Re: [PATCH v2 8/9] hw/dma: Use dma_addr_t type definition when relevant

[Philippe Mathieu-Daudé]

On 1/10/22 09:49, David Hildenbrand wrote:
> On 04.01.22 09:54, Philippe Mathieu-Daudé wrote:
>> From: Philippe Mathieu-Daudé <philmd@redhat.com>
>>
>> Update the obvious places where dma_addr_t should be used
>> (instead of uint64_t, hwaddr, size_t, int32_t types).
>>
>> This allows to have &dma_addr_t type portable on 32/64-bit
>> hosts.
>>
>> Move QEMUSGList declaration after dma_addr_t declaration
>> so this structure can use the new type.
>>
>> Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
>> Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
>> ---
>>  include/sysemu/dma.h  | 22 +++++++++++-----------
>>  hw/nvme/ctrl.c        |  2 +-
>>  hw/rdma/rdma_utils.c  |  2 +-
>>  hw/scsi/megasas.c     | 10 +++++-----
>>  softmmu/dma-helpers.c |  6 +++---
>>  5 files changed, 21 insertions(+), 21 deletions(-)
>>
>> diff --git a/include/sysemu/dma.h b/include/sysemu/dma.h
>> index 0db2478a506..7a8ae4fcd0b 100644
>> --- a/include/sysemu/dma.h
>> +++ b/include/sysemu/dma.h
>> @@ -15,22 +15,11 @@
>>  #include "block/block.h"
>>  #include "block/accounting.h"
>>  
>> -typedef struct ScatterGatherEntry ScatterGatherEntry;
>> -
>>  typedef enum {
>>      DMA_DIRECTION_TO_DEVICE = 0,
>>      DMA_DIRECTION_FROM_DEVICE = 1,
>>  } DMADirection;
>>  
>> -struct QEMUSGList {
>> -    ScatterGatherEntry *sg;
>> -    int nsg;
>> -    int nalloc;
>> -    size_t size;
>> -    DeviceState *dev;
>> -    AddressSpace *as;
>> -};
>> -
>>  /*
>>   * When an IOMMU is present, bus addresses become distinct from
>>   * CPU/memory physical addresses and may be a different size.  Because
>> @@ -43,6 +32,17 @@ typedef uint64_t dma_addr_t;
>>  #define DMA_ADDR_BITS 64
>>  #define DMA_ADDR_FMT "%" PRIx64
>>  
>> +typedef struct ScatterGatherEntry ScatterGatherEntry;
>> +
>> +struct QEMUSGList {
>> +    ScatterGatherEntry *sg;
>> +    int nsg;
>> +    int nalloc;
>> +    dma_addr_t size;
>> +    DeviceState *dev;
>> +    AddressSpace *as;
>> +};
> 
> Changing one member while moving is sneaky. Why the move in this patch?

Because dma_addr_t is declared before QEMUSGList. I will add an
intermediate patch.

> Apart from that and Peters comment
> 
> Reviewed-by: David Hildenbrand <david@redhat.com>

Thanks (Peter's comment addressed).