gallium/u_transfer_helper: add util functions for doing deinterleaving during map

in some cases (e.g., zink and d3d12) we only want to split the depth and stencil
buffers when we're mapping them, and we can handle packed buffers in other cases,
so being able to reuse the u_transfer_helper functionality is still desired but
only if we can preserve the underlying buffer the rest of the time

Kenneth Graunke notes during post-review:
Vulkan reads/copies on packed Z24S8 only return depth, so we need to use separate
Z24 and S8 reads and do packing tricks.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5338>
diff --git a/src/gallium/auxiliary/util/u_transfer_helper.c b/src/gallium/auxiliary/util/u_transfer_helper.c
index 0ba8abd..0dc9ecf 100644
--- a/src/gallium/auxiliary/util/u_transfer_helper.c
+++ b/src/gallium/auxiliary/util/u_transfer_helper.c
@@ -351,7 +351,8 @@
              const struct pipe_box *box)
 {
    struct u_transfer_helper *helper = pctx->screen->transfer_helper;
-   struct u_transfer *trans = u_transfer(ptrans);
+   /* using the function here hits an assert for the deinterleave cases */
+   struct u_transfer *trans = (struct u_transfer *)ptrans;
    enum pipe_format iformat, format = ptrans->resource->format;
    unsigned width = box->width;
    unsigned height = box->height;
@@ -544,3 +545,125 @@
 {
    free(helper);
 }
+
+
+/* these two functions 'deinterleave' are meant to be used without the corresponding
+ * resource_create/destroy hooks, as they perform the interleaving on-the-fly
+ *
+ * drivers should expect to be passed the same buffer repeatedly with the format changed
+ * to indicate which component is being mapped
+ */
+void *
+u_transfer_helper_deinterleave_transfer_map(struct pipe_context *pctx,
+                                            struct pipe_resource *prsc,
+                                            unsigned level, unsigned usage,
+                                            const struct pipe_box *box,
+                                            struct pipe_transfer **pptrans)
+{
+   struct u_transfer_helper *helper = pctx->screen->transfer_helper;
+   struct u_transfer *trans;
+   struct pipe_transfer *ptrans;
+   enum pipe_format format = prsc->format;
+   unsigned width = box->width;
+   unsigned height = box->height;
+
+   if (!((helper->separate_stencil && util_format_is_depth_and_stencil(format)) ||
+       (format == PIPE_FORMAT_Z32_FLOAT_S8X24_UINT && helper->separate_z32s8)))
+      return helper->vtbl->transfer_map(pctx, prsc, level, usage, box, pptrans);
+
+   debug_assert(box->depth == 1);
+
+   trans = calloc(1, sizeof(*trans));
+   if (!trans)
+      return NULL;
+
+   ptrans = &trans->base;
+   pipe_resource_reference(&ptrans->resource, prsc);
+   ptrans->level = level;
+   ptrans->usage = usage;
+   ptrans->box   = *box;
+   ptrans->stride = util_format_get_stride(format, box->width);
+   ptrans->layer_stride = ptrans->stride * box->height;
+
+   trans->staging = malloc(ptrans->layer_stride);
+   if (!trans->staging)
+      goto fail;
+
+   trans->ptr = helper->vtbl->transfer_map(pctx, prsc, level, usage | PIPE_TRANSFER_DEPTH_ONLY, box,
+                                           &trans->trans);
+   if (!trans->ptr)
+      goto fail;
+
+   trans->ptr2 = helper->vtbl->transfer_map(pctx, prsc, level,
+                                            usage | PIPE_TRANSFER_STENCIL_ONLY, box, &trans->trans2);
+   if (needs_pack(usage)) {
+      switch (prsc->format) {
+      case PIPE_FORMAT_Z32_FLOAT_S8X24_UINT:
+         util_format_z32_float_s8x24_uint_pack_z_float(trans->staging,
+                                                       ptrans->stride,
+                                                       trans->ptr,
+                                                       trans->trans->stride,
+                                                       width, height);
+         util_format_z32_float_s8x24_uint_pack_s_8uint(trans->staging,
+                                                       ptrans->stride,
+                                                       trans->ptr2,
+                                                       trans->trans2->stride,
+                                                       width, height);
+         break;
+      case PIPE_FORMAT_Z24_UNORM_S8_UINT:
+         util_format_z24_unorm_s8_uint_pack_separate(trans->staging,
+                                                     ptrans->stride,
+                                                     trans->ptr,
+                                                     trans->trans->stride,
+                                                     trans->ptr2,
+                                                     trans->trans2->stride,
+                                                     width, height);
+         break;
+      default:
+         unreachable("Unexpected format");
+      }
+   }
+
+   *pptrans = ptrans;
+   return trans->staging;
+
+fail:
+   if (trans->trans)
+      helper->vtbl->transfer_unmap(pctx, trans->trans);
+   if (trans->trans2)
+      helper->vtbl->transfer_unmap(pctx, trans->trans2);
+   pipe_resource_reference(&ptrans->resource, NULL);
+   free(trans->staging);
+   free(trans);
+   return NULL;
+}
+
+void
+u_transfer_helper_deinterleave_transfer_unmap(struct pipe_context *pctx,
+                                              struct pipe_transfer *ptrans)
+{
+   struct u_transfer_helper *helper = pctx->screen->transfer_helper;
+   enum pipe_format format = ptrans->resource->format;
+
+   if ((helper->separate_stencil && util_format_is_depth_and_stencil(format)) ||
+       (format == PIPE_FORMAT_Z32_FLOAT_S8X24_UINT && helper->separate_z32s8)) {
+      struct u_transfer *trans = (struct u_transfer *)ptrans;
+
+      if (!(ptrans->usage & PIPE_TRANSFER_FLUSH_EXPLICIT)) {
+         struct pipe_box box;
+         u_box_2d(0, 0, ptrans->box.width, ptrans->box.height, &box);
+         flush_region(pctx, ptrans, &box);
+      }
+
+      helper->vtbl->transfer_unmap(pctx, trans->trans);
+      if (trans->trans2)
+         helper->vtbl->transfer_unmap(pctx, trans->trans2);
+
+      pipe_resource_reference(&ptrans->resource, NULL);
+
+      free(trans->staging);
+      free(trans);
+   } else {
+      helper->vtbl->transfer_unmap(pctx, ptrans);
+   }
+}
diff --git a/src/gallium/auxiliary/util/u_transfer_helper.h b/src/gallium/auxiliary/util/u_transfer_helper.h
index e27c357..1c866c6 100644
--- a/src/gallium/auxiliary/util/u_transfer_helper.h
+++ b/src/gallium/auxiliary/util/u_transfer_helper.h
@@ -129,6 +129,16 @@
 
 void u_transfer_helper_destroy(struct u_transfer_helper *helper);
 
+void *
+u_transfer_helper_deinterleave_transfer_map(struct pipe_context *pctx,
+                                            struct pipe_resource *prsc,
+                                            unsigned level, unsigned usage,
+                                            const struct pipe_box *box,
+                                            struct pipe_transfer **pptrans);
+
+void
+u_transfer_helper_deinterleave_transfer_unmap(struct pipe_context *pctx,
+                                              struct pipe_transfer *ptrans);
 #ifdef __cplusplus
 } // extern "C" {
 #endif