Re: PI patch against 2.6.16-rt9

[Thomas Gleixner]

On Tue, 2006-03-28 at 22:17 +0100, Esben Nielsen wrote:
> I think we talk about the situation

No, we talk about existing lock chains L(0) --> L(n).

>                         B locks 1            C locks 2       D locks 3
>                         B locks 2, boosts C and block
>       A locks 2
>       A is boost B
>       A drop it's spinlocks and is preempted
>                                              C unlocks 2 and auto unboosts
>                         B is running
>                         B locks 3, boosts C and blocks
>       A gets a CPU again
>       A boosts B
>       A boosts D
> 
> Is there anything wrong with that?
> And in the case where A==D there indeed is a deadlock which will be
> detected.

If you get to L(x) the underlying dependencies might have changed
already as well as the dependencies x ... n. We might get false
positives in the deadlock detection that way, as a deadlock is an
"atomic" state.

	tglx


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/