[lkml]   [2000]   [Jul]   [23]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
SubjectRe: scsi-destroyer.c to come...
Aaron Sethman <> said:
> On Sat, 22 Jul 2000, David Luyer wrote:
> > So what?
> >
> > It is _much_ easier to do "cat </dev/zero >/dev/sda"
> > and about as likely to be effective.

> But how often is a program with a bug in it likely to execute this
> command?

Scripts, run by root, which are then given the wrong output file? Does
happen... Linus himself fried a partition table by dialing /dev/hda once ;-)

> Lets say some program has a buffer overflow somewhere and it
> ends up clobbering the section of memory that holds the bits for the
> ioctl. Suddenly your data is gone and your left clueless. My question is
> what functionality are you going to lose by have some sanity checking in
> the kernel? Yes I know you can fiddle with /dev/mem etc..but this is know
> where near as likely to be triggered by a buggy program now is it?

Never heard of anything like that happening. Or an accidental buffer
overflow reflashing a BIOS, or the SCSI controller's BIOS, or anything like
that. The real (much more frequent, that is) problems are quite mundane:
Overflows/following the wrong pointer inside the kernel has scribbled over
the disk and forced two or three complete reinstalls here during 2.3, core
files generated by overflows have clogged / on occasion, 2.4.0-testX NFS
doesn't work right with Red Hat 6.2 init scripts/programs (nfslock
complains and doesn't start, and running NFS this way screws up Solaris 2.6
machines serving files so they have to be rebooted), a couple of CDs I've
got give SCSI I/O errors on my SS4, and from then on more and more
processes freeze until in some 2 minutes the machine is dead. Those are
problems that do matter to me, not a theoretical way to kill a disk, given
that there are literally thousands of ways of doing so already.
Horst von Brand
Casilla 9G, Vin~a del Mar, Chile +56 32 672616

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to
Please read the FAQ at

 \ /
  Last update: 2005-03-22 13:57    [W:0.090 / U:9.764 seconds]
©2003-2018 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site