Xen devel - May 2006 - [PATCH] clean up xen_mkbuildtree_pre (was Re: [Xen-ia64-devel] xen-devel VP status and xen-mkbuildtree-pre)

Many months ago when we merged the Xenlinux/ia64 changes into
the main Xen -sparse tree, there were a number of changes
to Linux arch-neutral -sparse files necessary for Xenlinux/x86
that broke Xenlinux/ia64.  To work around this, we added
a script (xen-mkbuildtree-pre) that, when building Xenlinux/ia64,
moved some files around to avoid these problems.  It was ugly,
but effective.  However, it messes with the build directory
in a way that the distros are not very happy with.

Since then, there''s been a lot of cleanup in Xenlinux(/x86)
and it appears there is only one remaining problem in Xenlinux
common code -- a one line change to sparse/mm/Kconfig; see
below.

Once this is in, xen-mkbuildtree-pre can be greatly trimmed
(though not yet completely eliminated).

Could you please apply the one-line change below to mm/Kconfig?
Yamahata-san is away for a few days so I can''t ask him
for a signed-off-by line... since it is trivial, you can
sign it yourself, or use mine:

Signed-off-by: Dan Magenheimer <dan.magenheimer@hp.com>

The attachment includes the cleanup so far in xen-mkbuildtree-pre,
which requires one other source code workaround to be removed
to successfully compile.  If you could also apply the attached patch,
it would re-enable tip to compile on ia64.

Apologies that I can''t provide a clean single patch.  I''m
working remotely and am having some file copy problems.

Thanks,
Dan
> -----Original Message-----
> From: Isaku Yamahata [mailto:yamahata@valinux.co.jp] 
> Sent: Tuesday, May 02, 2006 6:10 AM
> To: Magenheimer, Dan (HP Labs Fort Collins)
> Cc: xen-ia64-devel@lists.xensource.com
> Subject: [PATCH] clean up xen_mkbuildtree_pre (was Re: 
> [Xen-ia64-devel] xen-devel VP status and xen-mkbuildtree-pre)
> 
> On Mon, May 01, 2006 at 02:46:04PM -0700, Magenheimer, Dan 
> (HP Labs Fort Collins) wrote:
> 
> > Also, with all your VP changes, could you look at the
> > file -sparse/arch/ia64/xen-mkbuildtree-pre and see
> > how many files/dirs still need to be mv''d?  Perhaps we
> > are close enough now that we can insert some ifdef''s
> > directly into any remaining conflicting files rather
> > than rm/mv them?
> 
> I found that only linux-2.6-xen-sparse/mm/Kconfig is needed to be mv.
> for mm/Kconfig.
> The patch is attached.
> 
> Maybe Kconfig should be modified as follows.
> 
>  # Default to 4 for wider testing, though 8 might be more appropriate.
>  # ARM''s adjust_pte (unused if VIPT) depends on mm-wide 
> page_table_lock.
>  # PA-RISC 7xxx''s spinlock_t would enlarge struct page from 
> 32 to 44 bytes.
>  # XEN uses the mapping field on pagetable pages to store a pointer to
>  # the destructor.
>  #
>  config SPLIT_PTLOCK_CPUS
>         int
>         default "4096" if ARM && !CPU_CACHE_VIPT
>         default "4096" if PARISC && !PA20
> -       default "4096" if XEN
> +       default "4096" if X86_XEN || X86_64_XEN
>         default "4"
>  
> 
> 
> -- 
> yamahata
> 

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

On 2 May 2006, at 18:46, Magenheimer, Dan (HP Labs Fort Collins) wrote:
> Could you please apply the one-line change below to mm/Kconfig?
> Yamahata-san is away for a few days so I can''t ask him
> for a signed-off-by line... since it is trivial, you can
> sign it yourself, or use mine:
Are you sure that''s safe? I''m not sure why Christian disabled
per-pte
locks for XenLinux.

  -- Keir


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

On 3 May 2006, at 07:44, Keir Fraser wrote:
>> Could you please apply the one-line change below to mm/Kconfig?
>> Yamahata-san is away for a few days so I can''t ask him
>> for a signed-off-by line... since it is trivial, you can
>> sign it yourself, or use mine:
>
> Are you sure that''s safe? I''m not sure why Christian
disabled per-pte
> locks for XenLinux.
Ah ok, I see why and the reason is indeed x86-specific. I''ll apply your
patches.

  -- Keir


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