Re: [PATCH] PPC: Sync guest visible MMU state

From: Alexander Graf <agraf@suse.de>
To: Avi Kivity <avi@redhat.com>
Cc: kvm@vger.kernel.org, kvm-ppc <kvm-ppc@vger.kernel.org>,
	Benjamin Herrenschmidt <benh@kernel.crashing.org>,
	linuxppc-dev@ozlabs.org
Subject: Re: [PATCH] PPC: Sync guest visible MMU state
Date: Thu, 26 Nov 2009 13:46:52 +0100	[thread overview]
Message-ID: <4B0E78BC.5050901@suse.de> (raw)
In-Reply-To: <4B0E7738.5030506@redhat.com>

Avi Kivity wrote:
> On 11/26/2009 01:16 PM, Alexander Graf wrote:
>> Currently userspace has no chance to find out which virtual address
>> space we're
>> in and resolve addresses. While that is a big problem for migration,
>> it's also
>> unpleasent when debugging, as gdb and the monitor don't work on virtual
>> addresses.
>>
>> This patch exports enough of the MMU segment state to userspace to make
>> debugging work and thus also includes the groundwork for migration.
>>
>>
>>   struct kvm_sregs {
>> -    __u32 pvr;
>> -    char pad[1020];
>> +    union {
>> +        struct {
>> +            __u32 pvr;
>> +            __u64 sdr1;
>> +            struct {
>> +                struct {
>> +                    __u64 slbe;
>> +                    __u64 slbv;
>> +                } slb[64];
>> +            } ppc64;
>> +            struct {
>> +                __u32 sr[16];
>> +                __u64 ibat[8];
>> +                __u64 dbat[8];
>> +            } ppc32;
>> +        };
>> +        __u8 pad[1024];
>> +    };
>>   };
>>    
>
> Please avoid unnamed unions in user-visible headers - they're a gcc
> extension.
>
> Yes, we have them elsewhere, but let's not add to the pile.

I'm open to scalable suggestions that don't break existing userspace code.
If I name the union now, existing qemu code will break. If I align the
pad array manually I'll definitely mess up something.

Alex