IRC channel logs


back to list of logs

<copypaste>damo22: your gnu/hurd presentation at fosdem was so cool, best of luck with all that, using netbsd's rumpkernel stuff is a great ieda :)
<copypaste>( )
<damo22>copypaste: thanks, i didnt know i had a DOI
<copypaste>damo22: DOI?
<copypaste>oh! 10.5446/56910 (DOI)
<copypaste>i wonder if based on your talk it's possible to use more complex types of drivers?
<copypaste>perhaps wifi, perhaps esoteric things like SIM modems?
<copypaste>it is *net*bsd after all :-D
<damo22>yes it is already compiled to have librumpnet
<damo22>but no one has done the work to write a translator for that
<damo22>some work in netbsd is required to make rump a first class kernel target as well
<damo22>i think its just makefile foo
<copypaste>how about intel gfx? i just wonder how far this idea can be pushed.
<damo22>why not?
<copypaste>wow :) i will look more into if it is that powerful.
<copypaste>i mean, gpu's are very anal about timing.
<copypaste>and many other things.
<copypaste>(also a lot of them are just downright broken.)
<copypaste>(and require software with ring-0 access / realtime capability to fix the hardware.)
<damo22>linux is a different beast, there is no such thing as "capabilities" in hurd, everything is controlled by filesystem node permissions
<damo22>if you want to grant a user full access to a pci card, you can
<copypaste>huh, might help with a driver i'm trying to fix atm.
<damo22>for example, you could have a truely private network with the NIC isolated to a user
<copypaste>i have a webcam that the company refuses to tell me how to power. i don't really care as i hate video chats but i would like to fix it for the sake of doing so.
<copypaste>i've wanted exactly that function so i could slowly ramp the power output to see which gpio controls which PMIC line.
<damo22>you could have a user based usb stack
<copypaste>this is not usb.
<copypaste>it interfaces with the gpu, it sucks.
<copypaste>to do things like autofocus/skin softening. it is exposed via ACPI.
<copypaste>and PCI.
<copypaste>the problem is i know that one of the lines must be run over capacity. the manual explicitly says it gets very hot and not to touch it, and i looked up the led module and the PMIC does not support the 5 volts it wants lol.
<copypaste>so it's russian roulette. if i guess wrong which line gets the 5V, i will burn out a camera.
<damo22>why not run it in windows, in a vm with pci passthrough and dump the pci commands
<copypaste>pci passthrough, is that a thing?
<copypaste>thanks. you saved me a lot of time.
<copypaste>(and possibly saved a webcam)
<damo22>i dont know how to do it anymore, i forgot but it used to work
<copypaste>no that's ok i will look into
<copypaste>i was under impression only usb can be passed thr
<damo22>possibly with qemu
<copypaste>yes, that makes sense
<copypaste>it's this:
<copypaste>Multimedia controller: Intel Corporation Device 9a19 (rev 01)
<copypaste>Subsystem: Matsushita Electric Industrial Co., Ltd. Device 8338
<copypaste>workig on it now :)
<Zopolis4>Hurd had a 4-page article in Linux Format!
<Zopolis4>havent read it yet
<Zopolis4>but still
<Zopolis4>all press is good press
<Zopolis4>planning to watch the rumpkernel talk soon
<Zopolis4>i think once we get drivers sorted out, that will tip hurd over the edge
<Zopolis4>we'll go from something you hear about, and could theoretically boot in a vm if you wanted to, to an actual curiosity system that you can install on your hardware, and maybe even daily drive
<jpoiret>daily drive? that i'm not too sure :p