pocl: update to 1.8.

Notable User Facing Changes
---------------------------

- support for LLVM 13
- CMake: Inter-Procedural Optimization is enabled on code of runtime library
  (libpocl.so is compiled with -flto on systems that support it).
- LTTng tracing improved - more command types are traced, and also
  some synchronous API calls (like clCreateBuffer) are traced.
- poclcc, tests and examples can be disabled with CMake options
- Valgrind support improved by making Valgrind aware of pocl's
  reference counting of cl_* objects
- kernels which are called by kernels are now force-inlined
- Support for NetBSD.
- Support for Unix systems without libdl.
- PoCL can now (optionally) respond to SIGUSR2 by printing
  some live debug information.
- improved SPIR support for CUDA devices

Notable Bug Fixes
-----------------

- Fixed a potential crash on Unix systems without sysfs mounted.
- Fixed compilation errors when building on macOS.
  - Fixed POCL_FAST_INIT macro; POCL_INIT_LOCK must be invoked with only one argument.
  - Fix bin/poclcc to not depend on OpenCL 2.0 symbols
- Fixed miscompilation in kernel loops with multiple conditionals with barriers in them.

Other
-----
- Add cmake options PARALLEL_COMPILE_JOBS, PARALLEL_LINK_JOBS to
  use ninja's seperate compile and link job pools.

- Improve memory architecture, buffer migration and allocation.
  Buffers are now allocated on a device when first used
  (previously each buffer was allocated on every device in context).

- the single global LLVMContext was replaced with
  multiple LLVMContexts, one per OpenCL cl_context.
  OpenCL code can now be compiled in parallel
  when using separate cl_contexts. This feature
  is disabled by default since it significantly slowed
  down PyOpenCL. This should be resolved by separating
  LLVM compilation in their own threads in the future.

- a new OpenCL extension was added to PoCL: cl_pocl_content_size.
  The extension allows the user to give optimization hint to PoCL,
  which will be used internally by PoCL to optimize buffer transfers
  between multiple devices.

11 files changed, 10 insertions, 229 deletions

diff --git a/parallel/pocl/Makefile b/parallel/pocl/Makefile
index 9285e49fd7f..7d6267f9a31 100644
--- a/parallel/pocl/Makefile
+++ b/parallel/pocl/Makefile
@@ -1,7 +1,6 @@
-# $NetBSD: Makefile,v 1.5 2021/09/20 09:05:30 nia Exp $
+# $NetBSD: Makefile,v 1.6 2021/11/20 20:05:36 wiz Exp $
 
-DISTNAME=	pocl-1.7
-PKGREVISION=	2
+DISTNAME=	pocl-1.8
 CATEGORIES=	parallel
 MASTER_SITES=	http://portablecl.org/downloads/
 
diff --git a/parallel/pocl/distinfo b/parallel/pocl/distinfo
index b8e036dd16e..54d0eb96609 100644
--- a/parallel/pocl/distinfo
+++ b/parallel/pocl/distinfo
@@ -1,14 +1,6 @@
-$NetBSD: distinfo,v 1.5 2021/10/26 11:10:37 nia Exp $
+$NetBSD: distinfo,v 1.6 2021/11/20 20:05:36 wiz Exp $
 
-BLAKE2s (pocl-1.7.tar.gz) = b111e491fb84465f30bf56bb78c0bc05f1be26d7954c0891fda2e6990669e7cd
-SHA512 (pocl-1.7.tar.gz) = 23bb82b25a222186c001b2ed413f26253e1e0e16518e98c95173f51491e15d107e6ab1b334ffd490b41199743267d9ad64882a22fea327e17ec9604f9847fc38
-Size (pocl-1.7.tar.gz) = 1577676 bytes
-SHA1 (patch-CMakeLists.txt) = d722efb971d9ce35fd7dfe2e238cbce3660e1eb8
-SHA1 (patch-config.h.in.cmake) = 8f15bf68b3a2c2af00bc8b72f52b383182db6168
-SHA1 (patch-lib_CL_devices_basic_basic.c) = 0220af7515718d44eea38d3a14844c68f75acab4
-SHA1 (patch-lib_CL_devices_common.c) = 44717bf410b948f9b6cd72cbea83dd9214faaa5c
-SHA1 (patch-lib_CL_devices_cpuinfo.c) = 30d8a1103d9eef6fb5b89c1b604dd953828b605b
-SHA1 (patch-lib_CL_devices_devices.c) = 17a5bb001086befdf88452b7e1126d2f5f4b1fc1
-SHA1 (patch-lib_CL_devices_hsa_pocl-hsa.c) = 27acb108cb1e12e4ac0c82e497916f841edc8cb1
-SHA1 (patch-lib_CL_devices_pthread_pthread.c) = 9f169b9a0a1d70ef3efc52dcd7b47a70c364d5d2
-SHA1 (patch-lib_CL_pocl__timing.c) = 6f82ad0eaf22d511745b47bf26f3024f1a25cf6f
+BLAKE2s (pocl-1.8.tar.gz) = 98983933cf43c9010352ecc48a539bc068bc5d1fd6a2a0f37b2973d384a1e3e0
+SHA512 (pocl-1.8.tar.gz) = bcbb3fa3d2234d4c5b0c17863eba0bc4c8f13f863cc58cfd1de49e21fa7bf0aec82b81aec143c81885e3a39274c8ae783b2f03b9a12846e024204d6ed0e59a9d
+Size (pocl-1.8.tar.gz) = 1614545 bytes
+SHA1 (patch-lib_CL_devices_devices.c) = 59824f5f30083651cff47340be0895148836aee3
diff --git a/parallel/pocl/patches/patch-CMakeLists.txt b/parallel/pocl/patches/patch-CMakeLists.txt
deleted file mode 100644
index 7c217fdbc73..00000000000
--- a/parallel/pocl/patches/patch-CMakeLists.txt
+++ /dev/null
@@ -1,33 +0,0 @@
-$NetBSD: patch-CMakeLists.txt,v 1.2 2021/06/10 12:26:26 nia Exp $
-
-Fix detection of dlopen
-https://github.com/pocl/pocl/pull/950
-
---- CMakeLists.txt.orig	2021-05-19 08:12:19.000000000 +0000
-+++ CMakeLists.txt
-@@ -787,9 +787,7 @@ elseif(UNIX)
- 
-     find_library(DL_LIB "dl")
-     find_file(DL_H "dlfcn.h")
--    if(DL_LIB AND DL_H)
--      message(STATUS "libdl found")
--
-+    if(DL_H)
-       get_filename_component(DL_H_INCLUDE_DIR "${DL_H}" DIRECTORY)
-       string(FIND "${CMAKE_C_IMPLICIT_INCLUDE_DIRECTORIES}" "${DL_H_INCLUDE_DIR}" LTPOSITION)
-       # include the directory of dlfcn.h, if its not in the default system include dirs
-@@ -797,9 +795,11 @@ elseif(UNIX)
-       if((LTPOSITION LESS "0") AND (NOT CMAKE_CROSSCOMPILING))
-         include_directories("${DL_H_INCLUDE_DIR}")
-       endif()
--      set(HAVE_LIBDL ON CACHE BOOL "dlopen" FORCE)
--    else()
--      message(FATAL_ERROR "Could not find DL library!")
-+      set(HAVE_DLFCN_H ON)
-+    endif()
-+
-+    if(NOT DL_LIB)
-+      set(DL_LIB "")
-     endif()
- 
- else()
diff --git a/parallel/pocl/patches/patch-config.h.in.cmake b/parallel/pocl/patches/patch-config.h.in.cmake
deleted file mode 100644
index 7611e097fb2..00000000000
--- a/parallel/pocl/patches/patch-config.h.in.cmake
+++ /dev/null
@@ -1,25 +0,0 @@
-$NetBSD: patch-config.h.in.cmake,v 1.2 2021/06/10 12:26:26 nia Exp $
-
-Fix detection of dlopen
-https://github.com/pocl/pocl/pull/950
-
---- config.h.in.cmake.orig	2021-05-19 08:12:19.000000000 +0000
-+++ config.h.in.cmake
-@@ -35,6 +35,8 @@
- 
- #cmakedefine ENABLE_SPIRV
- 
-+#cmakedefine HAVE_DLFCN_H
-+
- #cmakedefine HAVE_FORK
- 
- #cmakedefine HAVE_VFORK
-@@ -57,8 +59,6 @@
- 
- #cmakedefine HAVE_LTTNG_UST
- 
--#cmakedefine HAVE_LIBDL
--
- #cmakedefine HAVE_OCL_ICD
- 
- #cmakedefine HAVE_POSIX_MEMALIGN
diff --git a/parallel/pocl/patches/patch-lib_CL_devices_basic_basic.c b/parallel/pocl/patches/patch-lib_CL_devices_basic_basic.c
deleted file mode 100644
index 413450090e0..00000000000
--- a/parallel/pocl/patches/patch-lib_CL_devices_basic_basic.c
+++ /dev/null
@@ -1,18 +0,0 @@
-$NetBSD: patch-lib_CL_devices_basic_basic.c,v 1.2 2021/06/10 12:26:26 nia Exp $
-
-Fix detection of dlopen
-https://github.com/pocl/pocl/pull/950
-
---- lib/CL/devices/basic/basic.c.orig	2021-05-19 08:12:19.000000000 +0000
-+++ lib/CL/devices/basic/basic.c
-@@ -50,8 +50,8 @@
- #include "pocl_llvm.h"
- #endif
- 
--#ifndef HAVE_LIBDL
--#error Basic driver requires DL library
-+#ifndef HAVE_DLFCN_H
-+#error Basic driver requires dlopen
- #endif
- 
- struct data {
diff --git a/parallel/pocl/patches/patch-lib_CL_devices_common.c b/parallel/pocl/patches/patch-lib_CL_devices_common.c
deleted file mode 100644
index 002fb0f9831..00000000000
--- a/parallel/pocl/patches/patch-lib_CL_devices_common.c
+++ /dev/null
@@ -1,16 +0,0 @@
-$NetBSD: patch-lib_CL_devices_common.c,v 1.2 2021/06/10 12:26:26 nia Exp $
-
-Fix detection of dlopen
-https://github.com/pocl/pocl/pull/950
-
---- lib/CL/devices/common.c.orig	2021-05-19 08:12:19.000000000 +0000
-+++ lib/CL/devices/common.c
-@@ -61,7 +61,7 @@
- #include <unistd.h>
- #endif
- 
--#ifdef HAVE_LIBDL
-+#ifdef HAVE_DLFCN_H
- #if defined(__APPLE__)
- #define _DARWIN_C_SOURCE
- #endif
diff --git a/parallel/pocl/patches/patch-lib_CL_devices_cpuinfo.c b/parallel/pocl/patches/patch-lib_CL_devices_cpuinfo.c
deleted file mode 100644
index aafd9c76126..00000000000
--- a/parallel/pocl/patches/patch-lib_CL_devices_cpuinfo.c
+++ /dev/null
@@ -1,24 +0,0 @@
-$NetBSD: patch-lib_CL_devices_cpuinfo.c,v 1.2 2021/06/10 12:26:26 nia Exp $
-
-Do not segfault if /sys/bus/pci/devices/0000:00:00.0/vendor does not exist.
-
-https://github.com/pocl/pocl/pull/948
-
---- lib/CL/devices/cpuinfo.c.orig	2021-05-19 08:12:19.000000000 +0000
-+++ lib/CL/devices/cpuinfo.c
-@@ -416,9 +416,12 @@ pocl_cpuinfo_get_cpu_name_and_vendor(cl_
-   if (!device->vendor_id)
-     {
-       f = fopen (pci_bus_root_vendor_file, "r");
--      num_read = fscanf (f, "%x", &device->vendor_id);
--      fclose (f);
--      /* no error checking, if it failed we just won't have the info */
-+      if (f)
-+        {
-+          /* no error checking, if it failed we just won't have the info */
-+          num_read = fscanf (f, "%x", &device->vendor_id);
-+          fclose (f);
-+        }
-     }
- }
- 
diff --git a/parallel/pocl/patches/patch-lib_CL_devices_devices.c b/parallel/pocl/patches/patch-lib_CL_devices_devices.c
index 15159773762..94422d2d84b 100644
--- a/parallel/pocl/patches/patch-lib_CL_devices_devices.c
+++ b/parallel/pocl/patches/patch-lib_CL_devices_devices.c
@@ -1,8 +1,8 @@
-$NetBSD: patch-lib_CL_devices_devices.c,v 1.1 2021/06/10 09:57:40 nia Exp $
+$NetBSD: patch-lib_CL_devices_devices.c,v 1.2 2021/11/20 20:05:37 wiz Exp $
 
 Add support for non-Linux unixes.
 
---- lib/CL/devices/devices.c.orig	2021-05-19 08:12:19.000000000 +0000
+--- lib/CL/devices/devices.c.orig	2021-10-12 14:33:15.000000000 +0000
 +++ lib/CL/devices/devices.c
 @@ -27,7 +27,7 @@
  #include <string.h>
@@ -13,54 +13,12 @@ Add support for non-Linux unixes.
  #include <limits.h>
  #include <signal.h>
  #include <stdio.h>
-@@ -70,7 +70,7 @@
- #define PATH_MAX 4096
+@@ -469,7 +469,7 @@ pocl_init_devices ()
  #endif
  
--#ifdef HAVE_LIBDL
-+#ifdef HAVE_DLFCN_H
- #if defined(__APPLE__)
- #define _DARWIN_C_SOURCE
- #endif
-@@ -303,7 +303,7 @@ str_toupper(char *out, const char *in)
-  */
  
- #ifdef ENABLE_HOST_CPU_DEVICES
 -#ifdef __linux__
 +#if !defined(_WIN32) && !defined(__APPLE__)
- #ifdef __x86_64__
- 
- #define DIV_OPCODE_SIZE 1
-@@ -335,7 +335,11 @@ sigfpe_signal_handler (int signo, siginf
- {
-   ucontext_t *uc;
-   uc = (ucontext_t *)data;
-+#ifdef __NetBSD__
-+  unsigned char *eip = (unsigned char *)(uc->uc_mcontext.__gregs[_REG_RIP]);
-+#else
-   unsigned char *eip = (unsigned char *)(uc->uc_mcontext.gregs[REG_RIP]);
-+#endif
- 
-   if ((signo == SIGFPE)
-       && ((si->si_code == FPE_INTDIV) || (si->si_code == FPE_INTOVF)))
-@@ -384,7 +388,11 @@ sigfpe_signal_handler (int signo, siginf
-             n += IP_RELATIVE_INDEXING;
-         }
- 
-+#ifdef __NetBSD__
-+      uc->uc_mcontext.__gregs[_REG_RIP] += n;
-+#else
-       uc->uc_mcontext.gregs[REG_RIP] += n;
-+#endif
-       return;
-     }
-   else
-@@ -546,7 +554,7 @@ pocl_init_devices ()
- 
  
  #ifdef ENABLE_HOST_CPU_DEVICES
--#ifdef __linux__
-+#if !defined(_WIN32) && !defined(__APPLE__)
- #ifdef __x86_64__
- 
    if (pocl_get_bool_option ("POCL_SIGFPE_HANDLER", 1))
diff --git a/parallel/pocl/patches/patch-lib_CL_devices_hsa_pocl-hsa.c b/parallel/pocl/patches/patch-lib_CL_devices_hsa_pocl-hsa.c
deleted file mode 100644
index 6fb8116f9ea..00000000000
--- a/parallel/pocl/patches/patch-lib_CL_devices_hsa_pocl-hsa.c
+++ /dev/null
@@ -1,18 +0,0 @@
-$NetBSD: patch-lib_CL_devices_hsa_pocl-hsa.c,v 1.2 2021/06/10 12:26:26 nia Exp $
-
-Fix detection of dlopen
-https://github.com/pocl/pocl/pull/950
-
---- lib/CL/devices/hsa/pocl-hsa.c.orig	2021-05-19 08:12:19.000000000 +0000
-+++ lib/CL/devices/hsa/pocl-hsa.c
-@@ -73,8 +73,8 @@
- 
- #endif
- 
--#ifndef HAVE_LIBDL
--#error HSA driver requires DL library
-+#ifndef HAVE_DLFCN_H
-+#error HSA driver requires dlopen
- #endif
- 
- #include "pocl-hsa.h"
diff --git a/parallel/pocl/patches/patch-lib_CL_devices_pthread_pthread.c b/parallel/pocl/patches/patch-lib_CL_devices_pthread_pthread.c
deleted file mode 100644
index 5054dc05bbb..00000000000
--- a/parallel/pocl/patches/patch-lib_CL_devices_pthread_pthread.c
+++ /dev/null
@@ -1,18 +0,0 @@
-$NetBSD: patch-lib_CL_devices_pthread_pthread.c,v 1.2 2021/06/10 12:26:26 nia Exp $
-
-Fix detection of dlopen
-https://github.com/pocl/pocl/pull/950
-
---- lib/CL/devices/pthread/pthread.c.orig	2021-05-19 08:12:19.000000000 +0000
-+++ lib/CL/devices/pthread/pthread.c
-@@ -51,8 +51,8 @@
- #include "pocl_util.h"
- #include "pocl_mem_management.h"
- 
--#ifndef HAVE_LIBDL
--#error Pthread driver requires DL library
-+#ifndef HAVE_DLFCN_H
-+#error Pthread driver requires dlopen
- #endif
- 
- #ifdef OCS_AVAILABLE
diff --git a/parallel/pocl/patches/patch-lib_CL_pocl__timing.c b/parallel/pocl/patches/patch-lib_CL_pocl__timing.c
deleted file mode 100644
index e3af1eb65b1..00000000000
--- a/parallel/pocl/patches/patch-lib_CL_pocl__timing.c
+++ /dev/null
@@ -1,16 +0,0 @@
-$NetBSD: patch-lib_CL_pocl__timing.c,v 1.2 2021/06/10 12:26:26 nia Exp $
-
-NetBSD does not have CLOCK_UPTIME_FAST.
-https://github.com/pocl/pocl/pull/949
-
---- lib/CL/pocl_timing.c.orig	2021-05-19 08:12:19.000000000 +0000
-+++ lib/CL/pocl_timing.c
-@@ -74,7 +74,7 @@ uint64_t pocl_gettimemono_ns() {
- #   warning Using clock_gettime with CLOCK_MONOTONIC for monotonic clocks
-   clock_gettime(CLOCK_MONOTONIC, &timespec);
- #  endif
--# elif defined(__DragonFly__) || defined(__FreeBSD__) || defined(__NetBSD__) || defined(__OpenBSD__) || defined(__FreeBSD_kernel__)
-+# elif defined(__DragonFly__) || defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
-   clock_gettime(CLOCK_UPTIME_FAST, &timespec);
- # else
- # warning Using clock_gettime with CLOCK_REALTIME for monotonic clocks