It seems weird to do this, since I expect macOS to not be affected by the issue you're referencing. I was going to try it for myself, but we already see from google/leveldb#634 (comment):

but on macOS, for example, it doesn't cause a crash.

Yes, This only seems to happen in Monterey.

No, I don't think that's true.

❯ sw_vers
ProductName:	macOS
ProductVersion:	12.0.1
BuildVersion:	21A559
❯ cat test.cpp
#include <dlfcn.h>
#include <string>
int main() {
    std::string s;
    dlopen("/opt/homebrew/lib/libleveldb.dylib", RTLD_NOW);
    return 0;
}
❯ make test
c++     test.cpp   -o test
❯ ./test; echo $?
0
❯ otool -L /opt/homebrew/lib/libleveldb.dylib
/opt/homebrew/lib/libleveldb.dylib:
	/opt/homebrew/opt/leveldb/lib/libleveldb.1.dylib (compatibility version 1.0.0, current version 1.23.0)
	/opt/homebrew/opt/snappy/lib/libsnappy.1.dylib (compatibility version 1.0.0, current version 1.1.9)
	/opt/homebrew/opt/gperftools/lib/libtcmalloc.4.dylib (compatibility version 10.0.0, current version 10.9.0)
	/usr/lib/libc++.1.dylib (compatibility version 1.0.0, current version 1200.3.0)
	/usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current version 1311.0.0)

Where I see this is here: https://gitlab.haskell.org/ghc/ghc/-/issues/20610

I wonder if that might be a different linker issue as it does successfully link and then crash.

It seems like your bug is something else entirely, as the reproducer in the issue you reference doesn't apply. I'm reluctant to apply a change to a formula that affects many other users when it's not at all clear that this is the right fix. Moreover, even if it were the right one, I don't know how this weighs against potentially breaking other users who rely on leveldb using tcmalloc.

I don't use GHC, so I can't really say what's happening with your crash. Does GHC call dlopen on libleveldb?

I don't use GHC, so I can't really say what's happening with your crash. Does GHC call dlopen on libleveldb?

Yes, GHC does call dlopen on libleveldb.

I see. But it seems the dlopen by itself isn't enough to cause a crash, so there's likely something more going on.

Does this fix your issue? gperftools/gperftools#1315