rpm/OpenShadingLanguage
SHA256
1
0
Fork 0
forked from pool/OpenShadingLanguage
Code Pull Requests Activity

Accepting request 926039 from graphics

Browse Source 
OBS-URL: https://build.opensuse.org/request/show/926039
OBS-URL: https://build.opensuse.org/package/show/openSUSE:Factory/OpenShadingLanguage?expand=0&rev=11
This commit is contained in:
Dominique Leuenberger 2021-10-28 16:31:16 +00:00 committed by Git OBS Bridge
commit e5adef7848
6 changed files with 261 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

207
1420.diff Normal file
View File

@ -0,0 +1,207 @@
Index: b/INSTALL.md
===================================================================
--- a/INSTALL.md
+++ b/INSTALL.md
@@ -16,7 +16,7 @@ NEW or CHANGED dependencies since the la
* A suitable C++11 compiler to build OSL itself, which may be any of:
- GCC 4.8.5 or newer (tested through gcc 11)
- - Clang 3.4 or newer (tested through clang 12)
+ - Clang 3.4 or newer (tested through clang 13)
- Microsoft Visual Studio 2015 or newer
- Intel C++ compiler icc version 13 (?) or newer
@@ -43,7 +43,7 @@ NEW or CHANGED dependencies since the la
DYLD_LIBRARY_PATH on OS X) and then OSL's build scripts will be able
to find it.
-* **[LLVM](http://www.llvm.org) 7, 8, 9, 10, 11, or 12**, including
+* **[LLVM](http://www.llvm.org) 7, 8, 9, 10, 11, 12, or 13**, including clang
clang libraries.
Note that LLVM 10+ is not compatible with C++11, and requires C++14 or
@@ -73,13 +73,13 @@ Build process
Here are the steps to check out, build, and test the OSL distribution:
-0. Install and build dependencies.
+1. Install and build dependencies.
-1. Check out a copy of the source code from the Git repository:
+2. Check out a copy of the source code from the Git repository:
git clone https://github.com/AcademySoftwareFoundation/OpenShadingLanguage.git osl
-2. Change to the distribution directory and 'make'
+3. Change to the distribution directory and 'make'
cd osl
make
@@ -95,7 +95,7 @@ Here are the steps to check out, build,
"make STOP_ON_WARNING=0", that create a build that will only stop for
full errors, not warnings.
-3. After compilation, you'll end up with a full OSL distribution in
+4. After compilation, you'll end up with a full OSL distribution in
dist/ARCH, where ARCH is the architecture you are building on, one of
"linux", "linux64", "macosx", "windows", or "windows64".
@@ -103,14 +103,14 @@ Here are the steps to check out, build,
instead type 'make debug' at the top level, you will end up with
a debug build (no optimization, full symbols) in "dist/ARCH.debug".
-4. Add the "dist/ARCH/bin" to your $PATH, and "dist/ARCH/lib" to your
+5. Add the "dist/ARCH/bin" to your $PATH, and "dist/ARCH/lib" to your
$LD_LIBRAY_PATH (or $DYLD_LIBRARY_PATH on OS X), or copy the contents
of those files to appropriate directories. Public include files
(those needed when building applications that incorporate OSL)
can be found in "dist/ARCH/include", and documentation can be found
in "dist/ARCH/share/doc".
-5. After building (and setting your library path), you can run the
+6. After building (and setting your library path), you can run the
test suite with:
make test
Index: b/src/include/OSL/llvm_util.h
===================================================================
--- a/src/include/OSL/llvm_util.h
+++ b/src/include/OSL/llvm_util.h
@@ -577,6 +577,9 @@ public:
llvm::Value *src, int srcalign, int len);
/// Dereference a pointer: return *ptr
+ /// type is the type of the thing being pointed to.
+ llvm::Value *op_load (llvm::Type* type, llvm::Value *ptr);
+ // Blind pointer version that's deprecated as of LLVM13:
llvm::Value *op_load (llvm::Value *ptr);
/// Store to a dereferenced pointer: *ptr = val
@@ -589,17 +592,25 @@ public:
/// Generate a GEP (get element pointer) where the element index is an
/// llvm::Value, which can be generated from either a constant or a
- /// runtime-computed integer element index.
+ /// runtime-computed integer element index. `type` is the type of the data
+ /// we're retrieving.
+ llvm::Value *GEP (llvm::Type* type, llvm::Value *ptr, llvm::Value *elem);
+ // Blind pointer version that's deprecated as of LLVM13:
llvm::Value *GEP (llvm::Value *ptr, llvm::Value *elem);
/// Generate a GEP (get element pointer) with an integer element
- /// offset.
+ /// offset. `type` is the type of the data we're retrieving.
+ llvm::Value *GEP (llvm::Type* type, llvm::Value *ptr, int elem);
+ // Blind pointer version that's deprecated as of LLVM13:
llvm::Value *GEP (llvm::Value *ptr, int elem);
/// Generate a GEP (get element pointer) with two integer element
/// offsets. This is just a special (and common) case of GEP where
/// we have a 2-level hierarchy and we have fixed element indices
- /// that are known at compile time.
+ /// that are known at compile time. `type` is the type of the data we're
+ /// retrieving.
+ llvm::Value *GEP (llvm::Type* type, llvm::Value *ptr, int elem1, int elem2);
+ // Blind pointer version that's deprecated as of LLVM13:
llvm::Value *GEP (llvm::Value *ptr, int elem1, int elem2);
// Arithmetic ops. It auto-detects the type (int vs float).
Index: b/src/liboslexec/llvm_util.cpp
===================================================================
--- a/src/liboslexec/llvm_util.cpp
+++ b/src/liboslexec/llvm_util.cpp
@@ -2592,9 +2592,17 @@ LLVM_Util::op_memcpy (llvm::Value *dst,
llvm::Value *
+LLVM_Util::op_load (llvm::Type* type, llvm::Value* ptr)
+{
+ return builder().CreateLoad (type, ptr);
+}
+
+
+
+llvm::Value *
LLVM_Util::op_load (llvm::Value *ptr)
{
- return builder().CreateLoad (ptr);
+ return op_load(ptr->getType()->getPointerElementType(), ptr);
}
@@ -2608,9 +2616,26 @@ LLVM_Util::op_store (llvm::Value *val, l
llvm::Value *
+LLVM_Util::GEP (llvm::Type* type, llvm::Value* ptr, llvm::Value* elem)
+{
+ return builder().CreateGEP(type, ptr, elem);
+}
+
+
+
+llvm::Value *
LLVM_Util::GEP (llvm::Value *ptr, llvm::Value *elem)
{
- return builder().CreateGEP (ptr, elem);
+ return GEP(ptr->getType()->getScalarType()->getPointerElementType(), ptr,
+ elem);
+}
+
+
+
+llvm::Value *
+LLVM_Util::GEP (llvm::Type* type, llvm::Value* ptr, int elem)
+{
+ return builder().CreateConstGEP1_32(type, ptr, elem);
}
@@ -2618,7 +2643,16 @@ LLVM_Util::GEP (llvm::Value *ptr, llvm::
llvm::Value *
LLVM_Util::GEP (llvm::Value *ptr, int elem)
{
- return builder().CreateConstGEP1_32 (ptr, elem);
+ return GEP(ptr->getType()->getScalarType()->getPointerElementType(), ptr,
+ elem);
+}
+
+
+
+llvm::Value *
+LLVM_Util::GEP(llvm::Type* type, llvm::Value* ptr, int elem1, int elem2)
+{
+ return builder().CreateConstGEP2_32 (type, ptr, elem1, elem2);
}
@@ -2626,7 +2660,8 @@ LLVM_Util::GEP (llvm::Value *ptr, int el
llvm::Value *
LLVM_Util::GEP (llvm::Value *ptr, int elem1, int elem2)
{
- return builder().CreateConstGEP2_32 (nullptr, ptr, elem1, elem2);
+ return GEP(ptr->getType()->getScalarType()->getPointerElementType(), ptr,
+ elem1, elem2);
}
Index: b/src/liboslexec/runtimeoptimize.cpp
===================================================================
--- a/src/liboslexec/runtimeoptimize.cpp
+++ b/src/liboslexec/runtimeoptimize.cpp
@@ -2301,7 +2301,6 @@ RuntimeOptimizer::optimize_instance ()
// passes, but we have a hard cutoff just to be sure we don't
// ever get into an infinite loop from an unforseen cycle where we
// end up inadvertently transforming A => B => A => etc.
- int totalchanged = 0;
int reallydone = 0; // Force a few passes after we think we're done
int npasses = shadingsys().opt_passes();
for (m_pass = 0; m_pass < npasses; ++m_pass) {
@@ -2362,7 +2361,6 @@ RuntimeOptimizer::optimize_instance ()
// If nothing changed, we're done optimizing. But wait, it may be
// that after re-tracking variable lifetimes, we can notice new
// optimizations! So force another pass, then we're really done.
- totalchanged += changed;
if (changed < 1) {
if (++reallydone > 3)
break;

33
8682211d0bfe5c4be63a4a003d06037ff9721e66.diff Normal file
View File

@ -0,0 +1,33 @@
diff --git a/src/liboslexec/llvm_util.cpp b/src/liboslexec/llvm_util.cpp
index 4bd0dca35..4d0e1752d 100644
--- a/src/liboslexec/llvm_util.cpp
+++ b/src/liboslexec/llvm_util.cpp
@@ -1407,7 +1407,9 @@ LLVM_Util::make_jit_execengine (std::string *err,
options.NoZerosInBSS = false;
options.GuaranteedTailCallOpt = false;
+#if OSL_LLVM_VERSION < 120
options.StackAlignmentOverride = 0;
+#endif
options.FunctionSections = true;
options.UseInitArray = false;
options.FloatABIType = llvm::FloatABI::Default;
@@ -5385,7 +5387,7 @@ void
LLVM_Util::write_bitcode_file (const char *filename, std::string *err)
{
std::error_code local_error;
- llvm::raw_fd_ostream out (filename, local_error, llvm::sys::fs::F_None);
+ llvm::raw_fd_ostream out (filename, local_error, llvm::sys::fs::OF_None);
if (! out.has_error()) {
llvm::WriteBitcodeToFile (*module(), out);
if (err && local_error)
@@ -5447,7 +5449,9 @@ LLVM_Util::ptx_compile_group (llvm::Module* lib_module, const std::string& name,
options.AllowFPOpFusion = llvm::FPOpFusion::Fast;
options.NoZerosInBSS = 0;
options.GuaranteedTailCallOpt = 0;
+#if OSL_LLVM_VERSION < 120
options.StackAlignmentOverride = 0;
+#endif
options.UseInitArray = 0;
llvm::TargetMachine* target_machine = llvm_target->createTargetMachine(

3
OpenShadingLanguage-1.11.14.2.tar.gz
View File

@ -1,3 +0,0 @@
version https://git-lfs.github.com/spec/v1
oid sha256:6dea570686740a5d5c81e595020f3fef5b8af7201a95460024c17297fb7eb718
size 14143213

3
OpenShadingLanguage-1.11.15.0.tar.gz Normal file
View File

@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:62482b4f1751cd209ad0dd82c2d1505b4921e851fc2a0eb105aba31a03bef9c3
size 14143075

12
OpenShadingLanguage.changes
View File

@ -1,3 +1,15 @@
-------------------------------------------------------------------
Sun Oct 17 15:57:25 UTC 2021 - Hans-Peter Jansen <hpj@urpla.net>
- Update to version 1.11.15.0
+ Changes for compatibility with OpenImageIO 2.3. #1393 #1388
+ Fix potential crashes (or at least undefined behavior) due to
missing
+ virtual destructor of internal Symbol type. #1397
- Apply 8682211d0bfe5c4be63a4a003d06037ff9721e66.diff and 1420.diff
in order to build with LLVM 13.
- Add partio library dependency
-------------------------------------------------------------------
Tue Aug 31 14:36:55 UTC 2021 - Hans-Peter Jansen <hpj@urpla.net>

7
OpenShadingLanguage.spec
View File

@ -23,7 +23,7 @@
%define oiio_major_minor_ver %(rpm -q --queryformat='%%{version}' OpenImageIO-devel | cut -d . -f 1-2)
Name: OpenShadingLanguage
Version: 1.11.14.2
Version: 1.11.15.0
Release: 0
Summary: A language for programmable shading
License: BSD-3-Clause
@ -31,6 +31,10 @@ Group: Productivity/Graphics/Other
URL: https://github.com/AcademySoftwareFoundation/OpenShadingLanguage
Source0: https://github.com/AcademySoftwareFoundation/OpenShadingLanguage/archive/v%{version}.tar.gz#/%{name}-%{version}.tar.gz
Source1: https://creativecommons.org/licenses/by/3.0/legalcode.txt#/CC-BY-3.0.txt
# PATCH-FIX-UPSTREAM: 8682211d0bfe5c4be63a4a003d06037ff9721e66.diff - fix build with LLVM >= 12
Patch0: 8682211d0bfe5c4be63a4a003d06037ff9721e66.diff
# PATCH-FIX-UPSTREAM: 1420.diff - fix build with LLVM >= 13
Patch1: 1420.diff
BuildRequires: OpenEXR-devel
BuildRequires: OpenImageIO
BuildRequires: bison
@ -41,6 +45,7 @@ BuildRequires: gcc-c++
BuildRequires: libboost_filesystem-devel
BuildRequires: libboost_system-devel
BuildRequires: libboost_thread-devel
BuildRequires: partio-devel
BuildRequires: pkg-config
BuildRequires: python3-devel
BuildRequires: python3-pybind11-devel