lkml.org 
[lkml]   [1998]   [Jun]   [19]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectImproving kernel stability and Linux usability with a shared boot option
To every Linux kernel user ...

As Linux kernel is getting mature, there are more and more
places where a choice as to be done between a more efficient
skeem or a more conservative one. Let's take a few exemples:

- using PIO or DMA in some network drivers
- using XT-PIC or IO-APIC on SMP boxes
- enabeling SCSI tag queuing and chosing the highest speed
on the SCSI bus or the more conservative 5Mb/s

It seems to me that time as now come (before 2.2 for sure)
to add a single boot option (witch could be called 'safe')
that would ask any part of the kernel and any driver to
use the most conservative skeem instead of the most efficient
one (when the difference is clear).

Moreover, we would ask to any part of the code that changed
it's behaviour due to this flag to display a kernel message
saying in witch readme file one can find additional
informations about the peticular alternative so that reading
the boot report allow the user to discover any choice he has
when operating the kernel on his peticular computer, and
what he has to read in order to operate it properly:

There would be several short term benefits:
- Easyer to get a stable Linux kernel for entry level users,
so better speading of Linux since any new user that fail
to install it's first kernel is a lost user ...
- Better reporting in kernel mail list since except Linus,
Alan, and a few others that seem to know everything
about the kernel, most advanced user could learn at once
all the possible kernel alernative on any peticular box, then
read the revelent README's and finaly search through using
individual boot parameters witch exact alternative get the
machine into troubles and report more precisely than
'kernel 2.1.x locks on my computer'

As a conclusion, i would say that the current way of releasing
frequent kernel works very fine if we change only one feature
in each realease and everybody checks each release, witch is
less and less possible since:
- as the kernel is getting bigger, and there are more and more
hackers, we need to change many things at once.
- since quite stable and full features kernel already exist,
most advanced users are jumping over several releases. This
is proved by the fact that many reports are not concerning
the latest release.

As a sumary, we (advanceds users) now need a more efficient
way to test changes in the kernel to better report to you
(hackers), or each serie will get longer and harder to get
stable than the previous one.

What's above is just a preposal.
It's now Linus time to speak.


Hubert Tonneau

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu

\
 
 \ /
  Last update: 2005-03-22 13:43    [W:0.028 / U:0.048 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site