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: Test failure, documentation bugs?


Luca Benini wrote:
O0 -fcprop-registers -fdefer-pop -fguess-branch-probability -fif-conversion -fif-conversion2 -floop-optimize -fmerge-constants -fthread-jumps -fmerge-constants -fdelayed-branch -fthread-jumps -funit-at-a-time -fexceptions -fnon-call-exception
...Using all -O1 flag in documentation work.

-O0 means no optimization at all. If you say "-O0 -feverything" then you still get no optimization, because you used -O0.


What you want to do here is use "-O1 -fno-cprop-registers ...". You can selectively disable optimizations to try to determine which one is causing trouble.

However, keep in mind that not all optimizations can be explicitly enabled or disabled. When you use -O1, you will get some basic optimizations always. There is no -fno-combine or -fno-cse option for instance. It is highly likely that if you do "-O1 -fno-everything" that the testcase will still fail. This trick of trying -f flags is mainly useful when something works at -O1 but does not work at -O2, or vice versa.
--
Jim Wilson, GNU Tools Support, http://www.SpecifixInc.com



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