**** BEGIN LOGGING AT Sun Feb 17 02:59:57 2013 Feb 17 18:05:44 does anyone know how to enable realtek 8187 in the kernel? Feb 17 18:05:55 I mean 8192 Feb 17 18:06:12 icemetal: what is the problem? which kernel and arch? Feb 17 18:06:24 2.6.39.3 Feb 17 18:06:27 imx Feb 17 18:06:29 arm Feb 17 18:06:45 otavio salvator? Feb 17 18:06:54 yes Feb 17 18:06:59 icemetal: nice to meet you Feb 17 18:07:03 icemetal: who are you? Feb 17 18:07:31 Andre, seen your name all over the yocto project :) Feb 17 18:07:48 nice too meet you too Feb 17 18:08:18 I have the netgear usb wifi adapter, but I can not find it the kernel Feb 17 18:08:18 icemetal: eheh I try to be helpful :-) But Yocto is plent of nice people. I am just another one who works with it :) Feb 17 18:08:36 you been I follow your guides Feb 17 18:08:40 very helpful Feb 17 18:08:54 icemetal: arm; but which arm? Feb 17 18:08:58 imx233 Feb 17 18:09:13 icemetal: use the 3.7 patch we have in meta-fsl-arm Feb 17 18:09:15 icemetal: easier Feb 17 18:09:36 icemetal: 2.6.35.3 does not provide the support for this chip Feb 17 18:09:38 I did downloaded 3.7.x Feb 17 18:10:07 icemetal: there's also a branch based on upcoming 3.8 for testing Feb 17 18:10:52 thanks, I will do that, but still looks to enable PCI Feb 17 18:11:11 is one of the dependency Feb 17 18:11:21 that's what I been trying to do Feb 17 18:13:27 you mean there is a patch on git://github.com/Freescale/meta-fsl-arm.git? Feb 17 18:13:31 icemetal: you maybe be choosing the wrong driver Feb 17 18:13:48 icemetal: what build system you're using ? (I hope Yocto) Feb 17 18:13:54 yes yocto Feb 17 18:14:00 follow your guide Feb 17 18:14:01 icemetal: this or in git.yoctoproject.org Feb 17 18:14:05 RP: Have you used perf recently? Feb 17 18:14:11 no I have to look there Feb 17 18:14:19 icemetal: github is a mirror Feb 17 18:15:01 I think is outdated Feb 17 18:15:21 last update on gihub was a month ago Feb 17 18:15:36 I just cloned about 3 days ago Feb 17 18:15:40 and that's the kernel I got Feb 17 18:15:50 2.6.39 Feb 17 18:15:56 depends on the branch Feb 17 18:16:06 danny yes, master yestarday I think Feb 17 18:16:17 I run this command MACHINE=imx233-olinuxino-micro . ./setup-environment build Feb 17 18:16:51 icemetal: so you're using the fsl-community-bsp. Good Feb 17 18:16:57 icemetal: so do: repo sync Feb 17 18:17:10 ok Feb 17 18:18:03 which branch you're using danny or master? Feb 17 18:18:22 denzil Feb 17 18:18:35 i think that's my problem Feb 17 18:19:19 icemetal: move to danny ;-) Feb 17 18:19:28 ok :) Feb 17 18:19:40 is that why I have an older kernel? Feb 17 18:22:46 icemetal: well Feb 17 18:23:11 icemetal: FSL official kernel is 2.6.35.3; so you need to setup Yocto to use the mainline one. Feb 17 18:23:34 icemetal: but it is quite easy to do. Feb 17 18:24:39 you mean to switch the the newest kernel is supported on FSL? Feb 17 18:25:07 but either way, when I downloaded the newest kernel from kernel.org Feb 17 18:25:42 I was not able to enable rtl8192 Feb 17 18:25:54 cause of the PCI dependency Feb 17 18:30:44 icemetal: if you use danny you can build linux-3.7 from it Feb 17 18:31:00 yes you are right Feb 17 18:31:38 icemetal: than you use -c menuconfig Feb 17 18:32:16 got it, but if you type rtl8192 is not easy to enable it Feb 17 18:38:55 icemetal: you're right; you won't be able to setup rtlwifi as it depends on pci Feb 17 18:39:03 icemetal: I checked the source Feb 17 18:39:49 I don't understand why since is a USB wifi module Feb 17 18:40:28 this USB wifi module is very popular cause is very small and some people got it working on the beagle board Feb 17 18:40:37 icemetal: rtlwifi depends on it; git might be a bug Feb 17 18:41:03 http://embeddedgeeks.wordpress.com/2012/01/03/belkin-micro-wifi-usb-dongle-on-the-beagleboard/ Feb 17 18:41:44 oh ok, I though I was something I was doing wrong cause I couldn't figure out how other people got it to enable it so easy Feb 17 18:41:56 * davest -> airport for week of conferences Feb 17 19:11:39 Well, I managed to kill a machine fairly badly trying to load-test the patches, but once I got it back, the pseudo patches seem to be working roughly as hoped. Feb 17 19:12:21 Time on a quad-core machine with 12GB of RAM for three simultaneous 4x4 builds of one of our filesystems was ~223 minutes pre-patch, ~153 post-patch. Feb 17 19:13:09 Seems that the difference is larger the heavier the load is, and also larger the less compiling is needed. Neither of these surprises me. Feb 17 19:21:42 btw, there's at least one bug in the code, (uninitialized variable) which may cause an occasional useless message to get sent, and there'll need to be a fix for Darwin because they don't have an off64_t. Feb 17 21:35:56 hello guys, i've generated the qt toolchain's through poky but can't compile my project... my qt warns me that "ABI detection failure" Feb 17 21:39:49 never hear of abi before Feb 17 21:41:01 icemetal abi is application binary interface Feb 17 21:41:08 icemetal something to do with linkers Feb 17 21:45:34 oh ok, thanks I'm just a newbie :) Feb 17 21:45:37 seebs: the autobuilder didn't seem to highlight any issue Feb 17 21:48:46 The big thing I'm slightly worried about is losing permissions or modes. Feb 17 21:49:54 Fun fact: My first pass of implementation work for pseudo didn't track set*id(), because it assumed everything was using chown(). Feb 17 21:50:12 This did not show up in testing because rpm2cpio ignores UIDs and just reports everything as root:root. Feb 17 21:53:25 Anyway, I did find an actual bug (the existing_entry flag for unlinkat was uninitialized if there was no entry, so it might or might not work). Running some more tests. Feb 17 21:53:58 Any numbers on how performance changes? Note also: Unless you add --enable-force-async to configure, the fsync() stuff isn't done. Feb 17 21:58:11 seebs: no numbers, the autobuilder doesn't really do that atm Feb 17 21:58:17 k Feb 17 21:58:30 seebs: was useful from a stressing it perspective - our usual image builds continue to work Feb 17 21:59:04 I think some of our build farms do, I'll see whether I can get a patch out for submission. Probably, in this case, want to add the new pseudo but not remove the 1.4.5 package right away. **** ENDING LOGGING AT Mon Feb 18 02:59:58 2013