Dave Airlie wrote:
> On Sat, Apr 3, 2010 at 11:09 PM, Matthew W. S. Bell
> wrote:
>> On Sat, 2010-04-03 at 08:49 +0100, Dave Airlie wrote:
>>> No, its "designed" as is. We can't change it now as its ABI. We make sure
>>> we only use 32-bit handles anyways.
>
> The thing is "unsigned int" is cor
On Sat, Apr 3, 2010 at 11:09 PM, Matthew W. S. Bell
wrote:
> On Sat, 2010-04-03 at 08:49 +0100, Dave Airlie wrote:
>> No, its "designed" as is. We can't change it now as its ABI. We make sure
>> we only use 32-bit handles anyways.
The thing is "unsigned int" is correct for the handles, they are d
On Sat, 2010-04-03 at 08:49 +0100, Dave Airlie wrote:
> No, its "designed" as is. We can't change it now as its ABI. We make sure
> we only use 32-bit handles anyways.
OK, is this documented anywhere, as I'd like to pull some of that into a
comment? (It appears, on a casual glance, that this type
On Sat, 2010-04-03 at 08:49 +0100, Dave Airlie wrote:
> No, its "designed" as is. We can't change it now as its ABI. We make sure
> we only use 32-bit handles anyways.
Sorry, the comment about the ABI is, of course, nonsense, as the
assumption is implicit in the kernel ABI.
Matthew
signature.
>
> drm_handle_t appears to be assigned values from void*. As such unsigned
> int is certainly not the same size on 64-bit. Convert to uintptr_t in
> all cases as it is defined for this purpose.
No, its "designed" as is. We can't change it now as its ABI. We make sure
we only use 32-bit handles
From d457b05faabde1a51b8e4b8f6fc13af9f07809f8 Mon Sep 17 00:00:00 2001
drm_handle_t appears to be assigned values from void*. As such unsigned
int is certainly not the same size on 64-bit. Convert to uintptr_t in
all cases as it is defined for this purpose.
I fear this patch changes ABI, but I am
