svn commit: r326809 - head/sys/dev/cardbus

[Alexey Dokuchaev]

On Thu, Dec 14, 2017 at 01:16:27AM +0700, Eugene Grosbein wrote:
> 14.12.2017 1:08, Alexey Dokuchaev пишет:
> > On Thu, Dec 14, 2017 at 12:48:17AM +0700, Eugene Grosbein wrote:
> >> ...
> >> I don't understand why you want to backout that. It changed nothing for
> >> amd64.  An i386 is still vulnerable to double faults just because of a
> >> network packet processing path can overflow kstack for the GENERIC kernel.
> > 
> > The reason (or shall I say concern) was raised by kib@, as quoted:
> > 
> >> Plain workstation use, like X11+browser+editor+some other programs easily
> >> allocates 1000+ threads.  It was still possible to use 32bit x86 for that,
> >> of course in max memory config without PAE, and without ZFS.  Add some
> >> load that involves network, for instance torrent client, to establish the
> >> pressure on KVA.
> >> I am almost sure that users would get troubles now.
> > 
> > I'm a plain workstation user with X11+browser+torrent+quake2+you-name-it
> > apps, PAE-less, ZFS-less, and I don't want to get troubles due to r326758.
> 
> Add kern.kstack_size=2 to /boot/loader.conf of your workstation now
> and you will notice no difference.

FreeBSD is largely known for its sane defaults.  Users should not touch
those config files, especially if things worked for years without any
/boot/loader.conf tweaking.  If there are cases of kernel stack pressure
like your script had revealed, they should be fixed, and defaults stay
as they are now.

One does not tweak FreeBSD, it just works. (tm)

./danfe