***Server sets mode: +nt
***Emulatorman_ is now known as Emulatorman
<damo22>youpi: in general, how will hurd solve the problem that devices are enumerated by userspace processes and thus can be attached to any filesystem node, how will downstream of that detect which devices are present in the system? <damo22>for example, you plug in a soundcard <damo22>more importantly, how will a process know which device node to use when trying to use a device <damo22>perhaps pci-arbiter needs to create the nodes where devices are attached, addressed by bus <youpi>damo22: I don't think we want to necessarily map enumeration to physical adressing <youpi>that's the same concern as systemd's so-called "stable" names for network devices <youpi>in practice what happens for now is that netdde numbers devices <youpi>a sound rump daemon could do the same <youpi>and have /dev/snd0, /dev/snd1 etc. be devnode pointing to the rump daemon, just like /dev/eth0 point to netdde atm <youpi>then we could think about the sound rump daemon expose a directory instead of just a master node <youpi>each entry being for a device <youpi>that part could be hardware-based enumerated <youpi>and something else create /dev/snd0 etc. symlinks to provide convenient shortcuts for the user <youpi>for now I'd say just proceed like for netdde <gnu_srs1>Some further results: bash-5.0# ldd /bin/fsysopts; ...; /bin/ldd: cannot make pipe for command substitution: Address family not supported by protocol (expected) <gnu_srs1>+ printf %s ''; + return 0; Segmentation fault; /libexec/runsystem: line 60: 10 Segmentation fault /bin/bash <gnu_srs1>youpi: Nowadays when fsck fails boot continues making diskcheck of / impossible. What changed?? ***GeneralDuke1 is now known as GeneralDuke
<jrtc27>gnu_srs: inability to make pipes is often symptomatic of /servers/socket/1 being borked <gnu_srs1>jrtc27: Not being able to created pipes is expected: this setup is pre-dev and pre-translators, directly booting from a cross build. ***Server sets mode: +nt
<gnu_srs1>youpi: (13:27:03) srs: youpi: Nowadays when fsck fails boot continues making diskcheck of / impossible. What changed?? <gnu_srs1>(13:27:48) srs: I have to check / from Linux. <youpi>gnu_srs1: that's handled by sysvinit IIRC <gnu_srs1>youpi: Which package, and how to find out? <youpi>probably the fsck scripts in /etc/init.d <youpi>where you can put echo statements to see what is happening