[prev in list] [next in list] [prev in thread] [next in thread] 

List:       gcc
Subject:    Re: Can gcc.dg/torture/pr67828.c be an infinite loop?
From:       Aldy Hernandez via Gcc <gcc () gcc ! gnu ! org>
Date:       2021-09-24 11:41:22
Message-ID: a2d12afe-5825-2ea4-c6af-c001ab2b6a43 () redhat ! com
[Download RAW message or body]



On 9/24/21 1:37 PM, David Brown wrote:
> On 24/09/2021 10:03, Aldy Hernandez via Gcc wrote:
>> Hi folks.
>>
>> My upcoming threading improvements turn the test below into an infinite
>> runtime loop:
>>
>> int a, b;
>> short c;
>>
>> int
>> main ()
>> {
>>    int j, d = 1;
>>    for (; c >= 0; c++)
>>      {
>> BODY:
>>        a = d;
>>        d = 0;
>>        if (b)
>>      {
>>        xprintf (0);
>>        if (j)
>>          xprintf (0);
>>      }
>>      }
>>    xprintf (d);
>>    exit (0);
>> }
>>
>> On the false edge out of if(b) we thread directly to BODY, eliding the
>> loop conditional, because we know that c>=0 because it could never
>> overflow.
>>
>> Since B is globally initialized to 0, this has the effect of turning the
>> test into an infinite loop.
>>
>> Is this correct, or did I miss something?
>> Aldy
>>
>>
> 
> I am wondering about the claim that you can use "b" being 0 to optimise
> the conditional.  If the compiler knows that this is the complete
> program, that is fair enough.  But since "b" is not static, another
> compilation unit could modify "b" before "main" is called.  (In C, it is
> legal for another part of the code to call main() - perhaps the
> implementation of xprintf does so.  And in C++, a global constructor
> could change "b" before main() starts.)

In this case, it was on a thread where we knew we came through the c>=0 
conditional and we hadn't left the function.

Be that as it may, this was something else entirely.  The problem was a 
typo in the path solver that was causing it to use an incorrect range, 
which was then causing the elision.  It had nothing to do with promotion 
rules.  My bad.

Aldy

[prev in list] [next in list] [prev in thread] [next in thread]