<div dir="ltr">I'm not completely sure what you're saying, but doesn't drm_base_types.h (now drm_basic_types.h) make things robust to header order? The patch is in another email. You can also see it here: <a href="https://github.com/jpark37/linux/commit/0cc8ae750bfd9eab7e31c7de6aa84f24682f4f18">https://github.com/jpark37/linux/commit/0cc8ae750bfd9eab7e31c7de6aa84f24682f4f18</a><div> </div><div>Thanks,</div><div>James</div></div> <div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Mon, Dec 7, 2020 at 12:51 AM Pekka Paalanen <<a href="mailto:ppaalanen@gmail.com" target="_blank">ppaalanen@gmail.com</a>> wrote: </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">On Fri, 4 Dec 2020 11:07:41 -0800 James Park <<a href="mailto:james.park@lagfreegames.com" target="_blank">james.park@lagfreegames.com</a>> wrote: > I could adjust the block to look like this: > > #ifdef DRM_FOURCC_STANDALONE > #if defined(__linux__) > #include <linux/types.h> > #else > #include <stdint.h> > typedef uint32_t __u32; > typedef uint64_t __u64; > #endif > #else > #include "drm.h" > #endif > > Alternatively, I could create a new common header to be included from both > drm.h and drm_fourcc.h, drm_base_types.h or something like that: > > #ifdef DRM_FOURCC_STANDALONE > #include "drm_base_types.h" > #else > #include "drm.h" > #endif Hi, my point is, any solution relying on DRM_FOURCC_STANDALONE will fail sometimes, because there is no reason why userspace would *not* #define DRM_FOURCC_STANDALONE. Hence, #ifdef DRM_FOURCC_STANDALONE is completely moot, you have to make the headers work in any include order when DRM_FOURCC_STANDALONE is defined anyway. Thanks. pq > On Fri, Dec 4, 2020 at 7:58 AM Daniel Vetter <<a href="mailto:daniel@ffwll.ch" target="_blank">daniel@ffwll.ch</a>> wrote: > > > On Fri, Dec 4, 2020 at 9:12 AM Pekka Paalanen <<a href="mailto:ppaalanen@gmail.com" target="_blank">ppaalanen@gmail.com</a>> wrote: > > > > > > On Thu, 3 Dec 2020 21:45:14 +0100 > > > Daniel Vetter <<a href="mailto:daniel@ffwll.ch" target="_blank">daniel@ffwll.ch</a>> wrote: > > > > > > > On Thu, Dec 3, 2020 at 7:55 PM James Park <<a href="mailto:james.park@lagfreegames.com" target="_blank">james.park@lagfreegames.com</a>> > > wrote: > > > > > > > > > > The trailing underscore for DRM_FOURCC_STANDALONE_ isn't > > > > > intentional, right? Should I put all the integer types, or just the > > > > > ones that are used in that file? > > > > > > > > Yeah that trailing _ just slipped in. And I'd just do the types > > > > already used. I don't think anything else than __u32 (for drm fourcc) > > > > and __u64 (for drm modifier) is needed. > > > > > > Hi, > > > > > > can that create conflicts if userspace first includes drm_fourcc.h and > > > then drm.h? > > > > > > I would find it natural to userspace have generic headers including > > > drm_fourcc.h and then DRM-specific C-files including drm.h as well > > > (through libdrm headers). I think Weston might already do this. > > > > > > The generic userspace (weston) header would obviously #define > > > DRM_FOURCC_STANDALONE, because it is used by non-DRM C-files as well. > > > > Hm yes that would break. I guess we could just include the linux types > > header for this. And I guess on windows you'd need to have that from > > somewhere. Or we just require that users of the standalone header pull > > the right header or defines in first? > > -Daniel > > -- > > Daniel Vetter > > Software Engineer, Intel Corporation > > <a href="http://blog.ffwll.ch" rel="noreferrer" target="_blank">http://blog.ffwll.ch</a> > > </blockquote></div>