This is the mail archive of the gcc-patches@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: [tuples][patch] Convert pass_ccp and pass_store_ccp to tuples


On 3/4/08 4:34 PM, Bill Maddox wrote:

See add_ssa_edge in tree-ssa-propagate.c.  This is the only place where the flag
is set true, and it guards pushing the edge onto the queue.  I
discovered this bug
by tracing the queue/dequeue operations extensively, and it was clear
what was going
on in this case.

But it doesn't matter in the case of PHI nodes, because those are *always* simulated.


Mainline does not have this code, so in theory you should be able to reproduce the bug there. There's nothing in the tuples code that should require changes in the propagation logic.

So, if this is needed in tuples, it should also be needed in mainline. Can you send me a test case?


Diego.



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