On Wed, 20 Feb 2013 15:36:58 -0800, Christoffer Dall
<cdall@xxxxxxxxxxxxxxx> wrote:
> On Wed, Feb 13, 2013 at 03:46:58PM +0000, Marc Zyngier wrote:
>> stage2_set_pte() does too many things:
>> - Walks the page table
>> - Allocating each level as it goes
>> - Tests for I/O mappings
>> - Set the new PTE and takes care of the TLB invalidation
>>
>> Introduce stage2_get_pte() that only takes care of the first two
>> points. stage2_set_pte() is modified to use stage2_get_pte().
>>
>> Signed-off-by: Marc Zyngier <marc.zyngier@xxxxxxx>
>> ---
>> arch/arm/kvm/mmu.c | 23 +++++++++++++++++------
>> 1 file changed, 17 insertions(+), 6 deletions(-)
>>
>> diff --git a/arch/arm/kvm/mmu.c b/arch/arm/kvm/mmu.c
>> index 2943163..cb2b42e 100644
>> --- a/arch/arm/kvm/mmu.c
>> +++ b/arch/arm/kvm/mmu.c
>> @@ -409,20 +409,19 @@ void kvm_free_stage2_pgd(struct kvm *kvm)
>> }
>>
>>
>> -static int stage2_set_pte(struct kvm *kvm, struct kvm_mmu_memory_cache
>> *cache,
>> - phys_addr_t addr, const pte_t *new_pte, bool iomap)
>> +static pte_t *stage2_get_pte(struct kvm *kvm, struct
>> kvm_mmu_memory_cache *cache,
>> + phys_addr_t addr)
>> {
>> pgd_t *pgd;
>> pud_t *pud;
>> pmd_t *pmd;
>> - pte_t *pte, old_pte;
>>
>> /* Create 2nd stage page table mapping - Level 1 */
>> pgd = kvm->arch.pgd + pgd_index(addr);
>> pud = pud_offset(pgd, addr);
>> if (pud_none(*pud)) {
>> if (!cache)
>> - return 0; /* ignore calls from kvm_set_spte_hva */
>> + return NULL; /* ignore calls from kvm_set_spte_hva */
>> pmd = mmu_memory_cache_alloc(cache);
>> pud_populate(NULL, pud, pmd);
>> get_page(virt_to_page(pud));
>> @@ -432,15 +431,27 @@ static int stage2_set_pte(struct kvm *kvm, struct
>> kvm_mmu_memory_cache *cache,
>
> This should call stage2_get_pmd as we discussed.
Agreed. I'll introduce that when I get to rework this patch series after
the merge window.
M.
--
Fast, cheap, reliable. Pick two.
_______________________________________________
kvmarm mailing list
kvmarm@xxxxxxxxxxxxxxxxxxxxx
https://lists.cs.columbia.edu/cucslists/listinfo/kvmarm
