[RFC, PATCH 4/5] Paravirt_ops drop internal patches.patch

[RFC, PATCH 4/5] Paravirt_ops drop internal patches.patch

[Zachary Amsden]

Some of these boot hooks do not need to be patched, in fact, it is better not
to patch them at all and thus keep them totally private.

Signed-off-by: Zachary Amsden <zach@vmware.com>

diff -r dba7b8b8e43d include/asm-i386/paravirt.h
--- a/include/asm-i386/paravirt.h	Thu Apr 19 16:12:02 2007 -0700
+++ b/include/asm-i386/paravirt.h	Thu Apr 19 16:16:19 2007 -0700
@@ -686,12 +686,12 @@ static inline unsigned long apic_read(un
 
 static inline void setup_boot_clock(void)
 {
-	PVOP_VCALL0(setup_boot_clock);
+	paravirt_ops.setup_boot_clock();
 }
 
 static inline void setup_secondary_clock(void)
 {
-	PVOP_VCALL0(setup_secondary_clock);
+	paravirt_ops.setup_secondary_clock();
 }
 #endif
 
@@ -711,7 +711,7 @@ static inline void startup_ipi_hook(int 
 static inline void startup_ipi_hook(int phys_apicid, unsigned long start_eip,
 				    unsigned long start_esp)
 {
-	PVOP_VCALL3(startup_ipi_hook, phys_apicid, start_eip, start_esp);
+	paravirt_ops.startup_ipi_hook(phys_apicid, start_eip, start_esp);
 }
 #endif

Re: [RFC, PATCH 4/5] Paravirt_ops drop internal patches.patch

[Jeremy Fitzhardinge]

Zachary Amsden wrote:
> Some of these boot hooks do not need to be patched, in fact, it is better not
> to patch them at all and thus keep them totally private.
>   

I think I generally backed off on patching init functions and the like. 
Are these really the only ones which need adjustment?  I bet things like
the descriptor functions could be indirect/private with very little impact.

    J

Re: [RFC, PATCH 4/5] Paravirt_ops drop internal patches.patch

[Zachary Amsden]

Jeremy Fitzhardinge wrote:
> Zachary Amsden wrote:
>   
>> Some of these boot hooks do not need to be patched, in fact, it is better not
>> to patch them at all and thus keep them totally private.
>>   
>>     
>
> I think I generally backed off on patching init functions and the like. 
> Are these really the only ones which need adjustment?  I bet things like
> the descriptor functions could be indirect/private with very little impact.
>   

Perhaps, but I'd rather not split hairs on it just yet - I actually have 
no idea exactly which paravirt-ops will be needed, and so far, my 
conclusions have led me to believe that a fare bit more than the obvious 
paravirt-ops need to be available to modules.  So I would rather just 
split off pure init functions, which really have no business being 
exported to modules.

Zach