Re: [PATCH 08/18] amd_iommu: Helper to decode size of page invalidation command

[Alejandro Jimenez <alejandro.j.jimenez@oracle.com>]

On 4/22/25 8:26 AM, Sairaj Kodilkar wrote:
> 
> 
> On 4/14/2025 7:32 AM, Alejandro Jimenez wrote:
>> The size of the region to invalidate depends on the S bit and address
>> encoded in the command. Add a helper to extract this information, which
>> will be used to sync shadow page tables in upcoming changes.
>>
>> Signed-off-by: Alejandro Jimenez <alejandro.j.jimenez@oracle.com>
>> ---
>>   hw/i386/amd_iommu.c | 34 ++++++++++++++++++++++++++++++++++
>>   hw/i386/amd_iommu.h |  4 ++++
>>   2 files changed, 38 insertions(+)
>>
>> diff --git a/hw/i386/amd_iommu.c b/hw/i386/amd_iommu.c
>> index 5f55be1f4d36..0af873b66a31 100644
>> --- a/hw/i386/amd_iommu.c
>> +++ b/hw/i386/amd_iommu.c


>> +static uint64_t __attribute__((unused))
>> +amdvi_decode_invalidation_size(hwaddr addr, uint16_t flags)
>> +{
>> +    uint64_t size = AMDVI_PAGE_SIZE;
>> +    uint8_t fzbit = 0;
>> +
>> +    if (flags & AMDVI_CMD_INVAL_IOMMU_PAGES_S) {
>> +        fzbit = cto64(addr | 0xFFF);
>> +
>> +        if (fzbit >= 51 || !addr) {
> 
> I am skeptical about the condition addr == 0 (!addr)
> 
> Consider the case where user wants to invalidate 8K size page, starting
> from address 0. It'll cause address field to be 0, right ? If so, then
> we should invalidate only 8K page not the entire address range.

You are right, I'll fix the issue (i.e. remove the !addr special case) 
in v2.
This was a remnant of an earlier prototype where I was decoding the 
address with a different method, but as you point out this special case 
is not needed anymore and would result in sub-optimal behavior.

Thank you!
Alejandro

> 
> Am I missing something here ?
> 
> Regards
> Sairaj