Re: KMSGDUMP: dump kernel messages to a diskette

Maciej W. Rozycki (macro@ds2.pg.gda.pl)
Fri, 9 Jul 1999 14:51:21 +0200 (MET DST)


On Thu, 8 Jul 1999, Dan Hollis wrote:

> > > md5 the important parts of memory, and then checksum verification before
> > > using them. quite simple, and 100% guarantees we don't use clobbered ram.
> > what if the md5-verifying function is clobbered itself ?
>
> Then the md5 will fail and you will kindly advise the user that a dump is
> not possible (or warn them it is extremely unwise).

If RAM is clobbered you cannot assure the check will fail for bogus data.
Then the code executed afterwards may trash data on a hard disk.

> - PC locks up, HARD, due to interaction between PCI cards and buggy
> drivers. no oops is possible since the harware is locked up.
> - Press PC "red reset button" for hardware reset.
> - PC enters minimal dump code using '286 return to real mode' trick.

This does not work. Asserting RESET causes one of the bits in the
keyboard controller to remember the fact and all PC BIOSes check for a
cold start at the very beginning of the execution and ignore the reset
code in this case. Also memory refresh subsystem may not work after
asserting RESET. What PCs are really missing is a HALT button or just a
single button that might be configured either for HALT or for RESET.

Well, all the problems we are writing abou at the moment are the results
of PCs being PCs and not workstations nor servers. Actually there may
exist some ia32-based workstations or servers but most products advertised
as such are still PCs.

-- 
+  Maciej W. Rozycki, Technical University of Gdansk, Poland   +
+--------------------------------------------------------------+
+        e-mail: macro@ds2.pg.gda.pl, PGP key available        +

- 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/