Reliable version?

[Mike Bilow]

root wrote in a message to Mike Bilow:

 r> Could someone (as Linux users) kindly inform me how I might
 r> go  about working around a kernel panic when I try to
 r> recompile the  kernel (2.0.0) on my laptop.  I get 'kernel
 r> paging errors' at  different places in the compilation
 r> process.  
 r> 1. Thinking it was hardware, I tried switching out a 8MB and
 r>     16MB memory module with no effect.
 r> 2. Thought bad kernel image, so I compiled an image using a
 r>     486 desktop, and moved the image over.  No difference.
 r> What should be my approach in tackling this?

The most likely cause of kernel panics is something to do with the device
drivers, usually one of them picking up what it thinks is its device that is
not really there.  Do not put unnecessary drivers into your built kernel.  What
is the brand and model of the laptop?  Is it APM-compliant?  Do you get
different behavior with PCMCIA cards inserted or removed?

 r> An easier question that I have is a sendmail phobia, ahem,
 r> problem; How do you map the (envelope?) From: addresses of
 r> outgoing mail from user1 at mydomain to user2 at my_isp's_domain. 
 r> (Having the correct  address facilitates replies from
 r> others.)  You use the DM in  sendmail.cf to masquerade the
 r> domain, but what about username  translation.  I assume it
 r> has something to do with (from sendmail.cf)     # Generics
 r> table (mapping outgoing addresses)
 r>     #Kgenerics dbm /etc/genericstable
 r> 1. ...but, I could be wrong about the approach
 r> 2. What is the structure of that file? Similar to
 r> /etc/aliases?    Documentation references July 94 issue of
 r> Linux Journal as    an excellent source, but I don't have
 r> it.
 r> Before I dig out the source, am I even headed in the right 
 r> direction?

The macros in sendmail.cf are fully configurable, but the traditional one for
username translation is P.  If you have rules already present in the default
sendmail.cf file, they will generally look for the P macro.  There will also
often be an annotated "DP" empty definition.  The macro is usually referenced
is some ruleset specifically invoked for this case:

# Rewrite addresses so they look like they are from a specific user id        
S41                                                                           
R$*                     $: < $P > $1      expand with user id macro
R< > $+                 $@ $1             no user id macro defined
R< $+ > $+ < @ $H >     $@ $P < @ $H >    user at host => newuser at host
R< $+ > $+              $@ $2             user id macro defined but no rewrite

This ought to give you some idea.
 
-- Mike