[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: general protection - kernel & programs
On 20-Aug-98 Shlomi Fish wrote:
> On Thu, 20 Aug 1998, Evgeny Stambulchik wrote:
>
>> Buffer overflows are problems of badly written *programs* which can
>> crash kernel by rewriting system areas of memory (if they run under
>> root UID, of course).
If a badly written program crash (ANY) kernel it is a kernel bug.
It has nothing to do with root/non-root priviledeges as kernel
code and data/stack are located in memory that is not mapped by
the hardware from the virtual memory of ordinary processes.
The only way a normal process can access the kernel is through
the system-call interface which traps into kernel mode. Any
failure on the part of the kernel to check correct parameters to
system-calls (not library functions) is KERNEL BUG.
I stronly stands behind the postulate (stated by Evgeny):
Kernel panic in normal OS (Unix/Linux etc.) is caused
ONLY by hardware malfunction OR kernel bugs.
I think many people are confusing buffer overflows in applications and
in the kernel. The security implications of root owned processes who
may be exploited to execute arbitrary commands are in the APPLICATION domain.
> Linux is a relatively new OS, with a "limited" user-base. From what I
> heard, OpenBSD which is based on an older source code back from the 70's,
> is more bug-free than Linux is. If it makes you happy, you can switch to
You are correct in the APPLICATION domain were things changes very fast.
If you'll read through the change logs of the 2.0.x kernel patches, you'll
find the there weren't as many severe (crashing) kernel bugs as you may think.
----------------------------------------------------------------
Oron Peled Voice/Fax: +972-4-8228492
oron@actcom.co.il http://www.actcom.co.il/~oron
Linux: Because a PC is a terrible thing to waste.