Re: [PATCH] vm_event: Implement ARM SMC events

From: Razvan Cojocaru <rcojocaru@bitdefender.com>
To: Jan Beulich <jbeulich@suse.com>, tamas@tklengyel.com
Cc: wei.liu2@citrix.com, stefano.stabellini@eu.citrix.com,
	andrew.cooper3@citrix.com, ian.jackson@eu.citrix.com,
	julien.grall@arm.com, xen-devel@lists.xenproject.org,
	keir@xen.org, tklengyel@sec.in.tum.de
Subject: Re: [PATCH] vm_event: Implement ARM SMC events
Date: Tue, 12 Apr 2016 08:35:26 +0300	[thread overview]
Message-ID: <570C891E.5090904@bitdefender.com> (raw)
In-Reply-To: <570C881A02000078000E6497@prv-mh.provo.novell.com>

On 04/12/16 07:31, Jan Beulich wrote:
>>>> Tamas K Lengyel <tamas@tklengyel.com> 04/11/16 9:47 PM >>>
>> --- a/xen/include/public/vm_event.h
>> +++ b/xen/include/public/vm_event.h
>> @@ -166,6 +166,31 @@ struct vm_event_regs_x86 {
>      >uint32_t _pad;
>  >};
>  >
>> +struct vm_event_regs_arm {
>> +    uint32_t r0;
>> +    uint32_t r1;
>> +    uint32_t r2;
>> +    uint32_t r3;
>> +    uint32_t r4;
>> +    uint32_t r5;
>> +    uint32_t r6;
>> +    uint32_t r7;
>> +    uint32_t r8;
>> +    uint32_t r9;
>> +    uint32_t r10;
>> +    uint32_t r11;
>> +    uint32_t r12;
>> +
>> +    uint32_t sp; /* r13 - SP: Valid for Hyp. frames only, o/w banked (see below) */
>> +    uint32_t lr; /* r14 - LR: Valid for Hyp. Same physical register as lr_usr. */
>> +
>> +    uint32_t cpsr; /* Return mode */
>> +    uint64_t pc;
> 
> Why uint64_t instead of uint32_t?
> 
>> +    uint64_t ttbr0;
>> +    uint64_t ttbr1;
>> +    uint32_t _pad;
>> +};
> 
> This padding field is pretty strange: Without the adjustment to pc there are 16
> 32-bit fields (not counting the padding one), so I don't see why padding would be
> needed at all. And with pc adjusted the padding field would surely better go
> ahead of the two remaining 64-bit ones.
> 
>> @@ -254,6 +279,7 @@ typedef struct vm_event_st {
>      >union {
>          >union {
>              >struct vm_event_regs_x86 x86;
>> +            struct vm_event_regs_arm arm;
>          >} regs;
>  
>          Does this alter the x86 ABI? Perhaps the ARM structure is small enough for
> this to not happen now, but what's the general idea about not breaking other
> arch'es ABIs when adding support for a new arch here?

I'd suggest modifying VM_EVENT_INTERFACE_VERSION whenever that becomes
the case.

Thanks,
Razvan

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