[PATCH] Help? The zext-sext.ll test

[Jakob Stoklund Olesen]

On May 30, 2013, at 9:04 AM, Tim Northover <tnorthover at apple.com> wrote:

> Hi,
> 
> Having made a couple of botched attempts to fix this test I thought I'd see what people actually know about it.
> 
> Basically, it seems that scheduling means that the exact sequence being checked for may not occur depending on the host CPU. It *looks* like the test isn't trying to check that so the attached patch just fixes it at one which works at the moment.
> 
> Does anyone have any better suggestions or other ideas?

Can you disable scheduling completely / use source-order scheduling?

/jakob