ikarso has quit [Quit: Connection closed for inactivity]
alan_o has joined #beagle
Guest62 has joined #beagle
Guest62 has quit [Client Quit]
CrazyEddy has quit [Ping timeout: 244 seconds]
CrazyEddy has joined #beagle
calrama has joined #beagle
calrama has quit [Quit: Client closed]
demirok has quit [Quit: Leaving.]
calrama has joined #beagle
<calrama>
Hey, so using the device tree to enable a (pwm) timer on the beaglebone ai seems a bust. Using the devicetree suggested by zmatt ( https://pastebin.com/raw/7rVAHE8V ) just gives me dmesg errors for any timer I try it with ( https://pastebin.com/raw/pW7QnYwb ).
v0n has joined #beagle
Posterdati has quit [Ping timeout: 255 seconds]
xet7 has joined #beagle
xet7 has quit [Ping timeout: 258 seconds]
Posterdati has joined #beagle
<zmatt>
huh, that's weird
<zmatt>
based on that kernel log I'm guessing you added declarations for all three?
<zmatt>
calrama: lemme take a quick peek at the driver
<calrama>
zmatt: yeah, I did that for debugging^^
<calrama>
What I'm currently trying is setting the registers manually via /dev/mem
<zmatt>
and you were using 4.14-ti kernel series right?
<calrama>
Correct.
<zmatt>
please don't
<calrama>
Yeah, it doesn't work, either.
<zmatt>
like, if you want to go the manual register setting route, use uio instead of /dev/mem
<calrama>
I can select the clock.
<zmatt>
don't try to mess with PRCM (power/reset/clock management) behind the kernel's back
<calrama>
But when I try configuring the timer itself, the kernel throws `Data Access in User mode during Functional access` errors.
<calrama>
I just tried the /dev/mem route since I found an old example for beaglebone black and tried adjusting that.
<calrama>
I don't really want to use it, either.
<zmatt>
yeah, because the module is disabled. uio allows a module's registers to be exported to userspace with the kernel's knowledge, so it'll keep the module enabled while the device is open
<zmatt>
but yeah it shouldn't be needed for this obviously
<calrama>
If you want, I can give you ssh access to the device^^
<zmatt>
I don't think that's useful right now... I first want to check the pwm driver ... if git ever finishes checking out this branch
<calrama>
Could you point me to those build scripts?
<zmatt>
well, the problem there is also with the issue itself... like, obviously it's desirable to get TIDL to work on a newer kernel, and if it were easily done then the TIDL images would already *have* a newer kernel ;)
<zmatt>
my notes on building a custom kernel: https://pastebin.com/eLhrp1Hg (in your case you'd want the ti-linux-4.14.y branch of the ti-linux-kernel-dev repo)
<calrama>
Well, there was no official word on if it's even still being worked on or not, hence the issue. If I got a reply like "nope, we don't have time to ever realistically suport that" I'd be fine at least knowing that^^