From 2aec563acf63a8f5356e491d0edb89a98be0a358 Mon Sep 17 00:00:00 2001
From: Karol Herbst <kherbst@redhat.com>
Date: Fri, 23 Aug 2024 21:56:43 +0200
Subject: [PATCH] rusticl/image: fix clEnqueueFillImage for CL_DEPTH

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30831>
---
 src/gallium/frontends/rusticl/api/memory.rs  | 17 +++++++++++++----
 src/gallium/frontends/rusticl/core/memory.rs |  9 ++-------
 2 files changed, 15 insertions(+), 11 deletions(-)

diff --git a/src/gallium/frontends/rusticl/api/memory.rs b/src/gallium/frontends/rusticl/api/memory.rs
index ee7b7bf74f4..bf050b6481e 100644
--- a/src/gallium/frontends/rusticl/api/memory.rs
+++ b/src/gallium/frontends/rusticl/api/memory.rs
@@ -1967,16 +1967,25 @@ fn enqueue_fill_image(
     // description for origin and region.
     validate_image_bounds(&i, origin, region)?;
 
-    // we have to copy memory and it's always a 4 component int value
-    // TODO but not for CL_DEPTH
-    let fill_color = unsafe { slice::from_raw_parts(fill_color.cast(), 4).to_vec() };
+    // The fill color is a single floating-point value if the channel order is CL_DEPTH. Otherwise,
+    // the fill color is a four component RGBA floating-point color value if the image channel data
+    // type is not an unnormalized signed or unsigned integer type, is a four component signed
+    // integer value if the image channel data type is an unnormalized signed integer type and is a
+    // four component unsigned integer value if the image channel data type is an unnormalized
+    // unsigned integer type.
+    let fill_color = if i.image_format.image_channel_order == CL_DEPTH {
+        [unsafe { fill_color.cast::<u32>().read() }, 0, 0, 0]
+    } else {
+        unsafe { fill_color.cast::<[u32; 4]>().read() }
+    };
+
     create_and_queue(
         q,
         CL_COMMAND_FILL_BUFFER,
         evs,
         event,
         false,
-        Box::new(move |q, ctx| i.fill(q, ctx, &fill_color, &origin, &region)),
+        Box::new(move |q, ctx| i.fill(q, ctx, fill_color, &origin, &region)),
     )
 
     //• CL_INVALID_IMAGE_SIZE if image dimensions (image width, height, specified or compute row and/or slice pitch) for image are not supported by device associated with queue.
diff --git a/src/gallium/frontends/rusticl/core/memory.rs b/src/gallium/frontends/rusticl/core/memory.rs
index f15b0cbb0e3..672d2208a5a 100644
--- a/src/gallium/frontends/rusticl/core/memory.rs
+++ b/src/gallium/frontends/rusticl/core/memory.rs
@@ -1293,7 +1293,7 @@ impl Image {
         &self,
         q: &Queue,
         ctx: &PipeContext,
-        pattern: &[u32],
+        pattern: [u32; 4],
         origin: &CLVec<usize>,
         region: &CLVec<usize>,
     ) -> CLResult<()> {
@@ -1301,19 +1301,14 @@ impl Image {
 
         // make sure we allocate multiples of 4 bytes so drivers don't read out of bounds or
         // unaligned.
-        // TODO: use div_ceil once it's available
         let pixel_size: usize = self.image_format.pixel_size().unwrap().into();
         let mut new_pattern: Vec<u32> = vec![0; pixel_size.div_ceil(size_of::<u32>())];
 
-        // we don't support CL_DEPTH for now
-        assert!(pattern.len() == 4);
-
         // SAFETY: pointers have to be valid for read/writes of exactly one pixel of their
         // respective format.
         // `new_pattern` has the correct size due to the `size` above.
         // `pattern` is validated through the CL API and allows undefined behavior if not followed
-        // by CL API rules. It's expected to be a 4 component array of 32 bit values, except for
-        // CL_DEPTH where it's just one value.
+        // by CL API rules.
         unsafe {
             util_format_pack_rgba(
                 self.pipe_format,