• [00:05:44] * NishanthMenon (n=gnat@nat/ti/x-08c52cf8c4660eb1) Quit ("I cna ytpe 300 wrods pre mniuet!!!")
  • [01:05:23] * jkridner|work (n=a0321898@nat/ti/x-2e59248470a12477) has joined #beagle
  • [02:32:25] * lgentili (n=lgentili@host27.190-224-254.telecom.net.ar) has joined #beagle
  • [02:50:35] * lgentili (n=lgentili@host27.190-224-254.telecom.net.ar) Quit (Remote closed the connection)
  • [03:05:32] * BThompson (n=BThompso@cpe-76-185-66-252.tx.res.rr.com) Quit ("Trillian (http://www.ceruleanstudios.com")
  • [06:02:01] * RogerMon1 (n=a0740758@nat/ti/x-28653c9c26d6389c) has joined #beagle
  • [06:06:00] * RogerMonk (n=a0740758@nat/ti/x-3897f0f5b7c9e142) Quit (Remote closed the connection)
  • [06:52:33] * RogerMon1 (n=a0740758@nat/ti/x-28653c9c26d6389c) Quit (Remote closed the connection)
  • [06:52:50] * RogerMonk (n=a0740758@nat/ti/x-564f9105845cada5) has joined #beagle
  • [07:08:47] * RogerMon1 (n=a0740758@nat/ti/x-b661ce95e7b678f5) has joined #beagle
  • [07:12:45] * RogerMonk (n=a0740758@nat/ti/x-564f9105845cada5) Quit (Remote closed the connection)
  • [08:39:47] <koen> 2.6.26rc1 with L2 enabled: http://amethyst.openembedded.net/~koen/beagleboard/uImage-2.6.25+2.6.26-rc1+git1abd28c86f1771d4f31fc73e6ef83fedaa8f4ec2-r9-beagleboard.bin
  • [08:48:09] * RogerMon1 (n=a0740758@nat/ti/x-b661ce95e7b678f5) Quit (Remote closed the connection)
  • [08:48:35] * RogerMonk (n=a0740758@nat/ti/x-9b1da7f623c5e418) has joined #beagle
  • [09:00:00] <RogerMonk> Crofton:: you there?
  • [09:31:39] * RogerMon1 (n=a0740758@nat/ti/x-b29d2baf3c366152) has joined #beagle
  • [09:32:15] <koen> RogerMon1: crofton in in VA, so he should be awake in an hour or two :)
  • [09:32:25] <koen> root@beagleboard:~# uname -a
  • [09:32:25] <koen> Linux beagleboard 2.6.26-rc1-omap1 #1 Wed May 7 10:25:34 CEST 2008 armv7l unknown unknown GNU/Linux
  • [09:36:07] * RogerMonk (n=a0740758@nat/ti/x-9b1da7f623c5e418) Quit (Remote closed the connection)
  • [09:46:31] <DJWillis> koen: nice job, now to patch up sound.
  • [09:49:05] <koen> DJWillis: something is really wrong with communication between the TWL and the omap :(
  • [09:49:18] <koen> I suspect that's the cause of all our USB troubles
  • [09:51:50] <zodttd> Hi DJWillis!
  • [10:01:36] <DJWillis> koen: urggg :(
  • [10:01:59] <DJWillis> Hey Zodttd, hows the EVM working out? I seem to keep missing you.
  • [10:07:51] <DJWillis> koen: I guess it's not a supprise that the EVM lacks EHCI and it seems a little 'WIP'.
  • [10:08:15] <zodttd> :P
  • [10:08:21] <zodttd> Its going better now
  • [10:08:29] <zodttd> I got multi press keys working
  • [10:09:25] <zodttd> I was wondering how to either scale/rotate the screen via hardware on the EVM or do something to set the fb to QVGA (240x320)
  • [10:09:39] <DJWillis> :), just hitting the framebuffer or using the framebuffer and scaler for visual output?
  • [10:09:52] <zodttd> hitting the framebuffer
  • [10:10:04] <zodttd> 640x480 with per pixel color conversion == 16ms writes
  • [10:10:47] <zodttd> I tried everything to get rotation to work too. I found a /sys/class/......./rotate file
  • [10:10:52] <DJWillis> zodttd: you can pass info to the video driver to set the screen as VGA/qVGA and set the rotation. Has to be passed into the kernel. Problem is it does not seem to work ;-)
  • [10:10:55] <zodttd> but changes wouldnt stick in it
  • [10:11:04] <zodttd> noticed ;P
  • [10:11:11] <zodttd> As I was told...Why no worky? :P
  • [10:11:30] <zodttd> ohhhh
  • [10:11:32] <zodttd> !
  • [10:11:40] <zodttd> My kernel just doesnt have it enabled?
  • [10:12:13] <DJWillis> I was looking at working that out but I really need to spend more time on the Pandora board then the EVM at the moment.
  • [10:12:25] <zodttd> Understandable ;)
  • [10:12:53] <zodttd> Are you working on getting OE on Pandora?
  • [10:13:47] <koen> isn't that just 5 minutes of work?
  • [10:13:50] <zodttd> koen: I noticed the TWL crashing all the time
  • [10:13:59] <zodttd> I found a way around it
  • [10:14:02] <koen> copy evm machine description, add pandora defconfig
  • [10:14:26] <zodttd> koen: Heh ;)
  • [10:14:49] <koen> OE is realy easy when adding new machines
  • [10:14:50] <DJWillis> koen: almost ;-), swap EVM for Beagle and your about there ;-)
  • [10:15:18] <koen> zodttd: what was your workaround for the TWL?
  • [10:15:26] <DJWillis> zodttd: and blame koen and sakoman for getting me into OE again, yep, I am using that to pull everything together.
  • [10:16:04] <zodttd> First make sure you dont raise the priority of processes (that was a insta crash for me during heavy CPU load)
  • [10:16:09] <zodttd> I use:
  • [10:16:09] <zodttd> echo -n 0 > /sys/power/fb_timeout_value
  • [10:16:09] <zodttd> echo -n 0 > /sys/power/cpuidle_deepest_state
  • [10:16:26] <zodttd> And if it does crash, it at least picks up where left off
  • [10:16:50] <zodttd> DJWillis: I didnt know you were into OE before? Cool :)
  • [10:17:37] <zodttd> koen: You were getting errors with TWL timeouts right?
  • [10:19:22] <koen> zodttd: among other things
  • [10:19:37] <koen> it main syndrom was "ehci crashes under high cpu load"
  • [10:19:57] <koen> with l2 enabled it becomes "i2c timouts when writing to twl"
  • [10:21:47] <koen> which of course I can't reproduce now...
  • [10:25:50] <zodttd> I got the latter
  • [10:26:13] <zodttd> It was pretty random for me but always when under a large cpu load
  • [10:27:01] <koen> I don't have fb support or cpuidle in the kernel :)
  • [10:28:46] <zodttd> :(
  • [10:34:18] <DJWillis> zodttd: interesting, I had some odd problems with ScummVM under high CPU load that I just parked, seem to be on a similar line.
  • [10:45:14] <koen> heh
  • [10:45:19] <koen> kernel without l2 cache:
  • [10:45:26] <koen> BENCHMARKs: VC: 193.856s VO: 0.153s A: 0.000s Sys: 2.718s = 196.727s
  • [10:45:32] <koen> kernel with l2 cache:
  • [10:45:39] <koen> BENCHMARKs: VC: 56.150s VO: 0.133s A: 0.000s Sys: 1.043s = 57.327s
  • [10:45:49] <koen> mplayer is 3.5 times faster
  • [10:45:55] * koen OMGWTFBBQs
  • [10:48:02] <koen> DJWillis, zodttd: try applying http://amethyst.openembedded.net/~koen/beagleboard/l2-cache.patch to your kernel for fun
  • [10:51:33] <koen> sorry, only 2.7 times, due to increased clockspeed
  • [11:14:53] * zodttd (n=recompil@user-142gtg7.cable.mindspring.com) Quit ()
  • [11:29:22] <Crofton> RogerMon1, ping
  • [11:31:32] * Crofton still needs coffee
  • [11:40:05] <koen> coffee, good idea!
  • [11:41:24] <koen> rather "brownish water from the machine behind me"
  • [11:41:35] * koen is now known as koen|uni
  • [11:41:48] <koen|uni> Crofton: did you get the IM about l2 cache?
  • [11:45:22] <RogerMon1> Hey mr C - how's things?
  • [11:45:35] <Crofton> apparently not, I gather you hve it working though
  • [11:45:57] <RogerMon1> We got ur u-boot working... - was a problem with our flash utility...
  • [11:45:58] <koen|uni> Crofton: I was suspicous about the IMs since you have been online non-stop for the past 4 days :)
  • [11:47:11] <koen|uni> Crofton: executive summary: mplayer is now 2.7 times faster in raw decode speed
  • [11:47:31] <koen|uni> 3.5 times if you bump speed to 500MHz
  • [11:48:53] <koen|uni> googling for "mplayer NEON armv7" is funny
  • [11:49:15] <koen|uni> first hit is the beagle mailing list, 2nd and 3rd are for the angstrom feeds
  • [11:49:40] <Crofton> ug
  • [11:50:16] <koen|uni> but still, this is CPU decode speed
  • [11:50:17] <Crofton> another thing I like about OE is that it is easy to move the build environment from my desktop to my laptop for travel :)
  • [11:50:26] <koen|uni> we want to use the friggin DSP for this kind of stuff!
  • [11:52:26] <koen|uni> Crofton: btw, I bumped linux-omap2 git again
  • [11:52:42] <koen|uni> Crofton: and activated the l2 patch
  • [11:54:32] <koen|uni> jkridner|work: I have a growing suspicion that all beagle stability problems are due to i2c bugs when communicating with the twl
  • [11:55:29] <jkridner> k. I don't think you are including USB EHCI in that though, right?
  • [11:56:10] <koen|uni> I think ehci is a side effect of twl power glitches
  • [11:56:24] <koen|uni> drat, I forgot to bring the beagle
  • [11:56:30] <jkridner> oh.
  • [11:56:38] * koen|uni is 20m from the lab with the scopes
  • [11:57:14] <jkridner> but, I thought with the .22 kernel we didn't have the TWL I2C issues that you've seen on git kernel boot.
  • [11:57:47] <koen|uni> it doesn't report the issues
  • [11:57:56] <koen|uni> so I can't say for sure were aren't seeing them
  • [11:59:46] <koen|uni> http://focus.ti.com/general/docs/wtbu/wtbuproductcontent.tsp?templateId=6123&navigationId=11989&contentId=4682 doesn't list the 35xx
  • [12:01:17] <koen|uni> anyway, I'm happy L2 is enabled now
  • [12:01:52] <koen|uni> so linux-omap git from OE now has USB host, client, SD and L2 cache support
  • [12:02:18] <koen|uni> hopefully sakoman gets some time to do twl4030 ASoC soon
  • [12:04:38] <koen|uni> that would be the low hanging fruit
  • [12:04:59] <koen|uni> the hard part is the DVI driver(s) and debugging the stability problems
  • [12:06:25] <RogerMon1> Crofton: Did you manage to spin me a new kernel yet with 80008000?
  • [12:08:05] <Crofton> it should be uploaded
  • [12:08:14] <koen|uni> http://amethyst.openembedded.net/~crofton/davinci-dvevm/uImage-2.6.24+2.6.25-rc8-git713dc561fa98efb6cbac3f641c43d30ce8b840dd-r1-davinci-dvevm.bin
  • [12:11:37] * BThompson (n=BThompso@nat/ti/x-8c3e32dee7e01e50) has joined #beagle
  • [12:16:58] * koen|uni wonders why musb is stable while ehci isn't
  • [12:17:17] <RogerMon1> Crofton: kernel booted, nfs mounted, udhcp started, nfs hang. Can you remind me the change to disable reinit of network when using nfs pls?
  • [12:17:37] <koen|uni> edit /etc/network/interfaces and remove eth0
  • [12:17:41] <koen|uni> (from the top of my head)
  • [12:17:43] <Crofton> right :)
  • [12:17:51] <Crofton> comment out auto eth0
  • [12:18:31] <koen|uni> 2.6.26rc1 feels faster than 2.6.22 for some reason
  • [12:18:42] <koen|uni> probably CFS
  • [12:20:30] <Crofton> cfs
  • [12:20:44] <Crofton> RogerMon1, does the u-boot work?
  • [12:21:18] <koen|uni> 13:45 < RogerMon1> We got ur u-boot working... - was a problem with our flash utility...
  • [12:21:30] <Crofton> ah thanks
  • [12:21:44] <Crofton> I was worried about that
  • [12:21:57] <Crofton> that was built from the u-boot release tarball
  • [12:22:02] <RogerMon1> The Angstrom Distribution davinci-dvevm ttyS0
  • [12:22:02] <RogerMon1> Angstrom 2008.1-test-20080430 davinci-dvevm ttyS0
  • [12:22:06] <RogerMon1> NICE!
  • [12:22:14] <Crofton> thanks for testing
  • [12:22:15] <RogerMon1> booted and logged inn
  • [12:22:28] <Crofton> I'm going to send an email to the DV list
  • [12:22:38] <RogerMon1> anything cool in the rootdisk here that I should test?
  • [12:22:54] <Crofton> every now and then people ask about replacement file systems ...
  • [12:23:01] <Crofton> are you on a network?
  • [12:23:04] <RogerMon1> yep
  • [12:23:10] <Crofton> ipkg update
  • [12:23:16] <Crofton> er opkg update
  • [12:23:37] <Crofton> I don't know if there is anything cool
  • [12:23:50] <Crofton> do you have something you want to try?
  • [12:24:26] <Crofton> The key point is you now run a very similar setup to the beagle
  • [12:26:10] <Crofton> you could try "opkg install python"
  • [12:26:25] <Crofton> and compare that to the time required to build python on your own :)
  • [12:28:19] * NishanthMenon (n=gnat@nat/ti/x-a981c6ac00cc9dc8) has joined #beagle
  • [12:28:28] <RogerMon1> thanks -how should proxy be set up for opkg?
  • [12:29:50] <RogerMon1> "Signature check failed?"
  • [12:29:55] <Crofton> thats ok
  • [12:30:06] <Crofton> look in /etc/opkg.conf i think
  • [12:30:12] <RogerMon1> Downloading http://www.angstrom-distribution.org/feeds/2008/ipk/glibc/armv5te/py
  • [12:30:12] <RogerMon1> thon/Packages.sig
  • [12:30:12] <RogerMon1> Signature check failed
  • [12:30:12] <RogerMon1> An error ocurred, return value: 14.
  • [12:30:26] <Crofton> signatures are new, we do not have infrastructure for signing thm
  • [12:31:37] <RogerMon1> Failed to download http://www.angstrom-distribution.org/feeds/2008/ipk/glibc/arm
  • [12:31:37] <RogerMon1> v5te/base/Packages.gz, error 0
  • [12:32:45] <koen|uni> http_proxy = <something> in /etc/opkg.conf
  • [12:33:58] <RogerMon1> etc/resolve.conf.... downloading ok now.
  • [12:34:03] <RogerMon1> thanks!
  • [12:35:15] <RogerMon1> is there an x11/OpenOffice/Minimo type build available for this target?
  • [12:35:28] <Crofton> probably
  • [12:35:34] <RogerMon1> v.cool
  • [12:35:46] <Crofton> I am not a screen person though
  • [12:35:49] <RogerMon1> it's going to be very handy being able to run the same env on beagle and dvevm
  • [12:35:56] <Crofton> I can build x11 image
  • [12:36:08] <RogerMon1> thanks - that'd be great.
  • [12:36:08] <Crofton> but I'm not certain congif files, if any will be correct
  • [12:37:09] <RogerMon1> should usb work in this build?
  • [12:39:01] <koen|uni> RogerMon1: opkg install abiword minimo xserver-kdrive-fbdev angstrom-x11-base-depends angstrom-gpe-task-base angstrom-gpe-task-settings
  • [12:39:16] <koen|uni> that will install an X11 base system with abiword and minimo
  • [12:39:24] <koen|uni> openoffice doesn't cross-compile
  • [12:40:19] <RogerMon1> Thanks - I'm still on : Downloading http://www.angstrom-distribution.org/feeds/2008/ipk/glibc/armv5te/base/Packages.gz... - how long would u expect this to take on 8Mbps download?
  • [12:40:33] * ali_as (n=as@ambix.plus.com) has joined #beagle
  • [12:42:32] <koen|uni> a few seconds
  • [12:42:37] <koen|uni> it has a progress bar
  • [12:42:52] <koen|uni> if you don't see the progress bar it is trying to contact the server
  • [12:44:14] <RogerMon1> hmm ok, so something wrong then.. no progress bar...
  • [12:44:27] <Crofton> ping www.yahoo.com
  • [12:54:20] <Crofton> jkridner, how's the talk on patch submission going :)
  • [12:54:40] <jkridner> slowly.
  • [12:54:47] <Crofton> NishanthMenon, just got dinged for line wrapped patches on the u-boot list :(
  • [12:54:57] <Crofton> Wolfgang can be a pain ...
  • [12:55:02] <jkridner> the challenge is getting Windows machines setup, for one.
  • [12:55:14] <jkridner> I sent out the prerequisites last night.
  • [12:55:30] <jkridner> I built up an Ubuntu-image with the git trees for the kernel and u-boot...
  • [12:55:41] <jkridner> but the image (compressed) was almost 2GB.
  • [12:55:49] <Crofton> vmware?
  • [12:55:52] <jkridner> It is difficult to serve up that much data.
  • [12:55:53] <jkridner> yes.
  • [12:56:11] <Crofton> vmware is good and bad
  • [12:56:33] <jkridner> can't live with it. can't live without it.
  • [12:56:39] <Crofton> exactly
  • [12:56:57] <jkridner> Wubi is interesting. I hope it works for most of the people in the class.
  • [12:57:42] <jkridner> but, I can't easily send out a pre-built Wubi installer image.
  • [13:00:04] * koen|uni is old fashioned and logs in via ssh into a linuxbox for linux stuff
  • [13:00:33] * ali_as_ (n=as@ambix.plus.com) Quit (Read error: 110 (Connection timed out))
  • [13:02:47] <jkridner> that's what I do most of the time too. copying files onto SD cards requires physical access.
  • [13:03:21] <koen|uni> that's why I complain if the beagle crashes hard and ctrl-a-a-f b doesn't work :)
  • [13:03:36] <koen|uni> (sysrq-b inside minicom inside screen)
  • [13:04:02] * Crofton tries to be pure, but uses windows for Combat Mission and watching cycling.tv
  • [13:04:03] <jkridner> also, I think that many people at TI don't have access to Linux machines to ssh into, but I could be wrong.
  • [13:04:50] <jkridner> :) I'm also a screen fan. #2 install after vim. :O
  • [13:05:25] <jkridner> I hate the ctrl-a binding though. Messes with "go to start of line".
  • [13:07:04] * horizon_ is now known as horizon
  • [13:23:06] <Crofton> I sort of wish the oprofile for arm7 went through the OMAP list
  • [13:33:29] <RogerMon1> Crofton: can't ping, behind firewall, looks like that's my problem. jkridner, did you get opkg working through TI firewall/proxy? any tips?
  • [13:35:10] <jkridner> yes. This is w/ Beagle?
  • [13:35:20] <Crofton> EVM :)
  • [13:35:24] <RogerMon1> nope davinci, but same build
  • [13:35:25] <Crofton> er Davinci
  • [13:36:13] <Crofton> we really shouldl talk about DV in #davinci :)
  • [13:36:13] <jkridner> export http_proxy=http://(your proxy):80/
  • [13:36:24] <jkridner> indeed. ;)
  • [13:36:30] <jkridner> I'm over there too.
  • [13:36:32] <RogerMon1> ok, will switch to #davinci
  • [13:36:45] * prpplague (n=dave@mail.americanmicrosystems.com) has joined #beagle
  • [13:42:16] <koen|uni> good to hear it works
  • [13:43:28] <Crofton> yeah, I'm going to post the location to the dv list
  • [13:57:38] * bazbell (n=a0192809@nat/ti/x-8669ff7392440e81) Quit (Remote closed the connection)
  • [14:09:51] <Crofton> koen|uni, what is the url for the feed browser again?
  • [14:11:16] <koen|uni> http://www.angstrom-distribution.org/repo/
  • [14:11:27] <koen|uni> it is a bit slow during this time of the day
  • [14:11:55] <koen|uni> .dev -> 2008.6, .stable -> 297.
  • [14:11:58] <koen|uni> 2007.12 feeds
  • [14:12:48] <NishanthMenon> Crofton: yes.. missed the mail.. just saw the responses. :)... i guess outlook is a pain :(.. i prefer thunderbird at home :) will post the workaround for outlook to the list..
  • [14:12:59] <Crofton> koen|uni, found it
  • [14:13:02] * NishanthMenon on a meeting binge :(
  • [14:13:26] <Crofton> hopefully they are useful meetings
  • [14:14:26] <NishanthMenon> Crofton: yep...
  • [14:21:27] <Crofton> wew, glad felipe's beagle is ok
  • [14:27:24] * bazbell (n=a0192809@nat/ti/x-030acda0f4fde1e6) has joined #beagle
  • [14:27:40] <NishanthMenon> http://www.flickr.com/photos/nishanthmenon/tags/opensource/ -> outlook 2003 configuration!
  • [14:28:24] <Crofton> rofl
  • [14:28:38] <Crofton> post that to the u-boot list :)
  • [14:28:44] <NishanthMenon> :D yep ;)
  • [15:02:44] * BThompson (n=BThompso@nat/ti/x-8c3e32dee7e01e50) Quit ("Trillian (http://www.ceruleanstudios.com")
  • [15:34:24] * jkridner|work (n=a0321898@nat/ti/x-2e59248470a12477) Quit (Remote closed the connection)
  • [15:57:55] * koen|uni is now known as koen
  • [16:13:43] <koen> gah, the sf.net mailarchive is atrocious
  • [16:16:11] * BThompson (n=BThompso@nat/ti/x-e415fa0d219e60f4) has joined #beagle
  • [16:16:53] * koen is getting cranky and needs a cookie
  • [16:24:30] <khilman> Crofton: thanks for the Angstrom images for davinci evm!
  • [16:51:56] * ldesnogu (n=ldesnogu@ven06-2-82-247-86-183.fbx.proxad.net) has joined #beagle
  • [16:58:41] * prpplague (n=dave@mail.americanmicrosystems.com) Quit ("Leaving")
  • [17:04:35] * jkridner|work (n=a0321898@nat/ti/x-4b52bca154fdcfde) has joined #beagle
  • [17:05:40] * DarrenEtheridge (n=a0867391@nat/ti/x-a41b91105021e36b) has joined #beagle
  • [17:07:31] <sakoman> Is the preliminary ASoC work that Mistral did available publicly? Just want to check before I start working on it myself.
  • [17:12:23] <jkridner|work> http://focus.ti.com/general/docs/wtbu/wtbusplashcontent.tsp?templateId=6123&contentId=4750#omap3530 just went live.
  • [17:13:08] <ldesnogu> nice :)
  • [17:13:35] <ldesnogu> it looks like only u-boot is different
  • [17:13:47] <ldesnogu> does it contain the infamous L2 stuff?
  • [17:13:51] * prpplague (n=dave@mail.americanmicrosystems.com) has joined #beagle
  • [17:15:46] <jkridner|work> I don't know exactly. I know that if you use it all to boot on the EVM that L2 is enabled.
  • [17:16:30] <jkridner|work> this site is just a minor help, really. what matters is the support via the git.
  • [17:17:07] <Crofton> khilman, your welcome
  • [17:17:11] <jkridner|work> but, for those who don't have an OMAP3 EVM, it gives them a way to look at the open source components.
  • [17:17:16] <Crofton> thanks for keeping git updated
  • [17:17:58] <jkridner|work> by "matters" above, I mean for Beagle and the open source community.
  • [17:19:32] <ldesnogu> I am lost... Is the wiki/BeagleSourceCode page up to date?
  • [17:19:38] <jkridner|work> khilman, I hope it is helpful to you though.
  • [17:20:07] <jkridner|work> the wiki/BeagleSourceCode represents the latest "release" for Beagle...
  • [17:20:13] <khilman> jkridner|work: yes, it is. thx
  • [17:20:15] <jkridner|work> it does not represent the tip of the development.
  • [17:20:21] <khilman> OMAP git is looking better for omap3evm, but missing network support for starters
  • [17:20:37] <khilman> I just got my evm plugged in yesterday, so just evaluating status of kernel for it
  • [17:21:32] <ldesnogu> is the OMAP kernel git head (or whatever it's called in git terminology) able to boot on Beagle?
  • [17:21:38] <Crofton> yes
  • [17:21:45] <ldesnogu> great :)
  • [17:25:09] <sakoman> khilman: I have patches for OMAP3 EVM network support
  • [17:25:30] <khilman> sakoman: I was just about to ask you
  • [17:25:36] <khilman> I assumed someone was working on that
  • [17:26:09] <sakoman> need to find where to submit them because linux-omap won't take smc911x driver changes
  • [17:26:22] <sakoman> It's on my todo list for today
  • [17:27:22] <khilman> sakoman: you could post the board support code to linux-omap and teh smc911x stuff there as well for RFC
  • [17:28:03] <khilman> then you should probably collaborate with Nishant Kamat on the smx911x stuff he did for Zoom, and submit it to linux-net
  • [17:28:17] <sakoman> Someone from TI posted an initial patch set a week or two ago and was politley told to take it elsewhere
  • [17:28:47] <sakoman> I think it *was* Nishant that did the submission
  • [17:29:31] <sakoman> In any event those patches need rework now since the latest git has merged some smc911x changes
  • [17:30:43] <khilman> right, it was NIshant
  • [17:31:16] <khilman> I agree with Tony that in order to go upstream, they need to be posted elsewhere,
  • [17:31:51] <khilman> but they are still useful to get others going early, so posting to linux-omap as RFC is appropriate
  • [17:31:59] <sakoman> Yeah, I agree too. Just makes the process that much slower since wew need to wait for th trickle down to happen
  • [17:32:07] * prpplague (n=dave@mail.americanmicrosystems.com) Quit ("Leaving")
  • [17:32:08] <khilman> agreed
  • [17:36:00] * likewise (n=likewise@82-171-51-231.ip.telfort.nl) has joined #beagle
  • [17:38:03] * prpplague (n=dave@mail.americanmicrosystems.com) has joined #beagle
  • [17:44:41] <Crofton> koen, ping
  • [17:45:12] * richardw (n=richardw@nat/ti/x-8656d9a82dad380b) has joined #beagle
  • [17:47:07] * Crofton is wondering what the chances are of creating a useful x11 image by typing "bitbake x11-image" for the DV EVM ...
  • [17:51:08] * RogerMon1 is interested too!
  • [17:51:23] <Crofton> I''l run it
  • [17:51:29] <RogerMon1> (but that's for #davinci :)
  • [17:51:34] <Crofton> yeah
  • [17:51:40] <Crofton> koen isn't there though :)
  • [17:51:55] <RogerMon1> well... omap is davinci device anyway....
  • [17:52:25] <Crofton> yeah
  • [17:52:35] <Crofton> of course we should all be in #ol :)
  • [17:52:51] <RogerMon1> #ol? what's that
  • [17:53:01] <Crofton> omap linux
  • [17:53:07] <Crofton> from the "good old days"
  • [18:41:44] * dirk2 (n=dirk@F317b.f.strato-dslnet.de) has joined #beagle
  • [18:41:53] <Crofton> gm dirk2
  • [18:41:58] <dirk2> hi
  • [18:42:31] <JoeBorn> dirk2: jkridner suggested I talk to you about our upcoming davinci product
  • [18:42:48] <dirk2> JoeBorn: Hi!
  • [18:42:55] <JoeBorn> hi there
  • [18:43:26] <dirk2> I read the IRC log from monday about this
  • [18:43:34] * likewise (n=likewise@82-171-51-231.ip.telfort.nl) Quit ()
  • [18:44:36] <JoeBorn> that had some links?
  • [18:44:47] <JoeBorn> I'm sorry I don't remember exactly what we discussed.
  • [18:45:39] <dirk2> Yes: http://uzq.qlnk.net/
  • [18:46:06] * likewise (n=likewise@82-171-51-231.ip.telfort.nl) has joined #beagle
  • [18:48:52] <Crofton> dsp_link_1_50 is the current version?
  • [18:54:19] <dirk2> Crofton, koen: I'm still thinking about the flaky git kernel
  • [18:54:49] <dirk2> http://groups.google.com/group/beagleboard/msg/a850602099d0aa8e
  • [18:56:06] <dirk2> Crofton, koen: To be on the safe side: Did you check your boards for HW erratas 2 & 4 http://elinux.org/BeagleBoard#Hardware_2 ?
  • [18:57:08] <Crofton> I should try the slow u-boot also
  • [18:59:01] <dirk2> Crofon: Are L1 & L2 & L3 on the bottom of your board labeled 102?
  • [18:59:25] * Crofton gets out his reading glasses
  • [19:00:12] <Crofton> yep
  • [19:00:29] <dirk2> Then it's okay :)
  • [19:00:47] <dirk2> It's always better to check ;)
  • [19:00:52] <Crofton> yeah
  • [19:00:59] <RogerMon1> dsplink 1.50 is the latest
  • [19:01:15] <Crofton> I need to start looking at getting OE to build that
  • [19:01:20] <Crofton> it will be a beast
  • [19:01:29] <RogerMon1> Crofton: what you planning to do with dsplink?
  • [19:01:31] <Crofton> I am tempted to patch in autotools
  • [19:01:36] <Crofton> I want it to work
  • [19:01:47] <RogerMon1> will u be using codec engine on top?
  • [19:01:51] <RogerMon1> (at the moment)
  • [19:01:57] <Crofton> then I need to learn, or get someone who already knows, to write some dsp code for sdr
  • [19:02:00] <Crofton> no
  • [19:02:03] <RogerMon1> ok
  • [19:03:34] <JoeBorn> dirk2: http://wiki.neurostechnology.com/index.php?title=OSD2.0_Development is another one.
  • [19:03:53] <JoeBorn> if you are interested in a dev board, let me know
  • [19:04:02] <JoeBorn> jborn at neurostechnology.com
  • [19:05:45] <dirk2> JoeBorn: Thanks! Looking at the link...
  • [19:13:40] * Ragha (n=Ragha@nat/ti/x-b65578fb20f52ba3) has joined #beagle
  • [19:13:50] <altex4> ping jkridner
  • [19:13:55] <Crofton> where did I get the fast u-boot from?
  • [19:14:07] <Crofton> Do I need to try the slow MLO also?
  • [19:14:09] <dirk2> altex4: Hi!
  • [19:15:39] <altex4> hey dirk2
  • [19:16:34] <dirk2> altex4: jkridner mentioned you will do some beagle blog/news work?
  • [19:17:18] <altex4> yep. right now i'm looking through the news proposals you sent me
  • [19:17:53] <dirk2> Ah, I just saw there are some news http://beagleboard.blogspot.com/
  • [19:20:29] <jkridner|work> hi altex4.
  • [19:21:12] <dirk2> altex4: Any special topics you like to write about? Maybe we here can help. Some days ago there was the idea about "Angstrom on Beagle"
  • [19:22:05] <dirk2> altex4: ... and using some pictures like http://scap.linuxtogo.org/files/5e0946883882a895576ef8cee2312597.png
  • [19:24:07] <altex4> just trying to make sense of it, but is the picture a print-screen running off of Beagle?
  • [19:28:24] <jkridner|work> altex4: can you come on downstairs. I want to show you Beagle running OE.
  • [19:28:31] <altex4> k
  • [19:29:14] <dirk2> ;) Yes, this is the easiest solution instead of writing lot of explanation ;)
  • [19:34:43] <dirk2> jkridner|work: I had to learn that Bealge runs Angstrom, built by OE. We should get this right in the blog ;)
  • [19:35:21] <Crofton> Next time the OE guys are together, we will have to discuss this over beer
  • [19:42:03] <dirk2> JoeBorn: Will send you a mail
  • [19:42:21] <Crofton> Angstrom, built by OpenEmbedded,
  • [19:54:13] * dirk2 (n=dirk@F317b.f.strato-dslnet.de) has left #beagle
  • [20:01:23] <jkridner|work> dirk2: thanks for making this clear.
  • [20:01:33] <jkridner|work> doh, missed him.
  • [20:01:49] <jkridner|work> didn't get to thank him for the help on the blog post either.
  • [20:23:55] * prpplague (n=dave@mail.americanmicrosystems.com) Quit ("Leaving")
  • [20:53:11] * RogerMonk (n=a0740758@nat/ti/x-cb87116114ae9221) has joined #beagle
  • [20:55:19] <Crofton> RogerMonk, x11 images going up now
  • [20:55:36] <Crofton> it would be great if you check the new console-image also ...
  • [20:59:01] * RogerMon2 (n=a0740758@nat/ti/x-8680799168238e41) has joined #beagle
  • [21:03:08] * RogerMonk (n=a0740758@nat/ti/x-cb87116114ae9221) Quit (Remote closed the connection)
  • [21:03:10] * RogerMon1 (n=a0740758@nat/ti/x-b29d2baf3c366152) Quit (Remote closed the connection)
  • [21:15:33] * BThompson (n=BThompso@nat/ti/x-e415fa0d219e60f4) Quit ("Trillian (http://www.ceruleanstudios.com")
  • [21:41:49] * khilman_ (n=khilman@deeprooted.net) has joined #beagle
  • [21:46:30] * khilman (n=khilman@deeprooted.net) Quit ("Leaving.")
  • [21:46:49] * khilman_ (n=khilman@deeprooted.net) has left #beagle
  • [21:50:50] * bazbell (n=a0192809@nat/ti/x-030acda0f4fde1e6) Quit (Remote closed the connection)
  • [21:56:43] * khilman (n=khilman@deeprooted.net) has joined #beagle
  • [21:57:09] <koen> Crofton: pong
  • [22:02:54] <Crofton> hmm
  • [22:03:01] <Crofton> x11 image built for evm
  • [22:03:12] <jkridner|work> OMAP35x EVM?
  • [22:03:18] <Crofton> er DV EVM :)
  • [22:03:29] <koen> that's not a surprise
  • [22:03:34] <jkridner|work> go over to #davinci! ;)
  • [22:03:44] <Crofton> I'm wondering what sort of customization would be needed to change from beagle to DV EVM
  • [22:03:45] <jkridner|work> just kidding, but it does make things confusing.
  • [22:03:59] * Crofton makes a desperate attempt to be on topic :)
  • [22:04:04] <Crofton> I know
  • [22:04:27] <jkridner|work> discussion also works a bit on #neuros.
  • [22:05:42] <koen> time to go zzzz in CET timezones
  • [22:05:44] <koen> 'night all
  • [22:05:51] <Crofton> gn
  • [22:06:02] <Crofton> I need to go back to moing
  • [22:06:17] <Crofton> jkridner, I built our SCA implementation for beagle
  • [22:06:24] <Crofton> sort of works
  • [22:06:38] <Crofton> I suspect I need to tell it not to use ARM fixed point stuff
  • [22:06:50] <Crofton> well, I need to debug
  • [22:07:56] <jkridner|work> SCA?
  • [22:08:11] <Crofton> at least I can power it from my laptop, reducing the number of power adapters needed
  • [22:08:36] <Crofton> http://en.wikipedia.org/wiki/Software_Communications_Architecture
  • [22:08:59] <Crofton> End result of software engineers telling radio engineers how to make radios
  • [22:11:28] * BThompson (n=BThompso@cpe-76-185-66-252.tx.res.rr.com) has joined #beagle
  • [22:13:54] <likewise> Crofton: that looks like a pretty high layer stack for a radio (XML even?)
  • [22:14:36] <likewise> Crofton: CORBA even...
  • [22:14:49] <Crofton> yeah
  • [22:15:05] * DarrenEtheridge (n=a0867391@nat/ti/x-a41b91105021e36b) has left #beagle
  • [22:18:14] <RogerMon2> koen you there?
  • [22:18:36] <RogerMon2> can you look at my dvevm bootlog? - http://pastebin.com/d1f5d4692
  • [22:19:47] <likewise> RogerMon2: Anything particular that caught your attention?
  • [22:20:18] <RogerMon2> can we switch to #davinci?
  • [22:20:24] <likewise> yup
  • [22:24:39] * zodttd (n=recompil@user-142gtg7.cable.mindspring.com) has joined #beagle
  • [22:26:13] * khilman (n=khilman@deeprooted.net) has left #beagle
  • [22:27:39] <zodttd> So whats going on in the land of OMAP
  • [22:28:34] <ldesnogu> Intel is buying TI to replace the Atom which is so bad, with the upcoming OMAP4
  • [22:28:50] <zodttd> heh :)
  • [22:28:54] <ldesnogu> (smiley implied :p)
  • [22:29:05] <zodttd> lol
  • [22:29:43] <zodttd> I heard koen enabled L2 cache on the EVM and/or Beagle and got a large performance boost...
  • [22:30:02] <ldesnogu> he go a nice speedup with mplayer
  • [22:30:10] <ldesnogu> more than 2x at the same frequency
  • [22:30:22] <zodttd> I could really use such a speedup
  • [22:30:36] <ldesnogu> you have the Pandora or the EVM?
  • [22:30:39] <zodttd> How can such a patch be applied on the EVM?
  • [22:30:45] <ldesnogu> ok :)
  • [22:31:06] <zodttd> <koen> DJWillis, zodttd: try applying http://amethyst.openembedded.net/~koen/beagleboard/l2-cache.patch to your kernel for fun
  • [22:31:22] <zodttd> I have the patch...but I never compiled the kernel for the EVM...
  • [22:31:37] <zodttd> So I guess I need to know where to start
  • [22:31:44] <ldesnogu> I am not sure the EVM doesn't have its L2 disabled if you use the default files
  • [22:31:57] <ldesnogu> others would have to answer that :)
  • [22:32:02] <zodttd> Hmm, can I cat /proc/cpuinfo for that info?
  • [22:32:05] <zodttd> oh ok
  • [22:32:10] <ldesnogu> no you can't
  • [22:32:14] <zodttd> doh
  • [22:32:39] <zodttd> jkridner is Jason right?
  • [22:32:52] <ldesnogu> either you have to dump some system registers (for that you need a module or hack the kernel as dirk2 did) or you run lmbench and analyze bandwidth results
  • [22:32:57] <ldesnogu> zodttd: yes
  • [22:33:40] <zodttd> I like the sound of getting access to hw regs, but probably easiest to just run lmbench for now
  • [22:35:48] <zodttd> Is lmbench precompiled for ARM (specifically the EVM) ?
  • [22:35:52] <ldesnogu> writing the cp15 registers can be dangerous :)
  • [22:36:07] <ldesnogu> I think Koen said it was in some OE repository
  • [22:36:18] <zodttd> aha ok
  • [22:36:33] <ldesnogu> http://www.bitmover.com/lmbench/lmbench3.tar.gz
  • [22:36:42] <ldesnogu> but you will have to compile it :)
  • [22:37:22] * likewise (n=likewise@82-171-51-231.ip.telfort.nl) Quit ()
  • [22:40:48] <richardw> ldesnogu, there are other benchmarks a bit more simple, like cachebench, or nbench, really STREAM is pretty easy also, all you need to do is tweak the timing function.
  • [22:41:45] <ldesnogu> the only one I used is STREAM, and as far as I remember, you would have to do multiple runs manually with different sizes to get an idea
  • [22:41:53] <zodttd> I was looking for something to determine L2 cache being enabled thats either really simple to compile (quick) or just plain plug in for the EVM
  • [22:42:25] <ldesnogu> zodttd: richardw is surely right, lmbench is overkill if you run it all
  • [22:42:37] <ldesnogu> however you can run just a part of it
  • [22:42:54] <zodttd> You think Mistral would have L2 disabled...wonder.
  • [22:43:01] <zodttd> ah
  • [22:50:18] <richardw> zodttd, depends on the kernel version. Many people leveraged the TI-3430-wtbu code though 2.6.22. When some moved ahead some details were dropped. When up-porting this kind of thing happens. You will find that kernel.org or omap-linux as exists in the wild will lead and lag in different areas. this is especially true on a new arm.
  • [22:51:48] <richardw> the armv6 kernel was broke in kernel.org for a year with respect to cache-alias issues. RMK new this and was phasing in fixes. It seems may people are unaware of this kind of thing and blindly feel everything is ok when its reall not.
  • [22:53:17] <ldesnogu> there is also the horror story of the write buffer disabled in some (all?) xscale due to an old silicon bug, not applicable to more recent xscale's and that had a huge impact on memory bandwidth
  • [22:55:25] <richardw> If your not applying kernel.org stability patches ex 2.6.25.2, then apply arm ltd patches, then apply omap-git patches, then taking some ti ones, what you end up is usually something no one would every go to production on. Playing and developing is one thing, and shipping is another. If you have an old chip many things are already made their way through natural selection into kernel.org, but if the chip is new you better be savy.
  • [22:56:48] <richardw> ldesnogu, I can believe that. I've seen that with the old 1510's (TI-ARM925), its probably still broke today in tree.
  • [22:59:51] <jkridner|work> zodttd: yes, I'm Jason Kridner.
  • [22:59:57] <zodttd> Ah hi :)
  • [23:00:02] <jkridner|work> :)
  • [23:00:23] <jkridner|work> you getting all your OMAP3 questions answered?
  • [23:00:26] <zodttd> jkridner: EVM is great so far :)
  • [23:00:32] <zodttd> Yeah, things are running really well
  • [23:00:41] <jkridner|work> great.
  • [23:00:53] <zodttd> All except rotation and scaling of course :P
  • [23:01:47] <jkridner|work> I'm expecting rotation to come out soon.
  • [23:01:54] <zodttd> I have to do colorspace conversion from BGR555 to RGB565 so rotation and scaling + per pixel color conversion at 640x480 landscape makes things into a 16ms blit
  • [23:02:14] <jkridner|work> how many pixels?
  • [23:02:19] <jkridner|work> per pixel?
  • [23:03:14] <zodttd> I'm taking a variable size chunk out of a buffer that has BGR555 color data in it, and rotating it and, depending on its size, scaling it to the EVM framebuffer
  • [23:03:58] <zodttd> I scale it to become fullscreen on the EVM, so 640x480
  • [23:04:48] <jkridner|work> have you dug into the kernel on the rotation aspect? i guess you are doing this all user mode, right?
  • [23:04:56] <zodttd> user mode, yes
  • [23:05:03] <jkridner|work> (no DMA hardware used)
  • [23:05:06] <zodttd> I havent touched the kernel yet
  • [23:05:41] <zodttd> But I'd like to do anything to get more performance out of the EVM today
  • [23:06:04] <zodttd> The only thing that really affects things is the blit though. Otherwise I'm good to go.
  • [23:06:32] <zodttd> But if blitting cant go any faster atm for me on the EVM, I'd go other routes
  • [23:07:19] <ldesnogu> the bottleneck is the blitting itself or all the steps you described?
  • [23:07:30] <jkridner|work> what does your blit look like? no alpha, right?
  • [23:07:39] <zodttd> No alpha
  • [23:08:37] <zodttd> Just a bitblit
  • [23:09:12] <zodttd> The bottleneck is the steps involved
  • [23:09:23] <ldesnogu> you are directly writing into FB memory in a linear way (increasing/descreasing addresses)?
  • [23:10:05] <ldesnogu> well if the steps are the limiting factor forget my question
  • [23:10:16] <zodttd> Aha, no I wasnt writing to the fb linearly, I was wondering if that will decrease performance significantly...I had a strong feeling it would...glad you brought that up
  • [23:10:48] <ldesnogu> it will degrade speed I guess, because it might kill the store buffer
  • [23:10:48] <zodttd> Since I was writing by scanline effectively
  • [23:10:58] <zodttd> I chose not to write in a linear fashion
  • [23:11:43] <ldesnogu> IIRC Koen showed some lmbench results with around 400 MB/s for writes
  • [23:11:56] <zodttd> Hmm
  • [23:12:05] <ldesnogu> if you are much lower than that, you should start looking more closely
  • [23:12:58] <zodttd> hmmm
  • [23:12:59] <zodttd> ok
  • [23:13:02] <ldesnogu> however the BW might be limited by the video chip accessing the FB at the same time for refreshing the screen
  • [23:13:12] <ldesnogu> is there double buffering?
  • [23:13:16] <zodttd> Would OpenGL help?
  • [23:13:24] <zodttd> Im just writing to a mmap'd FB
  • [23:13:32] <zodttd> fb0 to be precise
  • [23:13:46] <zodttd> Im not double buffering
  • [23:14:10] <ldesnogu> that's *might* be a problem
  • [23:14:39] <zodttd> should I write my rotated buffer to another buffer, then blit that to fb0?
  • [23:14:43] <ldesnogu> I have no HW, so I can't say for sure, and no one has ever answered my questions about the L3 interconnect :)
  • [23:14:53] <zodttd> hehehe
  • [23:15:10] <zodttd> if I knew I'd answer :P
  • [23:15:37] <ldesnogu> you could perhaps just try to fill the mmap'ed FB at max speed, and see what bandwidth you get
  • [23:15:45] <jkridner|work> I probably would too.
  • [23:16:08] <ldesnogu> then do the same in "standard memory"
  • [23:16:33] <jkridner|work> what are you calling "standard"
  • [23:16:38] <jkridner|work> non-fb?
  • [23:16:42] <ldesnogu> yes :)
  • [23:18:04] <zodttd> Probably easier to just do the double buffering and check the FPS on my app in this case heh
  • [23:18:16] <zodttd> since I check the blit ms
  • [23:18:37] <ldesnogu> (I am not sure there will be a difference since the FB is in RAM anyway)
  • [23:18:48] <zodttd> oh true
  • [23:19:46] <ldesnogu> anyway try to write linearly that should be better
  • [23:20:39] <zodttd> Ok will do...
  • [23:21:15] <zodttd> Any other ways to trick out this EVM ;P
  • [23:21:33] <ldesnogu> bug Imgtec to have HW specs :p
  • [23:21:56] <jkridner|work> ldesnogu: thanks for not saying TI. :)
  • [23:22:05] <zodttd> Heh
  • [23:22:24] <ldesnogu> jkridner|work: I have already asked here so I won't ask again...
  • [23:22:36] <ldesnogu> at least not until I get HW :)
  • [23:23:21] <ldesnogu> there is an imgtec guy on the Pandora forum
  • [23:23:37] <zodttd> (hehe)
  • [23:23:43] <ldesnogu> so perhaps they are working on their drivers for real...
  • [23:27:07] <ldesnogu> gn all
  • [23:27:10] * ldesnogu (n=ldesnogu@ven06-2-82-247-86-183.fbx.proxad.net) has left #beagle
  • [23:40:06] * NishanthMenon (n=gnat@nat/ti/x-a981c6ac00cc9dc8) Quit ("If your not living on the edge, you're taking up too much space")
  • [23:52:04] * richardw (n=richardw@nat/ti/x-8656d9a82dad380b) has left #beagle
  • [23:52:12] * RogerMonk (n=a0740758@nat/ti/x-9a82580c403ba0a0) has joined #beagle
  • [23:56:05] * RogerMon1 (n=a0740758@nat/ti/x-822582ba7410d9e6) has joined #beagle
  • [23:56:08] * RogerMon2 (n=a0740758@nat/ti/x-8680799168238e41) Quit (Remote closed the connection)