Re-enable the bc-gh OSS-Fuzz project?

[xingyaner]

The bc-gh project in OSS-Fuzz (projects/bc-gh) is currently disabled. It was disabled in the original integration commit a7c16d818 because of LeakSanitizer leak reports, which the commit message noted were not reproducible under Valgrind (see quoted message below).

I set disabled to true because LeakSanitizer claims the fuzzers have leaks. However, this is not true because bc and dc run without leaks under Valgrind. After some debugging, I can cause them to free the memory by putting an assert(false); after all of the frees, and it trips. But if I remove it, LeakSanitizer sees leaks. Whether it's PEBKAC (probably) or miscompilation (unlikely), I don't know, but I figured I'd commit the necessary stuff for the project. I'll try to figure out the problem later.

Separately, the OSS-Fuzz build itself had been broken: the build script called ./configure -Z, which version 7.2.0's configure.sh no longer supports (OSS-Fuzz mode is now controlled by the BC_ENABLE_OSSFUZZ macro). Removing the invalid flag also exposed a configure.sh bug — the multi-line test_target variable breaks the sed-based substring_replace() (sed: unterminated 's' command).

A PR in OSS-Fuzz now fixes the build/compile path: build_fuzzers succeeds and check_build passes. However, this only addresses compilation — it does not verify the LeakSanitizer leak reports that originally caused the project to be disabled.

Question
Would you like to restart the fuzzing of this project? If yes, I will remove disabled: true from projects/bc-gh/project.yaml in the OSS-Fuzz PR.

Additional context

• There is also an upstream bug in configure.sh worth fixing independently: the test_target variable contains a literal newline that breaks the sed-based replacement in scripts/functions.sh. If fixed upstream, the temporary workaround in the OSS-Fuzz build script can be removed.

[xingyaner]

google/oss-fuzz#15930

[gavinhoward]

I'd like to enable OSS-Fuzz, but I just don't have the bandwidth or ability to test it. I did commit c7f5c22 to make it easier to enable OSS-Fuzz, but if you want to enable it, you'll have to figure out how. I'll accept PRs if you do; I don't want to change the OSS-Fuzz configuration in the OSS-Fuzz repo (besides disabled: true to disabled: false), so let's make the changes here, and I'll help you convince the OSS-Fuzz project to change it.

[xingyaner]

I've submitted PR #104 with the fixes to get the ./configure -Z OSS-Fuzz build working. It makes -Z automatically enable memcheck, sets the bc.expr_exit/dc.expr_exit defaults to 0 in fuzz/OSS-Fuzz modes (to satisfy the BC_EXPR_EXIT == 0 assertion), and makes the fuzzers call bc_vm_atexit() to free the VM state and eliminate the LeakSanitizer reports.
I verified it locally, build_fuzzers and check_build pass for both AFL and libFuzzer (address sanitizer, x86_64), and the normal non-fuzz build shows no regression. Please take a look when you have time.