[LLVMdev] Is this a bug in clang?

[Jean-Daniel Dupas]

You should search "Nasal daemons" in Google.

Le 20 avr. 2011 à 04:40, Robby Findler a écrit :

> So... Are 40 and 41 the only legal behaviors or are there more?
> 
> Robby
> 
> On Tuesday, April 19, 2011, Ahmed Charles <ahmedcharles at gmail.com> wrote:
>> This code is undefined, meaning that all bets are off, don't do it.
>> I.e. It reads the value of I between two sequence points and uses it
>> for something other than determining the value written. From: Csaba
>> Raduly
>> Sent: Tuesday, April 19, 2011 3:44 AM
>> To: Joe Armstrong
>> Cc: llvmdev at cs.uiuc.edu
>> Subject: Re: [LLVMdev] Is this a bug in clang?
>> Hi Joe
>> 
>> On Tue, Apr 19, 2011 at 10:59 AM, Joe Armstrong <joearms at gmail.com> wrote:
>>> Hello,
>>> 
>>> Is this a bug in clang, or a bug in my thinking?
>>> 
>>> /Joe Armstrong
>>> 
>>> 
>>> 
>>> When I compile the following program I get different answers in clang and gcc.
>>> 
>>> int printf(const char * format, ...);
>> 
>> Nitpick: you shouldn't do that. Now you have a potential mismatch
>> between whet you compile and what you link to.
>> 
>>> 
>>> int main()
>>> {
>>>  int i, j;
>>>  i = 10;
>>>  j = i++ + 20 + i;
>>>  printf("j = %d\n",j);
>>>  return(0);
>>> }
>> 
>>> 
>>>  $ gcc bug2.c
>>>  $ ./a.out
>>>  j = 40
>>>  $ clang bug2.c
>>>  $ ./a.out
>>>  j = 41
>>> 
>>>  I think the correct answer is 41. If my understanding of C is correct
>>>  (which, or course, it might not be) the incremented value of i++ is
>>>  first made available at the next sequence point, which is after the
>>>  ';' at the end of the offending statement statement.
>>> 
>> 
>> There is no such guarantee. From the C99 standard,
>> 
>> 5.1.2.3 Program execution:
>> } 2 ... At certain specified points in the execution sequence called
>> sequence points, all side effects
>> } of previous evaluations shall be complete and no side effects of
>> subsequent evaluations
>> } shall have taken place.
>> 
>> The side effect is made available _no_later_ than the sequence point.
>> There is no guarantee that it would not be made available earlier.
>> 
>> An example from the standard:
>> 
>> #include <stdio.h>
>> int sum;
>> char *p;
>> /* ... */
>> sum = sum * 10 - ’0’ + (*p++ = getchar());
>> 
>> the expression statement is grouped as if it were written as
>> 
>> sum = (((sum * 10) - ’0’) + ((*(p++)) = (getchar())));
>> 
>> but the actual increment of p can occur at any time between the
>> previous sequence point and the next
>> sequence point (the ;), and the call to getchar can occur at any point
>> prior to the need of its returned
>> value.
>> /quote
>> 
>> Also,
>> 
>> 6.5 Expressions
>> } 2 Between the previous and next sequence point an object shall have
>> its stored value
>> }    modified at most once by the evaluation of an expression.
>> Furthermore, the prior value
>> }    shall be accessed only to determine the value to be stored.60)
>> 
>> Footnote 60:
>> } This paragraph renders undefined statement expressions such as
>> }  i = ++i + 1;
>> }  a[i++] = i;
>> } while allowing
>> }  i = i + 1;
>> }  a[i] = i;
>> 
>> 
>> If you are lucky, the behavior of your program is unspecified (each
>> implementation is required to behave consistently with itself, which
>> is why GCC and clang are allowed to produce different results).
>> 
>> If not, you invoked the dreaded undefined behavior.
>> 
>> Csaba
>> --
>> GCS a+ e++ d- C++ ULS$ L+$ !E- W++ P+++$ w++$ tv+ b++ DI D++ 5++
>> The Tao of math: The numbers you can count are not the real numbers.
>> Life is complex, with real and imaginary parts.
>> "Ok, it boots. Which means it must be bug-free and perfect. " -- Linus Torvalds
>> "People disagree with me. I just ignore them." -- Linus Torvalds
>> 
>> _______________________________________________
>> LLVM Developers mailing list
>> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>> 
>> _______________________________________________
>> LLVM Developers mailing list
>> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>> 
> 
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev

-- Jean-Daniel