Apart from malicious software (eg. virus) or a device driver (which has
access to the kernel layer), is it possible for a standard application
written in C++ / C#.NET / VB / Delphi / etc to crash and/or corrupt Windows
XP?

We had one report from a customer who said that due to an accidental coding
bug, they saw a "divide by zero" error which not only crashed our software
but immediately afterwards Windows rebooted and then failed to restarted
(kept locking up and rebooting).

If the report is accurate, can user mode level programs do this? Is there
any way to prevent this?

As a final note, I reproduced the "dive by zero" crash and fixed it.
Certainly didn't crash my computer, though. In fact, I've never seen Windows
XP do what is reported here.

Thanks,
Alain

The users report is inaccurate or missing some vital detail

I make no comment on your ommission of a default err handler within your app


"Alain Dekker" <(E-Mail Removed)> wrote in message
news:OwnqO%(E-Mail Removed)...
> Apart from malicious software (eg. virus) or a device driver (which has
> access to the kernel layer), is it possible for a standard application
> written in C++ / C#.NET / VB / Delphi / etc to crash and/or corrupt
> Windows XP?
>
> We had one report from a customer who said that due to an accidental
> coding bug, they saw a "divide by zero" error which not only crashed our
> software but immediately afterwards Windows rebooted and then failed to
> restarted (kept locking up and rebooting).
>
> If the report is accurate, can user mode level programs do this? Is there
> any way to prevent this?
>
> As a final note, I reproduced the "dive by zero" crash and fixed it.
> Certainly didn't crash my computer, though. In fact, I've never seen
> Windows XP do what is reported here.
>
> Thanks,
> Alain
>

On Oct 27, 12:18*pm, "Alain Dekker" <abdek...@NOSPAM.fsmail.net>
wrote:
> Apart from malicious software (eg. virus) or a device driver (which has
> access to the kernel layer), is it possible for a standard application
> written in C++ / C#.NET / VB / Delphi / etc to crash and/or corrupt Windows
> XP?
>
> We had one report from a customer who said that due to an accidental coding
> bug, they saw a "divide by zero" error which not only crashed our software
> but immediately afterwards Windows rebooted and then failed to restarted
> (kept locking up and rebooting).
>
> If the report is accurate, can user mode level programs do this? Is there
> any way to prevent this?
>
> As a final note, I reproduced the "dive by zero" crash and fixed it.
> Certainly didn't crash my computer, though. In fact, I've never seen Windows
> XP do what is reported here.
>
> Thanks,
> Alain

Everything is possible. No two Windows XP are exactly the same. You
might be able to reproduce the error but you probably do not have the
same XP set up as the problematic PC.

You might be right that the report is missing some vital piece of info, but
the question is: Is it even possible for badly written code to corrupt
and/or crash Windows itself? I'm only calling standard Windows APIs in C++ /
Delphi so I hope not!

And yes, the programming bug and lack of default error handler was my bad...

Alain

"DL" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed)...
> The users report is inaccurate or missing some vital detail
>
> I make no comment on your ommission of a default err handler within your
> app
>
> "Alain Dekker" <(E-Mail Removed)> wrote in message
> news:OwnqO%(E-Mail Removed)...
>> Apart from malicious software (eg. virus) or a device driver (which has
>> access to the kernel layer), is it possible for a standard application
>> written in C++ / C#.NET / VB / Delphi / etc to crash and/or corrupt
>> Windows XP?
>>
>> We had one report from a customer who said that due to an accidental
>> coding bug, they saw a "divide by zero" error which not only crashed our
>> software but immediately afterwards Windows rebooted and then failed to
>> restarted (kept locking up and rebooting).
>>
>> If the report is accurate, can user mode level programs do this? Is there
>> any way to prevent this?
>>
>> As a final note, I reproduced the "dive by zero" crash and fixed it.
>> Certainly didn't crash my computer, though. In fact, I've never seen
>> Windows XP do what is reported here.
>>
>> Thanks,
>> Alain
>>
>
>

True. I'll try and set up a system like the customer reported and see if
this problems crashes the computer tomorrow.

However, your assertion that "anything is possible" is a bit worrying! I
thought just calling the odd Windows API should not be able to crash or
corrupt Windows itself, even if the programmer has done a bad (not
malicious) job.

Thanks

"smlunatick" <(E-Mail Removed)> wrote in message
news:5a9f11ac-454e-4229-8529-(E-Mail Removed)...
On Oct 27, 12:18 pm, "Alain Dekker" <abdek...@NOSPAM.fsmail.net>
wrote:
> Apart from malicious software (eg. virus) or a device driver (which has
> access to the kernel layer), is it possible for a standard application
> written in C++ / C#.NET / VB / Delphi / etc to crash and/or corrupt
> Windows
> XP?
>
> We had one report from a customer who said that due to an accidental
> coding
> bug, they saw a "divide by zero" error which not only crashed our software
> but immediately afterwards Windows rebooted and then failed to restarted
> (kept locking up and rebooting).
>
> If the report is accurate, can user mode level programs do this? Is there
> any way to prevent this?
>
> As a final note, I reproduced the "dive by zero" crash and fixed it.
> Certainly didn't crash my computer, though. In fact, I've never seen
> Windows
> XP do what is reported here.
>
> Thanks,
> Alain

Everything is possible. No two Windows XP are exactly the same. You
might be able to reproduce the error but you probably do not have the
same XP set up as the problematic PC.

Alain Dekker wrote:
> Apart from malicious software (eg. virus) or a device driver (which
> has access to the kernel layer), is it possible for a standard
> application written in C++ / C#.NET / VB / Delphi / etc to crash
> and/or corrupt Windows XP?
>
> We had one report from a customer who said that due to an accidental
> coding bug, they saw a "divide by zero" error which not only crashed
> our software but immediately afterwards Windows rebooted and then
> failed to restarted (kept locking up and rebooting).
>
> If the report is accurate, can user mode level programs do this? Is
> there any way to prevent this?
>
> As a final note, I reproduced the "dive by zero" crash and fixed it.
> Certainly didn't crash my computer, though. In fact, I've never seen
> Windows XP do what is reported here.
>

XP, unlike Win98, is built on the NT model. As such, applications are
rigorously segregated from the operating system and a crash of an
application program should not affect the OS. An application program causing
a system crash is about as likely as a piece of software causing a hardware
failure. In theory, it IS possible; in practice, extremely doubtful.

I'd follow the medical school maxim: "If you hear hoofbeats, think horses,
not Zebras."

Not in my experience, but with a pc virtually anything is possible be it
software or hw related

"Alain Dekker" <(E-Mail Removed)> wrote in message
news:%(E-Mail Removed)...
> You might be right that the report is missing some vital piece of info,
> but the question is: Is it even possible for badly written code to corrupt
> and/or crash Windows itself? I'm only calling standard Windows APIs in C++
> / Delphi so I hope not!
>
> And yes, the programming bug and lack of default error handler was my
> bad...
>
> Alain
>
> "DL" <(E-Mail Removed)> wrote in message
> news:(E-Mail Removed)...
>> The users report is inaccurate or missing some vital detail
>>
>> I make no comment on your ommission of a default err handler within your
>> app
>>
>> "Alain Dekker" <(E-Mail Removed)> wrote in message
>> news:OwnqO%(E-Mail Removed)...
>>> Apart from malicious software (eg. virus) or a device driver (which has
>>> access to the kernel layer), is it possible for a standard application
>>> written in C++ / C#.NET / VB / Delphi / etc to crash and/or corrupt
>>> Windows XP?
>>>
>>> We had one report from a customer who said that due to an accidental
>>> coding bug, they saw a "divide by zero" error which not only crashed our
>>> software but immediately afterwards Windows rebooted and then failed to
>>> restarted (kept locking up and rebooting).
>>>
>>> If the report is accurate, can user mode level programs do this? Is
>>> there any way to prevent this?
>>>
>>> As a final note, I reproduced the "dive by zero" crash and fixed it.
>>> Certainly didn't crash my computer, though. In fact, I've never seen
>>> Windows XP do what is reported here.
>>>
>>> Thanks,
>>> Alain
>>>
>>
>>
>
>

) Thanks, sensible advice. I also find it hard to believe my code somehow
corrupted Windows. In our experience, if a customer has done something major
bad (maybe ram the machine into a wall when using a forklift truck) then
they'll try and cover their trails and claim the "machine just went bang" if
they think they can get away with it.


"HeyBub" <(E-Mail Removed)> wrote in message
news:%23UX%(E-Mail Removed)...
> Alain Dekker wrote:
>> Apart from malicious software (eg. virus) or a device driver (which
>> has access to the kernel layer), is it possible for a standard
>> application written in C++ / C#.NET / VB / Delphi / etc to crash
>> and/or corrupt Windows XP?
>>
>> We had one report from a customer who said that due to an accidental
>> coding bug, they saw a "divide by zero" error which not only crashed
>> our software but immediately afterwards Windows rebooted and then
>> failed to restarted (kept locking up and rebooting).
>>
>> If the report is accurate, can user mode level programs do this? Is
>> there any way to prevent this?
>>
>> As a final note, I reproduced the "dive by zero" crash and fixed it.
>> Certainly didn't crash my computer, though. In fact, I've never seen
>> Windows XP do what is reported here.
>>
>
> XP, unlike Win98, is built on the NT model. As such, applications are
> rigorously segregated from the operating system and a crash of an
> application program should not affect the OS. An application program
> causing a system crash is about as likely as a piece of software causing a
> hardware failure. In theory, it IS possible; in practice, extremely
> doubtful.
>
> I'd follow the medical school maxim: "If you hear hoofbeats, think horses,
> not Zebras."
>