As discussed in LLVMdev [1], due to the execution order of
doFinalization functions, the GC information were deleted before
AsmPrinter::doFinalization was executed. Thus, the
GCMetadataPrinter::finishAssembly was never called.
The attached patch fixes that by moving the code of the
GCInfoDeleter::doFinalization to Printer::doFinalization. Pedro (cc'ed)
has already taken a look at this and he also thinks that it is OK.
I also tried to add a simple unit-test by using the available Ocaml
garbage collector. I 'm not very familiar with the code of that
collector so I'm not 100% sure that this suffices as a test; for
example, I only test that for the X86-64 architecture, should I add
similar tests for all architectures?
If nobody has any objections, can someone commit this for me? Because
every GC should be unusable right now!
As discussed in LLVMdev [1], due to the execution order of
doFinalization functions, the GC information were deleted before
AsmPrinter::doFinalization was executed. Thus, the
GCMetadataPrinter::finishAssembly was never called.
The attached patch fixes that by moving the code of the
GCInfoDeleter::doFinalization to Printer::doFinalization. Pedro (cc'ed)
has already taken a look at this and he also thinks that it is OK.
I also tried to add a simple unit-test by using the available Ocaml
garbage collector. I 'm not very familiar with the code of that
collector so I'm not 100% sure that this suffices as a test; for
example, I only test that for the X86-64 architecture, should I add
similar tests for all architectures?
If nobody has any objections, can someone commit this for me? Because
every GC should be unusable right now!
Looks great, committed in r175528. Thanks for fixing this, the test case should make sure that it won't break as badly again.
