Revert "Add purgeObj() to remove the symbol table entries for an object"
authorSimon Peyton Jones <simonpj@microsoft.com>
Mon, 1 Dec 2014 22:39:12 +0000 (22:39 +0000)
committerSimon Peyton Jones <simonpj@microsoft.com>
Tue, 2 Dec 2014 10:22:07 +0000 (10:22 +0000)
commit7932b2adaecac6c86038176d909c20ad1b1f9604
tree95ad2fdd457a252603901e539fc4e910b8e49727
parent9437a24d36f248fcd0b2709ae417b999d6a17444
Revert "Add purgeObj() to remove the symbol table entries for an object"

This reverts commit 9e6e4796437a7fc23e83605a45db9b2663570123.

I reverted it because one of these two patches
  9e6e4796437a7fc23e83605a45db9b2663570123 Add purgeObj()
  b5e8b3b162b3ff15ae6caf1afc659565365f54a8 Make the linker API thread-safe

causes a seg-fault on Windows.  The seg-fault happens immediately
the linker is invoked, in ghci or in Template Haskell.

I believe that it is the "linker API thread-safe" commit that causes
the seg-fault; it happens even if the "purgeObj" commit alone is
reverted.  But since the two patches mess with the same code, to
revert the "linker API" patch I had revert both.
includes/rts/Linker.h
rts/Linker.c
testsuite/tests/rts/linker_unload.c
testsuite/tests/rts/linker_unload.stdout