This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Bootstrap Failure on i686-pc-linux-gnu



On Wed, 1 Dec 2004, Richard Earnshaw wrote:


This is PR 18746 for which I had exposed a latent bug and submitted
a patch for <http://gcc.gnu.org/ml/gcc-patches/2004-12/msg00004.html>.


This isn't my speciality area, but I'm not sure I believe that patch.

ENABLE_CHECKING should *never* be changing the checked structures.
Calling a different routine here is just papering over a failure of that
fundamental requirement.

Normally i'd agree, but this happens because the updating of statement operands is done lazily. Checking that get_stmt_operands in the verification didn't actually do anything would be
1. Somewhat ugly to code
2. Very expensive to verify

His patch is correct because it actually forces the rescanning to happen at the right time, so that the verify_ssa's call to get_stmt_operands now does nothing again.

--Dan

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]