On 09/09/2016 10:56 PM, Segher Boessenkool wrote:
On Fri, Sep 09, 2016 at 12:57:32PM -0600, Jeff Law wrote:
I think the lack of test coverage is something we'll want to address.
Building and running the compiler, the various target libraries, and the
testsuite is more than enough coverage for correctness in my opinion --
I'd argue against that:
- a lot of compiler bugs for passes like this show up only for unusual
CFGs that you don't really get with normal input code.
- separate shrink wrapping explicitly tries to modify infrequent
paths, which may not even get exercised.
Also, unless I've missed it you've not provided any numbers about how
often this triggers, let's say on a cc1 build.
For these reasons I thought having self-tests would be a good idea, but
I was informed it's a waste of time, so I guess not.
Bernd
