[PATCH v5 3/4] KVM: PPC: epapr: install ev_idle hcall for e500 guest

[tiejun.chen]

Liu Yu-B13201 wrote:
> 
>> -----Original Message-----
>> From: tiejun.chen [mailto:tiejun.chen at windriver.com]
>> Sent: Tuesday, February 21, 2012 6:54 PM
>> To: Liu Yu-B13201
>> Cc: agraf at suse.de; kvm-ppc at vger.kernel.org; kvm at vger.kernel.org;
>> linuxppc-dev at ozlabs.org; Wood Scott-B07421
>> Subject: Re: [PATCH v5 3/4] KVM: PPC: epapr: install ev_idle hcall for
>> e500 guest
>>
>> Liu Yu wrote:
>>> If the guest hypervisor node contains "has-idle" property.
>>>
>>> Signed-off-by: Liu Yu <yu.liu at freescale.com>
>>> ---
>>> v5: no change
>>>
>>>  arch/powerpc/kernel/epapr_hcalls.S   |   29
>> +++++++++++++++++++++++++++++
>>>  arch/powerpc/kernel/epapr_paravirt.c |   11 ++++++++++-
>>>  2 files changed, 39 insertions(+), 1 deletions(-)
>>>
>>> diff --git a/arch/powerpc/kernel/epapr_hcalls.S
>>> b/arch/powerpc/kernel/epapr_hcalls.S
>>> index 697b390..72fa234 100644
>>> --- a/arch/powerpc/kernel/epapr_hcalls.S
>>> +++ b/arch/powerpc/kernel/epapr_hcalls.S
>>> @@ -15,6 +15,35 @@
>>>  #include <asm/ppc_asm.h>
>>>  #include <asm/asm-offsets.h>
>>>
>>> +#define HC_VENDOR_EPAPR		(1 << 16)
>>> +#define HC_EV_IDLE		16
>> Why not use 'EV_IDLE' directly?
>>
>>> +
>>> +_GLOBAL(epapr_ev_idle)
>>> +epapr_ev_idle:
>>> +	rlwinm	r3,r1,0,0,31-THREAD_SHIFT	/* current thread_info */
>>> +	lwz	r4,TI_LOCAL_FLAGS(r3)	/* set napping bit */
>>> +	ori	r4,r4,_TLF_NAPPING	/* so when we take an exception */
>>> +	stw	r4,TI_LOCAL_FLAGS(r3)	/* it will return to our caller */
>>> +
>>> +	wrteei	1
>>> +
>>> +idle_loop:
>>> +	LOAD_REG_IMMEDIATE(r11, HC_VENDOR_EPAPR | HC_EV_IDLE)
>> And could this line be simplified as something like this:
>>
>> LOAD_REG_IMMEDIATE(r11, EV_HCALL_TOKEN(EV_IDLE))
>>
>> If so, even we can remove the previous HC_VENDOR_EPAPR definition as well.
>>
> 
> Because the epapr_hcalls.h contains C functions,
> so it cannot be included by assembly code.

These common definitions are already covered in epapr_hcalls.h, but looks you
redefine the same items many times, in kvm_para.h/epapr_hcalls.S. And I think
maybe we'll also reuse these generics elsewhere lately.

So can we limit that with __ASSEMBLY__ in epapr_hcalls.h? Or other way. If so it
makes our life easy in the future.

Tiejun