Is there some test that can be added that would find a regression?

It all seems to be working.
I'll have to work on the EXPECT_DEATH aspect as a follow up.

Compiles and tests pass on Fuchsia.
The changes themselves look good to me.

There seems to be something wrong on Fuchsia with it, regular processes are crashing. Looking into this.

In D93731#2482877, @pcc wrote:

It's ready to go in if you don't mind me breaking Fuchsia. Otherwise I can wait for a Fuchsia implementation of setMemoryPermission() from you.

In D93731#2482877, @pcc wrote:

It's ready to go in if you don't mind me breaking Fuchsia. Otherwise I can wait for a Fuchsia implementation of setMemoryPermission() from you.

Is this still WIP like the title says?

Roland committed his change for the issue, this is no longer needed.

Let's go with yours, better comments/whitespaces.

Thanks Mitch! Looks reasonable to me.

... and remove the unused field in the 32-bit primary. Sorry.

Correct copy/paste error in the 32-bit primary (!= => ==)

Anything I can help address to move forward?

Correct > to >=.

In D91824#2432269, @oontvoo wrote:

In D91824#2426891, @pcc wrote:

I think @cferris was asking whether old scudo was still being used by asan/hwasan. To which the answer is "no" I believe, the sanitizers have their own independent copy of the allocator.

Just to wrap up the threads here, the consensus is that all the non-standalone stuff under compiler-rt/scudo/* should simply be deleted? (along with their tests?)

Do you need me to commit this?

Updating the CL with another proposed fix: reseting the TLS data in
the uninit function called by the tests.

This is an arc fail on my side

Thank you, good catch!!

In D91824#2409331, @pcc wrote:

Can we not just drop support for Android in non-standalone Scudo? As far as I know it is only being used (if at all) on a limited set of platforms which does not include Android.

Addressing review comments.

Ping please!

Further simplify the release function.

And another, sorry for the back & forth, getting numbers as they come.

Restoring a couple of UNLIKELY in the loops.

Changing some lines around.

Re-add a comment that got removed.

Addressing Mitch's request.

Addressing Roland's comments.

Couple of small fixes.

Addressing review comments.

Thank you for all your work!
IIRC everything was working except as you point out some of the very Unixy tests.
Is your build creating a clang_rt.scudo_cxx library that also needs to be link as well? The C++ specific interceptors usually end up in a separate library that is not linked into the straight C programs, which is the most likely culprit for not having new/delete be intercepted.

Adding stdint.h to try and quell some clang-tidy warnings.

Address review comments.

clang-format a file that was missed.

Addressing review comments:

• tried to strike a middle ground between name suggestions, settled on: `void *reserveGuardedPool(size_t Size);

void allocateInGuardedPool(void *Ptr, size_t Size) const;
void deallocateInGuardedPool(void *Ptr, size_t Size) const;
void unreserveGuardedPool();`

• added comments and clarifications about API contracts
• used anonymous namespace instead of static

Correcting a potential conversion warning.

In D89984#2348398, @nikic wrote:

I can confirm that this builds fine for me.

In D89908#2348293, @nikic wrote:

I've reverted this in https://github.com/llvm/llvm-project/commit/04e42f62548d4c0367664188a938b609435718e2 due to a build failure with GCC. The code seems a bit sensitive, so I didn't want to introduce random casts.

Addressing review comments.

• Restoring the old constants because some test failed with mine.