This is the mail archive of the gcc-bugs@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]

[Bug tree-optimization/61607] DOM missed jump threading and destroyed loops


https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61607

Jeffrey A. Law <law at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |law at redhat dot com

--- Comment #7 from Jeffrey A. Law <law at redhat dot com> ---
In my testing, the missed thread is due to not following the chain of
SSA_NAME_VALUEs.  We ought to easily discover that the two conditionals outside
the loop are in fact equivalent and fully threadable.


With a quick hack to follow the chain of values we're able to trivially
discover both of the interesting jump threading opportunities:

k.c.079t.dom1:  Registering jump thread: (6, 7) incoming edge;  (7, 8) normal;
k.c.079t.dom1:  Registering jump thread: (5, 7) incoming edge;  (7, 9) normal;


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