Applied suggestion in comment (making RuntimeCallInserter use an RAII approach)

In D143565#4620489, @vitalybuka wrote:

Is this still relevant?

Updated to address comments.

• • Changed std::optional to use a bool instead
  • Removed changes for __asan_* calls that don't apply to Windows EH handling scenarios (module ctor/dtor, globals array)
• Added compiler-rt test
• Improved the test for more coverage

Rebased on main.

In D102633#2786288, @aganea wrote:

@dexonsmith Yes, using the Clang Tooling C++ API seems like a good option, however some logic could/would be needed from clang/tools/clang-scan-deps/ClangScanDeps.cpp. Using clang-scan-deps as-a-DLL seemed like the best option on the short term, since we're using its main() almost verbatim with only a few minor changes. We still need to solve a contention issue related to the Windows implementation of sys::fs::status(), but then once everything is sorted out we'll re-consider the usage of the DLL.

Followed @aganea's suggestion, it's indeed clearer, thanks!

Fixed clang-format

In D139632#3985341, @thakis wrote:

LG (except for thieta's nit.)

Do you have commit permissions, or do you need someone to land this for you? If the latter, which name and email should we use for landing it?

Ping!

Small update: improve comments, remove empty line modification

Added support for empty MBB: ensures 2-byte nop is inserted, to make the function patchable.
This should also support the requirement for the functions not to jump to the first instruction.

In D137642#3934867, @rnk wrote:

I was going to approve it as is, since it looks like a correct simplification.

Relatedly, @dblaikie was considering establishing the invariant that all empty functions have at least one trap instruction, although for this code the entry BB could conceivably be empty, and you should handle that.

Simplified init of config->argv

I found new related crashes, I will investigate.

Perform the argument list filtering in the driver, to use the result of parsed arguments.
Updated pdb-linker-module.test, checking that the object file is indeed discarded.

Added release note

Update: cleanup lambda (remove capture, fix const correctness)

LGTM
I was curious whether it would be good to remove the notion of GUID validity from the tests, since we now have support for collision whatever the GUIDs are.
I just noticed it, but it was already the case in D122372, so it might be for a separate patch if we want that.

Nice! It's much more straightforward now, the reverse iteration was pretty confusing.
Thanks!

Sorry, I flagged this as "accepted", but I'm not sure I am allowed to do that...

It is indeed much cleaner this way`, thanks!
The tests are not anymore the same for clang-cl as for clang, but I agree what needs testing is the clang-cl specific behaviors, not having the same tests redone as a clang-cl flavor.

I understand, making clang-cl -Wall behave differently than `/Wall' is indeed confusing.
We'll take different steps for setting up the warnings in our Build System.

Updated for comments from @rnk, except for the canonicalization part.

Ping!

Applied the suggestions from the parent patch D95099

Updated for suggested modifications

Removed the support for "--" command lines, it was extracted to patch D95099.
Applied suggested fixes.

Bugfixes on the use of the iterator.

Ping!
Would anybody have a chance to take a look at this? I'd like to submit this before the LLVM 12 branch, if possible? Thanks in advance!

Ping!

Updated the tests for clang-cl command lines:
I moved the input files to end of command line, with --, to avoid linux-style paths conflict with clang-cl arguments.

Ping!

Rebase after https://reviews.llvm.org/D92330.
Also, improved the detection of target obj file.

Removed /Fo option. -o is also supported by clang-cl and overrides /Fo options that are found earlier in the command line.

Update patch : clang-format fixes.

Simplified the RealFS initialization, as suggested

Removed use of ProxyFileSystem

Updated the patch.