Re: [Xen-devel] RFC: 32 bits as smallest atomic size. - Keir Fraser - com.xensource.lists.xen-devel

26 messages in com.xensource.lists.xen-develRe: [Xen-devel] RFC: 32 bits as small...

From	Sent On	Attachments
Jimi Xenidis	25 Mar 2005 13:24
Ronald G. Minnich	25 Mar 2005 13:30
Keir Fraser	25 Mar 2005 13:43
Jimi Xenidis	25 Mar 2005 15:48
Keir Fraser	25 Mar 2005 16:02
Jimi Xenidis	26 Mar 2005 06:12
Keir Fraser	26 Mar 2005 09:07
Ronald G. Minnich	26 Mar 2005 09:29
Jimi Xenidis	26 Mar 2005 09:31
Ronald G. Minnich	26 Mar 2005 09:36
Keir Fraser	26 Mar 2005 09:49
Keir Fraser	26 Mar 2005 09:55
Jimi Xenidis	26 Mar 2005 10:46
David Hopwood	26 Mar 2005 16:27
Ronald G. Minnich	26 Mar 2005 19:34
Keir Fraser	27 Mar 2005 02:49
Keir Fraser	27 Mar 2005 02:55
Keir Fraser	27 Mar 2005 03:18
David Hopwood	27 Mar 2005 07:39
Ronald G. Minnich	28 Mar 2005 07:34
Ronald G. Minnich	28 Mar 2005 07:37
Kurt Garloff	29 Mar 2005 12:15
Hollis Blanchard	29 Mar 2005 13:38
Ian Pratt	29 Mar 2005 14:00
Hollis Blanchard	29 Mar 2005 14:32
Jimi Xenidis	29 Mar 2005 19:24

Subject:	Re: [Xen-devel] RFC: 32 bits as smallest atomic size.
From:	Keir Fraser (Keir...@cl.cam.ac.uk)
Date:	03/25/2005 01:43:01 PM
List:	com.xensource.lists.xen-devel

I expect we can fix that up in the ppc macros: if the atomic access is sub-32-bit aligned then round the address down to 32-bit boundary and do a 32-bit cmpxchg.

-- Keir

On 25 Mar 2005, at 21:25, Jimi Xenidis wrote:

In PPC land we have an issue where cmpxchg operations can only be applied to strictly aligned 32 and 64 bit quantities.

We understand on x86 the use of the packed attribute has certain benefits, but in the common areas we believe that portability should be the primary concern in the common code.

I can only assume that much debate will come from this as we find alignment and size issues all over.

The particular offender at the moment is: include/public/grant_table.h flags 76 u16 flags;

It is used with atomic methods such as cmpxchg_user() and clear_bit().

Any thoughts on how to proceed with this would be appreciated.

-JX

BTW: is clear_bit() as implemeted for x86 have soem alignment issues? My expertise in this space is limited.