[llvm] r251664 - Revert "[ARM] Remove XFAIL on test/CodeGen/Generic/MachineBranchProb.ll"

Renato Golin via llvm-commits llvm-commits at lists.llvm.org
Fri Oct 30 10:34:24 PDT 2015


On 30 October 2015 at 17:19, Weiming Zhao <weimingz at codeaurora.org> wrote:
> Is there any way to check a patch on buildbot before committing it?

I meant, check the buildbots. If it's also failing there, then, the
XFAIL need to be removed (with an explanation, not just "because it
started passing", as that could be instability on the test).

If the buildbots are green, then you have to understand why they pass
for you and not on the bots, in which case you'll have to do that
*before* committing the XPASS.


> Currently, we're not sure why the check below fails on the buildbot:
> test/CodeGen/Generic/MachineBranchProb.ll:18:10: error: expected string not found in input
> ; CHECK: Successors according to CFG: BB#2(64) BB#4(21)

I don't know either. It could be many things, including bad test,
different configuration, different setup, etc.

We have a large collection of bots on AArch64/ARM/Thumb, NEON/VFP,
single-stage/self-hosting, so it's unlikely that any of these factors
are responsible for it passing. We're currently not testing autoconf
any more, so that could be a reason, if you're using it. Regardless,
would be good to know what the difference is, so that we can make it
pass on the public bots, too.

cheers,
--renato


More information about the llvm-commits mailing list