Prepare for the CMake plumbing to build the loader with MinGW#7
Closed
illwieckz wants to merge 45 commits into
Closed
Prepare for the CMake plumbing to build the loader with MinGW#7illwieckz wants to merge 45 commits into
illwieckz wants to merge 45 commits into
Conversation
Only for amd64 host and target. nacl_helper_bootstrap also builds. - Check in linux_syscall_support.h (like we did for Breakpad) - Add a hacky symlink so we don't have to include the repository's parent directory or name it "native_client" - Use LLVM from /usr/bin instead of Chromium toolchains - Remove sysroot flags (depend on host system instead) - Disable setup of NaCl-target toolchains in scons scripts
Disable GDB tests by default and when they are disabled don't try to look for a 2nd NaCl toolchain just to grab its GDB. GDB tests on Linux are broken for me anyway in upstream with the incompatibly old curses lib dependency issue.
With some minor fixes, the build is able to run and produce an irt_core binary when configured with the Saigo toolchain. Added documentation how to do this to the README. Still keeping the 'if nacltools' guards I added so that sel_ldr can be built without acquiring any NaCl toolchains.
The only known working implementation is for i686 MSVC. - amd64 MSVC was already disabled and marked at not working, - there is no MinGW implementation in the code.
…gs into raw strings
'pipes' module was removed from the standard library.
assertEquals and assertNotEquals were removed in Python 3.12.
Doesn't work on my Debian Trixie. Anyway I believe this seccomp BPF thing is an experiment that was never really productionized. After this everything passes for me with ./scons --mode=opt-host,nacl saigo=1 platform=x86-64 all_tests
Clearly the test shouldn't be disabled with Saigo for that condition. But I couldn't find any problem with running the test for any PNaCl configuration either (it always had a build id).
This fixes some test (if the repository dir is not named exactly native_client).
Check in a copy of the 2013-vintage Googletest which is specified in the DEPS file. Trying to use an external one would not work because an old version that supports old C++ standards is needed. Hook it up to the build system so that most tests work now. Imported from gtest revision 2d3543f81d6d4583332f8b60768ade18e0f96220: - LICENSE - src/ - *.h files within include/
Like upstream, you must pass a breakpad_tools_dir argument for the tests to run. Unlike upstream, the Breakpad #includes will be looked for also in the tools dir, instead of the Chromium code depot.
For Linux toolchains (except x86-64 where I previously commented it out), there were hard-coded sysroot flags referring to the paths of Chromium's bundled dependencies. Get rid of this. On my machine the cross toolchains work fine without it. But I added a a 'sysroot' command line argument, so if anyone needs it they can pass sysroot=... on the scons command line to configure a --sysroot flag.
This one only seems to be triggered when a test fails.
Use system qemu-armhf for the emulator for tests. The command here basically matches what the run_under_qemu_arm script (which can be seen at tools/trusted_cross_toolchains/qemu_tool_arm.sh) was doing. Also change the emulator in run.py.
This fixes a NaCl syscall that only exists for a test. Fixes the run_infoleak_test test target on ARM.
REGS_SAVER_FUNC is meant to work even if the stack pointer is misaligned, but the multi-register PUSH instructions crash if the stack is misaligned, as per the ARM documentation. I guess the old(?) bundled QEMU used by upstream doesn't implement this faithfully. Fix it by using a series of single-register pushes instead. Fixes the run_register_set_test test target on ARM.
Use `volatile` as optimization barrier for the value of the pointer (NOT for what's pointed at) like in Daemon's /injectFault segfault. fixup: crashing
This test is using some custom macros instead of googletest and ASSERT_EQ apparently double-evaluates the arguments if the assertion fails.
Use --orphan-handling=unwanted when linking nacl_helper_bootstrap which makes the linker discard any sections not explicitly mentioned in the linker script. This prevents the linker from producing a defective binary with overlapping PT_LOAD headers when using GCC on certain distros such as Ubuntu and Arch that have it configured to produce a .note.gnu.property section
illwieckz
force-pushed
the
illwieckz/mingw
branch
from
ba61063 to
e7a867b
Compare
illwieckz
force-pushed
the
illwieckz/mingw-1
branch
from
e52b0bd to
9bf3b1f
Compare
illwieckz
changed the title
illwieckz
changed the title
illwieckz
changed the title
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The only implemented build is CMake:
Requires: