docs: add release notes for 17.0.5

Signed-off-by: Andres Gomez <agomez@igalia.com>

VERSION

@@ -1 +1 @@
-17.0.4
+17.0.5

bin/.cherry-ignore

@@ -3,3 +3,48 @@ d49d275c415d60ae08dc3e52d8db11f19a44010f anv/blorp: Don't sanitize the swizzle f
 # The commit addressed an earlier commit 0567ab0407e which did not land in branch.
 # A stable specific backported patch will follow later ...
 bc5d587a80b64fb3e0a5ea8067e6317fbca2bbc5 radv: Invalidate L2 for TRANSFER_WRITE barriers
+# stable: the commit depends on earlier commit 656e30b6860 which did
+# 	  not land in branch, and some other dependencies.
+d2d6cf6c8387ac06c8559027f57683a61b48671b anv: Add the pci_id into the shader cache UUID
+# stable: 17.1 nomination only. Fixes earlier commit 0b613c20 which
+# 	  did not land in branch.
+d4601b0efc7f5e24e3f39fefa8e29e79560245ce freedreno: fix crash if ctx torn down with no rendering
+# stable: 17.1 nomination only. the commit depends on earlier commit
+# 	  7748c3f5eb1 which did not land in branch and the part that
+# 	  could be merged is not really providing anything relevant to
+# 	  stable.
+b81d85f1754928139f9f01474495e024946aa1b4 configure.ac: Fix typos.
+# extra: this reverts a commit in stable only because a new feature in
+# 	 master replaces it.
+e75001811e3b66986b4ede165a0fdde703d4f05b Revert "etnaviv: Cannot render to rb-swapped formats"
+# extra: the commit depends on earlier commits which did not land in
+#	 branch.
+dee31311eb024a636466e359b43d3a67b0135f32 Revert "i965/fs: Don't emit SEL instructions for type-converting MOVs."
+# fixes: finds 2 commit references but it is really only one, and only
+#        in master.
+2b3c490e23a6cd7f4f6bb489267c2b9b4fde5b20 radv: fix typo in a2b10g10r10 fast clear calculation
+# fixes: removes unused symbols but it doesn't really add nor fix
+#        anything relevant.
+9807e9dea67a60ab8ee04741c75b748eade582ee anv: remove unused anv_dispatch_table dtable
+# fixes: removes unused symbols but it doesn't really add nor fix
+#        anything relevant.
+320561bd831b81843e732d38d2b254aa3b25f2f6 radv: remove unused radv_dispatch_table dtable
+# fixes: it just makes a local function static.
+8b79f0ed0812a3a80dd91e47112fd722bd2d6978 radv: make radv_resolve_entrypoint static
+# fixes: reported only because referes to fixes in 2 commits, but they
+#        are not in stable.
+3f2cb699cfe0481f214c709b5a1375caf0a9c7fe android: vulkan: add support for libmesa_vulkan_util
+# fixes: reported only because referes to fixes in 2 commits, but they
+#        are not in stable.
+61c38d14b7b1d0ccacc440340ee0b1bc4454eca2 android: r600: fix libmesa_amd_common dependency
+# fixes: removes unused symbols but it doesn't really add nor fix
+#        anything relevant.
+b69a03e12aa38ed5d8b60a056630ed445f6097dc i965: remove dead brw_new_shader() declaration
+# fixes: it just cleans .gitignore.
+f282ace67862c0633d0a8135e4808867740d0d39 i965: remove i965_symbols_test reference from .gitignore
+# fixes: reported only because referes to fixes in 2 commits, but they
+#        are not in stable.
+15603055fb36a630cf3023219e9dd5c4a49c0596 anv: automake: ensure that the destination directory is created
+# fixes: reported only because referes to fixes in 2 commits, but they
+#        are not in stable.
+a6840efc096bd26e92bcee1a50515d845988a2f9 anv: provide required gem stubs for the tests

configure.ac

@@ -2394,7 +2394,9 @@ if test -n "$with_gallium_drivers"; then
         xvirgl)
             HAVE_GALLIUM_VIRGL=yes
             require_libdrm "virgl"
-            require_basic_egl "virgl"
+            if test "x$enable_egl" = xyes; then
+                require_basic_egl "virgl"
+            fi
             ;;
         *)
             AC_MSG_ERROR([Unknown Gallium driver: $driver])

docs/relnotes/17.0.4.html

@@ -31,7 +31,8 @@ because compatibility contexts are not supported.
 
 <h2>SHA256 checksums</h2>
 <pre>
-TBD
+c4c34ba05d48f76b45bc05bc4b6e9242077f403d63c4f0c355c7b07786de233e  mesa-17.0.4.tar.gz
+1269dc8545a193932a0779b2db5bce9be4a5f6813b98c38b93b372be8362a346  mesa-17.0.4.tar.xz
 </pre>
 
 

docs/relnotes/17.0.5.html

@@ -0,0 +1,143 @@
+<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
+<html lang="en">
+<head>
+  <meta http-equiv="content-type" content="text/html; charset=utf-8">
+  <title>Mesa Release Notes</title>
+  <link rel="stylesheet" type="text/css" href="../mesa.css">
+</head>
+<body>
+
+<div class="header">
+  <h1>The Mesa 3D Graphics Library</h1>
+</div>
+
+<iframe src="../contents.html"></iframe>
+<div class="content">
+
+<h1>Mesa 17.0.5 Release Notes / April 28, 2017</h1>
+
+<p>
+Mesa 17.0.5 is a bug fix release which fixes bugs found since the 17.0.4 release.
+</p>
+<p>
+Mesa 17.0.5 implements the OpenGL 4.5 API, but the version reported by
+glGetString(GL_VERSION) or glGetIntegerv(GL_MAJOR_VERSION) /
+glGetIntegerv(GL_MINOR_VERSION) depends on the particular driver being used.
+Some drivers don't support all the features required in OpenGL 4.5.  OpenGL
+4.5 is <strong>only</strong> available if requested at context creation
+because compatibility contexts are not supported.
+</p>
+
+
+<h2>SHA256 checksums</h2>
+<pre>
+TBD
+</pre>
+
+
+<h2>New features</h2>
+<p>None</p>
+
+
+<h2>Bug fixes</h2>
+
+<ul>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=97524">Bug 97524</a> - Samplers referring to the same texture unit with different types should raise GL_INVALID_OPERATION</li>
+
+</ul>
+
+
+<h2>Changes</h2>
+
+<p>Andres Gomez (16):</p>
+<ul>
+  <li>cherry-ignore: Add the pci_id into the shader cache UUID</li>
+  <li>cherry-ignore: fix crash if ctx torn down with no rendering</li>
+  <li>cherry-ignore: Fix typos.</li>
+  <li>cherry-ignore: Revert "etnaviv: Cannot render to rb-swapped formats"</li>
+  <li>cherry-ignore: Revert "i965/fs: Don't emit SEL instructions for type-converting MOVs."</li>
+  <li>cherry-ignore: fix typo in a2b10g10r10 fast clear calculation</li>
+  <li>cherry-ignore: remove unused anv_dispatch_table dtable</li>
+  <li>cherry-ignore: remove unused radv_dispatch_table dtable</li>
+  <li>cherry-ignore: make radv_resolve_entrypoint static</li>
+  <li>cherry-ignore: vulkan: add support for libmesa_vulkan_util</li>
+  <li>cherry-ignore: r600: fix libmesa_amd_common dependency</li>
+  <li>cherry-ignore: remove dead brw_new_shader() declaration</li>
+  <li>cherry-ignore: remove i965_symbols_test reference from .gitignore</li>
+  <li>cherry-ignore: automake: ensure that the destination directory is created</li>
+  <li>cherry-ignore: provide required gem stubs for the tests</li>
+  <li>Update version to 17.0.5</li>
+</ul>
+
+<p>Boyan Ding (2):</p>
+<ul>
+  <li>nvc0/ir: Properly handle a "split form" of predicate destination</li>
+  <li>nir: Destination component count of shader_clock intrinsic is 2</li>
+</ul>
+
+<p>Emil Velikov (5):</p>
+<ul>
+  <li>docs: add sha256 checksums for 17.0.4</li>
+  <li>winsys/sw/dri: don't use GNU void pointer arithmetic</li>
+  <li>st/clover: add space between &lt; and ::</li>
+  <li>configure.ac: check require_basic_egl only if egl enabled</li>
+  <li>st/mesa: automake: honour the vdpau header install location</li>
+</ul>
+
+<p>Francisco Jerez (2):</p>
+<ul>
+  <li>intel/fs: Use regs_written() in spilling cost heuristic for improved accuracy.</li>
+  <li>intel/fs: Take into account amount of data read in spilling cost heuristic.</li>
+</ul>
+
+<p>Grazvydas Ignotas (1):</p>
+<ul>
+  <li>radv: report timestampPeriod correctly</li>
+</ul>
+
+<p>Jason Ekstrand (5):</p>
+<ul>
+  <li>anv/blorp: Flush the texture cache in UpdateBuffer</li>
+  <li>anv/cmd_buffer: Flush the VF cache at the top of all primaries</li>
+  <li>anv/cmd_buffer: Always set up a null surface state</li>
+  <li>anv/cmd_buffer: Use the null surface state for ATTACHMENT_UNUSED</li>
+  <li>anv/blorp: Properly handle VK_ATTACHMENT_UNUSED</li>
+</ul>
+
+<p>Kenneth Graunke (1):</p>
+<ul>
+  <li>i965/vec4: Avoid reswizzling MACH instructions in opt_register_coalesce().</li>
+</ul>
+
+<p>Marek Olšák (1):</p>
+<ul>
+  <li>st/mesa: invalidate the readpix cache in st_indirect_draw_vbo</li>
+</ul>
+
+<p>Nanley Chery (1):</p>
+<ul>
+  <li>anv/cmd_buffer: Disable CCS on BDW input attachments</li>
+</ul>
+
+<p>Nicolai Hähnle (4):</p>
+<ul>
+  <li>mesa: fix remaining xfb prims check for GLES with multiple instances</li>
+  <li>mesa: extract need_xfb_remaining_prims_check</li>
+  <li>mesa: move glMultiDrawArrays to vbo and fix error handling</li>
+  <li>vbo: fix gl_DrawID handling in glMultiDrawArrays</li>
+</ul>
+
+<p>Rob Clark (1):</p>
+<ul>
+  <li>util/queue: don't hang at exit</li>
+</ul>
+
+<p>Timothy Arceri (1):</p>
+<ul>
+  <li>mesa: validate sampler type across the whole program</li>
+</ul>
+
+</div>
+</body>
+</html>

src/amd/vulkan/radv_device.c

@@ -570,7 +570,7 @@ void radv_GetPhysicalDeviceProperties(
 		.storageImageSampleCounts                 = VK_SAMPLE_COUNT_1_BIT,
 		.maxSampleMaskWords                       = 1,
 		.timestampComputeAndGraphics              = false,
-		.timestampPeriod                          = 100000.0 / pdevice->rad_info.clock_crystal_freq,
+		.timestampPeriod                          = 1000000.0 / pdevice->rad_info.clock_crystal_freq,
 		.maxClipDistances                         = 8,
 		.maxCullDistances                         = 8,
 		.maxCombinedClipAndCullDistances          = 8,

src/amd/vulkan/radv_radeon_winsys.h

@@ -94,7 +94,7 @@ struct radeon_info {
 	uint32_t                    compute_rings;
 	uint32_t                    vce_fw_version;
 	uint32_t                    vce_harvest_config;
-	uint32_t                    clock_crystal_freq;
+	uint32_t                    clock_crystal_freq; /* in kHz */
 
 	/* Kernel info. */
 	uint32_t                    drm_major; /* version */

src/compiler/glsl/glsl_to_nir.cpp

@@ -927,7 +927,8 @@ nir_visitor::visit(ir_call *ir)
          nir_builder_instr_insert(&b, &instr->instr);
          break;
       case nir_intrinsic_shader_clock:
-         nir_ssa_dest_init(&instr->instr, &instr->dest, 1, 32, NULL);
+         nir_ssa_dest_init(&instr->instr, &instr->dest, 2, 32, NULL);
+         instr->num_components = 2;
          nir_builder_instr_insert(&b, &instr->instr);
          break;
       case nir_intrinsic_store_ssbo: {

src/compiler/nir/nir_intrinsics.h

@@ -91,7 +91,7 @@ BARRIER(memory_barrier)
  * The latter can be used as code motion barrier, which is currently not
  * feasible with NIR.
  */
-INTRINSIC(shader_clock, 0, ARR(0), true, 1, 0, 0, xx, xx, xx, NIR_INTRINSIC_CAN_ELIMINATE)
+INTRINSIC(shader_clock, 0, ARR(0), true, 2, 0, 0, xx, xx, xx, NIR_INTRINSIC_CAN_ELIMINATE)
 
 /*
  * Memory barrier with semantics analogous to the compute shader

src/gallium/auxiliary/util/u_queue.c

@@ -294,9 +294,18 @@ util_queue_add_job(struct util_queue *queue,
    struct util_queue_job *ptr;
 
    assert(fence->signalled);
-   fence->signalled = false;
 
    pipe_mutex_lock(queue->lock);
+   if (queue->kill_threads) {
+      pipe_mutex_unlock(queue->lock);
+      /* well no good option here, but any leaks will be
+       * short-lived as things are shutting down..
+       */
+      return;
+   }
+
+   fence->signalled = false;
+
    assert(queue->num_queued >= 0 && queue->num_queued <= queue->max_jobs);
 
    /* if the queue is full, wait until there is space */

src/gallium/drivers/nouveau/codegen/nv50_ir_emit_nvc0.cpp

@@ -58,6 +58,7 @@ private:
    void setImmediateS8(const ValueRef&);
    void setSUConst16(const Instruction *, const int s);
    void setSUPred(const Instruction *, const int s);
+   void setPDSTL(const Instruction *, const int d);
 
    void emitCondCode(CondCode cc, int pos);
    void emitInterpMode(const Instruction *);
@@ -373,6 +374,16 @@ void CodeEmitterNVC0::setImmediateS8(const ValueRef &ref)
    code[0] |= (s8 >> 6) << 8;
 }
 
+void CodeEmitterNVC0::setPDSTL(const Instruction *i, const int d)
+{
+   assert(d < 0 || (i->defExists(d) && i->def(d).getFile() == FILE_PREDICATE));
+
+   uint32_t pred = d >= 0 ? DDATA(i->def(d)).id : 7;
+
+   code[0] |= (pred & 3) << 8;
+   code[1] |= (pred & 4) << (26 - 2);
+}
+
 void
 CodeEmitterNVC0::emitForm_A(const Instruction *i, uint64_t opc)
 {
@@ -1864,7 +1875,7 @@ CodeEmitterNVC0::emitSTORE(const Instruction *i)
       if (i->src(0).getFile() == FILE_MEMORY_SHARED &&
           i->subOp == NV50_IR_SUBOP_STORE_UNLOCKED) {
          assert(i->defExists(0));
-         defId(i->def(0), 8);
+         setPDSTL(i, 0);
       }
    }
 
@@ -1936,7 +1947,7 @@ CodeEmitterNVC0::emitLOAD(const Instruction *i)
 
    if (p >= 0) {
       if (targ->getChipset() >= NVISA_GK104_CHIPSET)
-         defId(i->def(p), 8);
+         setPDSTL(i, p);
       else
          defId(i->def(p), 32 + 18);
    }

src/gallium/state_trackers/clover/llvm/codegen.hpp

@@ -49,7 +49,7 @@ namespace clover {
       build_module_library(const ::llvm::Module &mod,
                            enum module::section::type section_type);
 
-      std::unique_ptr<::llvm::Module>
+      std::unique_ptr< ::llvm::Module>
       parse_module_library(const module &m, ::llvm::LLVMContext &ctx,
                            std::string &r_log);
 

src/gallium/state_trackers/clover/llvm/codegen/bitcode.cpp

@@ -94,7 +94,7 @@ clover::llvm::build_module_library(const ::llvm::Module &mod,
    return m;
 }
 
-std::unique_ptr<::llvm::Module>
+std::unique_ptr< ::llvm::Module>
 clover::llvm::parse_module_library(const module &m, ::llvm::LLVMContext &ctx,
                                    std::string &r_log) {
    auto mod = ::llvm::parseBitcodeFile(::llvm::MemoryBufferRef(
@@ -104,5 +104,5 @@ clover::llvm::parse_module_library(const module &m, ::llvm::LLVMContext &ctx,
          fail(r_log, error(CL_INVALID_PROGRAM), s);
       });
 
-   return std::unique_ptr<::llvm::Module>(std::move(*mod));
+   return std::unique_ptr< ::llvm::Module>(std::move(*mod));
 }

src/gallium/state_trackers/clover/llvm/codegen/common.cpp

@@ -121,12 +121,12 @@ namespace {
          } else {
             // Other types.
             const auto actual_type =
-               isa<::llvm::PointerType>(arg_type) && arg.hasByValAttr() ?
-               cast<::llvm::PointerType>(arg_type)->getElementType() : arg_type;
+               isa< ::llvm::PointerType>(arg_type) && arg.hasByValAttr() ?
+               cast< ::llvm::PointerType>(arg_type)->getElementType() : arg_type;
 
             if (actual_type->isPointerTy()) {
                const unsigned address_space =
-                  cast<::llvm::PointerType>(actual_type)->getAddressSpace();
+                  cast< ::llvm::PointerType>(actual_type)->getAddressSpace();
 
                if (address_space == address_spaces[clang::LangAS::opencl_local
                                                    - clang::LangAS::Offset]) {

src/gallium/state_trackers/clover/llvm/codegen/native.cpp

@@ -156,7 +156,7 @@ clover::llvm::print_module_native(const ::llvm::Module &mod,
                                   const target &target) {
    std::string log;
    try {
-      std::unique_ptr<::llvm::Module> cmod { CloneModule(&mod) };
+      std::unique_ptr< ::llvm::Module> cmod { CloneModule(&mod) };
       return as_string(emit_code(*cmod, target,
                                  TargetMachine::CGFT_AssemblyFile, log));
    } catch (...) {

src/gallium/state_trackers/clover/llvm/compat.hpp

@@ -118,18 +118,18 @@ namespace clover {
 #endif
          }
 
-         inline std::unique_ptr<::llvm::Linker>
+         inline std::unique_ptr< ::llvm::Linker>
          create_linker(::llvm::Module &mod) {
 #if HAVE_LLVM >= 0x0308
-            return std::unique_ptr<::llvm::Linker>(new ::llvm::Linker(mod));
+            return std::unique_ptr< ::llvm::Linker>(new ::llvm::Linker(mod));
 #else
-            return std::unique_ptr<::llvm::Linker>(new ::llvm::Linker(&mod));
+            return std::unique_ptr< ::llvm::Linker>(new ::llvm::Linker(&mod));
 #endif
          }
 
          inline bool
          link_in_module(::llvm::Linker &linker,
-                        std::unique_ptr<::llvm::Module> mod) {
+                        std::unique_ptr< ::llvm::Module> mod) {
 #if HAVE_LLVM >= 0x0308
             return linker.linkInModule(std::move(mod));
 #else

src/gallium/state_trackers/clover/llvm/metadata.hpp

@@ -51,7 +51,7 @@ namespace clover {
          is_kernel_node_for(const ::llvm::Function &f) {
             return [&](const ::llvm::MDNode *n) {
                using ::llvm::mdconst::dyn_extract;
-               return &f == dyn_extract<::llvm::Function>(n->getOperand(0));
+               return &f == dyn_extract< ::llvm::Function>(n->getOperand(0));
             };
          }
 
@@ -65,7 +65,7 @@ namespace clover {
 #endif
          }
 
-         inline iterator_range<::llvm::MDNode::op_iterator>
+         inline iterator_range< ::llvm::MDNode::op_iterator>
          get_kernel_metadata_operands(const ::llvm::Function &f,
                                       const std::string &name) {
 #if HAVE_LLVM >= 0x0309
@@ -79,11 +79,11 @@ namespace clover {
             const auto kernel_node = find(is_kernel_node_for(f),
                                           get_kernel_nodes(*f.getParent()));
 
-            const auto data_node = cast<::llvm::MDNode>(
+            const auto data_node = cast< ::llvm::MDNode>(
                find([&](const ::llvm::MDOperand &op) {
-                     if (auto m = dyn_cast<::llvm::MDNode>(op))
+                     if (auto m = dyn_cast< ::llvm::MDNode>(op))
                         if (m->getNumOperands())
-                           if (auto m_name = dyn_cast<::llvm::MDString>(
+                           if (auto m_name = dyn_cast< ::llvm::MDString>(
                                   m->getOperand(0).get()))
                               return m_name->getString() == name;
 
@@ -106,7 +106,7 @@ namespace clover {
       get_argument_metadata(const ::llvm::Function &f,
                             const ::llvm::Argument &arg,
                             const std::string &name) {
-         return ::llvm::cast<::llvm::MDString>(
+         return ::llvm::cast< ::llvm::MDString>(
                detail::get_kernel_metadata_operands(f, name)[arg.getArgNo()])
             ->getString();
       }

src/gallium/winsys/sw/dri/dri_sw_winsys.c

@@ -200,7 +200,7 @@ dri_sw_displaytarget_display(struct sw_winsys *ws,
 
    if (box) {
        void *data;
-       data = dri_sw_dt->data + (dri_sw_dt->stride * box->y) + box->x * blsize;
+       data = (char *)dri_sw_dt->data + (dri_sw_dt->stride * box->y) + box->x * blsize;
        dri_sw_ws->lf->put_image2(dri_drawable, data,
                                  box->x, box->y, box->width, box->height, dri_sw_dt->stride);
    } else {

src/intel/vulkan/anv_blorp.c

@@ -683,6 +683,11 @@ void anv_CmdUpdateBuffer(
 
    assert(max_update_size < MAX_SURFACE_DIM * 4);
 
+   /* We're about to read data that was written from the CPU.  Flush the
+    * texture cache so we don't get anything stale.
+    */
+   cmd_buffer->state.pending_pipe_bits |= ANV_PIPE_TEXTURE_CACHE_INVALIDATE_BIT;
+
    while (dataSize) {
       const uint32_t copy_size = MIN2(dataSize, max_update_size);
 
@@ -691,6 +696,9 @@ void anv_CmdUpdateBuffer(
 
       memcpy(tmp_data.map, pData, copy_size);
 
+      if (!cmd_buffer->device->info.has_llc)
+         anv_state_clflush(tmp_data);
+
       int bs = 16;
       bs = gcd_pow2_u64(bs, dstOffset);
       bs = gcd_pow2_u64(bs, copy_size);
@@ -1122,6 +1130,9 @@ anv_cmd_buffer_flush_attachments(struct anv_cmd_buffer *cmd_buffer,
 
    for (uint32_t i = 0; i < subpass->color_count; ++i) {
       uint32_t att = subpass->color_attachments[i];
+      if (att == VK_ATTACHMENT_UNUSED)
+         continue;
+
       assert(att < pass->attachment_count);
       if (attachment_needs_flush(cmd_buffer, &pass->attachments[att], stage)) {
          cmd_buffer->state.pending_pipe_bits |=
@@ -1149,14 +1160,19 @@ subpass_needs_clear(const struct anv_cmd_buffer *cmd_buffer)
 
    for (uint32_t i = 0; i < cmd_state->subpass->color_count; ++i) {
       uint32_t a = cmd_state->subpass->color_attachments[i];
+      if (a == VK_ATTACHMENT_UNUSED)
+         continue;
+
+      assert(a < cmd_state->pass->attachment_count);
       if (cmd_state->attachments[a].pending_clear_aspects) {
          return true;
       }
    }
 
-   if (ds != VK_ATTACHMENT_UNUSED &&
-       cmd_state->attachments[ds].pending_clear_aspects) {
-      return true;
+   if (ds != VK_ATTACHMENT_UNUSED) {
+      assert(ds < cmd_state->pass->attachment_count);
+      if (cmd_state->attachments[ds].pending_clear_aspects)
+         return true;
    }
 
    return false;
@@ -1188,6 +1204,10 @@ anv_cmd_buffer_clear_subpass(struct anv_cmd_buffer *cmd_buffer)
    struct anv_framebuffer *fb = cmd_buffer->state.framebuffer;
    for (uint32_t i = 0; i < cmd_state->subpass->color_count; ++i) {
       const uint32_t a = cmd_state->subpass->color_attachments[i];
+      if (a == VK_ATTACHMENT_UNUSED)
+         continue;
+
+      assert(a < cmd_state->pass->attachment_count);
       struct anv_attachment_state *att_state = &cmd_state->attachments[a];
 
       if (!att_state->pending_clear_aspects)
@@ -1246,6 +1266,7 @@ anv_cmd_buffer_clear_subpass(struct anv_cmd_buffer *cmd_buffer)
    }
 
    const uint32_t ds = cmd_state->subpass->depth_stencil_attachment;
+   assert(ds == VK_ATTACHMENT_UNUSED || ds < cmd_state->pass->attachment_count);
 
    if (ds != VK_ATTACHMENT_UNUSED &&
        cmd_state->attachments[ds].pending_clear_aspects) {
@@ -1551,8 +1572,12 @@ anv_cmd_buffer_resolve_subpass(struct anv_cmd_buffer *cmd_buffer)
    blorp_batch_init(&cmd_buffer->device->blorp, &batch, cmd_buffer, 0);
 
    for (uint32_t i = 0; i < subpass->color_count; ++i) {
-      ccs_resolve_attachment(cmd_buffer, &batch,
-                             subpass->color_attachments[i]);
+      const uint32_t att = subpass->color_attachments[i];
+      if (att == VK_ATTACHMENT_UNUSED)
+         continue;
+
+      assert(att < cmd_buffer->state.pass->attachment_count);
+      ccs_resolve_attachment(cmd_buffer, &batch, att);
    }
 
    anv_cmd_buffer_flush_attachments(cmd_buffer, SUBPASS_STAGE_DRAW);
@@ -1565,6 +1590,9 @@ anv_cmd_buffer_resolve_subpass(struct anv_cmd_buffer *cmd_buffer)
          if (dst_att == VK_ATTACHMENT_UNUSED)
             continue;
 
+         assert(src_att < cmd_buffer->state.pass->attachment_count);
+         assert(dst_att < cmd_buffer->state.pass->attachment_count);
+
          if (cmd_buffer->state.attachments[dst_att].pending_clear_aspects) {
             /* From the Vulkan 1.0 spec:
              *

src/intel/vulkan/genX_cmd_buffer.c

@@ -272,26 +272,21 @@ color_attachment_compute_aux_usage(struct anv_device *device,
       att_state->input_aux_usage = ISL_AUX_USAGE_CCS_E;
    } else if (att_state->fast_clear) {
       att_state->aux_usage = ISL_AUX_USAGE_CCS_D;
-      if (GEN_GEN >= 9) {
-         /* From the Sky Lake PRM, RENDER_SURFACE_STATE::AuxiliarySurfaceMode:
-          *
-          *    "If Number of Multisamples is MULTISAMPLECOUNT_1, AUX_CCS_D
-          *    setting is only allowed if Surface Format supported for Fast
-          *    Clear. In addition, if the surface is bound to the sampling
-          *    engine, Surface Format must be supported for Render Target
-          *    Compression for surfaces bound to the sampling engine."
-          *
-          * In other words, we can't sample from a fast-cleared image if it
-          * doesn't also support color compression.
-          */
-         att_state->input_aux_usage = ISL_AUX_USAGE_NONE;
-      } else if (GEN_GEN == 8) {
-         /* Broadwell can sample from fast-cleared images */
+      /* From the Sky Lake PRM, RENDER_SURFACE_STATE::AuxiliarySurfaceMode:
+       *
+       *    "If Number of Multisamples is MULTISAMPLECOUNT_1, AUX_CCS_D
+       *    setting is only allowed if Surface Format supported for Fast
+       *    Clear. In addition, if the surface is bound to the sampling
+       *    engine, Surface Format must be supported for Render Target
+       *    Compression for surfaces bound to the sampling engine."
+       *
+       * In other words, we can only sample from a fast-cleared image if it
+       * also supports color compression.
+       */
+      if (isl_format_supports_lossless_compression(&device->info, iview->isl.format))
          att_state->input_aux_usage = ISL_AUX_USAGE_CCS_D;
-      } else {
-         /* Ivy Bridge and Haswell cannot */
+      else
          att_state->input_aux_usage = ISL_AUX_USAGE_NONE;
-      }
    } else {
       att_state->aux_usage = ISL_AUX_USAGE_NONE;
       att_state->input_aux_usage = ISL_AUX_USAGE_NONE;
@@ -418,23 +413,15 @@ genX(cmd_buffer_setup_attachments)(struct anv_cmd_buffer *cmd_buffer,
       abort();
    }
 
-   bool need_null_state = false;
-   unsigned num_states = 0;
+   /* Reserve one for the NULL state. */
+   unsigned num_states = 1;
    for (uint32_t i = 0; i < pass->attachment_count; ++i) {
-      if (vk_format_is_color(pass->attachments[i].format)) {
+      if (vk_format_is_color(pass->attachments[i].format))
          num_states++;
-      } else {
-         /* We need a null state for any depth-stencil-only subpasses.
-          * Importantly, this includes depth/stencil clears so we create one
-          * whenever we have depth or stencil
-          */
-         need_null_state = true;
-      }
 
       if (need_input_attachment_state(&pass->attachments[i]))
          num_states++;
    }
-   num_states += need_null_state;
 
    const uint32_t ss_stride = align_u32(isl_dev->ss.size, isl_dev->ss.align);
    state->render_pass_states =
@@ -444,11 +431,9 @@ genX(cmd_buffer_setup_attachments)(struct anv_cmd_buffer *cmd_buffer,
    struct anv_state next_state = state->render_pass_states;
    next_state.alloc_size = isl_dev->ss.size;
 
-   if (need_null_state) {
-      state->null_surface_state = next_state;
-      next_state.offset += ss_stride;
-      next_state.map += ss_stride;
-   }
+   state->null_surface_state = next_state;
+   next_state.offset += ss_stride;
+   next_state.map += ss_stride;
 
    for (uint32_t i = 0; i < pass->attachment_count; ++i) {
       if (vk_format_is_color(pass->attachments[i].format)) {
@@ -470,24 +455,22 @@ genX(cmd_buffer_setup_attachments)(struct anv_cmd_buffer *cmd_buffer,
       ANV_FROM_HANDLE(anv_framebuffer, framebuffer, begin->framebuffer);
       assert(pass->attachment_count == framebuffer->attachment_count);
 
-      if (need_null_state) {
-         struct GENX(RENDER_SURFACE_STATE) null_ss = {
-            .SurfaceType = SURFTYPE_NULL,
-            .SurfaceArray = framebuffer->layers > 0,
-            .SurfaceFormat = ISL_FORMAT_R8G8B8A8_UNORM,
+      struct GENX(RENDER_SURFACE_STATE) null_ss = {
+         .SurfaceType = SURFTYPE_NULL,
+         .SurfaceArray = framebuffer->layers > 0,
+         .SurfaceFormat = ISL_FORMAT_R8G8B8A8_UNORM,
 #if GEN_GEN >= 8
-            .TileMode = YMAJOR,
+         .TileMode = YMAJOR,
 #else
-            .TiledSurface = true,
+         .TiledSurface = true,
 #endif
-            .Width = framebuffer->width - 1,
-            .Height = framebuffer->height - 1,
-            .Depth = framebuffer->layers - 1,
-            .RenderTargetViewExtent = framebuffer->layers - 1,
-         };
-         GENX(RENDER_SURFACE_STATE_pack)(NULL, state->null_surface_state.map,
-                                         &null_ss);
-      }
+         .Width = framebuffer->width - 1,
+         .Height = framebuffer->height - 1,
+         .Depth = framebuffer->layers - 1,
+         .RenderTargetViewExtent = framebuffer->layers - 1,
+      };
+      GENX(RENDER_SURFACE_STATE_pack)(NULL, state->null_surface_state.map,
+                                      &null_ss);
 
       for (uint32_t i = 0; i < pass->attachment_count; ++i) {
          struct anv_render_pass_attachment *att = &pass->attachments[i];
@@ -603,6 +586,18 @@ genX(BeginCommandBuffer)(
 
    genX(cmd_buffer_emit_state_base_address)(cmd_buffer);
 
+   /* We sometimes store vertex data in the dynamic state buffer for blorp
+    * operations and our dynamic state stream may re-use data from previous
+    * command buffers.  In order to prevent stale cache data, we flush the VF
+    * cache.  We could do this on every blorp call but that's not really
+    * needed as all of the data will get written by the CPU prior to the GPU
+    * executing anything.  The chances are fairly high that they will use
+    * blorp at least once per primary command buffer so it shouldn't be
+    * wasted.
+    */
+   if (cmd_buffer->level == VK_COMMAND_BUFFER_LEVEL_PRIMARY)
+      cmd_buffer->state.pending_pipe_bits |= ANV_PIPE_VF_CACHE_INVALIDATE_BIT;
+
    if (cmd_buffer->usage_flags &
        VK_COMMAND_BUFFER_USAGE_RENDER_PASS_CONTINUE_BIT) {
       cmd_buffer->state.pass =
@@ -1153,7 +1148,18 @@ emit_binding_table(struct anv_cmd_buffer *cmd_buffer,
          assert(binding->binding == 0);
          if (binding->index < subpass->color_count) {
             const unsigned att = subpass->color_attachments[binding->index];
-            surface_state = cmd_buffer->state.attachments[att].color_rt_state;
+
+            /* From the Vulkan 1.0.46 spec:
+             *
+             *    "If any color or depth/stencil attachments are
+             *    VK_ATTACHMENT_UNUSED, then no writes occur for those
+             *    attachments."
+             */
+            if (att == VK_ATTACHMENT_UNUSED) {
+               surface_state = cmd_buffer->state.null_surface_state;
+            } else {
+               surface_state = cmd_buffer->state.attachments[att].color_rt_state;
+            }
          } else {
             surface_state = cmd_buffer->state.null_surface_state;
          }

src/mapi/glapi/gen/gl_API.xml

@@ -10190,7 +10190,7 @@
 </category>
 
 <category name="GL_EXT_multi_draw_arrays" number="148">
-    <function name="MultiDrawArraysEXT" es1="1.0" es2="2.0" alias="MultiDrawArrays">
+    <function name="MultiDrawArraysEXT" es1="1.0" es2="2.0" exec="dynamic" alias="MultiDrawArrays">
         <param name="mode" type="GLenum"/>
         <param name="first" type="const GLint *"/>
         <param name="count" type="const GLsizei *"/>

src/mesa/Makefile.am

@@ -96,6 +96,7 @@ endif
 
 AM_CPPFLAGS = $(DEFINES) $(INCLUDE_DIRS)
 AM_CFLAGS = \
+	$(VDPAU_CFLAGS) \
 	$(LLVM_CFLAGS) \
 	$(VISIBILITY_CFLAGS) \
 	$(MSVC2013_COMPAT_CFLAGS)

src/mesa/drivers/dri/i965/brw_fs_reg_allocate.cpp

@@ -822,12 +822,11 @@ fs_visitor::choose_spill_reg(struct ra_graph *g)
    foreach_block_and_inst(block, fs_inst, inst, cfg) {
       for (unsigned int i = 0; i < inst->sources; i++) {
 	 if (inst->src[i].file == VGRF)
-            spill_costs[inst->src[i].nr] += loop_scale;
+            spill_costs[inst->src[i].nr] += regs_read(inst, i) * loop_scale;
       }
 
       if (inst->dst.file == VGRF)
-         spill_costs[inst->dst.nr] += DIV_ROUND_UP(inst->size_written, REG_SIZE)
-                                      * loop_scale;
+         spill_costs[inst->dst.nr] += regs_written(inst) * loop_scale;
 
       switch (inst->opcode) {
 

src/mesa/drivers/dri/i965/brw_vec4.cpp

@@ -1066,6 +1066,13 @@ vec4_instruction::can_reswizzle(const struct gen_device_info *devinfo,
    if (devinfo->gen == 6 && is_math() && swizzle != BRW_SWIZZLE_XYZW)
       return false;
 
+   /* We can't swizzle implicit accumulator access.  We'd have to
+    * reswizzle the producer of the accumulator value in addition
+    * to the consumer (i.e. both MUL and MACH).  Just skip this.
+    */
+   if (reads_accumulator_implicitly())
+      return false;
+
    if (!can_do_writemask(devinfo) && dst_writemask != WRITEMASK_XYZW)
       return false;
 

src/mesa/main/api_validate.c

@@ -795,25 +795,10 @@ _mesa_validate_DrawRangeElements(struct gl_context *ctx, GLenum mode,
                                        "glDrawRangeElements");
 }
 
+
 static bool
-validate_draw_arrays(struct gl_context *ctx, const char *func,
-                     GLenum mode, GLsizei count, GLsizei numInstances)
+need_xfb_remaining_prims_check(const struct gl_context *ctx)
 {
-   struct gl_transform_feedback_object *xfb_obj
-      = ctx->TransformFeedback.CurrentObject;
-   FLUSH_CURRENT(ctx, 0);
-
-   if (count < 0) {
-      _mesa_error(ctx, GL_INVALID_VALUE, "%s(count)", func);
-      return false;
-   }
-
-   if (!_mesa_valid_prim_mode(ctx, mode, func))
-      return false;
-
-   if (!check_valid_to_render(ctx, func))
-      return false;
-
    /* From the GLES3 specification, section 2.14.2 (Transform Feedback
     * Primitive Capture):
     *
@@ -841,10 +826,33 @@ validate_draw_arrays(struct gl_context *ctx, const char *func,
     *     is removed and replaced with the GL behavior (primitives are not
     *     written and the corresponding counter is not updated)..."
     */
-   if (_mesa_is_gles3(ctx) && _mesa_is_xfb_active_and_unpaused(ctx) &&
-       !_mesa_has_OES_geometry_shader(ctx) &&
-       !_mesa_has_OES_tessellation_shader(ctx)) {
-      size_t prim_count = vbo_count_tessellated_primitives(mode, count, 1);
+   return _mesa_is_gles3(ctx) && _mesa_is_xfb_active_and_unpaused(ctx) &&
+          !_mesa_has_OES_geometry_shader(ctx) &&
+          !_mesa_has_OES_tessellation_shader(ctx);
+}
+
+
+static bool
+validate_draw_arrays(struct gl_context *ctx, const char *func,
+                     GLenum mode, GLsizei count, GLsizei numInstances)
+{
+   FLUSH_CURRENT(ctx, 0);
+
+   if (count < 0) {
+      _mesa_error(ctx, GL_INVALID_VALUE, "%s(count)", func);
+      return false;
+   }
+
+   if (!_mesa_valid_prim_mode(ctx, mode, func))
+      return false;
+
+   if (!check_valid_to_render(ctx, func))
+      return false;
+
+   if (need_xfb_remaining_prims_check(ctx)) {
+      struct gl_transform_feedback_object *xfb_obj
+         = ctx->TransformFeedback.CurrentObject;
+      size_t prim_count = vbo_count_tessellated_primitives(mode, count, numInstances);
       if (xfb_obj->GlesRemainingPrims < prim_count) {
          _mesa_error(ctx, GL_INVALID_OPERATION,
                      "%s(exceeds transform feedback size)", func);
@@ -892,6 +900,60 @@ _mesa_validate_DrawArraysInstanced(struct gl_context *ctx, GLenum mode, GLint fi
 }
 
 
+/**
+ * Called to error check the function parameters.
+ *
+ * Note that glMultiDrawArrays is not part of GLES, so there's limited scope
+ * for sharing code with the validation of glDrawArrays.
+ */
+bool
+_mesa_validate_MultiDrawArrays(struct gl_context *ctx, GLenum mode,
+                               const GLsizei *count, GLsizei primcount)
+{
+   int i;
+
+   FLUSH_CURRENT(ctx, 0);
+
+   if (!_mesa_valid_prim_mode(ctx, mode, "glMultiDrawArrays"))
+      return false;
+
+   if (!check_valid_to_render(ctx, "glMultiDrawArrays"))
+      return false;
+
+   if (primcount < 0) {
+      _mesa_error(ctx, GL_INVALID_VALUE, "glMultiDrawArrays(primcount=%d)",
+                  primcount);
+      return false;
+   }
+
+   for (i = 0; i < primcount; ++i) {
+      if (count[i] < 0) {
+         _mesa_error(ctx, GL_INVALID_VALUE, "glMultiDrawArrays(count[%d]=%d)",
+                     i, count[i]);
+         return false;
+      }
+   }
+
+   if (need_xfb_remaining_prims_check(ctx)) {
+      struct gl_transform_feedback_object *xfb_obj
+         = ctx->TransformFeedback.CurrentObject;
+      size_t xfb_prim_count = 0;
+
+      for (i = 0; i < primcount; ++i)
+         xfb_prim_count += vbo_count_tessellated_primitives(mode, count[i], 1);
+
+      if (xfb_obj->GlesRemainingPrims < xfb_prim_count) {
+         _mesa_error(ctx, GL_INVALID_OPERATION,
+                     "glMultiDrawArrays(exceeds transform feedback size)");
+         return false;
+      }
+      xfb_obj->GlesRemainingPrims -= xfb_prim_count;
+   }
+
+   return true;
+}
+
+
 GLboolean
 _mesa_validate_DrawElementsInstanced(struct gl_context *ctx,
                                      GLenum mode, GLsizei count, GLenum type,

src/mesa/main/api_validate.h

@@ -48,6 +48,10 @@ _mesa_valid_prim_mode(struct gl_context *ctx, GLenum mode, const char *name);
 extern GLboolean
 _mesa_validate_DrawArrays(struct gl_context *ctx, GLenum mode, GLsizei count);
 
+extern bool
+_mesa_validate_MultiDrawArrays(struct gl_context *ctx, GLenum mode,
+                               const GLsizei *count, GLsizei primcount);
+
 extern GLboolean
 _mesa_validate_DrawElements(struct gl_context *ctx,
 			    GLenum mode, GLsizei count, GLenum type,

src/mesa/main/uniform_query.cpp

@@ -856,6 +856,8 @@ _mesa_uniform(struct gl_context *ctx, struct gl_shader_program *shProg,
     */
    if (uni->type->is_sampler()) {
       bool flushed = false;
+      shProg->SamplersValidated = GL_TRUE;
+
       for (int i = 0; i < MESA_SHADER_STAGES; i++) {
 	 struct gl_linked_shader *const sh = shProg->_LinkedShaders[i];
 

src/mesa/main/uniforms.c

@@ -68,15 +68,14 @@ _mesa_update_shader_textures_used(struct gl_shader_program *shProg,
 				  struct gl_program *prog)
 {
    GLbitfield mask = prog->SamplersUsed;
-   struct gl_linked_shader *shader =
-      shProg->_LinkedShaders[_mesa_program_enum_to_shader_stage(prog->Target)];
+   gl_shader_stage prog_stage =
+      _mesa_program_enum_to_shader_stage(prog->Target);
+   struct gl_linked_shader *shader = shProg->_LinkedShaders[prog_stage];
 
    assert(shader);
 
    memset(prog->TexturesUsed, 0, sizeof(prog->TexturesUsed));
 
-   shProg->SamplersValidated = GL_TRUE;
-
    while (mask) {
       const int s = u_bit_scan(&mask);
       GLuint unit = prog->SamplerUnits[s];
@@ -92,8 +91,20 @@ _mesa_update_shader_textures_used(struct gl_shader_program *shProg,
        *     types pointing to the same texture image unit within a program
        *     object."
        */
-      if (prog->TexturesUsed[unit] & ~(1 << tgt))
-         shProg->SamplersValidated = GL_FALSE;
+      unsigned stages_mask = shProg->data->linked_stages;
+      while (stages_mask) {
+         const int stage = u_bit_scan(&stages_mask);
+
+         /* Skip validation if we are yet to update textures used in this
+          * stage.
+          */
+         if (prog_stage < stage)
+            break;
+
+         struct gl_program *glprog = shProg->_LinkedShaders[stage]->Program;
+         if (glprog->TexturesUsed[unit] & ~(1 << tgt))
+            shProg->SamplersValidated = GL_FALSE;
+      }
 
       prog->TexturesUsed[unit] |= (1 << tgt);
    }

src/mesa/main/varray.c

@@ -1546,24 +1546,6 @@ _mesa_UnlockArraysEXT( void )
 }
 
 
-/* GL_EXT_multi_draw_arrays */
-void GLAPIENTRY
-_mesa_MultiDrawArrays( GLenum mode, const GLint *first,
-                          const GLsizei *count, GLsizei primcount )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   GLint i;
-
-   FLUSH_VERTICES(ctx, 0);
-
-   for (i = 0; i < primcount; i++) {
-      if (count[i] > 0) {
-         CALL_DrawArrays(ctx->CurrentDispatch, (mode, first[i], count[i]));
-      }
-   }
-}
-
-
 /* GL_IBM_multimode_draw_arrays */
 void GLAPIENTRY
 _mesa_MultiModeDrawArraysIBM( const GLenum * mode, const GLint * first,

src/mesa/program/ir_to_mesa.cpp

@@ -3060,6 +3060,11 @@ _mesa_glsl_link_shader(struct gl_context *ctx, struct gl_shader_program *prog)
    }
 
    if (prog->data->LinkStatus) {
+      /* Reset sampler validated to true, validation happens via the
+       * LinkShader call below.
+       */
+      prog->SamplersValidated = GL_TRUE;
+
       if (!ctx->Driver.LinkShader(ctx, prog)) {
          prog->data->LinkStatus = GL_FALSE;
       }

src/mesa/state_tracker/st_draw.c

@@ -301,6 +301,8 @@ st_indirect_draw_vbo(struct gl_context *ctx,
    assert(ctx->NewState == 0x0);
    assert(stride);
 
+   st_invalidate_readpix_cache(st);
+
    /* Validate state. */
    if ((st->dirty | ctx->NewDriverState) & ST_PIPELINE_RENDER_STATE_MASK ||
        st->gfx_shaders_may_be_dirty) {

src/mesa/vbo/vbo_exec_array.c

@@ -404,7 +404,8 @@ vbo_bind_arrays(struct gl_context *ctx)
  */
 static void
 vbo_draw_arrays(struct gl_context *ctx, GLenum mode, GLint start,
-                GLsizei count, GLuint numInstances, GLuint baseInstance)
+                GLsizei count, GLuint numInstances, GLuint baseInstance,
+                GLuint drawID)
 {
    struct vbo_context *vbo = vbo_context(ctx);
    struct _mesa_prim prim[2];
@@ -418,6 +419,7 @@ vbo_draw_arrays(struct gl_context *ctx, GLenum mode, GLint start,
    prim[0].mode = mode;
    prim[0].num_instances = numInstances;
    prim[0].base_instance = baseInstance;
+   prim[0].draw_id = drawID;
    prim[0].is_indirect = 0;
 
    /* Implement the primitive restart index */
@@ -614,7 +616,7 @@ vbo_exec_DrawArrays(GLenum mode, GLint start, GLsizei count)
    if (0)
       check_draw_arrays_data(ctx, start, count);
 
-   vbo_draw_arrays(ctx, mode, start, count, 1, 0);
+   vbo_draw_arrays(ctx, mode, start, count, 1, 0, 0);
 
    if (0)
       print_draw_arrays(ctx, mode, start, count);
@@ -642,7 +644,7 @@ vbo_exec_DrawArraysInstanced(GLenum mode, GLint start, GLsizei count,
    if (0)
       check_draw_arrays_data(ctx, start, count);
 
-   vbo_draw_arrays(ctx, mode, start, count, numInstances, 0);
+   vbo_draw_arrays(ctx, mode, start, count, numInstances, 0, 0);
 
    if (0)
       print_draw_arrays(ctx, mode, start, count);
@@ -672,13 +674,52 @@ vbo_exec_DrawArraysInstancedBaseInstance(GLenum mode, GLint first,
    if (0)
       check_draw_arrays_data(ctx, first, count);
 
-   vbo_draw_arrays(ctx, mode, first, count, numInstances, baseInstance);
+   vbo_draw_arrays(ctx, mode, first, count, numInstances, baseInstance, 0);
 
    if (0)
       print_draw_arrays(ctx, mode, first, count);
 }
 
 
+/**
+ * Called from glMultiDrawArrays when in immediate mode.
+ */
+static void GLAPIENTRY
+vbo_exec_MultiDrawArrays(GLenum mode, const GLint *first,
+                         const GLsizei *count, GLsizei primcount)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   GLint i;
+
+   if (MESA_VERBOSE & VERBOSE_DRAW)
+      _mesa_debug(ctx,
+                  "glMultiDrawArrays(%s, %p, %p, %d)\n",
+                  _mesa_enum_to_string(mode), first, count, primcount);
+
+   if (!_mesa_validate_MultiDrawArrays(ctx, mode, count, primcount))
+      return;
+
+   for (i = 0; i < primcount; i++) {
+      if (count[i] > 0) {
+         if (0)
+            check_draw_arrays_data(ctx, first[i], count[i]);
+
+         /* The GL_ARB_shader_draw_parameters spec adds the following after the
+          * pseudo-code describing glMultiDrawArrays:
+          *
+          *    "The index of the draw (<i> in the above pseudo-code) may be
+          *     read by a vertex shader as <gl_DrawIDARB>, as described in
+          *     Section 11.1.3.9."
+          */
+         vbo_draw_arrays(ctx, mode, first[i], count[i], 1, 0, i);
+
+         if (0)
+            print_draw_arrays(ctx, mode, first[i], count[i]);
+      }
+   }
+}
+
+
 
 /**
  * Map GL_ELEMENT_ARRAY_BUFFER and print contents.
@@ -1272,7 +1313,7 @@ vbo_draw_transform_feedback(struct gl_context *ctx, GLenum mode,
         !_mesa_all_varyings_in_vbos(ctx->Array.VAO))) {
       GLsizei n =
          ctx->Driver.GetTransformFeedbackVertexCount(ctx, obj, stream);
-      vbo_draw_arrays(ctx, mode, 0, n, numInstances, 0);
+      vbo_draw_arrays(ctx, mode, 0, n, numInstances, 0, 0);
       return;
    }
 
@@ -1683,6 +1724,7 @@ vbo_initialize_exec_dispatch(const struct gl_context *ctx,
       SET_DrawRangeElements(exec, vbo_exec_DrawRangeElements);
    }
 
+   SET_MultiDrawArrays(exec, vbo_exec_MultiDrawArrays);
    SET_MultiDrawElementsEXT(exec, vbo_exec_MultiDrawElements);
 
    if (ctx->API == API_OPENGL_COMPAT) {

src/mesa/vbo/vbo_save_api.c

@@ -1176,6 +1176,40 @@ _save_OBE_DrawArrays(GLenum mode, GLint start, GLsizei count)
 }
 
 
+static void GLAPIENTRY
+_save_OBE_MultiDrawArrays(GLenum mode, const GLint *first,
+                          const GLsizei *count, GLsizei primcount)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   GLint i;
+
+   if (!_mesa_is_valid_prim_mode(ctx, mode)) {
+      _mesa_compile_error(ctx, GL_INVALID_ENUM, "glMultiDrawArrays(mode)");
+      return;
+   }
+
+   if (primcount < 0) {
+      _mesa_compile_error(ctx, GL_INVALID_VALUE,
+                          "glMultiDrawArrays(primcount<0)");
+      return;
+   }
+
+   for (i = 0; i < primcount; i++) {
+      if (count[i] < 0) {
+         _mesa_compile_error(ctx, GL_INVALID_VALUE,
+                             "glMultiDrawArrays(count[i]<0)");
+         return;
+      }
+   }
+
+   for (i = 0; i < primcount; i++) {
+      if (count[i] > 0) {
+         _save_OBE_DrawArrays(mode, first[i], count[i]);
+      }
+   }
+}
+
+
 /* Could do better by copying the arrays and element list intact and
  * then emitting an indexed prim at runtime.
  */
@@ -1484,6 +1518,7 @@ vbo_initialize_save_dispatch(const struct gl_context *ctx,
                              struct _glapi_table *exec)
 {
    SET_DrawArrays(exec, _save_OBE_DrawArrays);
+   SET_MultiDrawArrays(exec, _save_OBE_MultiDrawArrays);
    SET_DrawElements(exec, _save_OBE_DrawElements);
    SET_DrawElementsBaseVertex(exec, _save_OBE_DrawElementsBaseVertex);
    SET_DrawRangeElements(exec, _save_OBE_DrawRangeElements);