***Server sets mode: +nt
***Server sets mode: +nt
<janneke>hmm, if i change in grub.cfg: "module /hurd/ext3fs.static ... '${root}'" to ... '(hd0,msdos1)' <janneke>then i get "(hd0,msdos1)": Gratuitous error <janneke>i'm wondering what's going on, exactly how/what '${root}' expands to? <damo22>or are you working on ext3 suppot <janneke>but it looks like the hurd does not find the right partition <janneke>i'm trying several new (for me) things on bare iron <damo22> --exec-server-task='${exec-task}' -T typed '${root}' <--- ${root} here is something like device:/dev/sd0s2 <damo22>i cant recall if the "/dev/" is required <damo22>im trying to make hurd boot with rumpdisk translator instead of gnumach disk drivers <janneke>i'm "playing" with the ${root} parameter because a previous error (with dofault '${root}') says: <damo22>we're not far off userspace disk drives <janneke>bad magic number 0 (should be 0xef53) ... but i have two partitions, and tune2fs -l disagrees; none of them is 0 <janneke>damo22: yeah, i wanted to ask you about that yesterday <damo22>uh, maybe you didnt specify the partition <janneke>sounds exciting, but i'm unfamiliar with rump, what it means... <damo22>where X is the disk number and Y is the partition number <damo22>it may also be hdXsY instead if its IDE <damo22>rump is an interface to NetBSD kernel as a unikernel that can run as a userspace process <janneke>so...from a distance i see no difference with the current ext2 file server, that's also user space? *janneke misses something real important, i'm sure! <damo22>the rump driver will be running in userspace but it will appear as a regular hurd store <damo22>instead of accessing the disk via gnumach driver <damo22>so its basically doing PCI access in userspace, which can be managed by a server <damo22>but its a bit tricky because then it needs an IRQ handler to be installed from userspace <janneke>and then we don't need the ext2fs.static server anymore at all? <damo22>we still need that for filesystems <damo22>otherwise youre just on a block device <janneke>that's a higher layer, rump/you're talking about the underlying device access? <damo22>ext2fs.static can use the block device that rumpdisk will expose ***p2-mate_ is now known as p2-mate
<damo22>instead of ext2fs using the block device that the kernel exposes <damo22>but one day we can replace ext2fs server with a rump filesystem implementation as well <damo22>the benefit is, you get all the other filesystems that netbsd supports <damo22>usb, networking, audio to follow :) <janneke>could well fix the what i'm looking at <janneke>for the archives, using "device:hd0s2" gives the same as '${root}' <janneke>start ext2fs: ext2fs: device:hd0s2: panic: get_hypermetadata: bad magic number 0 (should be 0xef53) <janneke>so it looks like "someone" reads the partition table wrong <damo22>try part:2: device:hd0 <--- with no space after the 2: <janneke>no space... so => part:2:device:hd0 ? <damo22>but on my screen it converts :d into an emoji <damo22>what is the full line of command that includes {root} <janneke>hmm, ext2fs: part:2:device:hd0: Invalid argumnent <janneke>yes it has ... --store-type=typed --x-xattr-translator-records part:2:device:hd0 '$(task-create)' '$(task-resume)' <janneke>(let me look-up full module boot line) <damo22>multiboot /boot/gnumach-1.8-486.gz root=device:sd0s2 <janneke>ah...i've only been playing with the ext2fs.static line <janneke>for gnumach, i have multiboot /gnu/store/xxbi2zw00342gnan6wwmrlsk104hs50v-gnumach-1.8-1.097f9cf/boot/gnumach root=device:hd0s2 --root=hurd --system=/gnu/store/f3ny5r6xpqncmw9j7sgmwahfq41r6ykk-system --load=/gnu/store/f3ny5r6xpqncmw9j7sgmwahfq41r6ykk-system/boot <damo22>module /hurd/ext2fs.static ext2fs --multiboot-command-line='${kernel-command-line}' --host-priv-port='${host-port}' --device-master-port='${device-port}' --exec-server-task='${exec-task}' -T typed '${root}' '$(task-create)' '$(task-resume)' <janneke>yes, i have (as the non-hardcoded versions that i'm playing with): <janneke> module /gnu/store/wkxn4idl9g7rkbm5aq7412ppb0pr07fy-hurd-0.9-1.91a5167/hurd/ext2fs.static ext2fs --multiboot-command-line='${kernel-command-line}' --host-priv-port='${host-port}' --device-master-port='${device-port}' --exec-server-task='${exec-task}' --store-type=typed --x-xattr-translator-records '${root}' '$(task-create)' '$(task-resume)' <damo22>maybe --x-xattr etc needs to be moved before the --store-type <damo22>i usually use -T typed '${root}' <janneke>i don't think so...this all works great in a VM <janneke>i'm trying the same thing on bare iron the past days <janneke>previously, i used one partition only /dev/sda1 <janneke>installed guix gnu/linux and converted that partition in-place to the hurd <damo22>maybe your disk controller is AHCI <almuhs>hi. Are there any way to install libreoffice in Debian GNU/Hurd? <damo22>almuhs: have a look in the repo, is it there? <almuhs>there are libreoffice-common, but not libreoffice <damo22>maybe we need to focus on drivers for hw before using it as a proper word processor <damo22>if hurd booted on native hw i might switch to it and start caring about desktop packages <almuhs>the main advantage of Hurd over other microkernel-based operating systems is that doesn't requires virtualization to run most GNU/Linux applications <almuhs>so, I want to test some popular applications in Hurd <almuhs>I need to know more advantages of Hurd over other operating systems, to write this in my academic article <damo22>almuhs: i think the main advantages of a microkernel model is that you can run drivers as user processes so if your driver crashes you can recover without a kernel panic <damo22>drivers become individualised into separate processes running in userspace, kernel does very little except memory management and processing messages <almuhs>damo22: yes, but all microkernel OS has this characteristic, is not? <damo22>you mean the advantages of hurd over other microkernel systems? <damo22>i dont even know any other mainstream microkernel systems <almuhs>searching in internet, I found many systems based in seL4 or L4-family microkernels <damo22>the main advantage i would say is that its GNU <3 <almuhs>yes, by this reason Hurd keeps compatibility with Linux applications <almuhs>because the GNU environment is the same <almuhs>but I need more reasons. Maybe namespaces? <damo22>thats not what i mean, i mean its made by the same people who invented the rest of the free tools <almuhs>but I need a more "commercial" reason <damo22>its not a commercial product its a free as in freedom product <damo22>its purpose is to provide a system for everyone to use such that they are not restricted by a commercial interest <damo22>one of the big reasons i am working on this project is because i feel like it is helping unchain the world from harmful technology (usually driven by big companies) <damo22>its not necessarily a "technically" better product than something else, its not even the point <damo22>i want to help the people who started this fight <damo22>because they had the right idea in the beginning, and it still applies today <almuhs>maybe, the main reason about Hurd is simply that is a complete operating system, not only a microkernel <almuhs>seL4 doesn't seems to have any mature operating system, only specific applications ***Emulatorman_ is now known as Emulatorman
***DNS is now known as snd