Re: [PATCH] drm/panthor: use defines for sync flags

2024-10-29 Thread Boris Brezillon
On Tue, 29 Oct 2024 10:46:29 +0100 Erik Faye-Lund wrote: > Enums are always signed, and assigning 1u << 31 to it invokes > implementation defined behavior. It's not a great idea to depend on this > in the UAPI, and it turns out no other UAPI does either. > > So let's do what other UAPI does, and

Re: [PATCH] drm/panthor: use defines for sync flags

2024-10-29 Thread Mihail Atanassov
Hi Erik, On 29/10/2024 09:46, Erik Faye-Lund wrote: Enums are always signed, and assigning 1u << 31 to it invokes implementation defined behavior. It's not a great idea to depend on this in the UAPI, and it turns out no other UAPI does either. So let's do what other UAPI does, and use defines i

Re: [PATCH] drm/panthor: use defines for sync flags

2024-10-29 Thread Erik Faye-Lund
On Tue, 2024-10-29 at 10:15 +, Liviu Dudau wrote: > On Tue, Oct 29, 2024 at 10:46:29AM +0100, Erik Faye-Lund wrote: > > Enums are always signed, and assigning 1u << 31 to it invokes > > implementation defined behavior. It's not a great idea to depend on > > this > > in the UAPI, and it turns ou

Re: [PATCH] drm/panthor: use defines for sync flags

2024-10-29 Thread Liviu Dudau
On Tue, Oct 29, 2024 at 10:46:29AM +0100, Erik Faye-Lund wrote: > Enums are always signed, and assigning 1u << 31 to it invokes > implementation defined behavior. It's not a great idea to depend on this > in the UAPI, and it turns out no other UAPI does either. > > So let's do what other UAPI does