I like the idea of this, and fits what I had in mind for the original D49466 review. I want to take a closer look, but right now LGTM.

In D49466#1760860, @MaskRay wrote:

Add back remapDIPath that was unintentionally deleted by D69213, caught by a test.

Small adjustment of the code

Ping?

In D56769#1624345, @Lekensteyn wrote:

Should this be abandoned for D49466?

In D49466#1681534, @phosek wrote:

@dankm is it OK if we take over this change to push it forward?

• Address feedback from @Lekensteyn

Whoops. There are extra changes here I didn't mean to submit :/

• Address feedback from @Lekensteyn

In D65556#1610001, @lebedev.ri wrote:

Please fix patch title and description

In D49466#1586140, @manojgupta wrote:

@dankm are you still working on this patch?

Latest changes. I've been sitting on these for months, so I don't remember all that changed. The path remapping contract changed somewhat, and it's now based on the git monorepo.

Nail down the contract a bit.
Add a flag for strict path matching.
I think there's still unresolved issues with tests.

Attempt to document and test the desired interface.

Move trailing path separator stripping back to Clang.

Rewrote and simplified the replace_path_prefix function, and added a bunch of tests to catch corner cases I noticed during testing.

Sure, I'll (eventually) make a separate review.

Undo accidental change.

Update style.

Enforce path mapping. This requires LLVM review D56769.

Restored original test case file names.

In D49466#1354468, @Lekensteyn wrote:

Could you add more tests to check the error message for bad options (missing =):

-fdebug-prefix-map=bad
-fmacro-prefix-map=bad
-ffile-prefix-map=bad

renamed err_drv_invalid_argument_to_prefix_map to err_drv_invalid_argument_to_option
added more frontend tests for macro-prefix-map and file-prefix-map.

Made diagnostics for file-prefix-map display the actual option name.

In D49466#1354238, @Lekensteyn wrote:

It would be nice to have this for Clang 8.0, the branch date is within 5 days :)

In D49466#1353437, @alxu wrote:

FYI, according to my comment on D49652, assuming I checked it correctly, gcc applies the maps in reverse order of command line specification, not sorted order. It seems unlikely that anyone is actually depending on the order though.

Added unit tests for the prefix remapping.

In D49466#1296790, @ChristophBerg wrote:

PostgreSQL 11 is now using LLVM to do JITing of SQL expressions. We'd need this feature to strip the build directory off the .bc bitcode files so the .deb packages build reproducibly.
@dankm: Are you still working on this? What can we do to help getting this move forward?

Address some of the comments by erichkeane and joerg.

The changes to CMakeLists.txt are to ensure that the shared runtime doesn't link libc before libthr. If that happens then the pthread implementation doesn't get intercepted properly.