IRC channel logs

<damo22>youpi: !

<damo22>long time no see

<wleslie>I think that's just the netsplit (but if not, hi youpi & damo!)

<damo22>hi

<damo22>i need to polish my patches again and resubmit

<damo22>but rump is working with the arbiter

***Server sets mode: +nt

***illustris_ is now known as illustris

***Whistler_ is now known as Whistler

<youpi>yes that was just the netsplit, but yes that's also a comeback after months of work overload

<youpi>not really lightload, but still I can afford reading irc a bit

<azeem_>wb

***azeem_ is now known as azeem

<damo22>wb

<youpi>thanks :)

<gnu_srs1>wb

<wleslie>I'm glad it's starting to ease up

<damo22>i was losing motivation to plow through the rump stuff but since it is almost working i can see light

<wleslie>rump is always best slow-cooked

<damo22>everything worked in my test boot, all the tricky dance to make the servers talk to each other and resuming tasks etc

<wleslie>nice work on getting your disk back up

<damo22>i just need to clean up some stuff and find a way to fake uid=0

<damo22>since there is no auth server or proc

<wleslie>who's checking?

<damo22>the netfs for pci wants to look up a protid

<damo22>but before the root filesystem exists

<wleslie>isn't the statically linked root filesystem the first server that gets started? hmm

<damo22>s/look up/be passed

<damo22>yes but rumpdisk.static is looking up a netfs node in the pci arbiter

<damo22>to access the pci virtual filesystem

<youpi>wleslie: not any more

<youpi>it now needs the disk driver to be started before

<youpi>and the disk driver needs the pci arbiter before

<wleslie>superb, you have been busy

<wleslie>can you query the mach device list directly?

<wleslie>or is the interface too different

<damo22>its not a mach device in gnumach

<youpi>damo22: btw I was wondering whether pci-arbiter really needs libmachdev, you could just add the device demuxer and directly handle the RPC

<youpi>rather than having to handle conflicts between libmachdev and libnetfs

*wleslie <- dinner()

<damo22>youpi: machdev has handy way to inject the translator into place

<youpi>ok

<damo22>i would be reimplementing things

<youpi>yup

<youpi>if it was just for the demuxer that'd be easy, but the port dance is not :)

<damo22>yeah what ive got so far is good, i think it needs your suggestions all round

<damo22>which you sent already

<damo22>when i get disk stable, i will boot my x230 native and develop on real hw

<damo22>in theory it should work

<damo22>theres a bug in my change for libpciaccess, it crashes on native hw when there is a different list of devices

<damo22>possibly duplicates?

<damo22>maybe the bug is in pci-arbiter

<damo22> https://i2.paste.pics/9969e5f6e308f927183d7d675f1f2aa3.png

<damo22>thats what my boot looks like on real hw

<damo22>there are duplicate entries in the pci access list

<damo22>and then it fails to probe them due to mismatching refcounts probably

<damo22>ive seen it enumerate duplicates before on different hw

***roptat is now known as roptat_

***roptat_ is now known as roptat

***DNS is now known as SND

***justan0theruser is now known as justanotheruser

IRC channel logs

2021-03-04.log