lkml.org 
[lkml]   [1996]   [May]   [14]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: On the topic of noisy drivers...
On May 9, 12:23am, Chris Evans wrote:
> > I'll state this once more for a final bit of clarity. I have absolutely no
> > qualms with printing out large amounts of debugging information on bootup.
>
> Good. I like to have my Linux system reboot as infequently as possible (ie
> a week is too short an uptime), but when it does it's nice to see all the
> nice features of your various hardware being identified. What is annoying
> you? If your logfiles get too big, process and erase the useless bits once
> in a while.

Chris, I'll go over this once more for you. I think that I have explained this
well enough in past posts, so I feel that I am, now, going out of my way to
explain my position to someone who isn't listening.

> > Since it's for debugging, it's for people with knowlege of the system.
Since
> > the knowlegable people are going to read it, it would be fine to compact it
> > some or shorten some of the comments and spacing. Cryptic is fine. Taking
a
> > whole line to say that a SCSI device is a direct-access SCSI-1 CCS drive is
> > rediculious. My final point is that some of us have limited options when
it
> > comes to scrollback. If the 'debugging' information scrolls of the screen
> > before it can be read, it's not too useful for debugging, is it?
>
> No. Unless you have the sense to run sysklogd. Or use the 'dmesg'
> command. The buffer for kernel messages can be increased if you have >4k
> of messages upon bootup (or is the ring bufer more than 4k in recent
> kernels anyway?)

What if the boot doesn't complete? If the system can't finish booting, how am
I going to run syslog or syslogd? There is no logging of boot messages until
after the boot has finished. Here's the scenario:

I reboot the machine. Meaningful messages get printed. They scroll off
because the scsi code (all levels including the driver) decide to print out
poorly formatted init messages. There is no way to scroll back to see the
previous messages as the machine has a Herc mono card--which has no scrollback.
If the boot now fails, I will never be able to log the ring buffer to disk.
If I can't do that, I will never be able to read the init messages which might
tell me what happened durring boot.

> How are people supposed to see their hardware identified correctly if
> messages are cryptic? Give a good reason why verbose messages are a bad
> thing, and I might have sympathy!

Chris, reread my previous message. If you still can't see why this is needed,
I'll be glad to explain to you, in private email why this is necessary.

Your first question regarding how people are to identify their hardware is an
obvious straw man and I won't address it.

Cheers,
David


\
 
 \ /
  Last update: 2005-03-22 13:37    [W:0.032 / U:1.956 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site