12 05:24:54 which is build/tmp for me Jun 12 05:25:37 try setting TMPDIR in local.conf to that Jun 12 05:26:02 but it's already using that TMPDIR, for both local.confs Jun 12 05:31:30 Sirfred|lunch: pong Jun 12 05:38:37 Sirfred|lunch: please send me your patch and a patch against the current .bb file and I'll have a look at it later Jun 12 05:38:49 congrats, btw. Jun 12 05:39:07 mickey|bbl: Send where? Jun 12 05:42:10 pb_, flashing openslug with patched ifupdown now Jun 12 05:43:44 Sirfred|lunch: mickeyl@handhelds.org - need to do some housework now Jun 12 05:45:50 pb_, looks like it worked! Jun 12 05:50:47 pb_, OOPS - the nslu2 repos hadn't pulled that patch yet - I saw one go by but it must have been the old one - built on a different machine Jun 12 05:50:54 pb_, so it's still untested Jun 12 05:53:12 rebuilding... Jun 12 05:53:50 looks like xtroke.org is down Jun 12 05:53:55 xstroke.org Jun 12 05:54:23 but oesources had it :-) Jun 12 06:04:40 flashing again... Jun 12 06:06:27 03zecke 07 * r1.3558 10openembedded/packages/qpe-gaim/qpe-libgaim_1.3.1.bb: Jun 12 06:06:27 QPE Gaim: Jun 12 06:06:27 Update from gaim-1.3.0 to gaim-1.3.1 to fix two security Jun 12 06:06:27 issues Jun 12 06:09:30 pb_, OK, now I am confident idupdown patch is really tested and it appears to work Jun 12 06:12:51 03RP 07 * r1.3508.1.1 10openembedded/packages/oprofile/ (6 files in 3 dirs): Add oprofile_cvs + upgrade oprofile-0.8 to 0.9 Jun 12 06:15:06 RP: having fun with nwfpe Jun 12 06:15:08 ? Jun 12 06:15:57 koen: I've build and installed a famailiar image for my Z do I'll try and find out what the problem is... Jun 12 06:16:53 Of course my build partition just ran out of diskspace so I'm having to rearrange everything... Jun 12 06:21:51 I just gained 20GB by cleaning my multimachi build Jun 12 06:22:53 My builddir is on a 20GB partition and I have mutliple builds, none of which I can delete Jun 12 06:23:40 Hmmm. Make that a 30GB partition :-/ Jun 12 06:25:05 I need a larger drive for my laptop because of oe Jun 12 06:26:18 jacques: or create a bbclass to delete the workdir after building Jun 12 06:29:25 and watch pkgconfig blow up in your face Jun 12 06:29:59 minimo takes a while to build, doesn't it :-) Jun 12 06:30:56 koen: luckily most of the Opie apps do not use pkg-config anyway Jun 12 06:31:10 koen: the rest needs to be fixed, I'm sure one can automate it using mono Jun 12 06:31:33 anyway, pkgconfig rocks Jun 12 06:32:01 the c++ rewrite does rock indeed Jun 12 06:32:18 jacques: rock Jun 12 06:32:34 :-) Jun 12 06:33:11 hmmm, I just thought of something...... Jun 12 06:33:20 uh-oh Jun 12 06:33:52 nope, seems fine Jun 12 06:34:08 at one point we had a conditional to not include the patch for NSLU2 Jun 12 06:34:19 but that's not there, and I recall seeing the patch applied by oe Jun 12 06:35:27 Applying patch ifupdown-spurious-environ.patch Jun 12 06:35:27 patching file networking/ifupdown.c Jun 12 06:35:30 from the log Jun 12 06:37:47 ok, very good Jun 12 06:38:20 argh, my gpe-image build crapped out because it couldn't download libosip Jun 12 06:38:53 hho. gives a 404 Not Found for it Jun 12 06:39:06 as does oesources Jun 12 06:39:50 man, was the hh.o download section totally reorganized? Jun 12 06:40:21 jacques: http://familiar.handhelds.org/source/v0.8.2/ Jun 12 06:40:34 pb is glibc 2.3.5 now building for arm? Jun 12 06:40:44 koen, thanks :-) Jun 12 06:44:58 jacques: or http://familiar.handhelds.org/snapshots/v0.8.3/sources/ for more up to date stuff Jun 12 06:46:02 woglinde: yes Jun 12 06:46:33 pb oh will try it with the simpad Jun 12 06:46:44 pb_, I'll try for armeb :-P Jun 12 06:46:56 very good Jun 12 06:49:06 as for the handhelds.org download section, yeah, you probably need to remove "pub/" from the URI. Jun 12 06:49:55 pb_, I think it needs more than that :-\ Jun 12 06:53:46 jacques: ah Jun 12 06:54:37 http://handhelds.org/packages/libosip/libosip-0.9.7-hh2.tar.gz is the correct url Jun 12 06:54:51 it looks like the "linux/" directory has also gone away Jun 12 06:56:40 I was wondering, what configuration do I need to do to build an image for say a h2200 ? is ipaq-pxa-2.6 good enough ? Jun 12 06:57:18 I'm not sure. I think either that one, or h3900 (with KERNEL=kernel26) should probably be ok. Jun 12 06:57:29 ok, I'll check the wiki too Jun 12 06:57:53 good plan Jun 12 06:58:00 what's the fastest way to try to build glibc 3.3.5 ? can I specify a version to the bitbake command ? Jun 12 06:58:17 2.3.5 that is Jun 12 06:58:19 bitbake -b oe/packages/glibc/glibc_cvs.bb Jun 12 06:58:50 thanks Jun 12 06:59:01 or "bitbake glibc_2.3.5+cvs20050612" might work as well. Jun 12 06:59:17 oh, probably need - rather than _ Jun 12 07:00:47 it's doing the cvs co now Jun 12 07:09:57 I'm not able to connect using ssh openzaurus version Jun 12 07:10:35 pb_: Which maths tests was it you were recomending to test the nwfpe code with? Jun 12 07:11:17 It says me that no auth methods could be used. Jun 12 07:12:07 * Sirfred|760 is too confortable on his sofa to go for his laptop Jun 12 07:12:29 * CoreDump|home knows the feeling Jun 12 07:13:14 CoreDump|home: ;-) Jun 12 07:13:44 yeah thats feeling is great Jun 12 07:14:31 Unfortunately, there's only shit on the tv Jun 12 07:18:40 Am I right to assume that only collie has a real RTC and all other devices lose the time on a simple reboot? Jun 12 07:19:07 ugh s/real// Jun 12 07:21:13 glibc 3.3.5 is on the do_populate_staging step, so it looks like it built Jun 12 07:21:26 you mean 2.34 Jun 12 07:21:28 you mean 2.3.5 Jun 12 07:24:14 CoreDump|home: StrongARM and PXA have a RTC Jun 12 07:24:27 zecke|assignment: thanks Jun 12 07:24:28 CoreDump|home: the PXA RTC can not be buffered Jun 12 07:25:05 (battery buffered) Jun 12 07:25:50 woglinde, yeah :-) Jun 12 07:26:07 hms Jun 12 07:26:18 why I dont have that bitbake file Jun 12 07:26:28 it's the cvs one Jun 12 07:26:38 glibc-2.3.5+cvs20050612-r0 Jun 12 07:28:30 oehm I only have 2.3.3+cvs20050420.bb nad bk pull dont give me the new one Jun 12 07:30:50 It seems that on opie-console the supposed alt-key is mapped as ctrl-c. Is this correct? Jun 12 07:31:53 Sirfred|760: It shouldn't be but it wouldn't surprise me... Jun 12 07:32:02 oehm jacques hm were did pb the file put? Jun 12 07:32:06 woglinde, it uses the glibc_cvs.bb Jun 12 07:32:57 of course, the real test will be if gcc builds against glibc 2.3.5 Jun 12 07:33:58 RP: How the keyboard mapping works now on oe? If I load a new keyboard map with loadkeys should these changes be reflected automatically on opie? Jun 12 07:34:16 RP: s/oe/qte Jun 12 07:34:37 Sirfred|760: no a restart of Qt/E is needed Jun 12 07:35:12 Sirfred|760: correct - listen to zecke|assignment as he wrote the code ;-) Jun 12 07:35:19 zecke: I tried to map alt + e to eacute but it only works on the console Jun 12 07:35:38 RP:Thanks. ;-) Jun 12 07:36:27 zecke|assignment: On qt/e I'm getting a plain 'e' Jun 12 07:37:10 * Sirfred|760 prays for nick completion in opie-irc Jun 12 07:41:11 zecke|assignment: I'm using : alt keycode 18 = eacute Jun 12 07:41:53 Sirfred|760: does your font have a eacute? Jun 12 07:42:27 Sirfred|760: alternatively install opie keyview, it will show you keycode and unicode value of what you type Jun 12 07:42:32 zecke|assignment: I think so. Jun 12 07:49:05 zecke|assignment: the generated code is the same for (alt + e) and for (e) Jun 12 07:49:23 ok Jun 12 07:49:31 so there is a bug Jun 12 07:51:53 jques can ypu give me please your glibc-2.3.5.bb? Jun 12 07:53:22 jaques can ypu give me please your glibc-2.3.5.bb? Jun 12 07:54:03 woglinde: drunk? stoned? Jun 12 07:54:21 jaques maybee drunken from yesterday Jun 12 07:54:33 lol Jun 12 07:55:32 aehm Jun 12 07:55:35 yes Jun 12 07:55:52 woglinde, there is no glibc-2.3.5.bb, it uses the glibc_cvs.bb Jun 12 07:56:28 jaques aeh right, damned Jun 12 07:56:44 I am drunken I guess Jun 12 07:58:37 woglinde: You're not sure? We are. ;-) Jun 12 07:59:35 The icon toolbar on konqueror-embedded on my c760 looks too pixelized. Jun 12 07:59:48 The question is: are different icons used for different screen resolutions? Jun 12 08:00:17 Or the same icons are used for the 320x240 resolution PDAs and 640x480 ones? Jun 12 08:00:36 when I forced the glibc 2.3.5 build it worked, but now that I want to built an image with it, and set PREFERRED_PROVIDER_glibc, it says: NOTE: preferred version glibc-2.3.5+cvs20050612 of glibc not available Jun 12 08:01:09 jaquces it looks after the .bb file Jun 12 08:01:24 what could be overriding my PREFERRED_PROVIDER_glibc setting? Jun 12 08:01:33 do you mean PREFERRED_VERSION? Jun 12 08:01:43 afaik, there is only one provider for glibc Jun 12 08:02:39 judging from the message you pasted, you have a stray "glibc-" on the front of the preferred version string. Jun 12 08:02:44 pb hm what Entry we have to make in local.conf? Jun 12 08:02:53 i.e. you need: PREFERRED_VERSION_glibc = "2.3.5+cvs..." Jun 12 08:02:57 not Jun 12 08:03:06 PREFERRED_VERSION_glibc = "glibc-2.3.5+cvs..." Jun 12 08:03:26 pb hm Jun 12 08:03:46 pb_, yeah I said it wrong above, I have PREFERRED_VERSION_glibc = "glibc_2.3.5+cvs20050612" Jun 12 08:03:56 okay Jun 12 08:03:59 and it even says: NOTE: preferred version glibc-2.3.5+cvs20050612 of glibc not available Jun 12 08:04:02 remove that glibc_ and you shoukd be in business Jun 12 08:04:24 oops OK Jun 12 08:04:33 next try *sigh* Jun 12 08:05:06 also, you need to make sure that your CVSDATE (or CVSDATE_glibc) is set to 20050612 Jun 12 08:05:20 otherwise it won't match with the preferred string. Jun 12 08:05:29 ah, OK I'll do that next Jun 12 08:05:37 if this fails Jun 12 08:05:48 pb oh I see Jun 12 08:06:02 if it used the right date when you did "bitbake -b", that's probably set correctly. Jun 12 08:06:13 pb_, it did indeed Jun 12 08:06:24 pb thanks Jun 12 08:06:53 ah, kergoth awakes Jun 12 08:07:19 moin kergoth Jun 12 08:07:27 hey Jun 12 08:07:55 grr, poxy mythtv doesn't compile with gcc 4.0 Jun 12 08:08:04 I guess I'll have to go back to 3.4. Jun 12 08:08:22 pb_, you really want to move to gcc4 don't you? :-P Jun 12 08:09:07 yeah, I guess I am a closet gentoo user Jun 12 08:09:22 pb lol Jun 12 08:09:23 lol Jun 12 08:09:38 gentoo isn't even on 3.4 yet for non ~arch users Jun 12 08:09:56 OK oe is building the glibc I want it to now :-) Jun 12 08:10:00 very good Jun 12 08:10:41 dmaned Jun 12 08:10:44 preferred version glibc-2.3.5+cvs20050612 of glibc not available Jun 12 08:10:59 woglinde: doh, you did the same thing as jacques Jun 12 08:11:06 its not my day Jun 12 08:11:14 inspect your PREFERRED_VERSION_glibc line very carefully Jun 12 08:11:20 CVSDATE_glibc = "20050612" Jun 12 08:11:23 you need to remove all instances of the word "glibc" on the right hand side of the equals sign Jun 12 08:11:36 oh no Jun 12 08:11:41 right Jun 12 08:11:45 made sense Jun 12 08:11:52 I am getting drunken anymore Jun 12 08:11:59 heh Jun 12 08:41:40 well gcc 3.4.4 built against glibc 2.3.5 - a very good sign Jun 12 08:41:47 (for armeb) Jun 12 08:44:22 hm Jun 12 08:44:37 maybee trying nptl nexttime Jun 12 08:57:18 pb_: is the ARM ABI documented at arm.org? Jun 12 09:07:04 pb_, OK I booted the glibc 2.3.5 image on armeb, and it looks like it's missing the -fp patch Jun 12 09:07:21 it's behaving exactly like 2.3.2 did before the patch Jun 12 09:10:20 for example, bb top shoes: Jun 12 09:10:38 Load average: 0.00, -88326953556570877642275462878422061082632219594780949187754761537640248521182567140181047418108358710352477575822170393005911236564509685439907111763935233071588777251203231054004997849088.00, -0.00 Jun 12 09:13:09 (I meant the glibc-fp-byteorder.patch ) Jun 12 09:16:48 I'll see if I can get it to apply to 2.3.5 Jun 12 09:26:54 zecke|assignment: yes Jun 12 09:27:00 er, arm.com, not arm.org, but yes Jun 12 09:27:06 jacques: ah, right Jun 12 09:27:23 Is there some howto for creating your own .bb packages? Jun 12 09:27:42 ascent: 2400+ examples Jun 12 09:29:33 pb_, I just modified that patch (they moved arm into ports) and am building an image with it. Jun 12 09:29:41 great Jun 12 09:30:00 it does apply BTW Jun 12 09:30:37 righto Jun 12 09:32:40 if it works, I'll check that in to the cvs tree. Jun 12 09:33:28 pb_, you'll be the first to know :-) Jun 12 09:34:01 :-) Jun 12 09:45:32 AttributeError: EventException instance has no attribute 'args' Jun 12 09:45:39 How can I prevent that ? Jun 12 09:45:57 I modified opie-image.bb to my own, included hostap-utils, and it failed with this error Jun 12 09:46:31 it also says something about dependencies, but they are all in the deploy/ipk dir already Jun 12 10:19:34 ascent: is there a bitbake file and line number associated with the "no attribute args"? that's a bug in bitbake, not your own .bb file. (Although it might have been something weird in your .bb file that triggered it.) Jun 12 10:20:05 It seems I forgot to set some variable.. it seems to compile now. Jun 12 10:20:34 I don't really get it, but I took opie-image, and I put my packages I wanted in it in DEPENDs and RDEPENDS, but it needed to be in IPKG_INSTALL too.. or something.. :) Jun 12 10:21:09 ah, alsa is still broken Jun 12 10:22:26 ERROR: Cannot satisfy the following dependencies for alsa-utils: Jun 12 10:22:26 ${ALSA_PACKAGES} Jun 12 10:28:52 alsa-tools broken too.. Jun 12 10:29:03 patches welcome Jun 12 10:29:27 heheh :) Jun 12 10:29:28 nice try! Jun 12 10:30:05 you could wave a copy of the gpl in front of the people with a working image Jun 12 10:30:25 ehm? Jun 12 10:30:50 you are obliged to document your buildsystem too Jun 12 10:30:55 :) Jun 12 10:31:09 Well, with packages broken, my image won't be finished I guess :) Jun 12 10:31:15 I'll try to compile it later. Jun 12 11:30:55 whee Jun 12 11:30:57 got alsa patched Jun 12 11:35:21 wheee I got a gpe image Jun 12 11:35:28 I got opie! :) Jun 12 11:35:38 With some modifications Jun 12 11:35:59 Where can I submit a patch? Jun 12 11:36:11 Or, is there anyone who can tell me if my patch makes sense? Jun 12 11:48:01 Ciao all Jun 12 12:14:36 patches -> bugzilla || oe@handhelds.org Jun 12 13:37:41 03pb 07 * r1.3561 10openembedded/packages/xfonts/ (xfonts-xorg_6.8.bb lexer.patch): fix up xfonts-xorg build breakage Jun 12 14:10:15 Night Jun 12 14:10:26 'night Sirfred|movie Jun 12 14:27:44 hi. can someone explain the meaning of this python line? Jun 12 14:28:10 DEPENDS_prepend="${@["ipkg-utils-native ", ""][(bb.data.getVar('PACKAGES', d, 1) == '')]}" Jun 12 14:29:11 something similar seems to do what i need in another context Jun 12 14:29:17 but i dont understand it Jun 12 14:30:49 consider this Jun 12 14:30:56 a = True Jun 12 14:31:00 b = False Jun 12 14:31:12 c = ["False", "True"] Jun 12 14:31:21 now c[a] returns "True" Jun 12 14:31:24 and c[b] returns "False" Jun 12 14:31:32 because True and False evaluate to 0 and 1 Jun 12 14:31:46 this is a concise way to emulate the ? expression in C Jun 12 14:31:54 actually, it's an operator Jun 12 14:32:40 ah so if PACKAGES is empty then "ipkg-utils-native"? Jun 12 14:32:40 we could write the same with an 'if' expression, but for ${@ multi line statements suck a bit Jun 12 14:32:49 right Jun 12 14:32:57 no Jun 12 14:33:00 the other way round Jun 12 14:33:09 if packages is not empty, then we depend on ipkg-utils-native Jun 12 14:33:23 ok, other way round Jun 12 14:33:24 (of course, because we need it to package things up) Jun 12 14:33:58 strange syntax ;) but i dont know anything about python... Jun 12 14:34:12 normally i wouldn't write things like that Jun 12 14:34:14 mickey|bbl: hey Jun 12 14:34:26 but as i've said, it's because multiline statments suck in ${@ Jun 12 14:34:27 g'night Jun 12 14:34:33 hi zecke and c'ya ;) Jun 12 14:34:37 'night mickey|bbl Jun 12 14:35:03 thanks mickey Jun 12 14:35:52 'night mickey|zzZZzz Jun 12 14:36:10 03mickeyl 07 * r1.3556.1.1 10openembedded/packages/python/python-pygtk_2.2.0.bb: pygtk rdepends on python-shell Jun 12 14:36:20 ooooh Jun 12 14:36:26 sleep commiting Jun 12 14:36:31 koen: I've done some testing and I think my nwfpe patch is ok Jun 12 14:36:52 RP: very cool Jun 12 14:36:53 koen: The glibc math tests don't pass but then they don't before the patch ;-) Jun 12 14:37:51 :) Jun 12 14:37:52 koen: The one thing that does happen is more exceptions get raised as some were being ignored before. This is probably a feature though... Jun 12 14:38:33 "Thanks Richard." Jun 12 14:38:52 koen: I know :-/ Jun 12 14:38:55 RMK must really like you Jun 12 14:39:03 I'll post that I think its sane and put it into the patch system... Jun 12 14:39:30 Its not just me who things rmk is being a little funny then? :) Jun 12 14:39:32 * koen gets his soldering iron and puts in a fpu Jun 12 14:40:09 RP: check the war between france and RMK about serial port naming Jun 12 14:40:28 koen: I've seen it before... Jun 12 14:40:32 scary :) Jun 12 14:40:38 yeah Jun 12 14:40:54 at least now I know who to blame for ttySA0 Jun 12 14:41:52 All the serial ports should share the same device space and there should be a function sharing the device numbers out IMO Jun 12 14:42:52 At least we never managed ttyPXA0 :) Jun 12 14:42:55 Maybe I should suggest that... Jun 12 14:43:00 NOOOOOOOO!!!!! Jun 12 14:43:05 hehe :) Jun 12 14:43:22 or should that be bwahahahaha :) Jun 12 14:43:55 now that's scary Jun 12 14:45:58 * koen wonders who came up the the "advanced risc machine" which doesn't even have an fpu Jun 12 14:46:25 koen: do you know von Neumann? Jun 12 14:46:40 not personally Jun 12 14:46:52 koen: An fpu kind of goes against the idea of risc Jun 12 14:46:54 koen: "floating point math is for people too stupid to move the ','" Jun 12 14:46:54 but it would have been nice Jun 12 14:47:02 heh Jun 12 14:47:24 What worries me is the amount of floating point calls the system makes. What on earth is it doing to need to make so many of them... Jun 12 14:48:22 I thought it might make a handful of them. I had to turn off logging as the system was never going to boot with it enabled.... Jun 12 14:56:21 ok, cu, good night Jun 12 14:59:09 RP: ouch Jun 12 15:01:48 night Jun 12 15:29:09 'night all Jun 12 16:38:32 hi Jun 12 16:38:47 this may be a bit lame, but i'm asking... is anyone awake? Jun 12 16:39:07 yes Jun 12 16:39:14 :) Jun 12 16:39:23 :) Jun 12 16:39:51 i have been pulling my hair out, very frustrated, i want to sleuth whether or not my SL-C3000 is broken or not Jun 12 16:40:03 symptomes are no working usb host mode Jun 12 16:40:50 That sucks :( Jun 12 17:01:14 usbd mode works, i can connect the zaurus (cacko 3kb1 rom) to a computer Jun 12 17:01:29 kared: what sort of development are you interested in? Jun 12 17:02:21 at the moment i am just setting up a home server on a linksys nslu2, moving services off another box in the hope to decomission it and save some power Jun 12 17:11:52 but interested in just embedded programming in general, I write poker machine games for Konami Jun 12 17:20:51 kawaii Jun 12 18:28:32 kared: i did a similar thing with my wrt54gs Jun 12 18:28:46 it uses very little power now, i am happy for that Jun 12 22:06:02 any collie hackers awake? Jun 12 22:07:33 I build a custom image and my CF works, but my SD is not recognized. I have the binary module in there, what else do I need? I can mount the card myself and it works fine, but it doesn't mount on insert... Jun 12 23:06:42 hi all **** ENDING LOGGING AT Sun Jun 12 23:59:57 2005