Re: cli/sti
Systemkennung Linux (linux@mailhost.uni-koblenz.de)
Thu, 31 Jul 1997 01:39:55 +0200 (MET DST)
> Is there a general effort underway to squash cli/sti pairs?
> I ask because I think there should be. The code should restore
> the flags if it is going to be re-entrant.
>
> Example:
> The loop driver assumes some kernel code is re-entrant, when it
> definitely isn't (cli/sti pairs get executed). The bug strikes
> mostly on a loaded system, so it was terribly hard to track this
> bug down. Changing the code to restore the flags works fine, as
> the code involved is otherwise re-entrant.
>
> I wonder how much other code in the kernel makes assumptions about
> the interrupt state.
In adittion I'd like to see save_flags(); cli(); sequences to be
replaced by save_and_cli() calls. I've already done this for the
MIPS specific code. On MIPS this macro results in slightly shorter
code that keeps the pipeline happier and therefore is faster.
Ralf