Prev: net: mac8390 - Sort out memory/MMIO accesses and casts (was: Re: drivers/net/mac8390.c: Remove useless memcpy casting)
Next: Sony vaio serie F PCIe PME service failure.
From: Avi Kivity on 23 May 2010 10:00
On 05/23/2010 03:14 PM, Xiao Guangrong wrote:
> In current code, shadow page can become asynchronous only if one
> shadow page for a gfn, this rule is too strict, in fact, we can
> let all last mapping page(i.e, it's the pte page) become unsync,
> and sync them at invlpg or flush tlb time.
>
> This patch allow more page become asynchronous at gfn mapping time
>
>
> +
> +static void kvm_unsync_pages(struct kvm_vcpu *vcpu, gfn_t gfn)
> {
> - unsigned index;
> struct hlist_head *bucket;
> struct kvm_mmu_page *s;
> struct hlist_node *node, *n;
> + unsigned index;
>
> - index = kvm_page_table_hashfn(sp->gfn);
> + index = kvm_page_table_hashfn(gfn);
> bucket =&vcpu->kvm->arch.mmu_page_hash[index];
> - /* don't unsync if pagetable is shadowed with multiple roles */
> +
> hlist_for_each_entry_safe(s, node, n, bucket, hash_link) {
> - if (s->gfn != sp->gfn || s->role.direct)
> + if (s->gfn != gfn || s->role.direct || s->unsync)
> continue;
>

role.invalid?

> - if (s->role.word != sp->role.word)
> - return 1;
> + WARN_ON(s->role.level != PT_PAGE_TABLE_LEVEL);
> + __kvm_unsync_page(vcpu, s);
> }
> - trace_kvm_mmu_unsync_page(sp);
> - ++vcpu->kvm->stat.mmu_unsync;
> - sp->unsync = 1;
> -
> - kvm_mmu_mark_parents_unsync(sp);
> -
> - mmu_convert_notrap(sp);
> - return 0;
> }
>
> static int mmu_need_write_protect(struct kvm_vcpu *vcpu, gfn_t gfn,
> bool can_unsync)
> {
> - struct kvm_mmu_page *shadow;
> + unsigned index;
> + struct hlist_head *bucket;
> + struct kvm_mmu_page *s;
> + struct hlist_node *node, *n;
> + bool need_unsync = false;
> +
> + index = kvm_page_table_hashfn(gfn);
> + bucket =&vcpu->kvm->arch.mmu_page_hash[index];
> + hlist_for_each_entry_safe(s, node, n, bucket, hash_link) {
> + if (s->gfn != gfn || s->role.direct)
> + continue;
>
> - shadow = kvm_mmu_lookup_page(vcpu->kvm, gfn);
> - if (shadow) {
> - if (shadow->role.level != PT_PAGE_TABLE_LEVEL)
> + if (s->role.level != PT_PAGE_TABLE_LEVEL)
> return 1;
>

role.invalid?

> - if (shadow->unsync)
> - return 0;
> - if (can_unsync&& oos_shadow)
> - return kvm_unsync_page(vcpu, shadow);
> - return 1;
> +
> + if (!need_unsync&& !s->unsync) {
> + if (!can_unsync || !oos_shadow)
> + return 1;
> + need_unsync = true;
> + }
> }
> + if (need_unsync)
> + kvm_unsync_pages(vcpu, gfn);
> return 0;
> }
>
>

Looks good, I'm just uncertain about role.invalid handling. What's the
reasoning here?

--
error compiling committee.c: too many arguments to function

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
From: Xiao Guangrong on 23 May 2010 22:10


Avi Kivity wrote:

>> + if (need_unsync)
>> + kvm_unsync_pages(vcpu, gfn);
>> return 0;
>> }
>>
>>
>
> Looks good, I'm just uncertain about role.invalid handling. What's the
> reasoning here?
>

Avi,

Thanks for your reply.

We no need worry about 'role.invalid' here, since we only allow the PTE shadow
pages(role.level == 1) become unsync, and in current code, 'role.invalid' is only
used for root shadow pages.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
From: Avi Kivity on 24 May 2010 02:30
On 05/24/2010 05:03 AM, Xiao Guangrong wrote:
>
> Avi Kivity wrote:
>
>
>>> + if (need_unsync)
>>> + kvm_unsync_pages(vcpu, gfn);
>>> return 0;
>>> }
>>>
>>>
>>>
>> Looks good, I'm just uncertain about role.invalid handling. What's the
>> reasoning here?
>>
>>
> Avi,
>
> Thanks for your reply.
>
> We no need worry about 'role.invalid' here, since we only allow the PTE shadow
> pages(role.level == 1) become unsync, and in current code, 'role.invalid' is only
> used for root shadow pages.
>

Right, the invlpg change is not it yet. But I think it should be in
this patch; I don't like subtle dependencies, and it will make the
invplg patch simpler.

--
Do not meddle in the internals of kernels, for they are subtle and quick to panic.

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
From: Xiao Guangrong on 24 May 2010 03:20


Avi Kivity wrote:
> On 05/24/2010 05:03 AM, Xiao Guangrong wrote:
>>
>> Avi Kivity wrote:
>>
>>
>>>> + if (need_unsync)
>>>> + kvm_unsync_pages(vcpu, gfn);
>>>> return 0;
>>>> }
>>>>
>>>>
>>>>
>>> Looks good, I'm just uncertain about role.invalid handling. What's the
>>> reasoning here?
>>>
>>>
>> Avi,
>>
>> Thanks for your reply.
>>
>> We no need worry about 'role.invalid' here, since we only allow the
>> PTE shadow
>> pages(role.level == 1) become unsync, and in current code,
>> 'role.invalid' is only
>> used for root shadow pages.
>>
>
> Right, the invlpg change is not it yet. But I think it should be in
> this patch; I don't like subtle dependencies, and it will make the
> invplg patch simpler.
>

OK, i'll fix those two patches, thanks
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
 | 
Pages: 1
Prev: net: mac8390 - Sort out memory/MMIO accesses and casts (was: Re: drivers/net/mac8390.c: Remove useless memcpy casting)
Next: Sony vaio serie F PCIe PME service failure.