**** BEGIN LOGGING AT Wed Dec 14 02:59:57 2011 Dec 14 03:04:04 FWIW, keyboard repeat on tty is fine on an external USB keyboard Dec 14 03:04:14 So it must be something in the kb controller Dec 14 03:05:38 Feels weird to have a full-width keyboard for the first time in years Dec 14 03:05:54 heh I'd believe it Dec 14 03:06:01 the EC driver is hacky as shit Dec 14 03:06:11 EC? Dec 14 03:06:40 event controller Dec 14 03:07:06 controlls the keyboard and mouse Dec 14 03:08:17 So like evdev Dec 14 03:09:07 no Dec 14 03:09:07 cause it is hardware Dec 14 03:09:25 Righto Dec 14 03:10:08 the asusec is the controller for the keyboard and mouse Dec 14 03:10:15 Ah, that fucker Dec 14 03:10:30 and a few other little things Dec 14 03:10:37 Quite often in the previous install it would spew printks so fast I couldn't do anything but hard-reboot Dec 14 03:12:53 I was wondering what asusec was for. Dec 14 03:12:58 Nice to know. Dec 14 03:13:31 the asusec does a little more Dec 14 03:13:40 but that is the main thing Dec 14 03:19:39 yup Dec 14 03:20:43 Does tf101 have a hardware AES module? And if so, does ubuntu leverage it? Dec 14 03:21:03 Some of my programs (e.g. mutt) seem to be hanging a lot, and my first guess is they're churning TLS Dec 14 03:21:55 Hm, or it could just be that python is really crap Dec 14 03:21:59 10602 twb 20 0 8800 3568 1680 R 2 0.5 0:00.09 python -c import netrc; print netrc.netrc().authenticators('imap.cyber.com.au')[0] Dec 14 03:22:46 yes, and not really at this point I guess Dec 14 03:22:50 never tried Dec 14 03:23:06 but there is hardware aes Dec 14 03:24:57 Since bt isn't working atm I'm doing this: update-rc.d bsp-tf101 disable Dec 14 03:25:03 So at least I can halt reliably for now Dec 14 03:25:17 heh Dec 14 03:25:24 fair enough Dec 14 03:25:35 the other stuff in the bsp is for CrOS kernel so it wont affect you Dec 14 03:25:42 So next week when I come in yelling that it isn't starting, that is why ;-) Dec 14 03:26:12 haha ok Dec 14 04:20:54 lilstevie: is there a way to detect (programmatically) if the dock is attached? Dec 14 04:24:40 I'd assume the extra devices show up in /sys somewhere? Dec 14 04:24:56 http://paste.ubuntu.com/769703/ little /etc/cron.hourly/battery job I wrote to get a feel for battery consumption rate Dec 14 04:25:14 infinity: unfortunately no, they still appear and if you cat them it complains to all the login ttys Dec 14 04:25:33 infinity: so right now I think that cron job will print gank all over my screen occasionally Dec 14 04:26:18 (I also have a battery monitor in .screenrc, but that's not logging, so I can't look at the rate-of-change.) Dec 14 04:29:06 tegra-i2c tegra-i2c.3: no acknowledge; tps6586x 4-0034: failed reading at 0x20; Failed to set dvfs regulator vdd_cpu; Failed to set regulator vdd_cpu for clock cpu to 1000 mV; cpu-tegra: Failed to set cpu frequency to 1000000 kHz Dec 14 04:29:19 ^^ any idea whart those dmesg's are about? It sounds like it's failing to save my battery Dec 14 04:34:39 lilstevie: btw why was apmd installed instead of acpid? Dec 14 04:34:49 lilstevie: is apmd an OMAPism? Dec 14 04:35:08 And more importantly, is suspend-to-RAM working yet? Dec 14 04:48:19 there is no acpi on arm Dec 14 04:48:30 and no suspend functions do not work Dec 14 04:48:52 well more correctly waking up from suspend functions does not work properly Dec 14 04:57:12 lilstevie: well, acpi_listen can see hardware keys being pressed Dec 14 04:57:20 That's how I'm triggering shutdowns right now Dec 14 04:57:46 (re suspend -- OK, thanks.) Dec 14 04:58:51 lilstevie: Hrm, suspend is working on the ac100. Perhaps some bits need merging between trees? Dec 14 04:59:07 twb: my point is acpi is not really all there for arm Dec 14 04:59:11 infinity: no Dec 14 04:59:24 infinity: already spoke to ogra et al. with that Dec 14 04:59:32 the ac100 used some nvec stuff for it Dec 14 04:59:37 we don't use nvec Dec 14 04:59:44 Ah. Dec 14 04:59:48 lilstevie: OK, hardware-wise I see your point. From a practical perspective, should I uninstall acpid and install something "better"? Dec 14 05:00:12 and, the issue with suspend is the framebuffer Dec 14 05:00:13 (The goal being to respond to power/volup/voldn and kb media buttons like brightup/brightdn) Dec 14 05:00:29 twb: evdev does that :p Dec 14 05:00:55 I mean without resorting to gnome-power-manager or kpowerwhatever Dec 14 05:01:13 oh i see Dec 14 05:01:17 In the x86 world you just install acpid and acpi-support, but for some reason oneiric doesn't even seem to *have* an acpi-support package Dec 14 05:01:19 well that is a bit more difficult Dec 14 05:01:35 and yes, that is because we don't have acpi tables Dec 14 05:01:39 lilstevie: right now acpid is WORKING, which is partly why I'm confused. Dec 14 05:01:43 as in, not at all Dec 14 05:01:59 we have a kernel maintained device tree Dec 14 05:02:05 which will fill in a few of the blanks Dec 14 05:02:09 but not enough Dec 14 05:02:38 twb@elba[Desktop]$ config.guess ==> armv7l-unknown-linux-gnu Dec 14 05:02:38 twb@elba[Desktop]$ acpi_listen ==> video/brightnessup BRTUP 00000086 00000000 Dec 14 05:02:43 infinity: anyway, after suspend, the framebuffer keeps getting suspended Dec 14 05:03:46 infinity: under CrOS_2.6.38 it is worse :p it doesn't even wake up Dec 14 05:04:52 and twb right, there are some standard features which will Dec 14 05:05:21 Ah, OK Dec 14 05:05:23 Does android actually suspend the tf though when you leave it or press the power button? Dec 14 05:05:26 So the kernel is basically faking it Dec 14 05:05:35 TheMuso: yes Dec 14 05:05:48 Ah ok. Dec 14 05:05:54 Very responsive in coming back then. Dec 14 05:06:15 TheMuso: the difference is minimal Dec 14 05:06:33 sometimes when you wake it up it is instant, that did not go all the way to LP0 Dec 14 05:06:51 when there is about 1-1.5sec lag, that is waking up from WB0/LP0 Dec 14 05:08:13 ah ok. Dec 14 05:09:03 Are they registers or what? Dec 14 05:09:42 Hmm, http://www.faqs.org/patents/app/20090204837 ? Dec 14 05:10:37 kinda Dec 14 05:10:42 PMIC Dec 14 05:10:57 which has control registers Dec 14 05:12:47 btw, can I use any old mains<-->usb power brick? Dec 14 05:12:53 no Dec 14 05:12:59 Bugger Dec 14 05:13:28 I want a second one to leave in the office rather than carrying it with me every day Dec 14 05:14:28 how typical Dec 14 05:14:34 trying to debug the suspend issue Dec 14 05:14:44 cant trig the damn bug Dec 14 05:15:00 actually Dec 14 05:15:04 thats a good thing Dec 14 05:15:08 lol Dec 14 05:15:16 this is the first time I have done it dockless Dec 14 05:15:27 maybe the bug is in the dock resume :p Dec 14 05:15:28 THat means something. Dec 14 05:15:38 Or something USB. Dec 14 05:15:42 WHich doesn't surprise me. Dec 14 05:15:50 hah Dec 14 05:16:07 What interconnect does the dock use anyway? I just kinda assumed it was power + usb Dec 14 05:16:21 a very complex interconnect Dec 14 05:16:23 plus a few other random pins, of course Dec 14 05:16:38 lilstevie: custom asus one, or custom tegra one? Dec 14 05:16:47 usb+gpio+power+i2c+i2s Dec 14 05:16:58 (As if USB isn't complex enough interconnect on its own :P) Dec 14 05:17:31 the keyboard runs off a combination of GPU+i2c Dec 14 05:17:58 er Dec 14 05:18:00 GPIO Dec 14 05:18:04 heh Dec 14 05:18:37 Oh wow, I thought the keyboard would be USB. Dec 14 05:19:22 the mouse is i2c and i2s Dec 14 05:19:48 Even more weird. Dec 14 05:20:02 well not really Dec 14 05:20:12 mouse as in touchpad? Dec 14 05:20:15 it implements a ps/2 like interface over that Dec 14 05:20:20 yes Dec 14 05:20:20 I guess they wanted to keep the USB free for the USB ports. Dec 14 05:20:36 well technically the USB controller is in the dock ;) Dec 14 05:20:46 Yeah, another UBhub... Dec 14 05:20:49 if I was doing it, I'd have just run the keyboard and touchpad as USB off a UHCI controller in the dock Dec 14 05:20:51 usb hub... Dec 14 05:21:30 the keyboard is based of nvex Dec 14 05:21:33 nvec* Dec 14 05:21:33 How many ports can your typical UHCI controller have soldered onto it? Four? Eight? Dec 14 05:21:52 gpio is the primary interface Dec 14 05:21:56 If it was four you could just have two ports on the dock, plus one for the kb and one for the tp Dec 14 05:23:13 TheMuso: can you tack on to that blueprint that I need to reverse the off state for the lid switch Dec 14 05:23:15 :p Dec 14 05:23:35 and also, right on queue Dec 14 05:23:42 lilstevie: What does it have to do with that daemon Dec 14 05:23:45 dock+suspend == bad times Dec 14 05:23:52 TheMuso: nothing, just it is dock related :p Dec 14 05:24:10 lilstevie: Fair enough. You hsould have write access to the whiteboard of the blueprint... Dec 14 05:24:14 But I can add it. Dec 14 05:24:21 yeah I just CBF :p Dec 14 05:24:32 I really should do a kernel blueprint Dec 14 05:25:02 Yeah that would be good, I would like to know what you have planned. Dec 14 05:25:13 heh :) Dec 14 05:26:00 that kinterruptive stuff also is interconnected with the dock :( Dec 14 05:26:17 no trace of it in my dmesg, connect dock, bam Dec 14 05:26:25 lilstevie: added. Dec 14 05:26:30 ta Dec 14 05:27:07 ok, so will not suspend with the usb cord plugged in :p Dec 14 05:29:14 lol I love it when one gets side tracked from something else because the point of discussion is interesting enough to dig deeper into a problem. :) Dec 14 05:30:41 heh :p Dec 14 05:30:56 well I have been meaning to dig deeper into suspend Dec 14 05:31:13 cause man will that push batt life to extreme levels Dec 14 05:31:21 Yep. Dec 14 05:34:15 f*** it Dec 14 05:34:21 I cant trigger the damn bug again Dec 14 05:34:48 this is annoying, it is what happens every time Dec 14 05:35:13 heh Dec 14 05:35:19 and when i do trigger it, I am nowhere near a computer that will allow me to remote access it and get detailed logs rather than a 500ms flash Dec 14 05:36:32 holy crap Dec 14 05:36:37 I don't believe it Dec 14 05:36:43 I want to run sleepd Dec 14 05:36:58 plugging in the usb cable, hooked to the computer stops the bug Dec 14 05:37:01 Which basically just does pm-suspend if it hasn't seen keyboard input for ten minutes Dec 14 05:37:05 unplug, bang it is back Dec 14 05:37:28 Actually I should install it now and just configure it to turn off the backlight Dec 14 05:37:40 Since that's like 30% of the power drain Dec 14 05:38:23 lilstevie: weird. Dec 14 05:38:28 Anyway, gotta run. Dec 14 05:38:59 TheMuso: weird is right, plug usb cable in, resume bug goes away Dec 14 05:39:25 unplug it, within 30s bug starts back Dec 14 05:39:30 later TheMuso Dec 14 05:46:41 got it Dec 14 05:47:17 TheMuso: when you are around; figured out why it locks when the usb plug is in Dec 14 05:48:12 TheMuso: when the usb plug is in it prevents suspend. when unplugged it gets into a suspend/wake loop Dec 14 05:48:48 now just need to figure out why wake from suspend goes into a suspend/wake loop **** ENDING LOGGING AT Wed Dec 14 06:20:29 2011 **** BEGIN LOGGING AT Wed Dec 14 06:22:15 2011 Dec 14 06:33:39 mplayer says: mplayer: Symbol `ff_codec_bmp_tags' has different size in shared object, consider re-linking Dec 14 06:34:50 sudo mplayer http://radio1.internode.on.net:8000/117 ==> mplayer: relocation error: mplayer: symbol __aeabi_f2ulz, version LIBAVCODEC_53 not defined in file libavcodec.so.53 with link time reference Dec 14 06:35:04 >sad face< Dec 14 06:40:04 Hmm, that's mplayer1 Dec 14 06:40:09 * twb tries mplayer2 Dec 14 06:43:39 Same issue with mplayer2 Dec 14 06:47:27 ok I am kinda getting a little confused Dec 14 06:47:32 twb: which releasE? Dec 14 06:47:42 also mplayer would be terrible on unaccelerated hw Dec 14 06:47:56 there was a compiler issues which had libav emitting extra symbols on oneiric IIRC Dec 14 06:48:00 I only want audio Dec 14 06:48:07 It's just my scripts are set up to assume mplayer atm Dec 14 06:48:08 micahg: this is on oneiric Dec 14 06:48:30 ah, mplayer might need a rebuild if it's failing Dec 14 06:48:38 I could probably learn xmms2 or something but today ICBF Dec 14 06:48:43 ok, so this is what is happening Dec 14 06:49:27 lilstevie: won't *all* video be sucky without hw accel -- not just mplayer? Dec 14 06:49:29 as soon as the wake lock is removed (by unplugging the usb cable) suspend: sys_sync; suspend: enter suspend Dec 14 06:49:38 twb: actually Dec 14 06:49:49 all video will be sucky even with acceleration Dec 14 06:50:06 Why? Dec 14 06:50:09 until you start doing processing on the AVP Dec 14 06:50:16 cause tegra sucks ass Dec 14 06:50:20 k Dec 14 06:50:46 My simple-minded brain reduces that down to "hardware acceleration" Dec 14 06:51:16 ok, so basically as soon as every last piece of hw has come back from suspend; it calls back to suspend 0.o Dec 14 06:51:18 no idea why Dec 14 06:51:39 twb: it is a little more complex than that, the GPU is weak Dec 14 06:51:40 to save power duh :P Dec 14 06:52:06 twb: yeah but it resumes from the recalled suspend after 0.310sec Dec 14 06:52:09 eh I am coming from matrox and intel Dec 14 06:52:15 ^^measured by the kernel timer Dec 14 06:52:27 twb: no what i mean is the tegra GPU is really weak Dec 14 06:52:36 even weaker than an i915? Dec 14 06:52:37 unless video gets preprocessed by the AVP Dec 14 06:52:37 Wow Dec 14 06:53:05 hum, ok Dec 14 06:53:08 for mpeg vid Dec 14 06:53:31 thats what all those .axf files are in /lib on android Dec 14 06:53:43 they are programs for the avp to aid decoding Dec 14 06:54:08 presumably that's only useful if you're watching video, tho Dec 14 06:54:16 not e.g. playing doom Dec 14 06:55:03 yes Dec 14 06:55:06 correct Dec 14 06:55:26 Righto Dec 14 07:02:01 btw for some reason when I rebooted out of android, the caps lock light was on, and it won't turn off Dec 14 07:02:13 (Which is a surprise to me; I didn't even know there WAS a light.) Dec 14 07:02:29 lol Dec 14 07:16:10 micahg: there's this libavcodec-extra-53 that Provides libavcodec53, is it also likely to be buggered? Dec 14 07:16:28 twb: idk Dec 14 07:16:39 * twb tries Dec 14 07:18:22 answer: it does Dec 14 07:18:37 Er, "libavcodec-extra-53 is also buggered" Dec 14 07:21:51 hm Dec 14 07:22:10 http://pastie.org/3014429 Dec 14 07:23:04 lilstevie: how do you get that out? netconsole? Dec 14 07:23:18 (assume you're still banging against suspend issues.) Dec 14 07:23:48 because once you plug in the USB it triggers wake lock Dec 14 07:24:08 Ah Dec 14 07:24:16 it seems to fully wake up Dec 14 07:24:26 then bam out of nowhere suspends again Dec 14 07:25:01 Does the suspend key have a down and an up event? Dec 14 07:25:01 which is what that portion of the dmesg is Dec 14 07:25:12 it isn't that :p Dec 14 07:25:14 Maybe it suspends before the up even arrives, so it thinks you're still holding it down Dec 14 07:25:16 OK Dec 14 07:25:23 I called suspend from the power menu :) Dec 14 07:25:44 What happens if you throw away the GUI and call it from /sys ? ;-P Dec 14 07:26:10 no idea really :p Dec 14 07:26:34 the only way I know to suspend from /sys isn't in the kernel Dec 14 07:27:48 Might be /proc I forget Dec 14 07:30:32 Something like echo mem > /sys/power/state Dec 14 07:31:12 yeah Dec 14 07:31:14 And if you cat it it tells you someting like "swap mem" to mean it supports suspend to both disk and ram Dec 14 07:31:41 Of course that's JUST suspending, and pm-utils and friends do all sorts of crazy shit in addition to that to work around failtastic hw Dec 14 07:31:48 only supports mem Dec 14 07:32:01 like halting bt and turning all the LEDs off and rmmoding things Dec 14 07:32:39 It wouldn't surprise me in the slightest if one of those things was actually CAUSING the resume problems Dec 14 07:33:06 lol Dec 14 07:33:19 I will try it Dec 14 07:33:20 :) Dec 14 07:33:27 just need to call reboot Dec 14 07:33:35 I bet you get new and excitingly different problems Dec 14 07:33:44 something has to be calling suspend Dec 14 07:33:48 Just remember to turn off the GUI and esp. gnome-power-manager first Dec 14 07:33:55 heh Dec 14 07:34:09 service lightdm stop Dec 14 07:34:12 problem solved Dec 14 07:34:39 You could just "stop lightdm" Dec 14 07:37:09 fwiw identical result Dec 14 07:37:17 well surface is identical Dec 14 07:37:34 will check out the dmesg in 2 secs Dec 14 07:38:21 k Dec 14 07:43:43 yep same result Dec 14 07:44:13 Aw Dec 14 07:44:35 Well at least we know it's not any of the helper code now Dec 14 07:44:43 yeah Dec 14 07:44:51 which I already figured anyway :) Dec 14 07:45:27 the most I can see at the moment is a kernel owned task is calling the suspend Dec 14 07:45:36 as part of the end of waking up from suspend Dec 14 07:45:50 So-rry Dec 14 07:45:56 :p Dec 14 07:46:02 You can just tell me to STFU if I'm not helping Dec 14 07:46:03 worth a try though to confirm Dec 14 07:47:10 STFU every little idea helps Dec 14 07:48:20 maybe... it only happens when you hold the tablet upside down! Dec 14 07:48:40 lol Dec 14 07:49:27 well I may have been misinterpreting the log Dec 14 08:02:30 WTF Dec 14 08:02:43 festival --tts book, half a sentence in it goes "=== PAUSE ===" Dec 14 08:02:48 That has never happened before Dec 14 08:06:39 Goddammit, no bedtime stories for me tonight Dec 14 08:06:56 The actual thing pausing is aplay AFAICT Dec 14 08:09:10 Hmm, why does free -m say I only have 728MB of RAM? Dec 14 08:09:44 Surely it doesn't need >256MB reserved for the GPU or something? :-/ Dec 14 08:09:56 Anyway, home time. Dec 14 10:43:29 <_Thomas> Hmm, I have this usb wifi stick that works with open access points, but when I try to connect to encrypted ones, it fails to connect Dec 14 10:43:46 <_Thomas> (I'm trying to connect using the network wizzard) Dec 14 10:43:56 <_Thomas> erh, Network Manager Dec 14 10:46:48 is that with one of the official ubuntu images from cdimage ? Dec 14 10:48:41 <_Thomas> ogra_: No it's linaro's ubuntu-based image, but it's using most of it's packages from the ubuntu pool Dec 14 10:49:27 that doesnt matter, they are built differently (using metapackages instaed of tasks which can result in different package/dependency selection) Dec 14 10:49:45 so you might miss something Dec 14 10:49:58 do you see any messages in dmesg ? Dec 14 11:00:05 <_Thomas> ogra_: yes, I got some messages in dmesg Dec 14 11:00:17 pastebin them Dec 14 11:00:41 it could also be a missing kernel config option or so Dec 14 11:00:49 or missing firmware etc Dec 14 11:01:06 btw, what HW is that ? Dec 14 11:03:34 <_Thomas> ogra_: http://pastebin.com/Qwrs3a3e Dec 14 11:03:47 <_Thomas> rtl8187 Dec 14 11:04:44 do you see it loading the right firmware (thats not in your dmesg, a full copy would have been better than a snippet) Dec 14 11:04:57 and do you have linux-firmware installed ? Dec 14 11:05:31 <_Thomas> I can yank out the stick, and put it back in again Dec 14 11:05:56 yeah, check if you see anything about firmware and also check if you have that package installed Dec 14 11:06:14 and again, what hardware is that ? Dec 14 11:07:46 <_Thomas> RTL8187, as I said above :) Dec 14 11:08:32 _Thomas: which board are you using? Dec 14 11:08:46 kernel source tree etc Dec 14 11:09:06 right Dec 14 11:09:22 <_Thomas> my own board, kernel source tree is linux-samsung tree Dec 14 11:10:01 ah right. you make the libEGL for that not yet released samsung board? Dec 14 11:10:06 <_Thomas> :) Dec 14 11:11:22 it is the Origen board? Dec 14 11:12:26 <_Thomas> xranby: No, but it's the same AP Dec 14 11:13:09 _Thomas: can you check if the same bug exist on the Origen boarD? Dec 14 11:13:39 * ogra_ would guess its either a missing kernel option or some package missing Dec 14 11:14:25 <_Thomas> xranby: Kind of hard, since I don't have any origen boards Dec 14 11:15:33 <_Thomas> ogra_: Maybe, but I don't see what could be missing from the kernel Dec 14 11:16:42 i dont either since i dont know your kernel ... the ubuntu kernels have a good bunch of common options they all share and that specifically is true for external devices like USB ... Dec 14 11:17:41 (and independent of arch etc) Dec 14 11:18:23 <_Thomas> But could this be USB related, given that the device is detected, and works with open APs? Dec 14 11:18:41 unlikely ... but it could for example be driver related Dec 14 11:18:54 <_Thomas> And with the regards to cfg80211, I've enabled almost everything (only debugging parts not enabled) Dec 14 11:19:07 probably you should :) Dec 14 11:19:26 might get you some more insight (even though it seriously spams dmesg) Dec 14 11:20:22 <_Thomas> with regards to the driver itself, there was no options to choose from Dec 14 11:20:28 <_Thomas> other than mode / integrated / none Dec 14 11:20:32 <_Thomas> mode=module Dec 14 11:20:39 <_Thomas> and I chose to compile as module Dec 14 11:20:40 no, but there might be patches we have in ubuntu and you are missing Dec 14 11:20:52 <_Thomas> that is true Dec 14 11:20:59 <_Thomas> I'm running 3.2-rc1 Dec 14 11:21:00 what kernel version is that ? Dec 14 11:21:08 <_Thomas> fairly recent Dec 14 11:21:12 ah, same we have for panda and omap3 Dec 14 11:21:20 that should do Dec 14 11:21:41 not sure about possible patches though, #ubuntu-kernel might be able to talk about these Dec 14 11:22:15 <_Thomas> I tested the same USB-stick on regular ubuntu 11.04 (x86, that is) Dec 14 11:22:21 <_Thomas> with same issue Dec 14 11:22:31 aha Dec 14 11:22:41 <_Thomas> I had forgot about that until now :D Dec 14 11:25:45 <_Thomas> Should all sticks download firmware? Dec 14 11:25:52 <_Thomas> because there's nothing about firmware in the loggs Dec 14 11:26:01 <_Thomas> not even about failed download/upload or anything Dec 14 11:26:05 it might not report it if its successfull Dec 14 11:26:34 <_Thomas> ok, so it's safe to rule out firmware, then Dec 14 11:26:51 ogra@horus:~$ ls /lib/firmware/rtlwifi/ Dec 14 11:26:51 rtl8192cfw.bin rtl8192cufw.bin rtl8192defw.bin rtl8192sefw.bin rtl8712u.bin Dec 14 11:27:02 thats what i have on my machine in oneiric Dec 14 11:27:22 <_Thomas> no 8187 there Dec 14 11:27:38 no, but often firmware files apply to multiple devices Dec 14 11:28:56 if you actually have the same prob on other HW with the same device i would go to #ubuntu-kernel and ask there Dec 14 11:30:08 oh, and there is bug 802902 Dec 14 11:30:09 Launchpad bug 802902 in linux "Realtek rtl8187b wireless chipset slow speed" [Undecided,Confirmed] https://launchpad.net/bugs/802902 Dec 14 11:30:22 the last comment is intresting Dec 14 11:35:30 <_Thomas> I don't have the b-version of the chipset, it seems Dec 14 12:22:04 doko, just playing with the ac100 you gave me .... as i thought, the kbd cable was only half plugged in Dec 14 12:22:30 armhf feels quite a bit snappier on the desktop i must say Dec 14 12:27:14 ogra_: heh yeah hf is amazing Dec 14 12:27:30 is libreoffice working on hf yet? Dec 14 12:27:37 nope, didnt build yet Dec 14 12:27:43 patches accepted Dec 14 12:27:43 ah ok Dec 14 12:27:46 ;) Dec 14 12:28:01 ogra_, including the mmap fix? Dec 14 12:29:16 doko, no idea if janimo included that in the precise upload from yesterday Dec 14 12:29:32 i'll upgrade to the 3.0 kernel soon Dec 14 12:29:44 janimo, ^^^ ? Dec 14 12:29:47 for now i want to see how much rendering improves by using hf Dec 14 12:30:07 firefox seems to scroll a lot smoother Dec 14 12:39:26 lool, could you care about the flash-installer merge? Dec 14 12:39:55 doko, we are still discussiong it Dec 14 12:40:06 its not clear yet if we will switch at all Dec 14 12:40:08 ahh, ok, so not mine Dec 14 12:40:24 oh, wait you mean adobe flash? Dec 14 12:40:27 or flash-kernel Dec 14 12:42:18 (flash-installer is adobe, no ?) Dec 14 13:34:35 infinity, are you around today or still ill ? Dec 14 13:35:14 * ogra_ is about to roll an nvidia-tegra driver for precise restricted/multiverse and could need a package review in NEW then Dec 14 13:45:29 ogra_: have you obtained a armhf version? Dec 14 13:45:34 no Dec 14 13:45:45 up to nvidia to provide one Dec 14 13:46:04 i dont see aaronp in #ac100 anymore, else i would have asked him Dec 14 14:14:23 armhf rocks :) Dec 14 14:16:00 heh Dec 14 14:16:22 ogra: yes yes i find it nifty Dec 14 14:16:47 now a binary driver would really be lovely Dec 14 14:17:36 ogra: right now im also using 256kb default stack by setting * soft stack 256 in /etc/security/limits.conf Dec 14 14:17:56 and my armhf system ticks on fine without any zram swap tricks Dec 14 14:18:33 just that line ? Dec 14 14:18:36 * soft stack 256 Dec 14 14:18:38 ? Dec 14 14:18:40 yes Dec 14 14:18:46 * ogra adds it Dec 14 14:18:56 anything i need ot restart for it to take effect ? Dec 14 14:19:04 it makes pthread's use 256kb stack by default instead of 8Mb for each thread Dec 14 14:19:21 well, i guess i'll reboot Dec 14 14:24:04 in my humble optinion: since a ubuntu unity2d system + browser open runs around 321 running threads the default pthread stack size do matter Dec 14 14:24:19 ogra: how does things look on your side? Dec 14 14:24:40 what should be the difference with thet security conf setting Dec 14 14:24:51 there is no feelable difference it seems Dec 14 14:25:08 newly started programs threads should use less memory fore each thread by default Dec 14 14:25:37 programs can still opt in for more stack for each thread.. as i have understood it Dec 14 14:25:47 hm. k Dec 14 14:25:57 we'll see over time i guess Dec 14 14:26:47 hmm, actually seems to eat less ram Dec 14 14:26:54 ogra: now ulimit -a should report stack size 256kb Dec 14 14:27:00 yes, thats the idea Dec 14 14:27:09 programs start to use less ram automagically Dec 14 14:27:10 it does indeed Dec 14 14:27:32 i wonder what the implications would be if we set it by default Dec 14 14:29:33 dont set it too low. if the default are 64kb then unity will not work around 128kb jsome woud have trouble unzipping stuff 256 looks fine for most applications that i have tested Dec 14 14:30:05 k Dec 14 14:30:23 well, what i mean is ... there doesnt seem to be any default value in the file Dec 14 14:30:37 and i wonder if there is a reason for this Dec 14 14:31:00 the default have increased during the last years, i think the default are set in eglibc source Dec 14 14:31:30 ah, not a kernel thing then Dec 14 14:31:31 some years ago the default thread size on x86 was 2Mb now its 8Mb for most architectures Dec 14 14:31:36 ogra, xranby: does changing the stack limit make much of a difference in practice? I though the memory wasn't really allocated in advance... this just controls how much VMA space is reserved for stack growth Dec 14 14:31:41 doko, any thoughts on the above Dec 14 14:32:02 dmart, well, htop shows a *lot* less ram used Dec 14 14:32:25 with the stuff i have open atm i would usually hit 400M or so Dec 14 14:32:34 and it only reports about 280 Dec 14 14:33:08 Interesting. What is that actually measuring, though? Dec 14 14:33:24 htop ... i think it sums up RES Dec 14 14:33:47 not sure though Dec 14 14:34:47