tech-kern archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: PCIVERBOSE and kern/38579

On Sun, May 04, 2008 at 12:37:32PM -0700, Tim Rightnour wrote:
> On 04-May-2008 Andrew Doran wrote:
> > pci_subr.o is now quite large. I think it's time to retire PCIVERBOSE and
> > give the job of identification to pcictl, in userspace. It could do with a
> > way to operate on all busses at once but otherwise, it seems to do the job
> > already.
> Please no.  I make frequent use of that functionality when porting to a new
> machine.  It also makes my life much easier to just ask a user for a dmesg, 
> and
> have the descriptions right there.

I like the idea of teaching userland how to understand locators. If we do 
that, would it be sufficient for you if we had a script you feed the dmesg 
to and it adds the full strings to it?

> Additionally, as a user, I like to see that stuff come across the screen when
> booting.  I know instantly if something is not right.

Hmmm.... Ok, running this into the ground, another option might be to 
support custom lists. When you make a custom kernel, you can add a subset 
of the list (say all the stuff that showed up in dmesg last boot). So all 
the things you have in your system have nice names, and other stuff 
doesn't. If things start showing up cryptic, you know something changed..

> I don't want to have to sit around all the time looking up numbers on a chart
> while trying to figure out what is wrong with this or that device attachment. 
> I strongly object.  Pcictl is cute, assuming I can get to userland.

As above, I think having an annotating script would be a good idea.

Take care,


Attachment: pgpuaBBETANET.pgp
Description: PGP signature

Home | Main Index | Thread Index | Old Index