lkml.org 
[lkml]   [1999]   [Oct]   [25]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: /proc/pci unknown devices


On Mon, 25 Oct 1999, Dan Hollis wrote:

> On Mon, 25 Oct 1999, Gerard Roudier wrote:
> > On Mon, 25 Oct 1999, Dan Hollis wrote:
> > > Which programs depend on /proc/pci ?
> > User's habits. ;-)
> > You seem to want to kill /proc/pci.
>
> Yup ;)
>
> > At a rough approximate, such a table without any king of space
> > optimization should not be greater than 16KB. Even twice this value does
> > not look unacceptable bloat to me, given the actual size of any kernel
> > ready to run for common configurations and common bloatwares we are using
> > from userland (Netscape, Emacs, Gcc, Perl, Bash and others as the Java
> > monstrosity really donnot care of sparing these 16KB of memory you seem to
> > worry about).
>
> For those running embedded pci systems the extra 35kb of wasted text
> strings is not really nice. Especially if you are going to have to ROM the
> image. For some users this could make the difference between kernel too
> large for lilo, and one that works.

Indeed.

> I thought we were trying to avoid M$-thinking, or is redmond starting to
> influence developers, "16kb here, 32kb there, it won't matter".

May-be some kernel projects are still avoiding what you called
M$-thinking, but userland seems to have been definitely converted.

> At the very least this should be a kernel option,
> "legacy /proc/pci (obsolete)". Taking this choice out of users hands is
> *bad*.

The initial decision was indeed unclever, but once a feature has been made
available it gets hard to remove it.

> > If you really want to spare this fiew memory, may-be this information
> > should be made available as a data file than can be read from the
> > kernel, from a userland tool, and from some library function as well.
>
> Why does the *kernel* need access to descriptive text strings of devices
> in order to function?

Indeed, it does not.

Only PCI device drivers need to recognize the devices they want to attach,
and they didn't need human understandable text for that, thanks to the
clean PCI specifications.

> This is purely for user convenience, hence it can be in userspace with
> pciutils.

Agreed.

> Besides, lspci is 100,000x more verbose and useful than 'cat /proc/pci'.
> Is "user habit to type 'cat /proc/pci'" a good enough reason to keep it?

That is one of the possible question about /proc/pci and if decision is
taken to remove it, we must allow time to userland softwares that still
rely on this feature for being converted to the replacement feature (that
already exists). It is too late in my opinion for 2.4. May-be, this should
be announced at the starting of a development phase and not at the end.
Did I miss this announcement ?

By the way, if I read in a couple of days that /proc/pci is dropped, I
would not complain at all. ;-)

Gérard.


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:54    [from the cache]
©2003-2011 Jasper Spaans. Advertise on this site