| 00:00 | sickness | I've never even heard of the hi10p format =_) |
| 00:01 | newkind | it's mainly used for the anime right now |
| 00:02 | newkind | it's a 10 bit h.264 which needs some power to decode |
| 00:02 | newkind | as there are no hardware decoders existing for it |
| 00:02 | newkind | http://haruhichan.com/wpblog/205/hi10p-info-guide/ |
| 00:08 | jmontle | 00:08 * jmontleo loves distcc |
| 04:05 | billytwowilly | anybody noticing some audio sync issues on the new cubox-i? |
| 05:59 | MikeSeth | billytwowilly: you probably should post on forums |
| 06:13 | nikitis | Hello, how can I tell the status of my order? |
| 06:13 | nikitis | I got an email a month ago stating my i4 was possibly coming soon, but haven't heard anything since. |
| 06:18 | MikeSeth | nikitis: did it say what the estimated date was? |
| 06:18 | MikeSeth | mine was shipped a day before that date |
| 06:20 | jnettlet | nikitis, your best bet is to write to the customer support email on the website |
| 06:20 | nikitis | all i got was an email on Dec 8th, saying "The CuBox-i2Ultra and CuBox-i4Pro are, unfortunately, not shipping yet." |
| 06:21 | MikeSeth | yeah, email support |
| 06:21 | nikitis | So you guys received yours already? |
| 06:21 | MikeSeth | I received an i1 |
| 06:22 | MikeSeth | jnettlet: I just ran glxgears and it causes screen corruption |
| 06:22 | nikitis | oh wait found another email |
| 06:22 | nikitis | January 15th |
| 06:22 | MikeSeth | at first it just flickers and then the screen goes blank for several seconds |
| 06:23 | MikeSeth | this doesn't happen to anybody else right? |
| 06:23 | jnettlet | MikeSeth, possibly. I am still wrapping up work that has been done to stabilize the vivante kernel driver. But it is a binary blob so there is only so much we can do. |
| 06:23 | jnettlet | glxgears on my HB1 runs faster than e2sgears :-) |
| 06:23 | MikeSeth | jnettlet: I think it may have to do with my highly conservative memory allocation |
| 06:24 | MikeSeth | and, it goes away as soon as the window isn't visible anymore |
| 06:24 | jnettlet | check dmesg are you getting IPU errors? |
| 06:24 | MikeSeth | imx-ipuv3 imx-ipuv3.0: IPU Warning - IPU_INT_STAT_10 = 0x00080000 |
| 06:24 | MikeSeth | just one |
| 06:25 | nikitis | So for those who have theirs, have you tried to stream 1080p videos through it? how well does it work? |
| 06:25 | jnettlet | nikitis, which model did you order? |
| 06:25 | nikitis | i4 |
| 06:25 | nikitis | the beefiest one |
| 06:25 | jnettlet | no problems |
| 06:25 | nikitis | excellent |
| 06:25 | jnettlet | and it will only get better |
| 06:26 | nikitis | I'm big into streaming, but currently my setup uses too much power |
| 06:26 | nikitis | i'm hoping my i4 will replace with much lower power usage |
| 06:26 | MikeSeth | uh, it has like 3w peak draw? |
| 06:26 | nikitis | i have XBMC on an original xbox |
| 06:27 | nikitis | but it cannot stream 1080p |
| 06:27 | nikitis | only 420 |
| 06:27 | jnettlet | and uses way more power |
| 06:27 | MikeSeth | you can still use the xbox as a coffee table though |
| 06:27 | nikitis | i use a modified PS3 now |
| 06:27 | nikitis | i'm one of the original PS3 hackers who put Linux back on |
| 06:27 | jnettlet | nice |
| 06:28 | nikitis | and wrote debian / ubuntu installers for the PS3 |
| 06:28 | nikitis | https://github.com/nikitis/PS3-Debian-Installer |
| 06:29 | nikitis | i may write installers for i4 |
| 06:29 | nikitis | if it's needed |
| 06:29 | nikitis | i don't know what's needed atm cause i don't have mine, and haven't been keeping up |
| 06:29 | jnettlet | we can always use more help with distro installers |
| 06:30 | nikitis | yeah, i write mine by hand |
| 06:30 | nikitis | i tried to insert into repos, but ubuntu and debian both wouldn't accept PS3 installers |
| 06:30 | nikitis | Something about doing it all on my own without their help |
| 06:31 | MikeSeth | nikitis: my understanding is that you take an armhf root fs, put on a vivante-enabled kernel, add vivante userland libraries, then make an image of the fs to be writte to sd, and patch it with uboot |
| 06:31 | jnettlet | yeah well Sony weren't being super friendly to people trying to change their locked down firmware |
| 06:31 | MikeSeth | you don't install OSes 'on' cubox because there's nothing to install on in it |
| 06:32 | nikitis | MikeSeth: i'd probably try to figure out a way to install petitboot first, then port my installers for PS3 |
| 06:32 | nikitis | compile petitboot |
| 06:32 | nikitis | custom for i4 |
| 06:33 | nikitis | that gives me a basic shell with kernel, i could load vivante-kernel |
| 06:33 | nikitis | then you can run my scripts |
| 06:33 | nikitis | which install distro's |
| 06:33 | jnettlet | well vivante can be a separate module |
| 06:33 | jnettlet | it doesn't have to be built into the kernel |
| 06:34 | nikitis | with PS3, we had special compiled Linux kernel |
| 06:34 | nikitis | we grabbed basic default kernel src from kernel.org |
| 06:34 | MikeSeth | nikitis: I dunno if this is useful because you need at least one box with an mmc reader anyway |
| 06:34 | nikitis | then compiled PS3 drivers into it |
| 06:34 | MikeSeth | I mean beside the cubox |
| 06:34 | jnettlet | it used to be more difficult in older kernels like 3.0.x kernels due to memory management and allocation. With 3.10+ kernels CMA removes that problem mostly |
| 06:35 | jnettlet | In a perfect world we would start moving the FSL specific drivers into dkms modules so they can work on any upstream kernel. |
| 06:35 | nikitis | so I could probably grab basic default kernel from kernel.org, and compile vivante-kernel modules into it, then compile petitboot |
| 06:35 | jnettle | 06:35 * jnettlet has high hopes and we are far off from that point |
| 06:35 | MikeSeth | jnettlet: isn't there open source drivers being worked on? |
| 06:36 | nikitis | i'm not sure how uboot works |
| 06:36 | nikitis | can it give you a shell? |
| 06:36 | jnettlet | the basic 13.x kernel will boot on cubox-i thanks to _rmk_ but there is still work to be done there. You don't get "full" functionality |
| 06:36 | nikitis | as long as I can get a shell, even the most basic |
| 06:36 | nikitis | i can make a distro install on it |
| 06:36 | jnettlet | MikeSeth, yes the etnaviv project. I am almost to the point where I am declaring the vivante kernel driver "stable" and will shift focus back to the OSS drivers |
| 06:37 | MikeSeth | nikitis: there's a lot you can do through its own shell, like loading or copying images and examining all sorts of hardware information |
| 06:37 | jnettlet | nikitis, you can get that. |
| 06:37 | nikitis | hrm ok |
| 06:37 | MikeSeth | nikitis: if you want to switch to another bootloader the tricky part would be to replicate all the hardware initialization |
| 06:37 | nikitis | with shell are there basic packages? |
| 06:37 | jnettlet | nikitis, beware, you currently need different u-boot installs for the different models. We are hoping to have that fixed soonish |
| 06:38 | nikitis | the reason we used petitboot on PS3 was to get a shell running with very basic packages. 1 being debootstrap, 2, git-core (for cloning installs) |
| 06:39 | MikeSeth | nikitis: but PS3 has internal firmware and would initialize without a boot media, right? |
| 06:39 | nikitis | for some of the hardware yes |
| 06:40 | nikitis | the bus would halt though |
| 06:40 | nikitis | without boot media |
| 06:40 | nikitis | just sit there with green light |
| 06:40 | MikeSeth | my understanding is this isn't the case with imx, the bootloader is expected to initialize voltages, timings, and other hw-specific stuff before the kernel can be loaded, it basically acts like a BIOS |
| 06:41 | jnettlet | blah BIOS blah |
| 06:41 | jnettle | 06:41 * jnettlet is sick of that PC term |
| 06:42 | nikitis | the hard part for me would be to tie in the boot process with a shell that could either a.) compile packages needed, or b.) pre-compile a petitboot with driver to interface with boot process |
| 06:42 | MikeSeth | I was sick of it when Jordaine's book came out and that was like 25 years ago? |
| 06:42 | jnettlet | nikitis, what does petitboot use as a kernel? |
| 06:42 | MikeSeth | IIRC the original edition came with a prinout of PCjr BIOS or whatever |
| 06:42 | jnettle | 06:42 * jnettlet has no knowledge of the project |
| 06:42 | nikitis | it uses a basic linux kernel |
| 06:43 | nikitis | but you choose what packages to add |
| 06:43 | nikitis | when you compile petitboot |
| 06:43 | nikitis | and throw in drivers/modules etc |
| 06:43 | MikeSeth | nikitis: that's basically uboot + minimum rootfs |
| 06:43 | jnettlet | yeah, so you would just need a custom u-boot script that would load your kernel and environment. Easily doable |
| 06:43 | nikitis | does uboot have any functionality? |
| 06:43 | nikitis | or precompiled packages with it |
| 06:44 | MikeSeth | well its task is to initialize the hardware and boot the next step in the chain |
| 06:44 | jnettlet | if you can fit your environment into an initramfs image uboot can load a zImage and initramfs and boot into it |
| 06:44 | nikitis | would like gcc, but is too much to ask |
| 06:44 | MikeSeth | which is, for Linux, the OS kernel |
| 06:44 | nikitis | oh so it uses initramfs |
| 06:44 | nikitis | and zImage |
| 06:45 | nikitis | probably could work with that |
| 06:45 | jnettlet | well the latest does. I hate the old uImage stuff |
| 06:45 | MikeSeth | nikitis: it's a specialized bootloader for a wide array of embedded devices (200+) it has all sorts of tricks in the bag |
| 06:45 | jnettlet | you can netboot as well if you like |
| 06:45 | nikitis | i'm sure this will be doable then |
| 06:45 | nikitis | i'll port my installers over |
| 06:46 | MikeSeth | nikitis: I don't quite understand the strategy tbh. There isn't a benefit I can see in performing installating *on* a cubox |
| 06:46 | MikeSeth | installation* |
| 06:48 | nikitis | but i need an i4 first ;) |
| 06:48 | nikitis | Jan 15th |
| 06:48 | nikitis | THis is custom tailored to PS3, so it may seem a tad sloppy, but it's solid work with no bugs. https://github.com/nikitis/PS3-Debian-Installer/blob/master/debian-installer.sh |
| 06:48 | nikitis | especially the parted work |
| 06:48 | nikitis | but you have to understand the delecate situation the PS3 hdd is |
| 06:49 | MikeSeth | nikitis: but that's the thing, there isn't a storage in cubox, w/o an SD card it's a brick, which is why you literally have a bootstrapping problem |
| 06:49 | nikitis | well you could i suppose just load a distro already setup |
| 06:49 | nikitis | since it's mmc |
| 06:49 | nikitis | just use the SD Card as hdd |
| 06:49 | nikitis | but really installation is just for the first person to get the distro |
| 06:50 | nikitis | once it's setup, you could make an image of the setup |
| 06:50 | MikeSeth | that's more or less what they do today |
| 06:50 | MikeSeth | armhf -> local hacks -> fs image -> uboot patch |
| 06:50 | nikitis | then people can just dd the img onto the card |
| 06:51 | MikeSeth | then again, I am only thinking debian |
| 06:51 | MikeSeth | other OSes don't exactly exist in my world |
| 06:51 | nikitis | debian is a great base |
| 06:51 | nikitis | but lacks high end features |
| 06:51 | nikitis | i use mint |
| 06:52 | nikitis | which is ultimately based off of ubuntu, which is based off of debian |
| 06:52 | nikitis | mostly design features |
| 06:52 | nikitis | preference |
| 06:53 | MikeSeth | yeah, I'm a console nerd, if I have to click on something to get it working, I don't like it |
| 06:53 | nikitis | if given choice i'd use plain debian over RPM |
| 06:54 | nikitis | i'm already working on PS4 |
| 06:54 | MikeSeth | nikitis: btw look at this script by hste |
| 06:54 | MikeSeth | http://stende.no-ip.info/imx6/solidrun-hb/scripts/mk_jessie-lxde_hb.sh |
| 06:54 | MikeSeth | it hacks together a Debian Jessie rootfs |
| 06:54 | nikitis | http://palmettolan.com/pictures/ps4/MX25L25635FMI-10G_Removed.jpg |
| 06:54 | MikeSeth | time to go to work |
| 06:55 | nikitis | i'll check it out |
| 06:55 | MikeSeth | I'll be seeing what needs to be done to package vivante binaries into debs tonight |
| 06:55 | MikeSeth | and I'll prolly regenerate a Jessie image for 4Gb cards w/swap |
| 06:56 | MikeSeth | on 2Gb I ran out of space as soon as I installed emacs and iceweasel, durr |
| 06:56 | nikitis | yeah smallest computer i have now is RaspberryPi |
| 06:56 | nikitis | I'm using it headless now to use irssi to talk to you guys |
| 06:57 | nikitis | it just sucks video wise |
| 06:57 | nikitis | great headless box |
| 07:14 | jnettlet | nikitis, well any of the Cubox-i devices blow the raspberry pi away in performance |
| 07:15 | jnettlet | even the smallest single core with 512 is quite capable |
| 08:46 | jnettlet | okay merged the IMX 3.10.17 tree into linaro 3.10 lsk and the hotplugging is not spewing interrupts |
| 09:01 | hste_ | jnettlet: started work early today? |
| 09:01 | jnettlet | hste_, I like to get up early when it is short days so I can get a few hours in before the sun comes up. |
| 09:02 | hste_ | jnettlet: sun? it has been rainy weather here for many days |
| 09:02 | jnettlet | trying to figure out if the errors I am seeing now is because of mmc problems, or fs problems due to previous mmc problems |
| 09:03 | jnettlet | hste_, we have been off and on. We were going through a pretty dreary spell but have had a couple of nice days. |
| 09:03 | hste_ | jnettlet:: Do u know why I have to cut power to get c1 to boot after a reboot |
| 09:04 | jnettlet | hste_, it is a problem that seems to effect certain c1's. I have had two boards like that, and I guess you have one, but nobody else has seen the problem. |
| 09:04 | jnettlet | we are just lucky I guess :-) |
| 09:05 | jnettlet | hste_, you can wire up pin 7 of the 8 pin board next to the hdmi port and pull that low to cause a hardware reset instead of pulling the power. |
| 09:08 | hste_ | jnettlet:: I was trying stephans xbmc yocto version, but got a lot of those IPU locks when running 1080P and watch some movies. Using a fresh kernel from git |
| 09:09 | jnettlet | hste_, were the movies 1080p? |
| 09:10 | hste_ | jnettlet:: no it was some tv streams. Working ok on my gk802 |
| 09:10 | jnettlet | You need the to tweak the AXI bus QOS to give the IPU the priority on the HB1/C1 with 1080p |
| 09:10 | jnettlet | you probably still need the tweak. |
| 09:10 | jnettlet | but if you don't mind posting a link to the stream I can take a look. |
| 09:11 | hste_ | jnettlet:: it was some stream using tv3play plugins |
| 09:11 | jnettlet | hste_, okay I will install those and test later |
| 09:13 | hste_ | jnettlet:: I used this rootfs http://stende.no-ip.info/imx6/rootfs/3.0.35_4.1.0/xbmc_stephan_2013-12-20.tar.xz |
| 10:03 | dshankar | dv_: are you around? |
| 10:06 | dshankar | dv_: I'm trying to get VPU accelerated html5 video (webm/vp8 via webrtc) in chromium 31 on debian jessie (on gk802/imx6+vivante). google pulled up old IRC logs where you mentioned you had a chromium patch being pushed upstream that enabled this? any more info? |
| 10:14 | jnettlet | okay pulled in some more ext4 patches and things look to have finally stabilized |
| 10:24 | jnettlet | hmmm seeing some artifacting when running two simultaneos glesv2 apps |
| 10:24 | jnettlet | first seems fine, second has artifacting. |
| 10:25 | jnettlet | definitely a bug in the vivante event handling :-( |
| 10:28 | jnettlet | hmmm that time it is okay |
| 10:28 | jnettlet | great transient bugs are always fun to fix |
| 10:53 | jnettlet | rabeeh, ping. I have a question about the usbotg poert |
| 10:53 | jnettlet | port |
| 10:53 | MikeSeth | jnettlet: artifacts within the render output itself right? |
| 10:53 | MikeSeth | not on the general screen? |
| 10:53 | jnettlet | MikeSeth, yep just the render itself |
| 10:56 | jnettlet | it is only very specific orders it happens in. Must be a memory alignment wierdness |
| 10:58 | jnettlet | not bad running glmark2-es2, es2gears, glmark2, glxgears all simultaneously. |
| 10:58 | jnettlet | not fast but not horrible |
| 10:59 | MikeSeth | hah, rabeeh is still slumbering |
| 11:05 | dv_ | dshankar: yes, I am not working at this atm, it was a company project |
| 11:05 | dv_ | might come up later again though. give me your mail address so I can keep you informed |
| 11:10 | jnettlet | dv_, the IPU supports vp8 decoding? |
| 11:10 | jnettlet | make that VPU |
| 11:10 | dv_ | yes |
| 11:14 | jnettlet | oh nice. |
| 11:28 | rabeeh | jnettlet: pong |
| 11:29 | jnettlet | rabeeh, can the usbotg port only run in host mode? |
| 11:30 | rabeeh | it can run both |
| 11:30 | rabeeh | but since the usb connection is only host; you can only connect host ports |
| 11:31 | rabeeh | there is a way to workaround that by faking USB OTG ID pin to make it float (doable in HB) |
| 11:31 | jnettlet | right |
| 11:31 | rabeeh | and then just hook anything you want |
| 11:32 | rabeeh | we use that in the device manufacturing |
| 11:32 | rabeeh | what exactly do you want to do? |
| 11:32 | jnettlet | Nothing debugging the usb ports on the new cubox i4pro and device-tree and making sure I wasn't missing something stupid. |
| 11:33 | rabeeh | ok; so cubox-i4pro is different story than HB |
| 11:33 | jnettlet | It seems as the top port is the standard usb port and the lower port is the usbotg port. |
| 11:33 | rabeeh | the USB OTG ID pin is physically connected to a pull down resistor that the default makes it USB host |
| 11:34 | rabeeh | in you really want to play with it you can change the USB OTG ID pin in the iomux to something unconnected and then just internally pull it up. |
| 11:34 | rabeeh | by internally pull it up i mean define the unconnected to anything new USB OTG ID pin pulled up by the pad itself |
| 11:35 | jnettlet | okay, so really it should not need a special device-tree entry if we just want it to work in the default Host mode |
| 11:36 | rabeeh | for cubox-i no |
| 11:36 | rabeeh | just define the correct USB OTG IT pin as in the 3.0.35 tree |
| 11:36 | jnettlet | yep will double-check that. |
| 11:37 | rabeeh | https://github.com/SolidRun/linux-imx6/blob/imx_3.0.35_4.1.0/arch/arm/mach-mx6/board-mx6q_cubox-i.h#L50 |
| 11:37 | rabeeh | USB OTG ID is connected to GPIO_1 |
| 11:38 | jnettlet | Other than that my linaro-lts tree with the freescale patchset merged and a few more patches is looking quite good. I was able to run 2 gles2 and 2 glx apps simultanenously without any glitching or lockups |
| 11:38 | jnettlet | I just need to rebase on the proper hummingboard-cuboxi patchsets that have gone upstream. |
| 11:39 | hste_ | rabeeh: what bootparameter do you think I should set for default on an uEnv.txt for installer-script? |
| 11:41 | rabeeh | jnettlet: sounds great. i can't wait until we upgrade from this oldie oldie kernel |
| 11:41 | jnettlet | Beware of static, but http://www.youtube.com/watch?v=dIoxg9fXRO8 |
| 11:42 | rabeeh | hste_: |
| 11:42 | rabeeh | # Create the boot uEnv.txt |
| 11:42 | rabeeh | cat > $MNT/boot/uEnv.txt < |
| 11:42 | rabeeh | bootfile=/boot/uImage |
| 11:42 | rabeeh | mmcargs=setenv bootargs root=/dev/mmcblk0p1 rootwait video=mxcfb0:dev=hdmi consoleblank=0 |
| 11:42 | rabeeh | EOF |
| 11:43 | hste_ | thanks |
| 11:43 | rabeeh | i can send you my installation script that takes your images with my kernel and creates an image |
| 11:43 | cbxbiker61_ | jnettlet, that looks good! |
| 11:43 | rabeeh | it creates a 2GByte image file using loopback device; does all the partitioning and spits out an image that is ready to be flashed |
| 11:43 | jnettlet | and even running all of those we got results of score 71 for glmark2-es2 25 FPS es2gears score 62 glmark 45fps for glxgears |
| 11:44 | hste_ | rabeeh: Have some here too http://stende.no-ip.info/imx6/solidrun-hb/scripts/ |
| 11:44 | jnettlet | sorry flip flop es2gears and glxgears scores |
| 11:45 | rabeeh | jnettlet: wow |
| 11:45 | rabeeh | the 64bit DDR made all the difference. |
| 11:46 | rabeeh | this is something very unique in the freescale implementation where most other SoCs stay with 32bit DDR |
| 11:46 | jnettlet | rabeeh definitely helps |
| 11:51 | hste_ | jnettlet: did u look at temp when running all that ? |
| 11:53 | hste_ | jnettlet: I'll try to run that on my gk802 when I get home to check what a normal bad designed imx6q device can do :) |
| 11:53 | MarcusVinter | Haha |
| 11:55 | MarcusVinter | Oh I was thinking of MK802. that things trash. |
| 12:00 | rabeeh | jnettlet: cat /sys/devices/virtual/thermal/thermal_zone0/temp |
| 12:00 | rabeeh | that's to get the die junction temperature |
| 12:01 | rabeeh | jnettlet: but you need to give it some time to warm up |
| 12:01 | rabeeh | in general it takes ~45minutes for the CuBox-i to get into it's max temperature |
| 12:22 | MarcusVinter | Any idea why Debian might boot and then hang on the dekstop after logging in? couldnt even do ctrl+alt+F* |
| 12:23 | cbxbiker61 | did you try to ping the device, or ssh? |
| 12:24 | MarcusVinter | Didnt try ssh. cant ping because dhcp assigned IP, router admin is locked so cant check IP there and dont have nmap installed. One min ill find a way into router. |
| 12:25 | cbxbiker61 | yeah, it's always nice to know if the device is "totally" dead, i.e. no ping, or it's an app lockup |
| 12:27 | hste_ | MarcusVinter: maybe it is frozen. You could try a reboot and do switch to console with ctrl-al-f1 before login in to detect ip |
| 12:27 | MarcusVinter | Oh it's responding to ping. How perculiar. |
| 12:28 | hste_ | MarcusVinter: does it answer on ssh? |
| 12:29 | MarcusVinter | Not on port 22. |
| 12:30 | MarcusVinter | Unless theres something wrong with my HID's but they have always been fine for many devices including this one. |
| 12:51 | MarcusVinter | Seems to have had a problem recognising my HID. I had to unplug it and plug it back in a few times. Strange. |
| 12:51 | MikeSeth | rabeeh: I'm considering an embedded linux course in John Bryce, think it's worth it? |
| 12:52 | MikeSeth | MarcusVinter: there were some complaints on forums that some HIDs misbehave |
| 12:54 | MarcusVinter | Ahh I'll take a look. Thanks Mike. |
| 12:54 | rabeeh | MikeSeth: i'v learned linux and embedded linux from the biggest master - www.google.com :) |
| 12:59 | rabeeh | MarcusVinter: which USB are you using? |
| 12:59 | rabeeh | lsusb please |
| 13:01 | MarcusVinter | PCPlay PS/2 Keyboard+mouse adapter. ID 13ba:0017 |
| 13:01 | jnettlet | rabeeh, I didn't check the temp. But that reminds me that the temp is reporting is broken :-) |
| 13:01 | jnettlet | I looked at xbmc last night and it was reporting thousands of degrees |
| 13:02 | MarcusVinter | My temp is 46 |
| 13:03 | MikeSeth | rabeeh: oh by the way, you mentioned on forums volunteership for vivante .deb packaging, I'll be looking at it tonight |
| 13:05 | hste_ | jnettlet: that sounds a little bit high :) |
| 13:06 | jnettlet | yep right now it is reading 36781 |
| 13:07 | jnettlet | oh I bet it is not being converted. probably should be 36.781 |
| 13:08 | jnettlet | seeing as the outside of the box is 31.1 |
| 13:11 | jnettlet | ah yes it is in millicelsius |
| 13:12 | jnettlet | I guess we need to tell xbmc that our temp reporting is different |
| 13:21 | jnettlet | rabeeh, there is really no heat build up. cpu is really only running at about 40-50% |
| 13:22 | jnettlet | added neverball running in the background and that can't keep up |
| 13:23 | rabeeh | MikeSeth: that would be really helpful |
| 13:23 | MikeSeth | rabeeh: yeah, as soon as I identify all components and their dependencies.. |
| 13:23 | jnettlet | rabeeh, how long is sufficient to get it to warm up? |
| 13:24 | rabeeh | jnettlet: our hammering test showed 35-45 minutes |
| 13:25 | rabeeh | but i wouldn't be surprised if you get low temperatures; that's CuBox-i design |
| 13:25 | MikeSeth | rabeeh: btw I had a strange glitch, not sure what to think of it, first time I botted the jessi image it allocated 256M to the GPU and the calibration showed 800 bogomips; I rebooted with gpumem=64M and now its 2k bogomips or so |
| 13:25 | rabeeh | though 31.1 on the outer of the box is low; which means that you are not hammering the box high enough |
| 13:25 | MikeSeth | dunno what to think of it |
| 13:25 | MikeSeth | the 4.3 android was very slow and apparently dropping keyboard interrupts |
| 13:26 | jnettlet | rabeeh, well right now this test is very GPU bound |
| 13:26 | MikeSeth | at first boot it was a bit similar, jittery IO and 10 second hangs |
| 13:26 | jnettlet | I seem to be hovering at 45C |
| 13:27 | rabeeh | jnettlet: ok. that sounds that you are only hammering the gpu |
| 13:27 | rabeeh | just memterster 100M > /dev/null & few times to keep the other processors busy too |
| 13:27 | rabeeh | and send with the wifi random packets; and it's enough just to keep the ethernet link at 1Gbps |
| 13:29 | MikeSeth | i use ping -f to saturate |
| 13:31 | MarcusVinter | What's a good wifi network manager? network-manager? |
| 13:31 | jnettlet | network-manager if you don't mind the dependencies, wicd if you don't mind a bit of configuration, there is also conman |
| 13:31 | jnettlet | which desktop? |
| 13:31 | MikeSeth | MarcusVinter: if nothing exotic, just write a wpa_supplicant config file manually |
| 13:32 | MarcusVinter | I will be using this image for other users that will use the device at other locations. So really I want the most user friendly version. |
| 13:32 | MarcusVinter | wicd sounds good. I've used that before. |
| 13:33 | MarcusVinter | Although annoyingly you have the specify the encryption type even though it detects them. |
| 13:33 | jnettlet | MarcusVinter, which desktop? |
| 13:33 | jnettlet | wicd incorporates with XFCE the best. |
| 13:34 | MarcusVinter | not sure, it came with the debian image, doesnt look like XFCE but i might be wrong. |
| 13:34 | MikeSeth | MarcusVinter: nm-applet will work |
| 13:34 | MikeSeth | MarcusVinter: the jessi image is xfce, the alternate is lxde |
| 13:35 | jnettlet | hmmm. just got the i4pro to spit these out. imx-ipuv3 2400000.ipu: IPU Warning - IPU_INT_STAT_5 = 0x08800000 |
| 13:35 | jnettlet | imx-ipuv3 2400000.ipu: IPU Warning - IPU_INT_STAT_10 = 0x00080000 |
| 13:35 | MarcusVinter | Ahhh okay. thanks. |
| 13:37 | MikeSeth | jnettlet: 04:56 imx-ipuv3 imx-ipuv3.0: IPU Warning - IPU_INT_STAT_10 = 0x00080000 |
| 13:38 | MikeSeth | when running glxgears with reduced gpu allocation.. which tbh I am no longer sure is there |
| 13:41 | hste_ | MarcusVinter: The lxde version has network manager installed. |
| 13:41 | rabeeh | jnettlet: do you have the ipu to ddr transactions marked with qos=0xf? |
| 13:41 | jnettlet | rabeeh, not for the i4pro |
| 13:41 | rabeeh | devmem 0x020e0018 32 0xffffffff |
| 13:41 | rabeeh | try it |
| 13:42 | jnettlet | okay |
| 13:42 | rabeeh | or - |
| 13:42 | rabeeh | devmem2 0x020e0018 w 0xffffffff |
| 13:42 | rabeeh | MikeSeth: that's the same for you |
| 13:49 | jnettlet | rabeeh, that seems better. |
| 13:49 | jnettlet | should I add that to be configurable by device tree? |
| 13:50 | rabeeh | jnettlet: you can either do it in the kernel (that's what i do in the 3.0.35) or can be permanently added to u-boot |
| 13:51 | jnettlet | good question, is there any time when that configuration is not desirable? Probably a headless situation |
| 13:53 | rabeeh | if the ipu doesn't do any transactions towards the ddr; then this setting shouldn't be affecting anything |
| 13:55 | hste_ | rabeeh: how do u add it in kernel? |
| 13:56 | rabeeh | hste_: https://github.com/SolidRun/linux-imx6/blob/imx_3.0.35_4.1.0/arch/arm/mach-mx6/board-mx6q_microsom.c#L724 |
| 13:57 | MikeSeth | rabeeh: I will, not physically over there |
| 14:50 | MarcusVinter | What the best way to pair bluetooth HIDs? Is there a nice tool for it? |
| 14:53 | jnettlet | Blueman for non-Gnome |
| 15:05 | MarcusVinter | does XFCE run on gnome? |
| 15:06 | jnettlet | XFCE is it's own desktop. It is based on GTK |
| 15:07 | MarcusVinter | Oh... |
| 15:08 | MarcusVinter | Also is there a nice easy way to expand the partition? I have an 8gb sd and i would like to expand the current 4gb partition |
| 15:08 | jnettlet | resize2fs |
| 15:08 | MarcusVinter | thank you. ill man it |
| 15:11 | MarcusVinter | Damn, it says the file system is already blah blah long. |
| 15:12 | jnettlet | MarcusVinter, you need to resize the physical partition first. |
| 15:12 | jnettlet | so use fdisk or gdisk for that depending on the type of partition table you have |
| 15:12 | jnettlet | then make the partition larger on the device, then use resize2fs to grow the filesystem |
| 15:14 | MarcusVinter | Thank you. |
| 15:14 | MarcusVinter | I might remove the sd and just use gparted lol |
| 15:17 | MarcusVinter | There we go, changed from 2gb to 15gb :D I forgot the MMC is 16gb |
| 15:21 | MarcusVinter | If I needed to shrink it down again to dd it to a file to clone it later, is this possible? I dont need to know how, just if its doable. |
| 15:21 | MarcusVinter | Thanks |
| 15:23 | jnettlet | that is a bit harder but you can do it. |
| 15:23 | MarcusVinter | Thanks. |
| 15:24 | MarcusVinter | And does anyone know off the top of their head the wifi chipset? |
| 15:25 | MarcusVinter | I tried lspci but it doesnt work |
| 15:27 | MarcusVinter | looks like a broadcom |
| 15:29 | MarcusVinter | Cool it supports packet injection. |
| 15:37 | MarcusVinter | wlan tx power is 1496 wtf... |
| 15:39 | MarcusVinter | usually 20 dbi |
| 15:49 | jmontleo | MarcusVinter, gparted is a nice graphical way to do all you want if you have another desktop/laptop with a card reader you can use |
| 15:49 | jmontleo | grow, shrink, resize2fs, etc. |
| 15:50 | jnettlet | jmontleo, has your sdhc been stable since that device-tree change was made? |
| 15:50 | MarcusVinter | Yeah its working now. I used gparted on my linux dev pc to resize it nice and easily. |
| 15:51 | jmontleo | jnettlet, I just finished getting a new kernel built with modules and all over night; seems to be working ok so far though |
| 15:51 | jmontle | 15:51 * jmontleo needed/wanted all the iptables/selinux stuff |
| 15:52 | jnettlet | jmontleo, great! just making sure it wasn't an issue that needed additional looking into |
| 15:52 | jmontleo | nope, so far so good :) I change the dtb to use imx6q instead of imx6dl and I have 4 cores as well |
| 15:52 | jnettlet | although I need to figure out why UHS-1 mode isn't working |
| 15:53 | jnettlet | great, and the clocking to work with the faster ddr bandwidth |
| 15:59 | jnettlet | oh yeah looking at the mmc code and the imx driver not setting the max-width is very broken |
| 16:01 | jnettlet | _rmk_, ^^ do you want a patch or should I submit it upstream? |
| 16:03 | jmontleo | thing of beauty: http://ur1.ca/gd9t2 |
| 16:07 | jnettlet | great |
| 16:13 | MarcusVinter | damn, blueman doesnt seem to find the bluetooth device. Do I have to bring it up first somehow? |
| 16:23 | jnettlet | MarcusVinter, is the bluez daemon running? |
| 16:23 | jnettlet | it may be called something different now |
| 16:25 | MarcusVinter | Theres one called bluetooth that i started. |
| 16:30 | jnettlet | and hcitool dev finds the device? |
| 16:39 | jnettlet | jmontleo, actually it is strange device-tree already has bus-width = <4>; set. Wonder why it is not being inherited properly |
| 16:46 | MarcusVinter | It think bluez is not starting because it doesnt detect the bluetooth adapter. |
| 16:48 | jnettlet | MarcusVinter, but does hcitool dev? |
| 16:49 | MarcusVinter | Lists no devices, i tried that :( |
| 16:51 | jnettlet | yeah, so either bluetooth isn't compiled into your kernel, or you are missing the firmware to enable it |
| 16:52 | MarcusVinter | :'( Y gooby. |
| 16:52 | MarcusVinter | okay. |
| 16:52 | MarcusVinter | I can forget bluetooth for now then. USB it shall be. |
| 16:55 | jnettlet | _rmk_, actually don't worry about it. the driver actually needs a patch |
| 16:59 | jmontleo | jnettlet, is that inherited from one of the dts(i) files? I see it set in several of the dtsi files for others, but not in the imx6dl-hummingbird or the other included ones |
| 17:01 | MikeSeth | rabeeh: can I feel the boot process page on the wiki? it's currently empty |
| 17:02 | MikeSeth | s/feel/fill/ |
| 17:36 | xraxor | anyone know what type of Optical cable works on the i4_ro? |
| 17:36 | xraxor | i4Pro |
| 17:48 | jnettlet | jmontleo, yeah it gets inherited |
| 17:48 | rabeeh | xraxor: the cubox-i optical port is a bit inside the box |
| 17:48 | rabeeh | there are cables (that we have on our website) that works as is |
| 17:50 | MikeSeth | uhm, how well used is NEON in debian armhf and linux kernel? |
| 17:51 | MikeSeth | debian wiki says gcc support is poor and NEON optimizations are hacked in manually |
| 17:52 | rabeeh | MikeSeth: the kernel doesn't have any neon code |
| 17:52 | MikeSeth | rabeeh: anything in it that would get a performance bonus if it did? |
| 17:53 | MikeSeth | maybe kernel memory allocation, virtual memory etc, hmm |
| 17:53 | rabeeh | there were discussions few months ago to get neon inside the kernel to perform some btrfs optimizations |
| 17:53 | jnettlet | rabeeh, not in 3.0.35, but I think newer ones have NEON based mem* functions. |
| 17:53 | rabeeh | like for xor |
| 17:54 | jnettlet | or at least there was talk and patches floating around |
| 17:54 | MikeSeth | btw, why are we stuck with 3.0.35? |
| 17:54 | MikeSeth | fsl code drop? |
| 17:54 | hste | jnettlet: |
| 17:54 | rabeeh | MikeSeth: yes - we are bound to freescale releases |
| 17:54 | hste | jnettlet: xbmc worked ok for me on 720P |
| 17:55 | MikeSeth | rabeeh: which are binary only, yes? |
| 17:55 | jnettlet | hste, yeah 720p is no problem for the C1, 1080p is really pushing it |
| 18:07 | MikeSeth | ooh qemu support for armhf |
| 18:07 | MikeSeth | niiiice |
| 18:17 | rabeeh | MikeSeth: you can now your favourite Windows 95 on CuBox-i :) |
| 18:17 | rabeeh | MikeSeth: actually if you can get it to run; please take some pictures and send it through... i'm eagered to see the performance of DOS applications on CuBox-i |
| 18:18 | MarcusVinter | Me too. |
| 18:18 | rabeeh | (like Hashavshivit that you are probably familiar with) |
| 18:19 | jnettlet | rabeeh, we played around with it for some OLPC deployments that had MSVB based educational apps. It wasn't pretty |
| 18:19 | rabeeh | jnettlet: running VB is not nice also on x86 |
| 18:20 | jnettlet | :-) |
| 18:21 | rabeeh | Waseem installed pacman on dos emulation on c1 |
| 18:21 | rabeeh | worked flawless |
| 18:21 | jnettlet | yes but we have GLESv2 acceleration |
| 18:24 | dv | 18:24 * dv_ cant concentrate, damn cold |
| 18:24 | dv_ | yes, there is glesv2 acceleration, which is why I think using stuff like glxgears on the hummingboard and the cubox-i is a mistake |
| 18:25 | dv_ | dont use a desktop compositor with needs glx on these machines. dont use a software stack which relies on glx and desktop GL. |
| 18:27 | jnettlet | dv_, well yes and no. I am very disappointed at how slow distros have been at switching their packages to enable support GLESv2 instead of GL for arm based distros. However there is a lot of software that only support GL, and having some hardware acceleration for that is better than software. |
| 18:27 | jnettlet | but it should not be used as a crutch to fix the problem. |
| 18:29 | dv_ | exactly |
| 18:32 | jnettlet | hahahha I see our the newer kernels are trying to use the -cd pinmux as the 1.8v pin that answers both of my problems |
| 18:33 | jnettlet | jmontleo, ^ |
| 18:34 | MikeSeth | hashavshevet should burn in hell |
| 18:34 | MikeSeth | pure evil |
| 18:34 | rabeeh | MikeSeth: most the accountants around don't think so :) |
| 18:34 | jnettlet | or the bus change for uhs2 is causing the -cd pin to fluctuate |
| 18:34 | MikeSeth | rabeeh: but they dont know anything and in many occasions they're barely even people |
| 18:34 | jnettlet | rabeeh, any input on the uhs1 setup? |
| 18:35 | MikeSet | 18:35 * MikeSeth murmurs "Bosa Nova" |
| 18:37 | MikeSeth | somebody stuck in the elevator |
| 18:37 | MikeSeth | gah |
| 18:41 | rabeeh | jnettlet: no inputs.. sorry |
| 18:41 | rabeeh | i'm swamped with HB |
| 18:42 | jnettlet | rabeeh, that is fine I will figure it out. |
| 18:42 | rabeeh | jnettlet: please excuse my short memory; what was the issue? |
| 18:42 | rabeeh | the write errors when the gpu hammers the memory? |
| 18:43 | rabeeh | because as i recall this wasn't uhs-1 issue |
| 18:44 | jnettlet | rabeeh, nope this is problems with uhs-1 card detection. I have narrowed it down to either the code is trying to use the CD pin as the 1.8v pin for uhs or the switch to uhs is causing the CD pin to fluctuate on and off constantly |
| 18:45 | rabeeh | jnettlet: HB or CBi? |
| 18:45 | jnettlet | rabeeh, CBi although I was seeing it on the HB as well |
| 18:46 | jnettlet | which would make me think 1) the code is using the CD pin for 1.8v's |
| 18:46 | jnettlet | I had set no-1-8-v in device-tree to fix the problem on the HB |
| 18:46 | rabeeh | HB doesn't have any 1.8v; it's all 3.3v |
| 18:47 | rabeeh | CBi has the 3.3v to 1.8v switchover |
| 18:48 | jnettlet | are there additional pins that need to be added to the usdh2 mux? |
| 18:51 | rabeeh | KEY_ROW1 is attached to SD2 vselect |
| 18:51 | rabeeh | the feature is not in the 3.0.35 kernel since it wasn't massively tested |
| 18:51 | jnettlet | oh perfect. thanks |
| 18:51 | rabeeh | :) |
| 18:52 | rabeeh | jnettlet: now it's perfect timing to let me know if HB must have have 3.3v to 1.8v switchover or not |
| 18:52 | rabeeh | and by must have i mean must have that without it the dt will never work :) |
| 19:00 | jmontleo | sounds like uhs-1 mode might be easy to fix? |
| 19:00 | jnettlet | jmontleo, I think I have a real fix now. just a bit more tesitng |
| 19:01 | jnettlet | yep thinks look good. |
| 19:01 | jnettlet | rabeeh, that was exactly what I needed thanks |
| 19:02 | jnettlet | jmontleo, which kernel are you using main? |
| 19:03 | rabeeh | jnettlet: as you said probably the 3.0.35 kernel doesn't even support uhs-1; and that's why it wasn't asking for 3.3v to 1.8v switchover |
| 19:04 | jnettlet | oh this makes the things much snappier |
| 19:04 | jnettlet | I don't get chromium giving me the "waiting for cache" message |
| 19:05 | jnettlet | now the question is, how much do I make jmontleo beg for the fix :-P |
| 19:05 | jmontleo | jnettlet, 3.13-rc7 |
| 19:06 | jmontleo | heh; where do I direct donations :D |
| 19:06 | rabeeh | jnettlet: please 'dd if=/dev/mmcblk0 of=/dev/null bs=1M count=1000' |
| 19:06 | jnettlet | jmontleo, okay let me get you a quick patch for testing |
| 19:07 | rabeeh | jnettlet: and this is the 32GByte Sandisk uhs-1 right? |
| 19:07 | jnettlet | rabeeh, 16GB uhs-1 |
| 19:09 | jnettlet | rabeeh, should that be pulled low or high by default? |
| 19:09 | rabeeh | it's pulled down externally on the board to select 3.3v |
| 19:09 | jnettlet | so it's default state should be down |
| 19:10 | rabeeh | pulled down by a 10kohm |
| 19:10 | rabeeh | well.. doesn't really matter |
| 19:13 | jnettlet | jmontleo, can you remove the previous change and try this http://ur1.ca/gdbna |
| 19:16 | rabeeh | jnettlet: but this is for the HB |
| 19:17 | rabeeh | i mean you patch is for HB; but HB doesn't (and won't) support 3.3v to 1.8v |
| 19:17 | rabeeh | or you are using the same kernel for both machines? |
| 19:25 | jnettlet | rabeeh, same kernel and device-tree for both machines right now. It needs to be split out but haven't gotten that far yet. |
| 19:26 | jnettlet | and I know that jmontleo is using that device-tree file right now :-) |
| 19:30 | rabeeh | any performance numbers? |
| 19:30 | rabeeh | dd if=/dev/mmcblk0 of=/dev/null bs=1M count=1000 |
| 19:31 | rabeeh | i wonder if the sandisk uhs-1 can go beyond 40-50MB/sec |
| 19:31 | jnettlet | still making sure this is actually enabling everything properly and not falling into some other fallback path |
| 19:33 | rabeeh | ok. |
| 19:33 | jnettlet | still getting more mmcqd cpu usage than I think is appropriate |
| 19:37 | jmontleo | jnettlet, it boots removing the bus-width and adding that but I still see this: [ 1.387880] sdhci-esdhc-imx 2194000.usdhc: could not get ultra high speed state, work on normal mode |
| 19:38 | jnettlet | jmontleo, that is okay for now. I am still working that part out. |
| 19:38 | jnettlet | but if it boots then that is the first step in the right direction |
| 19:39 | jmontleo | ok, cool. wasn't sure how much that was intended to fix |
| 19:41 | jnettlet | still weaving my way through the layers of mmc/sdhci |
| 19:43 | rabeeh | xbmc on android works accelerated |
| 19:44 | rabeeh | well.. not perfect but better than the software based decoding |
| 19:45 | hste | rabeeh: hve u tried stephans yocto build for xbmc? |
| 19:45 | rabeeh | the old ones? |
| 19:45 | rabeeh | stephan's work is included in geexbox |
| 19:46 | hste | yes his but yocto is also nice |
| 19:46 | rabeeh | i previously talked to him about this and semi convinced him to go away from it |
| 19:47 | rabeeh | the main reason is that the micro xbmc distros spend lots of time getting the stuff around xbmc supported |
| 19:47 | rabeeh | like fonts, hot plug of mouse / keyboard and tons other things |
| 19:48 | hste | I prefer the yocto build :) |
| 19:48 | rabeeh | why? |
| 19:48 | rabeeh | what is there that isn't in the others? |
| 19:48 | hste | its so easy to update. just sync and bitbake |
| 19:48 | rabeeh | maybe i'm missing something here; |
| 19:55 | hste | rabeeh: http://stephan-rafin.net/blog/2013/12/18/build-environment/ |
| 20:21 | jmontleo | jnettlet, not too worried about it, but i tried the usb ports out of morbid curiosity and neither seems to work. is that likely due to the dts, or have I broken it? |
| 20:21 | jnettlet | jmontleo, the top port should work |
| 20:28 | jmontleo | lsusb gives me unable to initialize libusb: -99 and usbv-devices returns nothing. no dmesg output when i plug anything into either port either. odd. maybe I botched the kernel I built |
| 20:30 | jnettlet | oh that seems broken |
| 21:21 | MikeSeth | how do you generate a mmc card image out of a base rootfs? partition & format a loop device and then dump it? |
| 21:24 | MikeSeth | oh, uboot mkimage |
| 22:00 | MikeSeth | I'm building uboot with gnueabihf instead of gnueabi, is that a problem? |
| 22:05 | jas-hacks | MikeSeth: it isn't a problem |
| 22:07 | MikeSeth | yeah, I figured.. hosed it a little but distclean helps |
| 22:08 | MikeSeth | stock armhf kernel will work, right? |
| 22:08 | MikeSeth | no special patches in rabeeh's tree? |
| 22:09 | jas-hacks | depending on the board you probably need rabeeh's tree |
| 22:09 | MikeSeth | i1 |
| 22:11 | MikeSet | 22:11 * MikeSeth clones rabeeh's tree anyway |
| 22:11 | jnettlet | http://makezine.com/magazine/building-an-open-source-laptop/ |
| 22:12 | jnettlet | MikeSeth, yes you need either rabeeh's or my my uboot tree |
| 22:12 | MikeSeth | what's in this that isn't in the stock tree? |
| 22:13 | MikeSeth | goddamn that laptop is ugly and awesome at the same time |
| 22:14 | jnettlet | MikeSeth, all the cubox-i and hummingboard ddr timings and configurations |
| 22:15 | MikeSeth | aren't those initialized in uboot? |
| 22:15 | MikeSeth | sorry if this is a dumb question |
| 22:15 | jnettlet | sorry are you building uboot or the kernel? |
| 22:15 | jnettlet | regardless you still need rabeeh's kernel as that has cubox specific configurations also |
| 22:15 | MikeSeth | the kernel |
| 22:16 | jnettlet | for the kernel if you use head you can get basic headless support. Otherwise you need rabeeh's kernel. |
| 22:16 | MikeSeth | I'm going through all the tribulations of minimum debian image to one with all available drivers |
| 22:17 | jnettlet | My 3.10 kernel is almost ready for consumption just nailing down the last remaining issues |
| 22:17 | jnettlet | Then you need rabeeh's 3.0.35 kernel for now |
| 22:17 | MikeSeth | 3.10 is without vivante driver code though, yes? |
| 22:19 | jnettlet | nope 3.10 will have full driver support |
| 22:22 | MikeSeth | jnettlet: how? I thought the current vivante drivers were compiled for 3.0.35? |
| 22:22 | caitlin56 | Anyone have a favorite "how-to" on setting up opencl development environment for the cubox-i4pro? |
| 22:24 | jnettlet | MikeSeth, it is not about kernel versions it is about driver versions |
| 22:26 | MikeSeth | but those are kernelspace drivers right? |
| 22:27 | jas-hacks | caitlin56: not sure if anyone has tried/tested opencl support |
| 22:27 | jnettlet | there are two parts. OSS kernel space drivers and binary userspace drivers |
| 22:27 | MikeSeth | jnettlet: ooooh |
| 22:28 | hste | MikeSeth: The kernel space driver is backported from 3.10 to 3.0.35_4.1.0 so its the same |
| 22:29 | MikeSeth | ok I think now the whole thing makes perfect sense |
| 22:29 | caitlin56 | jsa-hacks: so try freescale pages, take notes on what I have to do to make it work, and right the page myself? |
| 22:29 | MikeSeth | hste: I presume also the drivers that enable general non-accelerated imx6 video output? |
| 22:29 | MikeSeth | hste: or are those one and same? |
| 22:29 | hste | they match |
| 22:30 | jas-hacks | caitlin56: yes ;) |
| 22:30 | MikeSeth | if you guys want to drop some noob work on me, go straight ahead |
| 22:30 | MikeSeth | this is exciting |
| 22:31 | caitlin56 | jas-hacks: I'll drop back on channel periodically to let you know how it's going. |
| 22:31 | jas-hacks | caitlin56: yes do. |
| 22:47 | MikeSeth | mhmhrm, should ask rabeeh to enable IO accounting in his tree |
| 22:47 | MikeSeth | iotop doesn't work and I see this weirdest spike of reads in vmstat when firefox is idling |
| 22:47 | rabeeh | MikeSeth: which feature would that be? |
| 22:48 | MikeSeth | rabeeh: CONFIG_TASKSTATS, CONFIG_TASK_DELAY_ACCT, CONFIG_TASK_IO_ACCOUNTING |
| 22:48 | MikeSeth | that's assuming this actually works on arm |
| 22:50 | rabeeh | it's marked as experiemental on kernel 3.0.35 |
| 22:50 | rabeeh | so i won't enable it by default... sorry |
| 22:51 | MikeSeth | alright |
| 22:51 | MikeSeth | I'll enable it here to see what's up with firefox |
| 22:52 | MikeSeth | but I have this suspicion that there's some funny memory problem in i1s |