Re: [Xen-devel] trap bounce flags

[Keir Fraser <keir@xxxxxxxxxxxxx>]

On 25/4/07 11:33, "Jan Beulich" <jbeulich@xxxxxxxxxx> wrote:

> That's the alternative solution I considered. The preferable one is to do the
> compat/native distinction before the null check, and then be consistent with
> the rest of the code and check cs for 32-bit guest and eip for 64-bit ones.
> That's how I'm preparing a patch right now.

Attached is my own proposed patch which I think cleans up all the issues.
Checking just flags in asm and keeping the null-bounce check in
init_int80_direct_trap() seems fine to me.

 -- Keir

>>> - from the above, why is it that only the lower byte (if anything) needs
>>> clearing?
>> 
>> Really it's a one-byte field: it's consistently treated that way in asm
>> code. The upper byte is always zero. We should probably make the field
>> explicitly uint8_t. Agree?
> 
> Making it a uint8_t is fine. It is, however, far from being consistently
> handled
> in assembly code:
> x86_32/entry.S: 4 word refs and 3 byte refs
> x86_64/entry.S: 6 word refs, 3 byte refs, and one size-less ref
> x86_64/compat/entry.S: 4 word refs and 3 byte refs

Attachment: 00-fix-trapbounce
Description: Binary data

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel