On 09.02.2024 09:25, Roger Pau Monné wrote:
> On Mon, Feb 05, 2024 at 02:57:12PM +0100, Jan Beulich wrote:
>> ..., thus allowing it and qinval_device_iotlb_sync() to become static.
>> There's nothing x86-specific about the function anyway. While moving,
>> adjust types to better match ./CODING_STYLE (albeit use of fixed-width
>> types for parameters is retained to limit the effective change).
>>
>> Signed-off-by: Jan Beulich <[email protected]>
>
> Acked-by: Roger Pau Monné <[email protected]>
Thanks.
>> +static int dev_invalidate_iotlb(struct vtd_iommu *iommu, uint16_t did,
>> + paddr_t addr, unsigned int size_order,
>> + uint64_t type)
>> +{
>> + struct pci_dev *pdev, *temp;
>> + int ret = 0;
>> +
>> + if ( !ecap_dev_iotlb(iommu->ecap) )
>> + return ret;
>> +
>> + list_for_each_entry_safe( pdev, temp, &iommu->ats_devices, ats.list )
>> + {
>> + bool sbit;
>> + int rc = 0;
>> +
>> + switch ( type )
>> + {
>> + case DMA_TLB_DSI_FLUSH:
>> + if ( !device_in_domain(iommu, pdev, did) )
>> + break;
>> + /* fall through if DSI condition met */
>> + case DMA_TLB_GLOBAL_FLUSH:
>> + /* invalidate all translations: sbit=1,bit_63=0,bit[62:12]=1 */
>> + sbit = 1;
>> + addr = (~0UL << PAGE_SHIFT_4K) & 0x7FFFFFFFFFFFFFFF;
>
> Given the MISRA stuff, won't it be better to append 'UL' here while
> moving?
Sure, done.
Jan
