Trap -> page fault but error code == 0 ?

[JulienDarc]

Hello,

I understand that having an error ==0 when having a page fault exception, when in kernel space, is not a dangerous thing.
http://wiki.osdev.org/Exceptions#Page_Fault

The cpu just means : "hey OS, map that address for me and we are done; nothing dangerous happened."

This situation seems pretty clear but I have no one else to ask/confirm but you.

Am I right ? Did I overlook something ?

Thanks

[Techel]

That's right. A page fault by reading from unmapped memory.

[JulienDarc]

Thanks Roflo

[iansjack]

It's not necessarily a dangerous thing as long as you handle it. In other words, just the same as any other page fault.

But normally, although it obviously depends upon the design of your kernel, any page fault in kernel space is a sign that something unexpected has happened.

[JulienDarc]

Yes thanks a lot iansjack,

i am currently writing the routine to handle that.

Now that it is clear that there is nothing potentially malicious, i feel safer

OS development is a very rewarding process, that is amazing.
It helps improving the code on the application side. Should have started it earlier !

But I digress now,

Bye

Julien

[xenos]

I would say if it is unexpected, then it is also potentially dangerous, because it shows there must be a bug in your code which causes this unexpected behavior. It's quite possible that this bug will do no harm, but it may just as well crash your kernel - in any case, your kernel will not do what you expect, so you should in any case be alerted and figure out the reason for the page fault.

But if you expect a page fault, because you design your kernel in such a way that you map pages on demand, and this demand is indicated by a page fault, then of course it is not dangerous.