**** BEGIN LOGGING AT Wed Nov 28 02:59:59 2012
Nov 28 03:22:20 <diego> I am looking for some help with the first i2c bus on the BeagleBone.
Nov 28 03:24:40 <alan_o> diego: ok
Nov 28 03:25:48 <alan_o> diego: I probably won't be able to help since I haven't messed with i2c on the bone, but if you state your problem more specifically, there's a better chance someone else might
Nov 28 03:26:20 <diego> Hey Alan. I'm working on a school project and I need access to a second i2c bus. I spent a good portion of Thanksgiving trying to recompile the kernel. I failed.
Nov 28 03:26:43 <diego> So, I tapped off the U7 chip on the BB. The SDA and SCL lines.
Nov 28 03:27:06 <alan_o> Which one is u7?
Nov 28 03:27:12 <alan_o> you mean you soldered right to one of the chips?
Nov 28 03:28:30 <diego> Yes. I should state that I'm an EE student and part of my job is to solder under a microscope. So, that was the easy part for me. It's the Variable & MAC memory EEPROM.
Nov 28 03:28:54 <diego> So, I can see the device now using i2cdetect.
Nov 28 03:29:05 <alan_o> ok
Nov 28 03:29:09 <alan_o> sounds like the hard part is done :)
Nov 28 03:29:14 <alan_o> why didn't you use the bus on the connector?
Nov 28 03:29:19 <alan_o> address conflict?
Nov 28 03:30:05 <alan_o> looks like i2c1 and i2c2 are both on the expansion connector
Nov 28 03:30:42 <diego> But I get read/write errors when I run the C program. I wrote this C program. So I just need to change the file /dev/i2c-3 to i2c-1.
Nov 28 03:31:07 <diego> Yes, address conflicts.
Nov 28 03:32:06 <diego> i2c-1 is I2C_0 on the schematics. They don't have headers for those.
Nov 28 03:32:09 <alan_o> so I'm assuming you have two  devices of the same type. Can you access one of them and not the other?
Nov 28 03:32:50 <alan_o> or does it conflict with some other type of device?
Nov 28 03:33:21 <diego> Yes. The one on bus 3 I can access. The same device on bus 1 I can not access. No other conflicts.
Nov 28 03:33:29 <alan_o> hmm
Nov 28 03:33:31 <alan_o> that's strange
Nov 28 03:33:45 <alan_o> are the permissions the same on /dev/i2c*
Nov 28 03:33:46 <alan_o> ?
Nov 28 03:33:49 <alan_o> what eror are you getting?
Nov 28 03:33:57 <alan_o> can you perror()?
Nov 28 03:34:14 <diego> Yup. And I can see it with i2cdetect. So, I'm wondering if it's disabled on i2c-1.
Nov 28 03:34:27 <diego> Let me check. I think the permissions were the same.
Nov 28 03:35:00 <diego> Yes, permissions are the same.
Nov 28 03:35:11 <alan_o> diego: can you read the eeprom?
Nov 28 03:36:21 <alan_o> also, what is perror() telling you after the read/write that is failing?
Nov 28 03:36:31 <diego> I haven't tried that. I get a perror.. Write error: Remote I/O error and Read error: Remote I/O error.
Nov 28 03:36:45 <alan_o> I assume you didn't get the i2c lines crossed
Nov 28 03:37:05 <diego> Nah, cause i2cdetect sees the device. 68.
Nov 28 03:37:16 <alan_o> is the device bad?
Nov 28 03:37:19 <alan_o> hmm
Nov 28 03:37:35 <diego> Nope. Switched it to the third bus and it works fine.
Nov 28 03:39:10 <alan_o> diego: is there a termination issue?
Nov 28 03:39:21 <alan_o> since you're pulling it off another device?
Nov 28 03:39:41 <diego> Also, I wrote a quick C program that just opens and closes the /dev/i2c-1 file and it doesn't give any errors.
Nov 28 03:40:07 <diego> Mmmmm. I don't think i2c needs a termination?
Nov 28 03:42:45 <alan_o> diego: a quick google for i2c termination yields some interesting results
Nov 28 03:42:52 <alan_o> diego: I don't know the electrical details for sure
Nov 28 03:43:47 <diego> Let me take a look.
Nov 28 03:44:46 <alan_o> diego: check out the presentation and resources for "Board Bringup: You, Me and i2c" on this page http://elinux.org/ELCE_Europe_2012_Presentations
Nov 28 03:44:51 <alan_o> prpplague: did that
Nov 28 03:45:00 <alan_o> I mean, prpplague did that
Nov 28 03:45:14 <alan_o> he has some stuff in there about termination
Nov 28 03:48:10 <alan_o> diego: look at page 46 in the ELC-E slides on that page
Nov 28 03:48:29 <alan_o> he shows 1k pull-ups from scl and sca to +vdd
Nov 28 03:49:07 <alan_o> diego: I'm assuming those would be there, since the eeprom works
Nov 28 03:49:26 <diego> Yeah, I believe those are there. I'm trying to find the schematics from ciruito.
Nov 28 03:49:58 <alan_o> is it read error _and_ write error?
Nov 28 03:50:04 <alan_o> or just write?
Nov 28 03:50:11 <alan_o> and is it an eeprom?
Nov 28 03:50:30 <alan_o> meaning does the write-protect have to be set (or un-set)?
Nov 28 03:51:15 <diego> I'm trying to read from a gyro/accel. MPU6050.
Nov 28 03:51:19 <diego> Both read and write.
Nov 28 03:51:53 <alan_o> could the gyro be rejecting the write somehow?
Nov 28 03:52:21 <alan_o> not sure exactly the number of things that Remote I/O Error could be caused by
Nov 28 03:52:44 <diego> Yeah, 1.5k resistors are there.
Nov 28 03:53:03 <diego> I don't see why. I can plug it into the bus 3 and read/write just fine.
Nov 28 03:53:30 <alan_o> That's the only difference?
Nov 28 03:53:40 <alan_o> just scl and sda in different holes?
Nov 28 03:53:44 <prpplague> diego: many SoC's have internal resistors pull ups
Nov 28 03:53:53 <prpplague> diego: as well as many level shifters
Nov 28 03:54:02 <prpplague> diego: if you have too little the line will not pull high
Nov 28 03:54:13 <prpplague> diego: if you have too much, it won't be able to be forced low
Nov 28 03:54:36 <alan_o> oh
Nov 28 03:54:41 <diego> Mmmmmm. Yeah, I usually use a 3k or 5k resistor.
Nov 28 03:54:42 <prpplague> diego: the speed at which the i2c is configured will also be affected
Nov 28 03:55:04 <alan_o> prpplague: doesn't the stuff on the expansion connectors have level shifters to 3.3v?
Nov 28 03:55:05 <prpplague> diego: the rise/fall time will be affected by the value of the resistor
Nov 28 03:55:06 <diego> I've checked the speed. They are both the same.
Nov 28 03:55:18 <alan_o> prpplague: where the SOC is 1.8v?
Nov 28 03:55:21 <prpplague> alan_o: just speaking in general
Nov 28 03:55:23 <ds2> stop screwing with userland
Nov 28 03:55:27 <ds2> i2c does not require terminaltion
Nov 28 03:55:57 * prpplague looks 
Nov 28 03:55:58 <diego> It's connected to 3.3V.
Nov 28 03:56:02 <prpplague> did they say termination?
Nov 28 03:56:11 <prpplague> ahh right
Nov 28 03:56:18 <ds2> the am33x as bone configured has I/O at 3.3V
Nov 28 03:56:18 <prpplague> diego: it is not terminal
Nov 28 03:56:22 <prpplague> diego: termination
Nov 28 03:56:40 <ds2> I2C is open drain drivers, it needed to be treated as such.
Nov 28 03:57:05 <alan_o> ds2: I thought jkridner told me that was external to the SOC. maybe I misunderstood. I haven't looked it up
Nov 28 03:57:28 <ds2> look at the schematics
Nov 28 03:57:34 <ds2> there are no external level shifters
Nov 28 03:57:47 <ds2> the DM37x chips are 1.8V I/O
Nov 28 03:57:53 <alan_o> ds2: ok, I must have misunderstood him then.
Nov 28 03:57:56 <diego> The beaglebone has 1.5k resistors. The device I am using has 10k resistors built in to the board.
Nov 28 03:57:57 <ds2> some AM35x can be configured for either
Nov 28 03:58:14 <diego> I'm looking at the schmatics. It says 3.3V>
Nov 28 03:59:10 <alan_o> I wonder if the pull is different though.
Nov 28 03:59:23 <alan_o> on i2c0 vs i2c1 and 2
Nov 28 03:59:26 <alan_o> in the pinmux
Nov 28 03:59:36 <alan_o> diego: which kernel/
Nov 28 03:59:36 <alan_o> ?
Nov 28 03:59:57 <diego> ArchLinux 3.2.21-3
Nov 28 04:00:19 <alan_o> diego: hmm... i wonder if that's derived from the angstrom TI kernel
Nov 28 04:01:28 <prpplague> diego: most likely the internal pullups are configured differently
Nov 28 04:01:30 <diego> I have no idea on that one. At this point, I think the hardware is ok. I say this only cause I can see the device using i2cdetect. So I'm wondering if they have done something in the kernel that is stopping me.
Nov 28 04:03:39 <alan_o> I only see i2c1 and i2c2 in the board-am335xevm.c file
Nov 28 04:04:17 <alan_o> ugh... I hate this off-by-one stuff...
Nov 28 04:04:40 <alan_o> i2c1, in the kernel anyway, is the one that the board eeprom is connected to
Nov 28 04:06:24 <diego> Yeah, it threw me off. Yes, and i2c1 is I2C0 on the schematics. So if I do an i2cdetect on the /dev/i2c-1, I see the EEPROM, TPS65217, a few other devices, and my device 68.
Nov 28 04:06:25 <alan_o> diego: i2c1 is pulled odwn
Nov 28 04:06:45 <alan_o> and i2c2 is pulled up
Nov 28 04:06:54 <alan_o> in the 3.2 TI kernel
Nov 28 04:07:16 <alan_o> and I don't see anything about i2c3
Nov 28 04:07:48 <diego> We're talking about i2c1 on the OS which is I2C0 on the schematics?
Nov 28 04:08:04 <alan_o> well....
Nov 28 04:08:26 <diego> I don't see how that could be true. It is pulled up on the schemtics with 1.5k resistors.
Nov 28 04:08:54 <alan_o> I mean in the pinmux
Nov 28 04:10:02 <diego> Let me see if i remember how to check the pinmux and can doubele check that.
Nov 28 04:10:27 <alan_o> diego: it's in /sys, can't remember off hand exactly the path
Nov 28 04:10:35 <alan_o> I think it's omap_mux or omap_pinmux or something like that
Nov 28 04:10:39 <alan_o> I don't run that kernel
Nov 28 04:12:18 <diego> Yes, it has them as pulldown.
Nov 28 04:12:36 <alan_o> diego: and the other i2c where it works?
Nov 28 04:13:35 <diego> Huh?
Nov 28 04:13:47 <alan_o> you have 2 i2c buses, right?
Nov 28 04:13:53 <alan_o> one where it works, and one where it doesn't
Nov 28 04:14:04 <alan_o> what's the pinmux look like on each
Nov 28 04:14:10 <alan_o> ?
Nov 28 04:14:46 <alan_o> oh... hey, I see all 3 buses now... one is named funny in the board file, sorry
Nov 28 04:14:58 <alan_o> let me check what the pinmux looks like on 3 in the kernel
Nov 28 04:17:12 <alan_o> ok, i2c2 (/dev/i2c-3) is pulled up, that's what we saw before, and i2c0 (/dev/i2c-1) is pulled down.
Nov 28 04:18:16 <diego> Yes. I see the same thing. And I think I've changed it correctly. Let me verify.
Nov 28 04:18:18 <alan_o> actually wait.. sorry... it looks like i2c0 (/dev/i2c-1) is not pulled at all
Nov 28 04:18:26 <alan_o> dang it
Nov 28 04:18:29 <alan_o> scratch that
Nov 28 04:18:43 <alan_o> negative logic.
Nov 28 04:18:47 <alan_o> original was right
Nov 28 04:19:13 <alan_o> to be clear, I'm asserting: i2c2 (/dev/i2c-3) is pulled up, that's what we saw before, and i2c0 (/dev/i2c-1) is pulled down.
Nov 28 04:19:49 <diego> Correct. That is what I saw. So I've pulled the resistors rhight on i2c0.
Nov 28 04:20:20 <alan_o> diego: you mean you changed the pinmux?
Nov 28 04:20:32 <alan_o> or the physical resistors?
Nov 28 04:20:56 <diego> Yes. I orginally thought this was my problem, so I learned how to change the pinmux.
Nov 28 04:21:28 <alan_o> diego: it's still your problem, (in the sense that it's not _my_ problem) :-D
Nov 28 04:22:21 <alan_o> diego: ok, so can you verify that the pinmux is the same between both buses?
Nov 28 04:23:31 <diego> Ha. Yes, I have changed it to pull up. Just working on changing everything back to bus 1 for testing. Another minute or so.
Nov 28 04:26:10 <diego> I'm not so sure that was a good idea. It's taking forever to do a i2cdetect scan.
Nov 28 04:26:23 <alan_o> errrmm...
Nov 28 04:26:40 <alan_o> I wonder if that messed up the eeprom that's on the board somehow
Nov 28 04:26:51 <alan_o> I mean if that circuit is messing up with the pull in the other direction
Nov 28 04:27:01 <alan_o> I mean given the assumption that it was pulled that way for a reason.
Nov 28 04:27:37 <diego> Yeah. I'm thinking that maybe that was that way for a reason.
Nov 28 04:27:54 <alan_o> if you look at the bone SRM
Nov 28 04:27:58 <alan_o> page 52
Nov 28 04:28:07 <alan_o> shows the eeprom circuit
Nov 28 04:28:20 <alan_o> looks like it's direct connected
Nov 28 04:28:39 <alan_o> so mdp will hate me for this.....
Nov 28 04:28:47 <alan_o> ever think of i2c-gpio ?
Nov 28 04:28:49 * alan_o ducks
Nov 28 04:32:06 <diego> I2C0 (schematics) is connected to the EEPROM, TPS6512, and one or two more things. With a 1.5k resistor. I'm wondeing if introducing that 10k from the gyro is causing the resistance to be too low?
Nov 28 04:32:31 <diego> I spent about 2 days trying to recompile that kernel. I couldn't get it to work.
Nov 28 04:33:11 <alan_o> diego: not sure what recompiling the kernel will get you
Nov 28 04:33:37 <alan_o> diego: but... http://www.angstrom-distribution.org/building-angstrom
Nov 28 04:33:43 <ds2> rather then all this shannigans...
Nov 28 04:33:57 <ds2> why not just boot up kernel. and put a scope on SCL and SDA?
Nov 28 04:34:10 <alan_o> ds2: that's a great idea
Nov 28 04:34:12 <emeb_mac> software people. :)
Nov 28 04:34:22 <alan_o> emeb_mac: I know... they're the _worst_
Nov 28 04:34:26 <diego> Recompiling should give me access to the second bus. I know a lot of the kernels have that enabled now.
Nov 28 04:34:28 <ds2> if it is floating or grounded, you have a problem
Nov 28 04:35:13 <diego> I could run to school and try that. I wouldn't think that's the problem though. It can see the EEPROM chip on the bus.
Nov 28 04:35:20 <alan_o> ds2: it's pull-down. not sure how it's working at all then, but I haven't looked at the schematics
Nov 28 04:35:40 <ds2> the pull's in the pinmux are weak
Nov 28 04:35:46 <ds2> they can be overriden by external ones
Nov 28 04:36:01 <ds2> and 1.5K is a relatively strong pull for I2C so it is probally overriding the internal stuff
Nov 28 04:36:04 <alan_o> diego: are you asserting that you have only buses 0 and 2 (/dev/i2c-1 and /dev/i2c-3) ?
Nov 28 04:37:00 <diego> ds2, do you think 1.3k is strong enough? I'm saying 1.3k cause of the 1.5k and 10k in parallel. That's how it's connected right now.
Nov 28 04:37:12 <diego> That is correct alan.
Nov 28 04:38:36 <alan_o> diego: I see that now
Nov 28 04:38:38 <alan_o> argh
Nov 28 04:38:47 <alan_o> i2c1 isn't turned on for beaglebone
Nov 28 04:39:00 <ds2> 1.5K is too strong, IMO
Nov 28 04:39:08 <ds2> 4K7 should suffice
Nov 28 04:39:41 <ds2> stronger would be good if you have a long cable but that isn't the case here
Nov 28 04:40:51 <diego> Yes, i2c1 isn't on the headers. i2c2 is, but it requires a kernel recompile.
Nov 28 04:41:00 <diego> 4.7k?
Nov 28 04:41:12 <ds2> read the app notes from the I2C guys
Nov 28 04:41:22 <alan_o> diego: with the instructions at the angstrom website I linked before, you should be able to build the kernel..
Nov 28 04:41:37 <alan_o> give that a go, and there are people here who can help with that if needed
Nov 28 04:41:41 <alan_o> well... one guy :)
Nov 28 04:42:04 <alan_o> diego: (and that guy isn't me)
Nov 28 04:42:33 <diego> Yeah, I think they recomemded from 1.5k to 4.7k.
Nov 28 04:43:01 <diego> Hahaha. Yeah, I just don't have the time to really try that again.
Nov 28 04:43:32 <diego> Plus, at this point I really want to stick with Archlinux. We have done a lot of work already with that OS.
Nov 28 04:43:45 <alan_o> diego: can run arch with angstrom kernel, most likely
Nov 28 04:44:00 <alan_o> diego: no way to change the address of this gyro?
Nov 28 04:44:14 <alan_o> diego: what part is it?
Nov 28 04:44:37 <diego> I have no clue how to drop in the angstrom kernel and use that instead.
Nov 28 04:44:54 <diego> It has two addresses. But I have 3 devices that use the same address.
Nov 28 04:45:10 <diego> 2x MPU6050 and a RTC. All on 68.
Nov 28 04:45:20 <ds2> gyro?
Nov 28 04:45:26 <ds2> ewwwwwwwwwwww MPU6050
Nov 28 04:45:37 <diego> Yes, 3 axis gyro/accel. Ha. Not a fan?
Nov 28 04:45:41 <alan_o> ds2 hates everything :)
Nov 28 04:45:44 <ds2> not a fan of that part
Nov 28 04:45:52 <ds2> drivers are a pain on that part
Nov 28 04:46:18 <alan_o> I'm not buying any parts that ds2 doesn't like.
Nov 28 04:46:51 <diego> Drivers? They make drivers? I had to write all the i2c commands and convert an Arduino library over to use this part. And I'm an EE student. Not a programmer.
Nov 28 04:46:54 <ds2> why are you even using that part?
Nov 28 04:47:12 <ds2> do you need the low noise density on that or?
Nov 28 04:48:47 <diego> The project is as following... We're putting all sorts of sensors (gyro/accel, 1 axis accel, temperature, heart rate, etc) on both a vehcile and driver of an offroad car the Mechicals engineers build.
Nov 28 04:49:01 <diego> Then we are transmitting all the data to a pit computer.
Nov 28 04:49:38 <diego> We chose it cause cause we knew we could interface it with the Arduino. Then we decided to use the BB for all the processing and make it wireless.
Nov 28 04:52:01 <ds2> why not use a L3G4200?
Nov 28 04:52:09 <ds2> drivers for those abound and they are relatively painless
Nov 28 04:52:46 <diego> Honestly, cause I'm a student and didn't know better. There a million things I would of changed at this point.
Nov 28 04:53:19 <diego> But the project is due in a few weeks. They know we won't it have 100%. It was a big project for 2 EE's and a computer engineer to take on.
Nov 28 04:53:49 <ds2> don't seem that complex... 1 EE and a CE should bea ble to have it done in a few weeks
Nov 28 04:55:48 <diego> I wish. They don't teach much programming at our school. They focus on the power side. So I taught myself a lot of C in the past few months. The other EE had to learn to interface with a bluetooth device and reverse engineer the packet struture to get back heart rate, pulse, etc.
Nov 28 04:56:16 <diego> We had to build a circuit board to connect everything to. Pick out sensors. Build the device.
Nov 28 04:56:33 <diego> And our CE has been doing all the displaying.
Nov 28 04:57:18 <alan_o> diego: for where you're at right now, I see a few options
Nov 28 04:57:33 <alan_o> 1. figure out the pull problems on i2c0
Nov 28 04:57:47 <alan_o> 2. use i2c1 (requires kernel rebuild/modify)
Nov 28 04:58:02 <alan_o> 3. use spi-gpio (requires kernel rebuild/modify)
Nov 28 04:58:15 <ds2> and watch the configuration of the MPU6050
Nov 28 04:58:16 <alan_o> 4. use a different part
Nov 28 04:58:22 <ds2> there are black magic in that chip
Nov 28 04:59:25 <ds2> that part can become a I2C master or switch out I2C buses
Nov 28 04:59:58 <alan_o> diego: do what ds2 said too, and scope i2c0
Nov 28 05:01:23 <diego> I think I may run to work tomorrow and change out the 10k for 100M. That should be 1.5k. Or take it out competely.
Nov 28 05:01:30 <diego> Or learn to recompile the kernel.
Nov 28 05:02:31 <ds2> just pop the resistor
Nov 28 05:02:55 <diego> What should I look for if I connect the scope? I feel like if i2cdetect picks it up, it has to be see it. I think the resistance is too low and it can't keep up with the speed.
Nov 28 05:03:09 <ds2> what speed is the bus?
Nov 28 05:03:11 <ds2> 100KHz?
Nov 28 05:03:26 <diego> Yes.
Nov 28 05:04:13 <ds2> there might be registers on that part that isn't configured right
Nov 28 05:04:20 <ds2> you can get really really wierd on that part
Nov 28 05:04:58 <diego> But I can disconnect it from bus 1 and connect it to bus 3 and communicate just fine with it.
Nov 28 05:05:22 <diego> The only thing different is the pullup resistance at that point. Or something in the OS.
Nov 28 05:06:05 <ds2> reset timing matters
Nov 28 05:06:44 <diego> Reseting the timer where?
Nov 28 05:06:56 <beaglexmenthu> i need some help with the interrupt controller on the beagle xm
Nov 28 05:07:08 <ds2> reset timing
Nov 28 05:07:10 <ds2> not the timer
Nov 28 05:07:24 <ds2> I have seen behavior on that part where reset matters a lot
Nov 28 05:08:15 <diego> Sorry, I'm confused about what timing you are talking about?
Nov 28 05:08:23 <ds2> reset timing
Nov 28 05:08:27 <ds2> timing of the reset of the device
Nov 28 05:08:36 <ds2> that sucker has internal microcontrollers on there
Nov 28 05:09:14 <diego> Ahhhh. Gotcha. Yeah, the first thing I do when I enter the program I wrote is reset the timer and wait some time. Maybe bump that up?
Nov 28 05:09:31 <ds2> for some reason that doesn't always work
Nov 28 05:09:45 <ds2> I forgot the details on the 6050 but the 3050 had intersting 'stuff'
Nov 28 05:10:08 <ds2> there might be erratas
Nov 28 05:10:17 <diego> Gotcha. I can try bumping that up. I think it's set at 200uS right now.
Nov 28 05:10:45 <ds2> you could be doing something that is not well defined ude to errats
Nov 28 05:10:48 <beaglexmenthu> does anyone have good knowledge about the interrupt controller on the beagle xm?
Nov 28 05:10:51 <ds2> full docs for that chip is not available
Nov 28 05:11:16 <beaglexmenthu> i am running an embedded os in beagle xm rev a, and i am using u-boot to boot into the os.
Nov 28 05:11:30 <beaglexmenthu> but i can't enable interrupts!
Nov 28 05:11:33 <diego> Even though it works fine on the other bus?
Nov 28 05:11:34 <ds2> the interrupt controller is in the DM3730
Nov 28 05:11:46 <ds2> diego: yep. errata are funny like that
Nov 28 05:11:46 <beaglexmenthu> the code hangs at the "msr cpsr....
Nov 28 05:11:53 <beaglexmenthu> instruction
Nov 28 05:12:02 <ds2> mrc
Nov 28 05:12:03 <ds2> mcr
Nov 28 05:12:38 <beaglexmenthu> i can't reset the irq and fiq bits in the cpsr..
Nov 28 05:12:48 <beaglexmenthu> anyone has any idea why this might be happening?
Nov 28 05:14:00 <diego> Nope. I changed the time after the reset. It right away gives me Write Error: Remote I/O error.
Nov 28 05:14:17 <beaglexmenthu> anyone?
Nov 28 05:14:19 <diego> As soon as it sends the command.
Nov 28 05:14:45 <ds2> beaglexmenthu: are you in supervisor mode?
Nov 28 05:15:40 <ds2> diego: 'sends' the command?
Nov 28 05:15:57 <diego> [root@alarm i2c2]# ./gyro_accel  Full reset Write Error: Remote I/O error Write Error: Remote I/O error
Nov 28 05:16:26 <ds2> that doesn't mean anythin gother then to the author of the program
Nov 28 05:16:29 <beaglexmenthu> yes the os starts in supervisor mode
Nov 28 05:16:46 <beaglexmenthu> i also tried system mode but that does not help
Nov 28 05:17:01 <ds2> how are you trying to clear it?
Nov 28 05:17:21 <diego> Sorry. The program prints "Full reset" and then issues the command to reset it. As soon as the command is issued, it returns the error.
Nov 28 05:17:25 <beaglexmenthu> using the " msr " instruction
Nov 28 05:18:12 <ds2> what is 'issues the command'?
Nov 28 05:18:28 <ds2> IIRC, it doesn't take 'commands'. it takes registers to configure
Nov 28 05:19:14 <ds2> checked the Linux arm code?
Nov 28 05:21:50 <diego> I write to the power management register - device reset bit.
Nov 28 05:23:11 <ds2> okay, and how does that relate to the error?
Nov 28 05:25:03 <diego> I've created a i2c header file that excutes my commands. this file opens /dev/i2c-1, reads or writes to the device, and closes the file. It gives me the error while reading or writing.
Nov 28 05:27:04 <ds2> hmmmm
Nov 28 05:27:09 <diego> It can open and close the file without errors. But reading and writing gives me errors.
Nov 28 05:27:40 <ds2> hmmm
Nov 28 05:27:50 <ds2> look at it on the scope and see if there are electrical issues then
Nov 28 05:27:55 <ds2> i.e. run it, and do a capture
Nov 28 05:28:37 <diego> If I try that in the next few minutes, will you still be around?
Nov 28 05:29:22 <ds2> maybe
Nov 28 05:29:27 <ds2> doing a bunch of stuff
Nov 28 05:30:08 <diego> I can run to school. Or do this tomorrow. School is only a few minutes away.
Nov 28 05:30:54 <ds2> oh... you don't live in the lab? :D
Nov 28 05:31:07 <ka6sox> under the bench?
Nov 28 05:31:29 <diego> I'm not joking when I say my apartment is a block away from the Eng building. It's basically my second home.
Nov 28 05:31:58 <ds2> back when I was in school, I lived in the lab :(
Nov 28 05:32:29 <alan_o> What's your alma mater, ds2?
Nov 28 05:32:42 <ds2> some small school in SoCal
Nov 28 05:32:51 <ds2> and I used to live on campus too
Nov 28 05:34:35 <ka6sox> Tangerine Tech?
Nov 28 05:34:41 <diego> Well, ds2 I'll leave it up to you if you can keep helping me tonight. If not, it's off to bed.
Nov 28 05:35:01 <ds2> heh... sure
Nov 28 05:35:49 <alan_o> you guys have fun. I'm out.
Nov 28 05:35:57 <diego> Thanks for the help alan.
Nov 28 05:36:15 <alan_o> diego: no problem man. Wish I could have helped more, but ds2 knows way more than I do :)
Nov 28 05:36:33 <ds2> <--- clueless dummy
Nov 28 05:37:05 <alan_o> still managed to save my bacon the other night
Nov 28 05:37:13 <diego> Ha.
Nov 28 05:37:13 <alan_o> (tickless kernel)
Nov 28 05:40:52 <diego> Alright. Heading there now. Back in a few.
Nov 28 05:55:17 <diego> Back.
Nov 28 06:01:11 <dfarmer> Is there a trick to getting bitbake to perform deploy?  I'm following the example here: http://www.slimlogic.co.uk/2011/05/openembeddedangstrom-kernel-workflow/, and although I've forced recompile deploy doesn't think it needs to run (and -f doesn't seem to work with deploy) any other ideas?
Nov 28 06:10:28 <Pedroo> a
Nov 28 06:23:28 <diego> ds2, still up?
Nov 28 06:24:08 <diego> ds2, still up?
Nov 28 06:27:08 <ds2> ?
Nov 28 06:28:20 <diego> Well... When I connect the scope to the SCL line, I can't see the device on when I run i2cdetect.
Nov 28 06:28:59 <ds2> what kind of probe?
Nov 28 06:29:35 <diego> Standard tektronix 10x probe.
Nov 28 06:29:43 <ds2> Okay
Nov 28 06:29:46 <ds2> is it low or high?
Nov 28 06:30:06 <ds2> or what is the voltage on it
Nov 28 06:32:45 <diego> I don't remember. Even more odd, there is some kind of noise on the line. It looks like a sawtooth wave. Let me unplug everything real fast and try again.
Nov 28 06:34:22 <ds2> sawtooth?
Nov 28 06:34:33 <ds2> it almost seem like you don't have any pull ups on there
Nov 28 06:34:40 <diego> Yeah. Super odd.
Nov 28 06:34:52 <diego> Well, the board has a 1.5k and device has a 10k.
Nov 28 06:35:54 <exuvo> what speed are you running it at?
Nov 28 06:36:17 <diego> No joke, 6V peak to peak crappy looking sawtooth shows on both SDA and SCL.
Nov 28 06:36:23 <diego> 100kHz.
Nov 28 06:36:47 <ds2> 6V P-P?
Nov 28 06:36:56 <ds2> is this on a long long long cable?
Nov 28 06:37:40 <ds2> is it +3/-3?
Nov 28 06:37:49 <diego> Nope. Maybe 3 inches.
Nov 28 06:38:09 <ds2> sure probe is sane?
Nov 28 06:38:18 <ds2> does it look good if you connect it to the scope's cal signal?
Nov 28 06:38:20 <exuvo> you have enough pull up for that speed unless the capacitance is really high
Nov 28 06:38:44 <diego> Ok, I grounded the scope and that went away.
Nov 28 06:38:51 <ds2> @#%$#@$@#$@#%@$%$@#%$@#%%^#$%#@$%@#$$@#$@#$%@#$@#$
Nov 28 06:38:57 <ds2> is it high or low?
Nov 28 06:42:02 <diego> High on sda.
Nov 28 06:42:12 <diego> High on sda.
Nov 28 06:42:35 <diego> and slc.
Nov 28 06:43:55 <exuvo> what problem are you having with your i2c ?
Nov 28 06:45:36 <diego> I keep getting remote i/o errors.
Nov 28 06:46:19 <ds2> okay
Nov 28 06:46:27 <ds2> set the scope to capture on an edge
Nov 28 06:46:31 <ds2> then run your program
Nov 28 06:46:47 <diego> Currently doing that. Pulses look ok.
Nov 28 06:46:50 <ds2> use as slow of a rate as possible
Nov 28 06:46:56 <ds2> and capture a long train
Nov 28 06:47:50 <diego> Currently doing that. Pulses look ok.
Nov 28 06:48:02 <diego> Currently doing that. Pulses look ok.
Nov 28 06:52:37 <exuvo> I take it you don't have a logic analyzer at hand. Do you know if the error is only during writes/read/both ? im not used to working with i2c in linux but have used it a lot with avr and smaller arm chips
Nov 28 06:54:30 <diego> Sorry. Looks like wireless took a crap.
Nov 28 06:54:37 <diego> Ok, I have a capture.
Nov 28 06:55:53 <exuvo> repeat if you missed it: I take it you don't have a logic analyzer at hand. Do you know if the error is only during writes/read/both ? im not used to working with i2c in linux but have used it a lot with avr and smaller arm chips
Nov 28 06:56:27 <diego> Nope, just an Oscope. It gives me both read and write.
Nov 28 06:58:48 <exuvo> if you run it at a really slow speed does the problem go away? does it work sometimes and others not or is it always failing?
Nov 28 06:59:21 <diego> If I just run i2cdetect over and over, it can't always find the device. I say 1 of 4 times, it disappears.
Nov 28 06:59:44 <diego> Also if I connect the oscope to the SCL, it won't find the device at all.
Nov 28 07:03:37 <ds2> hmmmmm
Nov 28 07:03:39 <exuvo> if you are able to slow the speed down to only a few Hz it should be able to cope with the scope attached while also making it easier to see whats going on if its not speed related error.
Nov 28 07:03:50 <ds2> reduce the pull up to 4K7 and see
Nov 28 07:03:56 <ds2> could be the drivers are not strong enough
Nov 28 07:04:08 <ds2> how's the slew rate and ringing look?
Nov 28 07:04:49 <diego> I'm not sure I can go less than 100kHz.
Nov 28 07:05:39 <diego> Slew rate looks fine. Not sure what to look for ringing?
Nov 28 07:08:14 <exuvo> did i understand correctly that you have a 1.5k and a 10k pull up in parallel? if so that should be plenty for 100kHz, how long is the wireing?
Nov 28 07:08:53 <diego> A few inches.
Nov 28 07:12:20 <ds2> is this professionally assembled or you guys did the soldering yourself?
Nov 28 07:12:39 <diego> Straight from sparkfun
Nov 28 07:13:02 <exuvo> what part?
Nov 28 07:13:14 <diego> MPU6050 Gyro/accel
Nov 28 07:15:52 <exuvo> this one https://www.sparkfun.com/products/11028 ? then you would have soldered a connector or wire maybe check if any joint is bad
Nov 28 07:16:44 <diego> That's the part. It can't be that. If I switch i2c busses, it works fine.
Nov 28 07:17:18 <diego> Well, I take that back. I've had to solder onto the BB to get access to the 1st i2c bus.
Nov 28 07:23:08 <exuvo> which bus is not working, the one you soldered?
Nov 28 07:24:08 <diego> Yes. When I run i2cdetect, it finds the device. But when I run the program it gives me rea/write errors. But if I move everything to the bus with headers, it all works fine.
Nov 28 07:26:17 <exuvo> maybe its a bad joint, try redoing?
Nov 28 07:27:32 <exuvo> i assume the problem is not in initialization as it should be almost the same as the other bus that works
Nov 28 07:29:51 <diego> I don't think its the joint. I spent a few minutes tugging on wires while reading ohms across the wire. Never changes. .4Ohm
Nov 28 07:30:05 <diego> I think that's a safe assumption.
Nov 28 07:32:05 <diego> I'm out of ideas at this point. I might try taking out the 10k resistor tomorrow. Maybe redoing the soldering. But that looks fine.
Nov 28 07:32:25 <diego> Thanks for the help you guys.
Nov 28 08:05:38 <ds2> Hmmm you have both VDD and VDDIO wired, right?
Nov 28 08:46:44 <av500> gm
Nov 28 08:48:09 <panto> mourning
Nov 28 09:19:20 <av500> kalimera!
Nov 28 09:21:46 <KotH> callamari?
Nov 28 09:22:10 <keesj> Calimero!
Nov 28 09:24:15 <KotH> does anyone have that with original sound?
Nov 28 09:30:45 <keesj> in what language would that be?
Nov 28 09:32:45 <KotH> japanese
Nov 28 09:32:55 <KotH> what else could it be? ;)
Nov 28 09:32:59 <av500> italian
Nov 28 09:33:22 <KotH> italians dont make any good cartoons
Nov 28 09:33:32 <av500> yes, but calimero was italian first
Nov 28 09:33:37 <av500> then the japanese ripped it off
Nov 28 09:34:47 <KotH> well.. if you think that a soap adv is being ripped off...
Nov 28 09:35:41 <KotH> av500: http://wtfqrcodes.com/image/21028188922
Nov 28 09:35:48 <KotH> av500: do you use that on your kids?
Nov 28 09:41:43 <av500> KotH: you mean in case the leash breaks?
Nov 28 09:43:49 <panto> hey
Nov 28 09:48:20 <KotH> av500: i expect your children to be able to free themselves very soon
Nov 28 09:51:49 <KotH> lol.. professor bashing the tea party
Nov 28 09:53:51 <av500> easy prey
Nov 28 10:17:44 <KotH> yeah... but usually profs refrain from bashing
Nov 28 10:38:49 <KotH> and now bashing workoholics
Nov 28 10:57:43 <tstone> hi i have two beaglebones A6 and i have compiled the kernel from https://github.com/beagleboard/kernel.git. But i don't get a reliable ethernet connection. Had it once but now it just fails. Same software...
Nov 28 10:58:40 <panto> define reliable connection
Nov 28 10:58:45 <panto> and kernel version please
Nov 28 11:01:14 <tstone> panto: i could ping, its 3.2.32 which i thought would be stable
Nov 28 11:01:59 <panto> you still haven't defined what you mean by reliable
Nov 28 11:02:06 <panto> ping works, but what fails?
Nov 28 11:03:58 <tstone> panto: ping worked one time, but after that the boards do not see any network packages, dhclient fails, i have these "CPSW phy found: id is : 0x7c0f1\nPHY 0:01 not found" messages but these where also there in the working case...
Nov 28 11:04:54 <panto> hum
Nov 28 11:05:52 <panto> check out branch from linux-ti33x-psp-3.2.30-r17h+gitr720e07b4c1f687b61b147b31c698cb6816d72f01 git://github.com/koenkooi/linux.git
Nov 28 11:06:05 <panto> you could have some kind of wonky network interface
Nov 28 11:06:13 <tstone> panto: i just found out that by issuing a ifconfig eth0 down; ifconfig eth0 up i got a working connection ...
Nov 28 11:06:37 <panto> is it a gigabit switch you connect to?
Nov 28 11:06:40 <panto> and from which manuf?
Nov 28 11:07:09 <tstone> panto: don't know never had any problems...
Nov 28 11:08:34 <panto> welcome to the wonderful world of ethernet negotiation
Nov 28 11:08:50 <panto> got to head out
Nov 28 11:10:03 <tstone> panto: well it seems much like a driver problem to me, the ifconfig down/up seems to work everytime.
Nov 28 11:10:17 <panto> it could very well be
Nov 28 11:11:04 <panto> that's why you should either test that tree I pointed out, or something closer to mainline
Nov 28 11:11:48 <tstone> panto: tried lots of mainline stuff, failed to get mmc and network working... any pointers what to try?
Nov 28 11:12:12 <panto> on that tree I gave there's a 3-7-rebase-for-panto branch
Nov 28 11:12:26 <panto> try that, but remember it's using device tree
Nov 28 11:12:44 <tstone> panto: learn't that the hard way :-)
Nov 28 11:12:46 <panto> so you have to generate the uImage-dtb.am335x-bone target
Nov 28 11:13:10 <panto> that's pretty close to what I use, and most of everything works (sort of)
Nov 28 11:16:22 <panto> really have to go now
Nov 28 11:22:14 <av500> mru: http://www.sciencedaily.com/releases/2012/11/121127094111.htm
Nov 28 11:24:08 <mru> well, english is clearly a hybrid of many different things
Nov 28 11:24:13 <tstone> panto: will report back in a while... compiling and eating
Nov 28 11:28:54 <mru> also, in linguistics is easy to find specific examples to back up almost any theory
Nov 28 12:05:37 <bugra> can i run a jar file which is created on windows os, on beagleboard?
Nov 28 12:06:15 <mru> if you mean the java things, a jar file is just a zip file
Nov 28 12:06:27 <mru> and one does not "run" zip files
Nov 28 12:06:50 <mru> presumably it contains a java program, and this is what you wish to run
Nov 28 12:07:13 <mru> if that is the case, it doesn't matter where the zip was created
Nov 28 12:07:38 <mru> the java code needs to be compatible with your jvm and your system at large
Nov 28 12:07:43 <mru> which is rarely the case
Nov 28 12:07:53 <mru> write once, debug everywhere and all that
Nov 28 12:09:16 <bugra> "the java code needs to be compatible with your jvm" does this mean, both machines need to have same jdk?
Nov 28 12:09:24 <mru> no
Nov 28 12:18:44 <jacekowski> bugra: java as such is compatible with all JVMs in theory
Nov 28 12:19:01 <jacekowski> bugra: in practice, there is a lot of native OS dependant stuff in there
Nov 28 12:19:11 <jacekowski> bugra: if you are careful, you can have it compatible though
Nov 28 12:19:32 <av500> bugra: what does the program do?
Nov 28 12:20:07 <bugra> jacekowski: My main concern is, beagleboard has arm architecture
Nov 28 12:20:20 <mru> jacekowski: I have yet to see a non-trivial java program work outside the environment in which it was created
Nov 28 12:20:35 <bugra> so i think jar files i create on windows might be incompatible with beagleboard
Nov 28 12:20:48 <av500> bugra: what does the program do?
Nov 28 12:20:59 <av500> android x86 runs android arm SW just fine
Nov 28 12:21:04 <av500> if you stick with java
Nov 28 12:21:15 <jacekowski> bugra: as long as it's pure java without native code, it will run
Nov 28 12:21:20 <mru> lies
Nov 28 12:21:41 <bugra> av500: It will be pure java no native code
Nov 28 12:21:57 <av500> bugra: and, is yours native?
Nov 28 12:22:00 <av500> bugra: what does the program do?
Nov 28 12:23:57 <bugra> av500:i will plugin various sensors such as proximity sensors to beagleboard and display their value etc
Nov 28 12:24:13 <av500> aha
Nov 28 12:24:40 <mru> and how do you interface with those sensors from pure java, pray tell
Nov 28 12:25:02 <jacekowski> bugra: that will require native code
Nov 28 12:25:07 <av500> or sysfs
Nov 28 12:25:09 <av500> :)
Nov 28 12:25:09 <jacekowski> hmm
Nov 28 12:25:13 <jacekowski> yeah
Nov 28 12:25:19 <mru> it will require something system-dependent
Nov 28 12:25:23 <jacekowski> yeah
Nov 28 12:25:30 <jacekowski> it may be architecture independent
Nov 28 12:25:37 <jacekowski> but it will be os dependent
Nov 28 12:25:41 <jacekowski> dependant*
Nov 28 12:25:45 <jacekowski> e or a?
Nov 28 12:25:50 <av500> e
Nov 28 12:26:27 <mru> a pendant is something else
Nov 28 12:26:40 <bugra> i believe each sensor is mounted on a file under for example /dev/sensors/sensor1 (i dont recall exact name, sorry)
Nov 28 12:26:46 * av500 likes fondant
Nov 28 12:26:51 <bugra> so i will simply read from those files
Nov 28 12:27:05 <av500> that could work
Nov 28 12:27:07 <bugra> basic file operation
Nov 28 12:27:26 <jacekowski> though, you need some code in kernel to present those sensors under those device nodes
Nov 28 12:29:24 <bugra> jacekowski: I am not sure how it is handled in the kernel but in a tutorial i watched on yourtube, it was done with a single mount command
Nov 28 12:29:43 <bugra> jacekowski: then several files are created on mounted directory
Nov 28 12:30:02 <bugra> such as /dev/sensors/sensor1/value
Nov 28 12:30:49 <mru> that sounds more like sysfs
Nov 28 12:47:37 <bugra> mru: yes. i guess it is called sysfs. i found something similar to what i have in mind: http://bwgz57.wordpress.com/2012/04/01/beaglebone-how-hot-is-it/
Nov 28 12:48:31 <bugra> mru: example program simply reads /sys/devices/platform/tsc/ain2's value indefintely
Nov 28 12:53:11 <av500> bugra: no, only until the heat death of our sun
Nov 28 12:54:12 <bugra> av500: that is good enough for me :)
Nov 28 12:54:39 <mru> I had a sun workstation once, but it overheated
Nov 28 12:55:04 <panto> did it explode taking the solar system with it?
Nov 28 12:55:04 <av500> did it collapse into a white dwarf?
Nov 28 12:55:35 <mru> panto: I think the scsi controller was fried too
Nov 28 13:16:10 <av500> Crofton|work: ping?
Nov 28 13:30:14 <tstone> panto: linux-ti33x-psp-3.2.30-r17h+gitr720e07b4c1f687b61b147b31c698cb6816d72f01 as the 3.7-for-panto-rebase fail to compile for me, do you possibly have a .config for me?
Nov 28 13:31:18 <panto> isn't there a defconfig file in the root kernel directory?
Nov 28 13:31:25 <panto> cp defconfig .config && make oldconfig
Nov 28 13:33:06 <tstone> panto: ok will try these...
Nov 28 13:41:02 <mdp> "Grievre: gordonDrogon: I would definitely not want to bang them from userland in this case"
Nov 28 13:41:13 <mdp> hrm...#rpi matures
Nov 28 13:41:30 <av500> :)
Nov 28 14:33:19 <tstone> panto: it seems i am out of luck today. The 3.7-for-panto-rebase does not boot for me :-(
Nov 28 14:48:01 <tstone> panto: thanks , even if  i didn't succeed..
Nov 28 15:22:16 <alan_o> jkridner: it appears that the logging bot is down
Nov 28 15:22:47 <alan_o> http://www.beagleboard.org/irclogs/ doesn't have anything since 11/26 at 1304 UTC
Nov 28 15:23:14 <alan_o> unless my HTTP is being cached somewhere.....
Nov 28 15:23:50 * alan_o wants to know if ds2 helped diego get his i2c problem worked out last night after he went to bed.
Nov 28 15:25:54 <jkridner> alan_o: thanks
Nov 28 15:27:09 <av500> alan_o: http://pastebin.com/24dricNz
Nov 28 15:27:21 <av500> er, wait
Nov 28 15:27:31 <av500> now
Nov 28 15:27:37 <LetoThe2nd> jkridner: howdy
Nov 28 15:27:42 <alan_o> av500: thanks
Nov 28 15:27:46 <av500> np
Nov 28 15:27:58 <av500> that was in the scroll buffer
Nov 28 15:28:12 <LetoThe2nd> jkridner: got a second in private?
Nov 28 15:28:13 <jkridner> howdy LetoThe2nd
Nov 28 15:28:27 <av500> wer flüstert.....
Nov 28 15:28:27 <jkridner> as long as it isn't a support question. :)
Nov 28 15:28:36 <LetoThe2nd> jkridner: not at all.
Nov 28 15:28:36 <av500> LetoThe2nd: ask for teh codez!
Nov 28 15:29:01 <jkridner> the code is 7
Nov 28 15:29:01 <mdp> jkridner has the exact stepz
Nov 28 15:29:08 <LetoThe2nd> not 42?
Nov 28 15:29:33 <jkridner> that's just 6 x 9
Nov 28 15:29:41 <jkridner> hardly a secret.
Nov 28 15:29:54 <LetoThe2nd> hehe
Nov 28 15:30:33 <jkridner> LetoThe2nd: just got on a call. :(
Nov 28 15:30:41 <LetoThe2nd> jkridner: np, i'll be around.
Nov 28 15:31:25 <alan_o> mdp: jkridner has them? I've been barking up the wrong tree all along....
Nov 28 15:31:58 <alan_o> mdp: so my parents are hounding me about what I want for Christmas... I'm thinking an Openbench Logic Sniffer and some probes.
Nov 28 15:32:29 <alan_o> mdp: maybe some bone capes :)
Nov 28 15:32:31 <mdp> spend your own money on that...think big..like a Challenger SRT8 or similar fun car ;)
Nov 28 15:32:34 <av500> alan_o: no LEGO?
Nov 28 15:32:56 <alan_o> av500: hehe...
Nov 28 15:33:18 <mdp> alan_o, run, don't walk away from the seeedstudio probes for the OLS
Nov 28 15:33:55 <mdp> read bert's probe reviews on the sigrok.org site and be sure to get some good ones
Nov 28 15:34:35 <alan_o> mdp: ok, cool. I'll check that out. (I don't know much about scopes,probes, etc)
Nov 28 15:34:38 * mdp gives in and heads to the office
Nov 28 15:35:41 <mdp> if you breadboard a lot of stuff like I do...just getting a few strips of 20cm ribbon cable with female<->male and male<->male pins is really handy
Nov 28 15:36:00 <mdp> couple good ebay stores with that stuff stocked in .us for quick ship
Nov 28 15:36:32 <mdp> alan_o, same stuff you saw running the data-buses on my 6502 breadboard
Nov 28 15:36:39 <alan_o> mdp: yeah, I was envious of those on your demo
Nov 28 15:37:04 <alan_o> I usually just have a bunch of telephone station wire jammed in the holes
Nov 28 15:37:12 <mdp> hehe
Nov 28 15:37:34 * mdp really goes :(
Nov 28 15:39:13 <KotH> alan_o: if you get an openbench sniffer, please work on the software, please!
Nov 28 15:39:36 <alan_o> KotH: is it bad?
Nov 28 15:39:52 <alan_o> I have enough projects.....
Nov 28 15:40:00 <KotH> it's java crap that works, barely
Nov 28 15:40:13 <KotH> but the worst part is its intuitivity
Nov 28 15:40:28 <alan_o> KotH: ugh.... I don't understand why people make Java programs.
Nov 28 15:40:47 <KotH> i bought an openbench just to find out whether a 50usd LA is worth anything
Nov 28 15:40:49 <alan_o> There are so many better ways to make cross platform stuff.
Nov 28 15:41:00 <KotH> the hw is fine..more or less... but the software sucks balls the size of planets
Nov 28 15:41:14 <KotH> if i had the choice, i'd go for a sigma2
Nov 28 15:41:36 <KotH> costs only 4 times as much, but has a usable software, better triggers and....
Nov 28 15:41:44 <KotH> but no linux software though
Nov 28 15:41:56 <alan_o> It has a picture of Tux on the main page...
Nov 28 15:42:07 <KotH> alan_o: the biggest problem for me are actually the triggers.. they are very non-intuitive
Nov 28 15:42:31 <KotH> alan_o: the sigma2 software runs in wine... thats enough linux support
Nov 28 15:43:51 <alan_o> KotH: yeah, wine
Nov 28 15:46:17 <KotH> alan_o: oh.. and there specs for the logger are open, so there is support from the openbench guys as well
Nov 28 16:04:57 <panto> back
Nov 28 16:05:12 <mru> front
Nov 28 16:06:03 <ogra_> left
Nov 28 16:09:37 <LetoThe2nd> surround party?
Nov 28 16:10:48 <jkridner> LetoThe2nd: I'm back.
Nov 28 16:11:11 <LetoThe2nd> jkridner: great
Nov 28 16:11:13 <mru> jkridner: we need more people front, you can't all be back
Nov 28 16:14:27 <ogra_> for surround we also need a subwoofer
Nov 28 16:15:10 <mru> no, technically it's surround as long as you have at least one rear element
Nov 28 16:15:24 <mru> so 2f1r would count
Nov 28 16:15:49 <ogra_> yeah, but its just half the fun
Nov 28 16:16:11 <mru> subwoofers are great for annoying the downstairs neighbours
Nov 28 16:16:20 <ogra_> pink floyd did 2f1r for ages at their concerts
Nov 28 16:16:29 <ogra_> that too !
Nov 28 16:16:38 <mru> in my old flat, the ceiling lamps downstairs would shake if I turned up the volume
Nov 28 16:16:57 <ogra_> haha
Nov 28 16:27:04 <mranostay> neighbours suck
Nov 28 16:31:00 <KotH> i thought you dont swing that way?
Nov 28 16:32:42 * bradfa 's reset enforcer for bone testing, GO
Nov 28 16:34:46 <Russ> howdy neighbour!
Nov 28 16:48:29 <_av500_> SUBWOOFER!!!
Nov 28 16:52:28 <bradfa> reset enforcer works! 100ms warm software reset :)
Nov 28 16:53:09 <mranostay> bradfa: how goes the intern?
Nov 28 16:55:33 <bradfa> mranostay, intern left, I wanted to see what he'd done...
Nov 28 16:55:43 <bradfa> he was very successful in creating big git repo
Nov 28 16:55:48 <mranostay> uh oh
Nov 28 16:56:25 <mru> I'm running a script creating a git repo
Nov 28 16:56:34 <mru> it'll have ~350k commits when done
Nov 28 16:56:35 <bradfa> thankfully for me, it's all C# code so I don't have to touch it
Nov 28 16:56:51 <bradfa> mru, put on github, I will fork
Nov 28 16:58:47 <mru> 118k commits done
Nov 28 16:59:09 <mranostay> mru: you code fast
Nov 28 16:59:19 <mru> it's a script
Nov 28 16:59:25 <mru> importing a pile of data
Nov 28 17:00:40 <mru> repo size presently 3.7G
Nov 28 17:01:08 <bradfa> reset enforcer doesn't play super nice with fat fingers on the reset button...
Nov 28 17:01:20 <bradfa> skinny little fingers work OK
Nov 28 17:01:29 <bradfa> s/little/fast/
Nov 28 17:02:10 <_av500_> what does it do?
Nov 28 17:02:42 <bradfa> _av500_, one shot monostable oscillator driving a fet to pull down SYSRESET_n when SYSRESET_n has a falling edge
Nov 28 17:03:01 <bradfa> effectively, making stupid am335x reset line stay low for > 100 us like the SMSC PHY wants
Nov 28 17:03:12 <bradfa> unlike how am335x warm software reset wants
Nov 28 17:03:16 <bradfa> stupid reset circuit
Nov 28 17:03:38 <bradfa> all so blinkin' lights on Ethernet go the right way, every time
Nov 28 17:03:48 <bradfa> going on cape
Nov 28 17:04:40 <_av500_> ah
Nov 28 17:04:52 <_av500_> fixing silicon with more silicon
Nov 28 17:04:54 <bradfa> quick note going to ml shortly with part number
Nov 28 17:05:09 <bradfa> _av500_, that's how it should be!
Nov 28 17:05:10 <bradfa> :)
Nov 28 17:05:38 * _av500_ prefers SW fixes
Nov 28 17:05:53 <bradfa> _av500_, wonderful am335x silicon makes that not reasonable
Nov 28 17:06:02 <mru> if it can be fixed with a cap and a resistor, it's not really broken
Nov 28 17:06:16 <strata> C-Media usb sound cards (the cheap $2 ones) not getting any voltage across the 5v rails. they pulling too much? anyone have experience trying these on a beaglebone?
Nov 28 17:06:44 <strata> (i'm using external 5v power to the bone)
Nov 28 17:06:44 <mru> could be taking without asking
Nov 28 17:06:55 <mru> cheap usb gadgets sometimes do that
Nov 28 17:07:44 <strata> i read 4.89V across C133
Nov 28 17:08:58 <strata> across +5v/gnd on the pins of the usb connector of the gadget, 0v
Nov 28 17:09:57 <strata> plugged into a pc, 4.93V at the pins
Nov 28 17:10:13 <KotH> strata: that's within the specs
Nov 28 17:10:19 <strata> seems like the bone is JUST under voltage or something maybe?
Nov 28 17:10:23 <KotH> strata: usb specs require at least 4.75V
Nov 28 17:10:41 <KotH> or was it 4.5?
Nov 28 17:10:44 <KotH> something like that...
Nov 28 17:10:58 <strata> ah ok. could be something else then.
Nov 28 17:11:09 <mru> unfortunately usb specs don't seem to require things to actually work
Nov 28 17:11:16 <strata> maybe i have to turn the port on?
Nov 28 17:11:30 <strata> i'm in an odd mux mode right now because i've been doing some serial communications with a zipit
Nov 28 17:11:37 <KotH> strata: the specs dont tell you whether the cheap designer actually designed according to the specs
Nov 28 17:11:55 <strata> KotH: true. i'll be back in a while. figure this out later i guess :)
Nov 28 17:12:53 <KotH> strata: if it works in a pc and not on the beagle, i would first guess, that the rush in current of the gadget is too high, causing a voltage drop that brings it into an unknown condition
Nov 28 17:13:05 <KotH> strata: for more info, connect a beagle480 inbetween
Nov 28 18:07:40 <Crofton> anyone in a layoff meeting
Nov 28 18:08:38 <ds2> who is moving forward?
Nov 28 18:11:25 <emeb> opportunities for advancement
Nov 28 18:48:19 <djlewis> oddly enough i2c ie more popular today than opencv.. :)
Nov 28 18:48:31 <djlewis> on the google mail list
Nov 28 18:49:59 <paulk-desktop> hi
Nov 28 18:50:05 <mru> lo
Nov 28 18:50:11 <paulk-desktop> where are the Android 4.0 kernel sources for beagleboard located?
Nov 28 18:50:13 <djlewis> lo +1
Nov 28 18:51:56 <mdp> Crofton: yes, lots of people were in that "meeting" today
Nov 28 18:58:44 <mranostay> Crofton: i endorsed your trolling btw
Nov 28 19:26:59 <Crofton> awesome
Nov 28 19:27:16 <Crofton> mdp, I hope thinks work out OK for everyone invovled
Nov 28 19:28:19 <ds2> Crofton: mdp is involved?
Nov 28 19:28:30 <mdp> Crofton: unfortunately, it's so many people that I think it will not work out OK for many of them
Nov 28 19:28:45 <Crofton> long term .....
Nov 28 19:28:45 <mdp> ds2, Sitara-land has no reductions
Nov 28 19:29:04 <ds2> mdp: that's what I thought... it is purely your cousins
Nov 28 19:29:09 <Crofton> I ended up having to quit since they would not lay me off
Nov 28 19:29:17 <mdp> Crofton: yeah, I think it will be fine for all the good engineers caught in this
Nov 28 19:29:19 <wmat> ouch
Nov 28 19:29:23 <mdp> might even be a blessing
Nov 28 19:29:27 <Crofton> heh
Nov 28 19:29:28 <Crofton> yeah
Nov 28 19:29:32 <ds2> mdp: but you guys are not taking in people either, right?
Nov 28 19:30:04 <Crofton> I saw that with several guy sI knew, rough at first, but they ended up doing better
Nov 28 19:30:21 <mdp> ds2, well, we're actually looking for a security expert in applications engineering
Nov 28 19:30:32 <mdp> so I guess that means we're taking in people
Nov 28 19:31:09 <mdp> our previous guy left to finish his PhD
Nov 28 19:31:22 <ds2> i wonder if it was like this when the TI 486 was nuked
Nov 28 19:40:57 <strata> KotH: sure. because i have like 3 beagle 480's just laying around in my junk drawer.
Nov 28 19:42:09 <strata> i had 4, but one of them got smashed because it was in my back pocket with my oversized wallet and i sat down on it.
Nov 28 19:42:14 <Crofton> mranostay, thanks. Everyone in the room is helping
Nov 28 19:43:56 <KotH> strata: :-)
Nov 28 19:44:12 <KotH> strata: if you are doing anything with usb, an beagle480 is a must have
Nov 28 19:45:25 <strata> i'm not analyzing protocols here. i'll stick with my $15 dmm and some experimentation. not all of us here are rich pricks. to even suggest that? *laughs*
Nov 28 19:53:09 * KotH isnt rich
Nov 28 19:53:12 * KotH is a poor student
Nov 28 19:55:10 <KotH> and i wouldnt trust a $15 dmm.. it doesnt even work as a guestimeter
Nov 28 19:55:53 <mru> they can _usually_ give a reasonable indication of open/closed circuit
Nov 28 19:57:06 <KotH> yeah.. that's about what they can do
Nov 28 19:58:07 <mru> maybe even differentiate between 5V and 12V
Nov 28 19:59:12 <mru> that is, they give a roughly double reading for 12V compared to 5V
Nov 28 19:59:37 <KotH> my tounge can do that as well ;)
Nov 28 20:03:49 <woglinde> koth only student?
Nov 28 20:04:00 <woglinde> ah thats why you can chat the whole day at work
Nov 28 20:04:04 <mru> he's a fake student
Nov 28 20:04:12 <woglinde> like av500?
Nov 28 20:04:16 <KotH> nope
Nov 28 20:04:21 <KotH> i actually go to lectures
Nov 28 20:04:21 <mru> worse
Nov 28 20:04:29 <mru> he's studying psychology
Nov 28 20:04:44 <mru> I think he's only doing it for the girls
Nov 28 20:04:55 <ds2> lectures are overrated
Nov 28 20:05:12 <ds2> just skip them and attend only the ones from the headmaster/dean/etc
Nov 28 20:05:44 <KotH> ds2: doesnt work if you have just a couple of pp slides and an exam you have to pass
Nov 28 20:06:05 <mru> ds2: but what if the girls are at the lectures?
Nov 28 20:06:08 <KotH> and unlike anything hard science or engineering, there are hardly any books that you can use as reference
Nov 28 20:06:26 <KotH> mru: psst! dont blow my cover!
Nov 28 20:08:14 <ds2> mru: that'd be an exception
Nov 28 20:08:41 <mru> it's the only reason I can think of for studying psychology in the first place
Nov 28 20:09:11 <KotH> mru: pharmacy would be better actually..
Nov 28 20:09:18 <woglinde> mru dont the women which study psychology needing a therapy themself?
Nov 28 20:09:21 <KotH> better looking girls too
Nov 28 20:09:25 <ds2> isn't that stuff obsolete quicker then CS?
Nov 28 20:09:39 <mru> ds2: obsolete girls?
Nov 28 20:10:04 <mru> woglinde: maybe, haven't investigated the matter
Nov 28 20:10:10 <KotH> woglinde: only half of them do, the other half are the ones who failed the numerus clausus for medicine
Nov 28 20:10:19 <woglinde> koth haha
Nov 28 20:10:22 <ds2> either
Nov 28 20:10:44 <mru> one of the local barmaids is a psychology student...
Nov 28 20:10:53 <woglinde> ah barmaids again
Nov 28 20:11:38 <ds2> %#$#@!$#%@$ android dumps
Nov 28 20:11:50 <ds2> attributing things to a .so that isn't even within the .so
Nov 28 20:12:39 <KotH> mru: for my defence, i also take lectures in engineering
Nov 28 20:12:40 <ds2> n/m
Nov 28 20:12:47 <ds2> damn elf sections
Nov 28 20:12:48 <mru> probably the nearest symbol was in that .so
Nov 28 20:12:53 * KotH doesnt mind ds2 mumbling to himself
Nov 28 20:13:12 <ds2> mru: no, it turns out that address is PLTGOT
Nov 28 20:13:24 <ds2> i was just looking at objdump -d
Nov 28 20:13:34 <mru> I always use readelf
Nov 28 20:13:50 <mranostay> Crofton: wow you are more qualified in trolling than ARM now
Nov 28 20:15:05 <Crofton> rofl
Nov 28 20:15:11 <mru> mranostay: hm?
Nov 28 20:15:15 <Crofton> I just had a roomfull of people endorse me
Nov 28 20:16:17 <mranostay> Crofton: FPGAs coming up
Nov 28 20:19:17 <_av500_> Crofton: Thomas works for ubiquity?
Nov 28 20:19:57 <Crofton> I thought he was skiig
Nov 28 20:20:08 <_av500_> well, when not skiing :)
Nov 28 20:20:17 <_av500_> one of their network cameras is pretty cheap
Nov 28 20:20:29 <_av500_> pondering to buy one
Nov 28 20:20:50 <Crofton> Tsou?
Nov 28 20:21:02 <Crofton> lunch
Nov 28 20:21:05 <_av500_> yes
Nov 28 20:21:07 <_av500_> no?
Nov 28 20:21:22 <_av500_> somebody was working for them
Nov 28 20:31:47 <Crofton> no
Nov 28 20:32:42 <_av500_> ah
Nov 28 20:32:44 <_av500_> hmm
Nov 28 20:32:52 <mranostay> gah it sucks that you only use one channel of the ehrpwm with the current driver
Nov 28 20:33:04 <_av500_> Crofton: ok, never mind then :)
Nov 28 20:33:10 <mru> mranostay: bitbang the rest
Nov 28 20:33:29 <prpplague> mmmm someone said bitbang
Nov 28 20:33:42 <star_> hi beagleboard! quick question about the BOM for rev C — anyone here who can answer a hardware question?
Nov 28 20:33:49 <KotH> prpplague: do you have a highlight rule on that? :)
Nov 28 20:34:08 <KotH> star_: no, all hardware guys left with the second space ship
Nov 28 20:34:17 <prpplague> bitbang, jtag, lvds and beer
Nov 28 20:34:25 <KotH> star_: they were saying something about a space goat or so...
Nov 28 20:35:00 <star_> I'm trying to find online/decode the part number for the 2x17 header (F618-MG -D051-XX-CF358), so I can find the mating connector online (designing a beagleboard peripheral
Nov 28 20:36:25 <KotH> star_: isnt taht a standard 1.27mm pitch smd header pin?
Nov 28 20:37:01 <SilicaGel> 10 yrs ago, when I first started working here, I opened up a package of silica gel and poured it into a glass vial with a lid, and filled it with water
Nov 28 20:37:23 <KotH> why would you do that?
Nov 28 20:37:35 <KotH> impress the girls?
Nov 28 20:37:38 <SilicaGel> I can now report that the silica gel is still not at all gel-like, and once you dry it off, it doesn't feel any different.  Sadly, I forgot to weigh them before I immersed them, so I can't really tell how much moisture they absorbed.
Nov 28 20:37:39 <ds2> found it a bit dessicating?
Nov 28 20:37:58 <SilicaGel> I can, however, report that they smell absolutely AWFUL
Nov 28 20:38:03 <KotH> rotfl
Nov 28 20:38:16 <SilicaGel> Why?  I have no idea why I did this.  It was a decade ago!  How can you expect me to remember that
Nov 28 20:38:21 <SilicaGel> This is scienc!
Nov 28 20:38:22 <SilicaGel> e
Nov 28 20:39:24 <mdp> prpplague: I've graduated from bitbanging to bitbanding
Nov 28 20:39:26 <SilicaGel> Actually, I take it back, they do look a bit diferent.  They are almost like a clear sphere now, with a white cloudy center
Nov 28 20:39:29 <KotH> SilicaGel: http://www.smbc-comics.com/index.php?db=comics&id=1202
Nov 28 20:39:49 <mdp> mranostay, you have the source…patches are unwelcome
Nov 28 20:39:50 <SilicaGel> yeay!
Nov 28 20:39:50 <ds2> mdp: has atomicity got you in a bunch?
Nov 28 20:40:07 <mdp> ds2, indeed
Nov 28 20:40:38 <mdp> I had one brief technical discussion today, it was refreshing after all the negative energy here
Nov 28 20:41:01 <ds2> if you want technical discussions and don't mind am35x.... ;)
Nov 28 20:41:17 <SilicaGel> The "Do not Eat" warning is long, long gone.  Should I eat it?
Nov 28 20:42:07 <SilicaGel> Perhaps waht I should do is towel dry the beads, weigh them, then dry them out in a low temperature oven, then weigh them again
Nov 28 20:42:12 <KotH> SilicaGel: yes! try it! for science!
Nov 28 20:43:27 * mranostay notes all his friends are as weird as KotH 
Nov 28 20:43:34 * mranostay *sighs* :)
Nov 28 20:43:48 <ds2> problem with the peninsula?
Nov 28 20:43:48 <SilicaGel> Weird like a fox!
Nov 28 20:44:01 <Crofton> https://www.youtube.com/watch?v=P9xwlK3PUf8&feature=g-upl
Nov 28 20:44:42 <KotH> mranostay: i'm not weird!
Nov 28 20:44:50 <KotH> mranostay: i'm a perfectly normal swiss geek!
Nov 28 20:45:43 <KotH> Crofton: c'est quoi ça?
Nov 28 20:46:21 <Crofton> Fire
Nov 28 20:46:29 <KotH> er... yes..
Nov 28 20:46:34 <KotH> and what is being burned?
Nov 28 20:46:42 <Crofton> a ping pong ball
Nov 28 20:46:54 <Crofton> mranostay, I have Tizened on Castro Street
Nov 28 20:47:08 <mranostay> er not SF right? :)
Nov 28 20:47:19 <Crofton> right
Nov 28 20:47:45 <mranostay> Crofton: what block of Castro or do i have to find out?
Nov 28 20:47:55 <Crofton> you have to find it
Nov 28 20:48:13 <ds2> Crofton: you'll wave as the ark passes through?
Nov 28 20:48:21 <Crofton> yeah
Nov 28 20:48:26 <Crofton> this rain is making me sad
Nov 28 20:50:02 <mranostay> Crofton: turn that frown upside down!
Nov 28 20:51:23 <Crofton> hmm
Nov 28 20:52:04 <Crofton> am I still connected
Nov 28 20:52:11 <mranostay> Crofton: no
Nov 28 20:52:15 * KotH disconnects Crofton from reality
Nov 28 20:52:20 <mranostay> _av500_: post some cat pics for Crofton
Nov 28 20:52:35 <mranostay> KotH: like you are connected to that
Nov 28 20:53:32 <KotH> mranostay: i'm studying the axioms of my own selfbuild reality
Nov 28 20:53:45 <KotH> mranostay: in a way, i'm more connected to reality than you ever can hope to be
Nov 28 20:54:14 <mru> I made a graph: https://docs.google.com/spreadsheet/ccc?key=0AguHvNGaLXy9dHVTNzJWREhGS2tmQlM4dk5WbjZ2UlE#gid=2
Nov 28 20:54:38 <mranostay> i have no desire to be connected to reality... real life is boring :)
Nov 28 20:55:05 <mranostay> mru: is that #beagle usage?
Nov 28 20:55:09 <mru> yes
Nov 28 20:55:31 <mru> one chart per day, one rolling average over 30 days
Nov 28 20:56:10 <mru> I see a clear trend
Nov 28 20:56:22 <KotH> and exponential groth!
Nov 28 20:56:26 <KotH> growth*
Nov 28 20:56:35 <mranostay> bits has been banging more than ever
Nov 28 20:56:41 <mranostay> *have
Nov 28 20:58:14 <Crofton> mru, you need more to work on
Nov 28 20:58:46 <mru> that was just a trivial grep through the logs
Nov 28 20:59:54 <mranostay> bit bang!
Nov 28 21:00:23 * mranostay trys to inflate the numbers
Nov 28 21:01:05 <_av500_> mru: how can it be 0.5 bitbang???
Nov 28 21:01:19 <koen> only bang the 1s
Nov 28 21:01:21 <mru> _av500_: average over 30 days
Nov 28 21:01:24 <_av500_> is the bit banged or not?
Nov 28 21:01:41 <mru> there were two bits, of which one was banged
Nov 28 21:03:20 <_av500_> schroedingers bits
Nov 28 21:03:45 <mru> scrödinger's bat
Nov 28 21:04:05 <_av500_> #baturday
Nov 28 21:05:42 <mranostay> Crofton: we are past FPGAs!
Nov 28 21:22:46 <alan_o> mru: so did you do American date style, or did google detect my IP and show me what I wanted to see?
Nov 28 21:22:59 <mru> I didn't do anything
Nov 28 21:23:15 <alan_o> mru: in your graph
Nov 28 21:23:18 <mru> the text file I uploaded has y-m-d style
Nov 28 21:23:34 <alan_o> hmm.... it shows me mm/dd/yyyy
Nov 28 21:23:40 <mru> google detects that as a date and marks the cells as such
Nov 28 21:24:05 <alan_o> so google must be showing me what I want to see then
Nov 28 21:24:26 <mru> no, google simply uses american style
Nov 28 21:24:44 <alan_o> ok, so that was part of the question.
Nov 28 21:24:50 <KotH> alan_o: you dont want to use american style dates *jedi move*
Nov 28 21:25:17 <alan_o> KotH: I like the ISO, personally. No confusion.
Nov 28 21:25:35 <alan_o> KotH: and everything sorts it correctly
Nov 28 21:25:43 <KotH> alan_o: european has also no way for confusion :-)
Nov 28 21:26:03 <mru> european is iso
Nov 28 21:26:05 <alan_o> KotH: no more than American, you mean?
Nov 28 21:26:05 <KotH> alan_o: neither has japanese... unless you confuse it with usian style
Nov 28 21:26:17 <mru> british is d/m/y
Nov 28 21:26:25 <KotH> mru: nope, most people in europe use DD.MM.YYYY which is different from iso
Nov 28 21:26:26 <alan_o> only the British do that?
Nov 28 21:26:41 <alan_o> KotH: that's what I had thought
Nov 28 21:26:41 <KotH> alan_o: no japanese do the same
Nov 28 21:27:04 <KotH> er.. no japanese do different
Nov 28 21:27:05 <mru> sweden certainly uses y-m-d as standard
Nov 28 21:27:12 <KotH> interesting
Nov 28 21:27:14 <KotH> didnt know that
Nov 28 21:27:26 <KotH> then, i've never been in a country where bones are sexual objects
Nov 28 21:28:01 <_av500_> lol
Nov 28 21:28:02 <KotH> then, i probably should be asleep by now
Nov 28 21:28:18 <_av500_> sweden reminds me, sec
Nov 28 21:29:22 <_av500_> done
Nov 28 21:29:33 <KotH> aha
Nov 28 21:29:46 <KotH> havent thought about that
Nov 28 21:31:29 <KotH> night boys...
Nov 28 21:31:42 <KotH> keep the guns under your beds dry!
Nov 28 21:32:56 <wmat> d/m/y is canada as well
Nov 28 21:33:22 <mru> wmat: canada is technically part of britain :)
Nov 28 21:33:31 <wmat> mru: technically, it's not
Nov 28 21:33:39 <mru> royally, it is
Nov 28 21:34:04 <wmat> we do still have her highness on all of our money
Nov 28 21:34:21 <wmat> but we don't have to send them beaver pelts and wheat anymore
Nov 28 21:35:01 <_av500_> soon you will have to
Nov 28 21:35:08 <_av500_> as UK is not in the EU, we will not bail them out
Nov 28 21:35:44 * wmat goes to set some traps
Nov 28 21:35:59 <_av500_> TARPS
Nov 28 21:36:24 <djlewis> too big to fold my arse
Nov 28 21:38:10 <georgem> my wife is canadian
Nov 28 21:39:51 <wmat> she can stay then ;)
Nov 28 21:48:00 <georgem> Actually, according to her she's Canadien. heh
Nov 28 21:51:55 <Lizalfos> is the internal boot ROM for the beagleboard replacable?
Nov 28 21:52:09 <Lizalfos> the code that operates before xloader that is.
Nov 28 21:52:59 <Crofton> no
Nov 28 21:53:07 <Crofton> that is what ROM means
Nov 28 21:53:13 <Crofton> Read only
Nov 28 21:53:35 <Lizalfos> it's not stord in NAND?
Nov 28 21:53:40 <Lizalfos> *stored, excuse me.
Nov 28 21:54:03 <mru> then it would not be ROM
Nov 28 21:54:44 <Lizalfos> huh. alright.
Nov 28 21:55:27 <woglinde> *g*
Nov 28 21:56:08 <Lizalfos> I'm planning on using the beagleboard for bare metal programming, and was just wondering if I could overwrite the initial bootloader.
Nov 28 21:56:22 <woglinde> lol
Nov 28 21:56:50 <woglinde> you dont even know what ROM means and want to do bare metal programming?
Nov 28 21:56:53 <woglinde> good luck
Nov 28 21:57:08 <alan_o> knock it off guys
Nov 28 21:57:16 <alan_o> you all know dang well that ROM is often writeable
Nov 28 21:57:21 <Russ>  mru don't be hasty
Nov 28 21:57:28 <alan_o> like  EPROM and EEPROM
Nov 28 21:57:31 <Lizalfos> ^
Nov 28 21:57:35 <woglinde> I bet next question is "give me please exact steps"
Nov 28 21:57:40 <Lizalfos> no, actually.
Nov 28 21:57:54 <Russ> if he's ordering a few 100k or so units from TI, it probably is only a matter of negotiation as far as custom roms go
Nov 28 21:57:59 <Lizalfos> I'm just not familiar with the beagleboard and am wanting to pick a nice embedded system to toy around with for the winter months.
Nov 28 21:58:00 <mru> alan_o: ROM is never writable, might be _erasable_ though
Nov 28 21:58:01 <mru> hah
Nov 28 21:58:17 <Lizalfos> I thought it was stored in NAND.
Nov 28 21:58:22 <woglinde> lol
Nov 28 21:58:28 <mru> read the manual more carefully
Nov 28 21:58:37 <woglinde> Lizalfos better get an atmel
Nov 28 21:58:46 <Lizalfos> that's what I have on my bench currently.
Nov 28 21:58:49 <Russ> isn't there a slide on elinux.org that shows the boot process?
Nov 28 21:58:55 <alan_o> Lizalfos: some older beagleboards have NAND where the bootloader can be stored.
Nov 28 21:59:04 <Lizalfos> rev Bs or the A1/2?
Nov 28 21:59:19 <woglinde> b's and c's
Nov 28 21:59:30 <alan_o> that's non -xM
Nov 28 21:59:37 <alan_o> but some of the early -xM do also
Nov 28 21:59:40 <Lizalfos> that's what I was asking, non XM.
Nov 28 21:59:43 <alan_o> not a separate designation for those
Nov 28 21:59:44 <mru> alan_o: we don't talk about those
Nov 28 21:59:53 <alan_o> mru: hehe
Nov 28 22:00:00 <woglinde> alan_o they are legend
Nov 28 22:00:03 <woglinde> *g*
Nov 28 22:00:07 <Lizalfos> only spoken of in myth.
Nov 28 22:00:25 <Lizalfos> but no, I'm no newbie. just wanting to shop around for some new toys to buy myself. :)
Nov 28 22:00:36 <woglinde> hm
Nov 28 22:00:54 <Lizalfos> thought the beagleboard would be suitable since broadcom's chipset on the pi is an IP nightmare.
Nov 28 22:01:32 <Lizalfos> and powervr is sexy.
Nov 28 22:01:41 <woglinde> powervr is ip too
Nov 28 22:01:49 <Russ> why would anyone need to rewrite the rom?
Nov 28 22:02:01 <mru> Russ: to fix the bugs in it
Nov 28 22:02:03 <woglinde> russ as he wrote boring winter months
Nov 28 22:02:06 <Lizalfos> for shits and giggles.
Nov 28 22:02:22 <Russ> just lie to yourself and tell your self it doesn't exist
Nov 28 22:02:44 <woglinde> good night
Nov 28 22:03:06 <Lizalfos> the boot rom is spoken of only in myth, operating on the spiritual sillicon plane.
Nov 28 22:03:52 <Russ> rewrite MLO and be happy
Nov 28 22:04:05 <Lizalfos> the hell does MLO even stand for.
Nov 28 22:04:26 <Lizalfos> malloc's lame offspring?
Nov 28 22:04:39 <_av500_> +1
Nov 28 22:04:47 <Lizalfos> memory linked object?
Nov 28 22:04:55 <Crofton> mru, remind me what the a9 unaligned load for neon penalty is/
Nov 28 22:05:07 <ds2> Milo's Linux Output
Nov 28 22:05:16 <Lizalfos> hah.
Nov 28 22:05:18 <Russ> Crofton, 30 lashings
Nov 28 22:05:45 <ds2> Neon penality is proportional to distance from Vegas
Nov 28 22:06:04 <Lizalfos> what happens in cortex stays in cortex.
Nov 28 22:07:34 <mru> Crofton: depends on the address you load from
Nov 28 22:07:57 <mru> if you're crossing a page boundary, you could end up with two page faults
Nov 28 22:08:48 <mru> loading with an unaligned-capable instruction from an _aligned_ address takes one cycle more than the equivalent aligned-only instruction
Nov 28 22:08:59 <ds2> mru: do you know off hand if the VFP on the Cortex A's are a superset of the VFP on a Cortex-M4?
Nov 28 22:09:12 <mru> M4 does not have vfp
Nov 28 22:09:28 <ds2> the M4 has a totally different FP block?
Nov 28 22:10:26 <mru> but the instruction set is mostly a subset of vfp, yes
Nov 28 22:10:52 <ds2> 'k
Nov 28 22:11:13 <ds2> so a Cortex-A8 can completely simulate a M4 then?
Nov 28 22:11:19 <mru> no
Nov 28 22:11:46 <ds2> what is missing on the A8 side?
Nov 28 22:11:49 <mru> well, you can of course run a cortex-m4 simulator on an a8
Nov 28 22:12:01 <mru> but you can't expect to execute m4 code directly
Nov 28 22:12:36 <ds2> other then interrupt controllers, isn't it just the same thumb2 on both?
Nov 28 22:14:42 <mru> the entire interrupt model is different
Nov 28 22:15:31 <mru> obviously some sequences of instructions will behave the same
Nov 28 22:15:40 <ds2> yes, excluding that... application is the wrong term to use but that is the sort of stuff I am thinking of
Nov 28 22:15:56 <mru> but armv7-a and armv7-m have separate arch ref manuals
Nov 28 22:16:36 <ds2> i noticed
Nov 28 22:16:55 <_av500_> ds2: use a language that works on both, like java
Nov 28 22:17:51 <ds2> _av500_: u got a jvm for the M3?
Nov 28 22:21:05 <_av500_> it was M4 five minutes ago
Nov 28 22:21:15 <_av500_> if I say yes, will ask for M0
Nov 28 22:21:19 <ds2> fine, M4
Nov 28 22:21:25 <ds2> I can deal with that
Nov 28 22:21:42 <_av500_> no idea, did you ask Oracle?
Nov 28 22:22:10 <ds2> of course not. I am an engineer... not a high priest
Nov 28 22:22:20 <ds2> oracle is for ancient obsoletes
Nov 28 22:24:34 <mranostay> Oracle is evil
Nov 28 22:27:34 <mdp> M5>M4
Nov 28 22:40:29 <mranostay> _av500_: i demand some cat photos to lighten the day
Nov 28 22:41:45 <mdp> gah, stop it
Nov 28 22:42:10 <mranostay> mdp: stop what? :)
Nov 29 01:44:02 <Crofton> mru, does gcc 4.7 make you use fast-math to get neon?
Nov 29 01:45:21 <mru> depends on what you mean
Nov 29 01:45:54 <Crofton> so we are playing with some NEON stuff here
Nov 29 01:46:18 <mru> you should only use -ffast-math if you're certain the things it does are safe with your code
Nov 29 01:46:23 <Crofton> ans the guy doing the work said no neon until he turned on fast-math
Nov 29 01:46:30 <mru> bullshit
Nov 29 01:46:49 <Crofton> that is what I was was wondering
Nov 29 01:46:57 <mru> some of the options -ffast-math enables are required for auto-vectoriser to be useful in some cases
Nov 29 01:46:57 <ds2> how close is gcc 4.7 to what you can get with armcc?
Nov 29 01:47:04 <Crofton> basically, he is trying to do something with intrinsics
Nov 29 01:47:07 <mru> ds2: who cares?
Nov 29 01:47:21 <mru> ds2: both are a far cry from what you can do by hand
Nov 29 01:47:24 <Crofton> and without fast-math and -O something, the NEON disappered
Nov 29 01:47:34 <mru> I say bullshit
Nov 29 01:47:43 <Crofton> heh
Nov 29 01:47:55 <mru> intrinsics give you the instructions you say provided you have -mfpu=neon
Nov 29 01:48:00 <Crofton> I wish I had a small example
Nov 29 01:48:08 <Crofton> hmm,
Nov 29 01:48:16 <ds2> I thought armcc will do some basic stuff w/o using intrinsics?
Nov 29 01:48:22 <mru> ds2: as will gcc
Nov 29 01:48:30 <mru> some *very* basic stuff
Nov 29 01:48:35 <Crofton> we'll look further
Nov 29 01:48:48 <mru> in my testing, the auto-vectoriser does more harm than good
Nov 29 01:48:58 <ds2> I see
Nov 29 01:49:19 <tholm> hi all, I'm working with SPI, I need to do a pulse on Chip select per word by each TX, does the kernel support that option?
Nov 29 01:50:22 <mru> Crofton: you do want to enable some of the options -ffast-math turns on
Nov 29 01:51:49 <alan_o> tholm: CS should go down (active) and back up again (idle) around the call for each spi message
Nov 29 01:51:59 <tholm> yes alan_o
Nov 29 01:53:33 <alan_o> tholm: not sure what you mean
Nov 29 01:53:55 <alan_o> tholm: do you mean you need the CS to go down and back up again around each 4 bytes?
Nov 29 01:53:57 <mru> Crofton: but you have to evaluate each one and make sure you're not doing something it breaks
Nov 29 01:55:47 <Crofton> I am somewhat stumped by the compiler making NEON go away
Nov 29 01:56:12 <mru> either someone is full of shit, or there's something you're not telling me
Nov 29 01:57:14 <tholm> for example alan_o I need to send 0xD000 (16 bit) and I will receive a data, but CS to go up and back down with the second word
Nov 29 01:57:43 <alan_o> tholm: can you make a separate message for each time you need CS to go down and back up?
Nov 29 01:58:19 <alan_o> tholm: then call spi_sync() or whatever you're doing on both messages?
Nov 29 01:58:55 <tholm> hold on alan_o I'll pastebin the code
Nov 29 01:59:01 <Crofton> like I said, I am annoyed we do not have something easy to show this with
Nov 29 01:59:02 <alan_o> tholm: I haven't seen a magic option to make it do what you're saying, but I haven't looked for it either.
Nov 29 01:59:31 <mru> Crofton: if you can't come up with a simple demonstration, the cause is likely not simple either
Nov 29 02:00:04 <Crofton> hang on
Nov 29 02:00:13 <Crofton> we are going to pastebin
Nov 29 02:00:30 <Crofton> runexe, patebin the code and the compiler outpit
Nov 29 02:00:38 <Crofton> no barfing at the intrinsic
Nov 29 02:00:50 <Crofton> :)
Nov 29 02:00:51 <tholm> http://pastebin.com/tBhxehCf
Nov 29 02:01:07 <tholm> shut it Crofton :P
Nov 29 02:03:06 <alan_o> tholm: ok, so make each word a separate ioc_transfer.
Nov 29 02:03:11 <alan_o> tholm: that may not be enough though
Nov 29 02:03:20 <alan_o> might have to have a separate ioctl for each
Nov 29 02:03:32 <alan_o> tholm: I'm less familiar with the userspace interface.
Nov 29 02:03:47 <runexe> Crofton, mru: http://pastebin.com/mrYAfTVm <- C code (two for loops, only the first should emit NEON instructions)
Nov 29 02:04:39 <Crofton> can you pastebin the code built without O
Nov 29 02:04:56 <Crofton> and the code with O and the compielr args
Nov 29 02:05:01 <runexe> Working on it it, pastbin == teh slow (at least from these internets)
Nov 29 02:05:09 <mru> runexe: what are you seeing, what do you expect to see, and what would you like to see?
Nov 29 02:05:47 <mru> those v*_f32() intrinsics should either give the corresponding instruction or a compiler error
Nov 29 02:06:15 <runexe> Ok - with these compiler flags:
Nov 29 02:06:18 <runexe> arm-oe-linux-gnueabi-gcc -S -march=armv7-a -mthumb-interwork -mfloat-abi=softfp -mfpu=neon -mcpu=cortex-a9 -mtune=cortex-a9 -O3 -ftree-vectorize -ffast-math --sysroot=/opt/oecore-x86_64/sysroots/ettus_e200-oe-linux-gnueabi neon_test.c
Nov 29 02:06:23 <mru> for plain C code nothing is ever promised
Nov 29 02:06:30 <runexe> I get: http://pastebin.com/ivpHC0u6
Nov 29 02:06:46 <Crofton> mru, he is typing as fast as he can on his ridiculous heavy laptop
Nov 29 02:06:47 <runexe> Which includes the vld, vmul's that I'm looking for
Nov 29 02:07:43 <Crofton> he is cursing pastebin now
Nov 29 02:07:55 <mru> oh, I can do that all day
Nov 29 02:08:03 <mru> thankfully there are alternatives
Nov 29 02:08:12 <tholm> ok alan_o I think imma try to send only one word
Nov 29 02:08:35 <runexe> With: arm-oe-linux-gnueabi-gcc -S -march=armv7-a -mthumb-interwork -mfloat-abi=softfp -mfpu=neon -mcpu=cortex-a9 -mtune=cortex-a9 -O3 -ftree-vectorize --sysroot=/opt/oecore-x86_64/sysroots/ettus_e200-oe-linux-gnueabi neon_test.c
Nov 29 02:08:45 <runexe> (i.e. everything else the same, but no -ffast-math)
Nov 29 02:08:53 <tholm> actually that works but the performance is not that good
Nov 29 02:08:58 <mru> that code is not what produced that assembly
Nov 29 02:09:02 <Crofton> more swearing
Nov 29 02:09:23 <alan_o> tholm: looks from the spidev driver code that it cycles the CS one time per ioctl() call. Check it with scope (is what ds2 would say).
Nov 29 02:09:23 <Crofton> hmm, that is an intersting theory ....
Nov 29 02:09:27 <runexe> I get: http://pastebin.com/4BnyEJiX
Nov 29 02:10:07 <mru> oh, I see what's going on here
Nov 29 02:10:20 <mru> the result of the first loop is never used so it's dropped entirely
Nov 29 02:10:34 <mru> the second loop is partially auto-vectorised when -ffast-math is enabled
Nov 29 02:10:37 <tholm> ok imma try, thx alan_o
Nov 29 02:10:38 <mru> but not otherwise
Nov 29 02:10:41 <Crofton> thanks
Nov 29 02:10:56 <runexe> mru: Interesting
Nov 29 02:10:57 <alan_o> tholm: no problem.
Nov 29 02:12:42 <Crofton> runexe, is now contempplating the fact he made an error
Nov 29 02:13:12 <Crofton> more swearing
Nov 29 02:13:27 <Crofton> :)
Nov 29 02:13:57 <mru> gcc splits the loop in one part for a multiple of the vector size and a second part for any remaining tail
Nov 29 02:14:10 <mru> and _lots_ of surrounding junk code
Nov 29 02:14:26 <mru> the scalar one is also used if the pointers alias
Nov 29 02:14:36 <mru> you should restrict-qualify your pointers
Nov 29 02:14:49 <mru> assuming they do not alias
Nov 29 02:17:36 <mru> and if the pointers are 8/16-byte aligned, tell gcc about this using __builtin_assume_aligned()
Nov 29 02:18:17 <mru> and _never_ count on the auto-vectorisation doing what you want
Nov 29 02:18:40 <runexe> mru: Ah! I was under the impression I could not use intrinsics and pass along alignment info - thanks very much!
Nov 29 02:18:52 <mru> it's new in 4.7
Nov 29 02:18:59 <runexe> sweet
Nov 29 02:19:06 <mru> and intrinsics don't work before that anyway
Nov 29 02:19:23 <mru> it might compile, but you don't want to see the result
Nov 29 02:20:22 <mru> with 4.7 intrinsics can give pretty good code in simple cases
Nov 29 02:20:24 <Crofton> thinnk training wheels
Nov 29 02:20:54 <Crofton> the alternative is we take your brain and make milkshakes
Nov 29 02:20:57 <mru> if you start running out of registers, things go downhill quickly
Nov 29 02:21:18 <Crofton> most of the stuff we are looking at right now shouldn't run out of registers
Nov 29 02:21:27 <runexe> Output is looking *much* better now :)
Nov 29 02:21:41 <alan_o> mru: same when you start running out of tape
Nov 29 02:21:48 <mru> alan_o: yep
Nov 29 02:23:01 <mru> Crofton: runexe: just keep an eye on what the compiler spits out
Nov 29 02:23:20 <mru> if it starts doing things you can't easily follow, something is probably weird
Nov 29 02:23:40 <Crofton> yeah
Nov 29 02:23:49 <mru> and if you need help, my off-shore account is always accepting donations
Nov 29 02:24:19 <runexe> mru: Right :) - thanks very much!
Nov 29 02:56:02 <tholm> what does "!!" mean in C?
Nov 29 02:58:51 <mru> what does ! mean in C?
Nov 29 02:59:04 <tholm> nope !!
Nov 29 02:59:12 <mru> answer the question
Nov 29 02:59:21 <tholm> !!pointer->cs
Nov 29 02:59:27 <mru> humour me
Nov 29 02:59:31 <mru> what does ! mean?
**** ENDING LOGGING AT Thu Nov 29 02:59:59 2012