Re: Sysenter crash with Nested Task Bit set

[Jeremy Fitzhardinge <jeremy@goop.org>]

Linus Torvalds wrote:
> Final note (I promise): now that we save/restore eflags again, we 
> should probably revert the set_iopl_mask() in task switching too. However, 
> that apparently has some para-virtualization issues, so I'm going to 
> ignore that part from now.
>   

I'm wondering if we shouldn't have a (__)switch_to paravirt hook, so we 
can wrap the context switch in whatever we like.

> However, I'd really like people who know and care about the 
> paravirtualization to take a good long look at it: because right now, with 
> the addition of the eflags save/restore, the set_iopl_mask() in 
> __switch_to() is entirely useless for non-virtualized environments, afaik.
>   

Hm.  Zach removed the pushf/popf in switch_to this last Sept, with the 
comment "The pushf/popf in switch_to are ONLY used to switch IOPL. 
Making this explicit in C code is more clear.  This pushf/popf pair was 
added as a bugfix for leaking IOPL to unprivileged processes when using 
sysenter/sysexit based system calls (sysexit does not restore flags)."


> Zack added to the cc. Who else needs to know?
Rusty, Chris Wright and me.