<div dir="ltr"><div class="gmail_extra"><div class="gmail_quote">On Tue, Mar 21, 2017 at 1:45 PM, Ben Widawsky <<a href="mailto:ben@bwidawsk.net" target="_blank">ben@bwidawsk.net</a>> wrote: <blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">This patch begins introducing how we'll actually handle the potentially many modifiers coming in from the API, how we'll store them, and the structure in the code to support it. Prior to this patch, the Y-tiled modifier would be entirely ignored. It shouldn't actually be used until this point because we've not bumped the DRIimage extension version (which is a requirement to use modifiers). Measuring later in the series with kmscube: Linear: Read bandwidth: 1048.44 MiB/s Write bandwidth: 1483.17 MiB/s Y-tiled: Read bandwidth: 471.13 MiB/s Write bandwidth: 589.10 MiB/s Similar functionality was introduced and then reverted here: commit 6a0d036483caf87d43ebe2edd1905873446c9589 Author: Ben Widawsky <<a href="mailto:ben@bwidawsk.net">ben@bwidawsk.net</a>> Date: Thu Apr 21 20:14:58 2016 -0700 i965: Always use Y-tiled buffers on SKL+ v2: Use last set bit instead of first set bit in modifiers to address bug found by Daniel Stone. v3: Use the new priority modifier selection thing. This nullifies the bug fixed by v2 also. Signed-off-by: Ben Widawsky <<a href="mailto:benjamin.widawsky@intel.com">benjamin.widawsky@intel.com</a>> Reviewed-by: Eric Engestrom <<a href="mailto:eric.engestrom@imgtec.com">eric.engestrom@imgtec.com</a>> Acked-by: Daniel Stone <<a href="mailto:daniels@collabora.com">daniels@collabora.com</a>> Reviewed-by: Jason Ekstrand <<a href="mailto:jason@jlekstrand.net">jason@jlekstrand.net</a>> --- src/mesa/drivers/dri/i965/intel_screen.c | 36 ++++++++++++++++++++++++++++---- 1 file changed, 32 insertions(+), 4 deletions(-) diff --git a/src/mesa/drivers/dri/i965/intel_screen.c b/src/mesa/drivers/dri/i965/intel_screen.c index 8ba90f0a0c..8bb3aca946 100644 --- a/src/mesa/drivers/dri/i965/intel_screen.c +++ b/src/mesa/drivers/dri/i965/intel_screen.c @@ -23,6 +23,7 @@ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */ +#include <drm_fourcc.h> #include <errno.h> #include <time.h> #include <unistd.h> @@ -518,11 +519,13 @@ intel_destroy_image(__DRIimage *image) enum modifier_priority { MODIFIER_PRIORITY_INVALID = 0, MODIFIER_PRIORITY_LINEAR, + MODIFIER_PRIORITY_Y, }; const uint64_t priority_to_modifier[] = { [MODIFIER_PRIORITY_INVALID] = DRM_FORMAT_MOD_INVALID, [MODIFIER_PRIORITY_LINEAR] = DRM_FORMAT_MOD_LINEAR, + [MODIFIER_PRIORITY_Y] = I915_FORMAT_MOD_Y_TILED, }; static uint64_t @@ -530,11 +533,13 @@ select_best_modifier(struct gen_device_info *devinfo, const uint64_t *modifiers, const unsigned count) { - enum modifier_priority prio = MODIFIER_PRIORITY_INVALID; for (int i = 0; i < count; i++) { switch (modifiers[i]) { + case I915_FORMAT_MOD_Y_TILED: + prio = MAX2(prio, MODIFIER_PRIORITY_Y); + break; case DRM_FORMAT_MOD_LINEAR: prio = MAX2(prio, MODIFIER_PRIORITY_LINEAR); break; @@ -575,6 +580,9 @@ intel_create_image_common(__DRIscreen *dri_screen, case DRM_FORMAT_MOD_LINEAR: tiling = I915_TILING_NONE; break; + case I915_FORMAT_MOD_Y_TILED: + tiling = I915_TILING_Y; + break; case DRM_FORMAT_MOD_INVALID: if (modifiers) return NULL; @@ -628,8 +636,26 @@ intel_create_image_with_modifiers(__DRIscreen *dri_screen, const unsigned count, void *loaderPrivate) { - return intel_create_image_common(dri_screen, width, height, format, 0, NULL, - 0, loaderPrivate); + uint64_t local_mods[count]; + int local_count = 0; + + /* This compacts the actual modifiers to the ones we know how to handle */ + for (int i = 0; i < count; i++) { + switch (modifiers[i]) { + case I915_FORMAT_MOD_Y_TILED: + case DRM_FORMAT_MOD_LINEAR: + local_mods[local_count++] = modifiers[i]; + break; + case DRM_FORMAT_MOD_INVALID: + unreachable("Invalid modifiers specified\n"); </blockquote><div> </div><div>I don't think you want unreachable here. The higher layers only check modifier zero. They don't walk the entire list. </div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"> + default: + /* Modifiers from other vendors would land here. */ + break; + } + } </blockquote><div> </div><div>Also, what's the benifit of the local modifiers thing? The select_best_modifier can handle an arbitrarily long list so I don't see what this is gaining us. </div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"> + + return intel_create_image_common(dri_screen, width, height, format, 0, + local_mods, local_count, loaderPrivate); } static GLboolean @@ -1999,7 +2025,9 @@ intelAllocateBuffer(__DRIscreen *dri_screen, if (intelBuffer == NULL) return NULL; - /* The front and back buffers are color buffers, which are X tiled. */ + /* The front and back buffers are color buffers, which are X tiled. GEN9+ + * supports Y tiled and compressed buffers, but there is no way to plumb that + * through to here. */ uint32_t tiling = I915_TILING_X; unsigned long pitch; int cpp = format / 8; -- 2.12.0 </blockquote></div> </div></div>