[llvm] r219223 - Two case switch to select optimization

[Marcello Maggioni]

Hmm,

compiling your test I've noticed that even before entering the
SwitchToSelect optimization the PHI that breaks the module already has two
incoming edges for the same predecessor:

This is the IR just entering SwitchToSelect:

; Function Attrs: nounwind ssp uwtable

define i32 @fn1() #0 {

entry:

  %0 = load i32* @b, align 4

  %tobool = icmp eq i32 %0, 0

  br i1 %tobool, label %if.end3, label %if.then


if.then:                                          ; preds = %entry

  %1 = load i32* @a, align 4

  switch i32 %1, label %if.end3 [

    i32 5, label %return

    i32 0, label %return

  ]


if.end3:                                          ; preds = %if.then, %entry

  br label %return


return:                                           ; preds = %if.then,
%if.then, %if.end3

  %retval.0 = phi i32 [ 0, %if.end3 ], [ %1, %if.then ], [ %1, %if.then ]

  ret i32 %retval.0

}

Is this condition expected to be handled by the switch lowering? Or is the
switch formation that probably introduced this malformed?
Is pretty easily to workaround this in SwitchToSelect to reiterate Incoming
Edge deletion when the switch is removed until no more edges from the same
predecessor are found, but I guess if we have some other issue more up in
the pipeline.

2014-10-12 14:35 GMT-07:00 Marcello Maggioni <hayarms at gmail.com>:

> Thanks again Joerg, I missed the attachment!
>
> Marcello
>
> 2014-10-12 13:50 GMT-07:00 Joerg Sonnenberger <joerg at britannica.bec.de>:
>
> On Sun, Oct 12, 2014 at 01:09:25PM -0700, Marcello Maggioni wrote:
>> > Thanks Joerg for reverting and thanks Jiangning for reporting the
>> problem.
>> >
>> > Would it be possible to have the snippet of the code that is causing the
>> > issue or is it not publicly available? In order to be able to debug this
>> > more effectively and safely.
>>
>> See my earlier post in this thread for a test case.
>>
>> Joerg
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20141012/bb354386/attachment.html>