All of lore.kernel.org
 help / color / mirror / Atom feed
* difference between PGT_gdt_page and PGT_ldt_page
@ 2006-02-15  9:09 Jan Beulich
  2006-02-15  9:26 ` Keir Fraser
  0 siblings, 1 reply; 4+ messages in thread
From: Jan Beulich @ 2006-02-15  9:09 UTC (permalink / raw)
  To: xen-devel

Is there any particular reason why there cannot just be a PGT_descr_page, combining the two currently existing types?
Even if it's only a page, I'd want to avoid wasting memory for nothing, and i386's default_ldt could easily move into
the same page cpu_gdt_table lives in. Alternatively one could also use empty_zero_page here. Or do it like on x86-64,
where the symbol doesn't exist at all, and knowledge about it being all zeroes is used where needed. Jan

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: difference between PGT_gdt_page and PGT_ldt_page
  2006-02-15  9:09 difference between PGT_gdt_page and PGT_ldt_page Jan Beulich
@ 2006-02-15  9:26 ` Keir Fraser
  2006-02-15 11:22   ` Jan Beulich
  2006-02-15 11:23   ` Jan Beulich
  0 siblings, 2 replies; 4+ messages in thread
From: Keir Fraser @ 2006-02-15  9:26 UTC (permalink / raw)
  To: Jan Beulich; +Cc: xen-devel


On 15 Feb 2006, at 09:09, Jan Beulich wrote:

> Is there any particular reason why there cannot just be a 
> PGT_descr_page, combining the two currently existing types?
> Even if it's only a page, I'd want to avoid wasting memory for 
> nothing, and i386's default_ldt could easily move into
> the same page cpu_gdt_table lives in. Alternatively one could also use 
> empty_zero_page here. Or do it like on x86-64,
> where the symbol doesn't exist at all, and knowledge about it being 
> all zeroes is used where needed. Jan

If default_ldt can be replaced by empty_zero_page, why do we need a 
default LDT at all?

The page types could quite easily be merged. The main problem is that 
all descriptor tables need to be page aligned. We could get rid of that 
restriction for LDTs I think.

  -- Keir

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: difference between PGT_gdt_page and PGT_ldt_page
  2006-02-15  9:26 ` Keir Fraser
@ 2006-02-15 11:22   ` Jan Beulich
  2006-02-15 11:23   ` Jan Beulich
  1 sibling, 0 replies; 4+ messages in thread
From: Jan Beulich @ 2006-02-15 11:22 UTC (permalink / raw)
  To: Keir Fraser; +Cc: xen-devel

>If default_ldt can be replaced by empty_zero_page, why do we need a 
>default LDT at all?

That would be one of the implied questions. I assume this is done to minimize the delta to native code, which
needlessly has this, too.

Jan

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: difference between PGT_gdt_page and PGT_ldt_page
  2006-02-15  9:26 ` Keir Fraser
  2006-02-15 11:22   ` Jan Beulich
@ 2006-02-15 11:23   ` Jan Beulich
  1 sibling, 0 replies; 4+ messages in thread
From: Jan Beulich @ 2006-02-15 11:23 UTC (permalink / raw)
  To: Keir Fraser; +Cc: xen-devel

>The page types could quite easily be merged. The main problem is that 
>all descriptor tables need to be page aligned. We could get rid of that 
>restriction for LDTs I think.

Wouldn't that imply that you don't need a special PGT_ldt_page type then anymore?

Jan

^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2006-02-15 11:23 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2006-02-15  9:09 difference between PGT_gdt_page and PGT_ldt_page Jan Beulich
2006-02-15  9:26 ` Keir Fraser
2006-02-15 11:22   ` Jan Beulich
2006-02-15 11:23   ` Jan Beulich

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.