Re: Questions regarding a PANIC situation


"adamsville2k" <adamsville2k@xxxxxxxxxxx> wrote in message
news:4d70ebe3-97d6-4473-bd34-9d80dd47d0df@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
On 9 mai, 08:41, Andrew Smallshaw <andr...@xxxxxxxxxxxxxxxx> wrote:
On 2008-05-09, adamsville2k <adamsvill...@xxxxxxxxxxx> wrote:



Last week, we experienced 2 KERNEL PANIC error on our SCO Unix 5.0.7
server within a 2 hour span. So I called our sofware support which is
OGC using running on an Informix Database. Within 5 minutes I had as
answer that the problem was an Hardware issue. Since the server is
supported by another supplier, I then called them. They told me that
such of a problem could be software as well. Not only hardware like
the SCO site posts here.

More details would have been appropriate here - you usually have
some indication of the source of the panic. However, the source
is unlikely to be your application, unless possibly it is doing
something nasty with raw memory via /dev/mem or something of that
sort. Put simply, your application shouldn't be able to make the
OS panic. It should reject any invalid requests made by system
calls etc cleanly and without threatening the integrity of the
system. This is generally what happens in reality. To do otherwise
would be a security flaw in that it would present a potential DoS
vector.

So the source of the panic is somewhere within the OS kernel itself
which doesn't have the same level of protection as userland
applications. OSR507 itself is generally fairly stable and so I
would tend to agree with your application provider that this is
probably a hardware issue. Kernel mode drivers typically have
limited error recovery built in to them but this is designed for
errors in normally functioning kit. Since you can't in general
anticipate what problems defective kit might throw up the drivers
don't attempt to and panic if things get too confusing.

--
Andrew Smallshaw
andr...@xxxxxxxxxxxxxxxx

Thanks Andrew for replying. I know that there is few details here but
the point is to find out who's wrong and who's right! So if "software"
means SCO OS alone, it's fine to me. I just want ot make sure that
applications don't make the KERNEL to panic.

Adamsville2k

Ok, let's look at this logically. Has your application EVER caused the
system to 'panic' trap? My guess is 'NO'. Therefore, it is only logical to
look elsewhere. Most times, when we experience panics, it's the direct
result of something else having occurred ...... either the hardware has
malfunctioned, someone has changed something in the OS, or maybe the lights
flickered a few times. Like the others, there really isn't enough
information here to make a sound decision on just what caused the panic, but
it would be prudent to start looking at your hardware and make dead sure you
have adequately backed up the important files. Kernel panics have a nasty
habit of crashing systems.

JP


.

Relevant Pages

  • Re: Microsoft Word wont load on my iMac G4 after kernel panic
    ... that it is hardware related, but I've found that an archive and install ... fixes it for about another 6 mos to a year (and they the panics return). ... reported there was 6 weeks ago, and said there was a memory exception ... /Library/Logs contains your panic reports. ...
    (microsoft.public.mac.office.word)
  • Re: stable sata patch: panic at kernel boot (cant dump)
    ... DA> | kernel panic. ... DA> | Kernel paniced just after sio0/sio1, where basic RELENG_4 starts ata channel ... No, the system panics reliably, just after sio initializing (for me it seems ... I did not use hardware RAID, I use vinum over these 5 ...
    (freebsd-stable)
  • Re: kernal panics -> hardware issue?
    ... machine has crashed four times with kernel panics. ... Up until now I've been thinking the panics were simply bugs in the intel ... In my experience, hardware problems ... behavior as well as kernel crashes. ...
    (comp.sys.mac.system)
  • Re: 2.6.30-rc1 - nmi_watchdog broken?
    ... I've been having fun finding bugs in 30-rc1. ... I've not seen this type of problem on this hardware ... If I disable highres timers it panics. ... Logs of boots with and without highres timers inlined below. ...
    (Linux-Kernel)
  • Re: 2.6.30-rc1 - nmi_watchdog broken?
    ... I've been having fun finding bugs in 30-rc1. ... I've not seen this type of problem on this hardware ... available and the boot stops. ... If I disable highres timers it panics. ...
    (Linux-Kernel)