:: Re: [Dng] What if systemd infects t…
Top Page
Delete this message
Reply to this message
Author: Jude Nelson
Date:  
To: Gravis
CC: dng@lists.dyne.org
Subject: Re: [Dng] What if systemd infects the kernel?
> I have been aware of the pending assimilation of systemd into the kernel
since Linus dramatically rejected Kay Sievers' code last spring. Recently
there has been renewed chatter about the impending doom. But I'm not quite
clear how that would affect devuan. Hoping you can help me get a grip on
the situation:

As far as I know, systemd is *not* getting merged into the kernel. It's
not even being *considered* for merging.

Linus was mad at Kay Sievers for something completely unrelated. Kay had
introduced a regression in udev that could cause it to hang due to the way
it handled a driver's request for firmware (until the kernel timed the
firmware request out after 30 seconds). Instead of fixing the regression
(and it was a regression--he was using the *well documented* firmware
loader *incorrectly*), he went on to suggest that the kernel developers
were at fault for the situation by designing Linux's firmware loader the
way they did, instead of fixing the problem he created. The rest is
history.

As I said in another thread, kdbus as it is incarnated today (and this is
subject to change--it's been pretty much re-written each time Greg KH has
submitted it to lkml) requires some userspace help in setting it up. This
is only necessary if your kernel has kdbus and you have programs that want
to use it. kdbus by itself is just another way to move bytes between
processes, and it offers some semantics that aren't found in other existing
IPC mechanisms. It is *not at all* like dbus, any more than a UNIX domain
socket is like dbus or a pipe is like dbus. The only thing kdbus has in
common with dbus is that it shares part of the name.

-Jude




On Sat, Feb 28, 2015 at 11:03 PM, Gravis <ring3k@???> wrote:

> > I have been aware of the pending assimilation of systemd into the kernel
> since Linus dramatically rejected Kay Sievers' code last spring.
>
> a huge reason for it's rejection was it's lack of documentation. this
> situation hasnt changed much in regard to kdbus.
>
>
> > Recently there has been renewed chatter about the impending doom.
>
> it's extremely unlikely. however, if for some reason it does, it can
> be excluded if one chooses to exclude it. some of the stuff written
> for kdbus actually has been accepted (about six months ago), in
> particular the sealed memory file descriptor stuff. this actually is
> good code and is the reason that kdbus has zero copy. with this, we
> can now use unix domain sockets to pass memory with the same zero copy
> goodness as kdbus. the question remains now is what good is the rest
> of kdbus? since the documentation is lacking, we dont know and thus
> it's not getting into the kernel. if kdbus eventually gets into the
> kernel, it will have been well inspected and reviewed and deemed
> worthy of inclusion. this is the complete opposite of how systemd has
> been operating.
>
>
> > Why not one response?
>
> probably because gmail (among other services) is automatically marking
> your emails as spam since they are claimed to have been sent _by_
> yahoo's server but yahoo's server is denying it. either you are not
> using yahoo's server to actually send your emails or the mailing list
> server is misconfigured.
>
> --Gravis
>
>
> On Sat, Feb 28, 2015 at 8:07 PM, Go Linux <golinux@???> wrote:
> > On Sat, 2/28/15, Go Linux <golinux@???> wrote:
> >
> > Subject: What if systemd infects the kernel?
> > To: dng@???
> > Date: Saturday, February 28, 2015, 1:05 AM
> >
> > I have been aware of the pending assimilation of systemd into the kernel
> since Linus dramatically rejected Kay Sievers' code last spring. Recently
> there has been renewed chatter about the impending doom. But I'm not quite
> clear how that would affect devuan. Hoping you can help me get a grip on
> the situation:
> >
> > Would a systemd-infected kernel bring devuan to its knees?
> >
> > IOW will devuan require a systemd-free kernel to run properly?
> >
> > Would the VUAs be able to disinfect the kernel? Or is that something
> that would have to go through Linus?
> >
> > Is there any chance that the kernel devs would be willing to maintain
> two separate kernel versions?
> >
> > Or will devuan be up a creek if/when that happens?
> >
> > I'm assuming the VUAs have thought about this - I can't imagine they
> would they be going through this monumental effort only to be foiled by a
> systemd kernel - and that there is a solution. Please enlighten me. :)
> >
> > golinux
> >
> >
> > --------------------------------------------
> >
> > Either this is an incredibly stupid question or it's the elephant in the
> room. Why not one response? This inquiring mind would like to know.
> >
> > golinux
> >
> >
> > _______________________________________________
> > Dng mailing list
> > Dng@???
> > https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng
> _______________________________________________
> Dng mailing list
> Dng@???
> https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng
>