The reason for the error is that the Lock pages in memory policy did not have any allowed users. As documented at AdjustTokenPrivileges, use the Local Security Policy app to change the security policy.
Lock Pages in Memory Security Policy
Adding a User to the “Lock pages in memory” policy
The output I got once the privilege issue was addressed is in the snippet below. To find these message in the source code, run grep -Rin --include *.cpp "Failed to reserve and commit" ./src/.
OpenJDK 64-Bit Server VM warning: Failed to reserve and commit memory using large pages. req_addr: 0x0000000000000000 bytes: 251658240
[0.249s][info][gc,init] CardTable entry size: 512
OpenJDK 64-Bit Server VM warning: Failed to reserve and commit memory using large pages. req_addr: 0x0000000600000000 bytes: 8589934592
OpenJDK 64-Bit Server VM warning: Failed to reserve and commit memory using large pages. req_addr: 0x0000000000000000 bytes: 134217728
OpenJDK 64-Bit Server VM warning: Failed to reserve and commit memory using large pages. req_addr: 0x0000000000000000 bytes: 134217728
[0.374s][info][gc,init] Version: 19.0.1+10 (release)
[0.375s][info][gc,init] CPUs: 12 total, 12 available
[0.375s][info][gc,init] Memory: 32487M
[0.375s][info][gc,init] Large Page Support: Enabled
[0.376s][info][gc,init] NUMA Support: Disabled
[0.376s][info][gc,init] Compressed Oops: Enabled (Zero based)
[0.376s][info][gc,init] Heap Region Size: 4M
[0.377s][info][gc,init] Heap Min Capacity: 8G
[0.377s][info][gc,init] Heap Initial Capacity: 8G
[0.377s][info][gc,init] Heap Max Capacity: 8G
[0.378s][info][gc,init] Pre-touch: Disabled
[0.378s][info][gc,init] Parallel Workers: 10
[0.378s][info][gc,init] Concurrent Workers: 3
[0.379s][info][gc,init] Concurrent Refinement Workers: 10
[0.379s][info][gc,init] Periodic GC: Disabled
openjdk 19.0.1 2022-10-18
OpenJDK Runtime Environment Temurin-19.0.1+10 (build 19.0.1+10)
OpenJDK 64-Bit Server VM Temurin-19.0.1+10 (build 19.0.1+10, mixed mode, sharing)
Here is the call stack where the actual memory reservation happens (the call to the VirtualAlloc function):
jvm.dll!virtualAlloc(void * lpAddress, unsigned __int64 dwSize, unsigned long flAllocationType, unsigned long flProtect) C++
jvm.dll!reserve_large_pages_single_range(unsigned __int64 size, char * req_addr, bool exec) C++
jvm.dll!reserve_large_pages(unsigned __int64 size, char * req_addr, bool exec) C++
jvm.dll!os::pd_reserve_memory_special(unsigned __int64 bytes, unsigned __int64 alignment, unsigned __int64 page_size, char * addr, bool exec) C++
jvm.dll!os::reserve_memory_special(unsigned __int64 size, unsigned __int64 alignment, unsigned __int64 page_size, char * addr, bool executable) C++
jvm.dll!reserve_memory_special(char * requested_address, const unsigned __int64 size, const unsigned __int64 alignment, const unsigned __int64 page_size, bool exec) C++
jvm.dll!ReservedSpace::reserve(unsigned __int64 size, unsigned __int64 alignment, unsigned __int64 page_size, char * requested_address, bool executable) C++
jvm.dll!ReservedSpace::initialize(unsigned __int64 size, unsigned __int64 alignment, unsigned __int64 page_size, char * requested_address, bool executable) C++
jvm.dll!ReservedCodeSpace::ReservedCodeSpace(unsigned __int64 r_size, unsigned __int64 rs_align, unsigned __int64 rs_page_size) C++
jvm.dll!CodeCache::reserve_heap_memory(unsigned __int64 size) C++
jvm.dll!CodeCache::initialize_heaps() C++
jvm.dll!CodeCache::initialize() C++
jvm.dll!codeCache_init() C++
jvm.dll!init_globals() C++
jvm.dll!Threads::create_vm(JavaVMInitArgs * args, bool * canTryAgain) C++
jvm.dll!JNI_CreateJavaVM_inner(JavaVM_ * * vm, void * * penv, void * args) C++
jvm.dll!JNI_CreateJavaVM(JavaVM_ * * vm, void * * penv, void * args) C++
jli.dll!InitializeJVM(const JNIInvokeInterface_ * * * pvm, const JNINativeInterface_ * * * penv, InvocationFunctions * ifn) C
jli.dll!JavaMain(void * _args) C
jli.dll!ThreadJavaMain(void * args) C
The virtualAlloc function uses a logging wrapper to capture and log the result of the VirtualAlloc Windows function. It uses a separate logging tag (os). To capture both OS and pagesize logs to disk, use this command:
Here’s a snippet of the beginning of pagesize.txt.
[0.070s][trace][pagesize] Attempt special mapping: size: 240M, alignment: 2M
[0.071s][debug][pagesize] Reserving large pages in a single large chunk.
[0.320s][info ][pagesize] CodeHeap 'non-nmethods': min=4M max=6M base=0x000002a9c5c00000 page_size=2M size=6M
[0.320s][info ][pagesize] CodeHeap 'profiled nmethods': min=4M max=116M base=0x000002a9be800000 page_size=2M size=116M
[0.320s][info ][pagesize] CodeHeap 'non-profiled nmethods': min=4M max=118M base=0x000002a9c6200000 page_size=2M size=118M
[0.324s][trace][pagesize] Attempt special mapping: size: 8G, alignment: 4M
[0.324s][debug][pagesize] Reserving large pages in a single large chunk.
[0.948s][info ][pagesize] Heap: min=8G max=8G base=0x0000000600000000 page_size=4K size=8G
And below is the beginning of the os.txt log. Notice that the arguments to VirtualAlloc are logged, as well as its return value and the error code when a failure occurs. For example, the 2nd-to-last line shows that the 8GB allocation with large pages failed with error code 1450 (ERROR_NO_SYSTEM_RESOURCES aka Insufficient system resources exist to complete the requested service).
[0.052s][debug][os] Initial active processor count set to 12
[0.059s][trace][os] VirtualAlloc(0x0000000000000000, 8192, 2000, 4) returned 0x000002a9b5c60000.
[0.060s][trace][os] VirtualAlloc(0x000002a9b5c60000, 8192, 1000, 4) returned 0x000002a9b5c60000.
[0.060s][info ][os] SafePoint Polling address, bad (protected) page:0x000002a9b5c60000, good (unprotected) page:0x000002a9b5c61000
[0.065s][trace][os] VirtualAlloc(0x000000bc22600000, 16384, 1000, 4) returned 0x000000bc22600000.
[0.066s][info ][os] attempting shared library load of C:\java\binaries\jdk\x64\jdk-19.0.1+10\bin\java.dll
[0.069s][info ][os] shared library load of C:\java\binaries\jdk\x64\jdk-19.0.1+10\bin\java.dll was successful
[0.319s][trace][os] VirtualAlloc(0x0000000000000000, 251658240, 20003000, 40) returned 0x000002a9be800000.
[0.320s][trace][os] VirtualAlloc(0x0000000000000000, 65536, 2000, 4) returned 0x000002a9b5c80000.
[0.320s][trace][os] VirtualAlloc(0x000002a9b5c80000, 32768, 1000, 4) returned 0x000002a9b5c80000.
[0.320s][trace][os] VirtualAlloc(0x0000000000000000, 983040, 2000, 4) returned 0x000002a9b5c90000.
[0.320s][trace][os] VirtualAlloc(0x000002a9b5c90000, 32768, 1000, 4) returned 0x000002a9b5c90000.
[0.320s][trace][os] VirtualAlloc(0x0000000000000000, 983040, 2000, 4) returned 0x000002a9b5d80000.
[0.320s][trace][os] VirtualAlloc(0x000002a9b5d80000, 32768, 1000, 4) returned 0x000002a9b5d80000.
[0.944s][info ][os] VirtualAlloc(0x0000000600000000, 8589934592, 20003000, 4) failed (1450).
[0.947s][trace][os] VirtualAlloc(0x0000000600000000, 8589934592, 2000, 4) returned 0x0000000600000000.
The JBS bug made it sound like this feature worked in JDK 17 so I ran the same command using JDK 17.0.5.
I wanted to test some recent changes I was making in the OpenJDK repo. Running make test-tier1 failed because I did not specify the location of jtreg when I ran configure using this command on Windows or bash configure on my MacBook M1. I cleaned up the sample commands in the script to specify the --with-jtreg option as explained at jdk/testing.md at master · openjdk/jdk · GitHub.
Building target 'test-tier1' in configuration 'macosx-aarch64-server-release'
Test selection 'tier1', will run:
* jtreg:test/hotspot/jtreg:tier1
* jtreg:test/jdk:tier1
* jtreg:test/langtools:tier1
* jtreg:test/jaxp:tier1
* jtreg:test/lib-test:tier1
Error: jtreg framework is not found.
Please run configure using --with-jtreg.
RunTests.gmk:1027: *** Cannot continue. Stop.
make[2]: *** [test-tier1] Error 2
To run these tests on macOS, run bash configure --with-jtreg=/Users/saint/java/binaries/jtreg-7.1.1+1. configure does not like the ~/java/… path format for some reason. I also missed the fact that the Gtest suite is included in the tier1 tests. Therefore, I got errors like:
--------------------------------------------------
TEST: gtest/GTestWrapper.java
TEST JDK: /Users/saint/repos/java/forks/panama-foreign/build/macosx-aarch64-server-release/images/jdk
...
...
...=---==]=============
java.lang.Error: TESTBUG: the library has not been found in /Users/saint/repos/java/forks/panama-foreign/build/macosx-aarch64-server-release/images/test/hotspot/jtreg/native. Did you forget to use --with-gtest to configure?
at GTestWrapper.main(GTestWrapper.java:62)
at java.base/jdk.internal.reflect.DirectMethodHandleAccessor.invoke(DirectMethodHandleAccessor.java:104)
at java.base/java.lang.reflect.Method.invoke(Method.java:578)
at com.sun.javatest.regtest.agent.MainActionHelper$AgentVMRunnable.run(MainActionHelper.java:312)
at java.base/java.lang.Thread.run(Thread.java:1623)
JavaTest Message: Test threw exception: java.lang.Error
JavaTest Message: shutting down test
cd ~/repos
git clone -b release-1.8.1 https://github.com/google/googletest
cd ~/repos/java/forks/panama-foreign
bash configure --with-debug-level=slowdebug \
--with-jtreg=/Users/saint/java/binaries/jtreg-7.1.1+1 \
--with-gtest=/Users/saint/repos/googletest
make test-tier1
On Windows, I time the commands (out of my own curiosity) since they take much longer to run on my hardware:
cd /c/repos
git clone -b release-1.8.1 https://github.com/google/googletest
cd /cygdrive/c/java/forks/panama-foreign
time bash configure --with-debug-level=slowdebug \
--with-jtreg=/cygdrive/c/java/binaries/jtreg-7.1.1+1 \
--with-gtest=/cygdrive/c/repos/googletest
time make test-tier1
gtest Failure on macOS
make test-tier1 fails on macOS due to errors in the googletest sources. Here is a snippet of the configure output showing the C and C++ compiler versions in use:
configure: Using default toolchain clang (clang/LLVM)
checking for clang... /usr/bin/clang
checking resolved symbolic links for CC... no symlink
configure: Using clang C compiler version 13.1.6 [Apple clang version 13.1.6 (clang-1316.0.21.2.5) Target: arm64-apple-darwin21.2.0 Thread model: posix InstalledDir: /Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin]
checking whether the C compiler works... yes
checking for C compiler default output file name... a.out
checking for suffix of executables...
checking whether we are cross compiling... no
checking for suffix of object files... o
checking whether the compiler supports GNU C... yes
checking whether /usr/bin/clang accepts -g... yes
checking for /usr/bin/clang option to enable C11 features... none needed
checking for clang++... /usr/bin/clang++
checking resolved symbolic links for CXX... no symlink
configure: Using clang C++ compiler version 13.1.6 [Apple clang version 13.1.6 (clang-1316.0.21.2.5) Target: arm64-apple-darwin21.2.0 Thread model: posix InstalledDir: /Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin]
checking whether the compiler supports GNU C++... yes
checking whether /usr/bin/clang++ accepts -g... yes
checking for /usr/bin/clang++ option to enable C++11 features... none needed
checking how to run the C preprocessor... /usr/bin/clang -E
checking how to run the C++ preprocessor... /usr/bin/clang++ -E
configure: Using clang linker version 764 [@(#)PROGRAM:ld PROJECT:ld64-764]
checking for ar... /usr/bin/ar
The errors are about implicit copy constructors like in the example below. The build fails because there are too many errors (all related to this warning).
Creating hotspot/variant-server/libjvm/gtest/gtestLauncher from 1 file(s)
In file included from /Users/saint/repos/googletest/googlemock/src/gmock-all.cc:39:
In file included from /Users/saint/repos/googletest/googlemock/include/gmock/gmock.h:59:
/Users/saint/repos/googletest/googlemock/include/gmock/gmock-actions.h:484:3: error: definition of implicit copy constructor for 'PolymorphicAction<testing::internal::ReturnNullAction>' is deprecated because it has a user-declared copy assignment operator [-Werror,-Wdeprecated-copy]
GTEST_DISALLOW_ASSIGN_(PolymorphicAction);
^
/Users/saint/repos/googletest/googletest/include/gtest/internal/gtest-port.h:928:8: note: expanded from macro 'GTEST_DISALLOW_ASSIGN_'
void operator=(type const &) GTEST_CXX11_EQUALS_DELETE_
^
/Users/saint/repos/googletest/googlemock/include/gmock/gmock-actions.h:1125:10: note: in implicit copy constructor for 'testing::PolymorphicAction<testing::internal::ReturnNullAction>' first required here
return MakePolymorphicAction(internal::ReturnNullAction());
^
/Users/saint/repos/googletest/googlemock/include/gmock/gmock-actions.h:484:3: error: definition of implicit copy constructor for 'PolymorphicAction<testing::internal::ReturnVoidAction>' is deprecated because it has a user-declared copy assignment operator [-Werror,-Wdeprecated-copy]
GTEST_DISALLOW_ASSIGN_(PolymorphicAction);
^
/Users/saint/repos/googletest/googletest/include/gtest/internal/gtest-port.h:928:8: note: expanded from macro 'GTEST_DISALLOW_ASSIGN_'
void operator=(type const &) GTEST_CXX11_EQUALS_DELETE_
^
/Users/saint/repos/googletest/googlemock/include/gmock/gmock-actions.h:1130:10: note: in implicit copy constructor for 'testing::PolymorphicAction<testing::internal::ReturnVoidAction>' first required here
return MakePolymorphicAction(internal::ReturnVoidAction());
^
In file included from /Users/saint/repos/googletest/googlemock/src/gmock-all.cc:39:
In file included from /Users/saint/repos/googletest/googlemock/include/gmock/gmock.h:62:
In file included from /Users/saint/repos/googletest/googlemock/include/gmock/gmock-generated-function-mockers.h:44:
In file included from /Users/saint/repos/googletest/googlemock/include/gmock/gmock-spec-builders.h:71:
Creating hotspot/variant-server/libjvm/libgtest/libgtest.a from 1 file(s)
/Users/saint/repos/java/forks/panama-foreign/test/hotspot/gtest/gtestMain.cpp:233:7: error: no member named 'FLAGS_gtest_internal_run_death_test' in namespace 'testing::internal'; did you mean 'testing::FLAGS_gtest_internal_run_death_test'?
if (::testing::internal::GTEST_FLAG(internal_run_death_test).length() > 0) {
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
testing::FLAGS_gtest_internal_run_death_test
Looks like this will need some additional tweaks to get the macOS tests to run successfully. However, the tests on Windows x64 ran successfully and that was enough for what I was investigating.
Update: gtest Failure on Windows
I tried setting up a build environment on a new Windows machine and got this error about the gtest version from bash configure.
checking for gtest... /cygdrive/c/repos/googletest
configure: error: gtest at /cygdrive/c/repos/googletest does not seem to be version 1.8.1
configure exiting with result code 1
configure detects the googletest version by grepping the googletests CMakeLists.txt for GOOGLETEST_VERSION then using a regex to replace the whole line with the version number only.
grep GOOGLETEST_VERSION /cygdrive/c/repos/googletest/CMakeLists.txt | sed -E -e 's/set\(GOOGLETEST_VERSION (.*)\)/\1/'
The output is the string 1.9.0 as expected. Wondering if this is a line ending issue, I switch CMakeLists.txt to the Unix line endings using Notepad++. The new error below means that was indeed the issue!
git clone https://github.com/rstudio/rstudio
cd rstudio\dependencies\windows
The instructions for setting up a build environment have you run the Install-RStudio-Prereqs.ps1 script as an administrator. I do not like the way in which it dumps stuff all over the drive so I decided to fork the repo and adjust the installation locations. I first needed to enable powershell script execution as described by this article about Execution Policies – PowerShell seeing as running scripts is disabled on my system (Get-ExecutionPolicy currently outputs Restricted).
I used a virtual machine so I could easily revert to clean state. On this clean machine, the script successfully installs the prerequisites. The VS Build Tools 2019 are installed without the VS IDE and the VS Installer shows the corresponding stand-alone entry.
It then suggests installing Qt 5.12.8. When I saw this, I was a bit concerned about the commercial license requirements I ran into when Building ParaView. Fortunately, Qt 5.12 is still available in the Qt downloads archive.
I decided to use my physical machine on which I had already installed R so the first step was to uninstall R. Once that completed, I decided to try launching RStudio (2022.12.0+353) one more time since it wouldn’t have an R installation. Lo and behold, it now launched successfully!? Before proceeding, I decide to determine the state of my physical machine by listing all the installed products. I recall using wmic for this a while back. wmic /? now outputs a message that it is deprecated. windows – What can I do about “WMIC is deprecated”? – Stack Overflow says that the Common Information Model (CIM) is the way to go. Nonetheless, I can still get what I need using this command from the web:
wmic /output:initial-state.txt product get name, version
I decide to leave RStudio installed and proceed with installing the prereqs.
Set-ExecutionPolicy Bypass -Scope Process -Force; iex .\Install-RStudio-Prereqs.ps1
Installing Qt
I selected only the Qt * components and the sources under the Qt 5.12 node in the installer and Qt Creator in the Developer and Designer Tools node. This cut down on the size on disk of the Qt installation. Changes should be easy to make should any build errors pop up downstream.
Unfortunately, cmake ..\cpp -GNinja failed. Notice from the output though that the custom R location was detected just fine.
...
-- Configured to build DESKTOP
-- LIBR_HOME: C:/software/R/R-3.6.3
-- Found LibR: C:/software/R/R-3.6.3
-- Found R: C:/software/R/R-3.6.3
CMake Error at desktop/CMakeLists.txt:105 (message):
Did not find supported Qt SDK
-- Configuring incomplete, errors occurred!
See also "C:/repos/forks/rstudio/src/build/CMakeFiles/CMakeOutput.log".
See also "C:/repos/forks/rstudio/src/build/CMakeFiles/CMakeError.log".
The new maintenance tool looks similar to the one I used last summer. Makes sense since the version of Qt in use by rstudio is older. I needed to change the filters to show the MSVC 2017 64-bit component (as well as its peers that I didn’t install earlier). What a convoluted process! Nevertheless, this created the C:\software\Qt\Qt5.12.8\5.12.8\msvc2017_64 directory with the bin\qmake.exe file I was looking for earlier.
...
-- Version: 8.1.1
-- Build type: Debug
-- CXX_STANDARD: 11
-- Required features: cxx_variadic_templates
-- Configured to build DEVELOPMENT
-- Configured to build DESKTOP
-- LIBR_HOME: C:/software/R/R-3.6.3
-- Found R: C:/software/R/R-3.6.3
-- Found Qt: C:/software/Qt/Qt5.12.8/5.12.8/msvc2017_64/bin/qmake.exe
-- Configuring done
-- Generating done
-- Build files have been written to: C:/repos/forks/rstudio/src/build
I could then build the C++ sources by running ninja. This takes just under 5 minutes on my AMD Ryzen 7 5800X 8-Core system. Launching src\build\desktop\rstudio.exe fails due to missing Qt binaries (Qt5PrintSupportd.dll, Qt5WebEngineWidgetsd.dll, Qt5WebEngineCored.dll, Qt5WebChanneld.dll). Interestingly, none of these binaries are loaded in the public RStudio build’s rstudio.exe process.
These commands will deploy the necessary binaries into the rstudio desktop directory.
This is sufficient to launch RStudio but only the main menu is present in the RStudio window.
RStudio window after building locally.
A quick review of the launch instructions and I see that I didn’t need to copy these binaries: From command prompt, cd to the build location, and run rstudio.bat. This batch file updates the PATH to include the location of these Qt binaries!
@echo off
setlocal
set "RS_CRASH_HANDLER_PATH=C:/repos/forks/rstudio/src/cpp/../../dependencies/windows/crashpad-release/bin/crashpad_handler.exe"
set "QT_PLUGIN_PATH=C:/software/Qt/Qt5.12.8/5.12.8/msvc2017_64/plugins"
set "PATH=C:/software/Qt/Qt5.12.8/5.12.8/msvc2017_64/bin;%PATH%"
desktop\rstudio.exe
Building Java/Gwt
As evident from the post so far, I started by building C++ instead of Java/Gwt. Not sure if that’s why I got a blank window. The relationship between the C++ and Java/Gwt components is still a mystery to me. My first attempt at building Java/Gwt was to use ant desktop.
...
panmirror:
[echo] yarn location: ../../dependencies/common/node/16.14.0/node_modules/yarn/bin/yarn.cmd
[mkdir] Created dir: C:\repos\forks\rstudio\src\gwt\www\js\panmirror
[exec] 'node' is not recognized as an internal or external command,
[exec] operable program or batch file.
BUILD FAILED
C:\repos\forks\rstudio\src\gwt\build.xml:211: The following error occurred while executing this line:
C:\repos\forks\rstudio\src\gwt\build.xml:233: The following error occurred while executing this line:
C:\repos\forks\rstudio\src\gwt\build.xml:121: exec returned: 1
That failed because nodejs is not in the PATH. dir /s node.exe in the repo root shows it to be in dependencies\common\node\16.14.0. I manually update the path:
set "PATH=C:/repos/forks/rstudio/dependencies/common/node/16.14.0;%PATH%"
The build now proceeds (with lots of warnings) and launches the GWT Development Mode window below. The last message in the console is a notification that The code server is ready at http://127.0.0.1:9876/. Not sure what I expected, but this wasn’t it.
Launching rstudio in a new command prompt (by running rstudio.bat from the build location) now leads to a non-blank window:
Compiling rstudio
A few moments later (not sure exactly how long it took, but it was long enough for me to get a screenshot), RStudio loads!
RStudio Window
I notice that the About window cannot be moved outside the bounds of the parent window. This is some interesting rendering going on.
About RStudio
Notice rstudio.exe’s child process QtWebEngineProcessd.exe and ant.exe and its jdk1.8.0_211 java.exe child processes. A minidump of rstudio.exe shows many of the Qt DLLs loaded into it, unlike the public release rstudio.exe. Fascinating architecture all around.
Outstanding Items
How is Qt statically linked into the released RStudio?
I also follow the instructions and install MiKTeX and check for updates. The process is detailed at Install MiKTeX on Windows. I had some trouble getting the MiKTeX installer to work when installing for all users so I end up installing for just myself but place it in c:\software because I find the user-specific paths too obnoxious.
I was expecting to have TeXnicCenter installed but that is just an editor and it would have required MiKTeX as well. It is not required for building R but I install it anyway (since I usually typeset a lot of personal PDFs such as course assignments/reports).
Environment Setup
I run this command to ensure all programs can be found which make gcc pdflatex tar. However, I did this before step 6: Add gcc, MiKTeX and tar to the PATH and set one tar option.
/usr/bin/make
which: no gcc in (/usr/local/bin:/usr/bin:/bin:/opt/bin:/c/Windows/System32:/c/Windows:/c/Windows/System32/Wbem:/c/Windows/System32/WindowsPowerShell/v1.0/:/c/progra~1/git/cmd:/usr/bin/site_perl:/usr/bin/vendor_perl:/usr/bin/core_perl)
which: no pdflatex in (/usr/local/bin:/usr/bin:/bin:/opt/bin:/c/Windows/System32:/c/Windows:/c/Windows/System32/Wbem:/c/Windows/System32/WindowsPowerShell/v1.0/:/c/progra~1/git/cmd:/usr/bin/site_perl:/usr/bin/vendor_perl:/usr/bin/core_perl)
/usr/bin/tar
This output is interesting because it reminds me of my search for gcc in the MSYS environment when trying to build octave (see Building Octave on Windows)! This is the environment I use on my machine:
Running make check takes 3 minutes on my machine and ends with this output:
...
running tests of plotting Latin-1
expect failure or some differences if not in a Latin-1 or UTF-8 locale
running code in 'reg-plot-latin1.R' ... OK
comparing 'reg-plot-latin1.pdf' to './reg-plot-latin1.pdf.save' ... OK
running code in 'reg-S4.R' ... OK
comparing 'reg-S4.Rout' to './reg-S4.Rout.save' ... OK
running tests of Internet functions
running code in 'internet.R' ... OK
comparing 'internet.Rout' to './internet.Rout.save' ... OK
Launching R
Run R using the command $TOP_SRCDIR/bin/x64/Rgui.exe. Help > About displays this information for my local build.
I was recently checking out a data analytics course that uses R as one of its tools. It recommended using RStudio. Having never used R, I was intrigued so I started at Step 1 on the RStudio Desktop – Posit download: installing R. It links to The Comprehensive R Archive Network (rstudio.com), which has a link to the Windows download page. I installed R 4.2.2 on my desktop. Unfortunately, the web page uses frames, which makes it non-trivial to get the appropriate link to share here. Installing then launching RStudio presented this dialog:
Launching RStudio
Unfortunately, nothing happens after I click OK. There is no trace of the process in Task Manager either. I download the RStudio ZIP and see if perhaps it’s an installer issue.
Defender SmartScreen Warning about Unzipped rstudio.exe
Unfortunately, this doesn’t address the startup failure. I try one more thing: explicitly selecting the version of R that is used by RStudio.
Choosing a specific version of R
When this doesn’t work, I install R and RStudio on my Surface Pro and RStudio launches successfully! I wonder if I might need to build a local installation to dig deeper into this issue. I will update this post with the results of my investigation when it completes.
Disassemble the VarArgs function to simplify stepping through the code (this enables me to interpret the assembly instructions, mapping them to the source code):
cd build\windows-aarch64-server-slowdebug\support\test\jdk\jtreg\native\support\libVarArgs\
dumpbin /disasm /out:libVarArgs.asm libVarArgs.obj
dumpbin /all /out:libVarArgs.txt libVarArgs.obj
Now stepping through the code, we observe that the process terminates.
From the assembly, what appears to be happening is the switch statement is immediately jumping to the default case, which calls exit(-1). So, pretty simple test failure. Why did I think it was a crash? I assumed that a crash was the only reason the JVM would terminate prematurely but this was actually a clean exit, by design. Perhaps an assertion failure would have made the issue more visible.
I recently bought the Skytech Blaze 3.0 High-Performance PC Desktop, my first desktop purchase since 2010. The ASRock B550M-C motherboard caught my eye as I was looking around at the components in the desktop. Figured I might as well learn the basics of how motherboards are made (and in the process, a brief history of motherboards). Here are some videos I watched to learn about these topics.
cd \dev\repos
hg clone https://hg.octave.org/octave
Developer FAQ – Octave has a discourse forum (just like LLVM) and I’m realizing I need to jump into these forums and at least hear what’s happening. Instructions for building on Windows have a separate page 😀 Building on Microsoft Windows – Octave. MSYS2 is used for building natively, and since I used it to build Elmer, I might as well see how well it works for Octave.
The vastness of qt5 is the first thing that confronts me when I run this command. The install size of all the packages is about 3.5 GB. Installation takes about 7.5 minutes.
37 Members in Group mingw-w64-x86_64-qt5
cd /c/dev/repos/octave
./bootstrap
mkdir -p .build
cd .build
../configure --disable-docs ac_cv_search_tputs=-ltermcap
Digging into Configure Failures
Configure fails on my machine with this error the first time I run it:
configure: loading site script /etc/config.site
checking for a BSD-compatible install... /usr/bin/install -c
checking whether build environment is sane... yes
checking for a race-free mkdir -p... /usr/bin/mkdir -p
checking for gawk... gawk
checking whether make sets $(MAKE)... yes
checking whether make supports nested variables... yes
checking whether UID '197630' is supported by ustar format... yes
checking whether GID '197630' is supported by ustar format... yes
checking how to create a ustar tar archive... gnutar
checking whether make supports nested variables... (cached) yes
checking build system type... x86_64-w64-mingw32
checking host system type... x86_64-w64-mingw32
checking whether make supports the include directive... yes (GNU style)
checking for gcc... no
checking for cc... no
checking for cl.exe... no
checking for clang... no
configure: error: in `/c/dev/repos/octave/.build':
configure: error: no acceptable C compiler found in $PATH
See `config.log' for more details
I try updating the .bash_profile as suggested but this doesn’t help:
Unfortunately, gcc is still not found. I verified that gcc is indeed on disk C:\dev\software\msys64\mingw64\bin\gcc.exe using the path structure at Package: mingw-w64-x86_64-gcc – MSYS2 Packages. The top answer did suggest modifying the PATH but I’m perplexed at how the compiler was found in the ElmerFEM build environment since it also cannot find the gcc command.
So, the culprit turns out to be the fact that I was using the UCRT shell instead of the MINGW64 shell. I think this bit me with Elmer as well. Should have carefully reviewed that post (see the Custom Generator in MSYS section).
Building the Code
Using the correct MSYS terminal allows configure to work. It takes 4m:45s on my machine. Here is the summary after all the flags are displayed. This piques my curiosity about where Java methods are called by Octave but I’ll ignore it for now!
Default pager: less
gnuplot: gnuplot
Build Octave Qt GUI: yes (version: 5)
Build Java interface: no
Build static libraries: no
Build shared libraries: yes
Dynamic Linking API: LoadLibrary
Include support for GNU readline: yes
Use push parser in command line REPL: yes
64-bit array dims and indexing: yes
64-bit BLAS array dims and indexing: no
OpenMP SMP multithreading: yes
Truncate intermediate FP results: yes
Build cross tools: no
Build docs: no
configure: WARNING: JAVA_HOME environment variable not initialized. Auto-detection will proceed but is unreliable.
configure: WARNING: No Java executable found. Octave will not be able to call Java methods.
configure: WARNING: building documentation disabled; make dist will fail.
configure:
configure: NOTE: Libraries or auxiliary programs may be skipped if they are not found
configure: NOTE: OR if they are missing required features on your system.
Build Octave by running make with these options. The last few lines of output from a successful build are shown below as well. The build took 21min on my new desktop.
make all -j8
...
GEN libinterp/dldfcn/gzip.oct
GEN doc/interpreter/doc-cache
Octave successfully built. Now choose from the following:
./run-octave - to run in place to test before installing
make check - to run the tests
make install - to install (PREFIX=/mingw64)
HG ID for this build is "5744dac88986"
make[2]: Leaving directory '/c/dev/repos/octave/.build'
make[1]: Leaving directory '/c/dev/repos/octave/.build'
./run-octave launches the Octave command line. Since we built with Qt, we can launch the Octave GUI using this command:
./src/octave-gui --gui
Unfortunately, there is a segmentation fault when I close Octave! TODO: Why?
I was recently implementing matrix multiplication on the GPU (using CUDA). For my application, I was generating random numbers and generating statistics about the performance of matrix multiplication variants (e.g. using shared memory vs naive multiplication). Some of the results tended to differ from the CPU’s results. Therefore, I decided to use deterministic matrices for the inputs to ensure my algorithm is correct. What I needed was a neutral (3rd party) matrix multiplication algorithm. This seems like a job for MATLAB. Unfortunately, my license expired a few years ago. My robotics professor at the University of Washington was a fan of Octave because it is open source and free. Here is the script I created to generate matrices with the positive integers.
A = 1:10000;
B = 10001:20000;
A = reshape(A, [100,100]);
B = reshape(B, [100,100]);
A = transpose(A);
B = transpose(B);
C = A * B;
# format short;
save 'mmult100x100.txt' C;
Backstory
It has been a while since I used MATLAB. Here are the searches I used to create the script.
This process of using Nsight Compute to profile CUDA kernels is documented in detail at Nsight Compute :: Nsight Compute Documentation (nvidia.com). Here are the screenshots with the “quick start” steps without all the verbosity of the documentation.
