[Bug c/10719] invalid code generated (x86, "int $5") with __builtin_va_arg(va, char);

[Bug c/10719] invalid code generated (x86, "int $5") with __builtin_va_arg(va, char);

[pinskia at physics dot uc dot edu]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10719


pinskia at physics dot uc dot edu changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |kostya at tessart dot kiev
                   |                            |dot ua


------- Additional Comments From pinskia at physics dot uc dot edu  2003-07-08 13:07 -------
*** Bug 11464 has been marked as a duplicate of this bug. ***

[Bug c/10719] invalid code generated (x86, "int $5") with __builtin_va_arg(va, char);

[pinskia at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10719


pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |masoud dot kermani at
                   |                            |latticesemi dot com


------- Additional Comments From pinskia at gcc dot gnu dot org  2003-10-15 19:05 -------
*** Bug 12627 has been marked as a duplicate of this bug. ***

[Bug c/10719] invalid code generated (x86, "int $5") with __builtin_va_arg(va, char);

[pinskia at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10719



------- Additional Comments From pinskia at gcc dot gnu dot org  2003-10-15 19:16 -------
*** Bug 12627 has been marked as a duplicate of this bug. ***

[Bug c/10719] invalid code generated (x86, "int $5") with __builtin_va_arg(va, char);

[mkermani at latticesemi dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10719



------- Additional Comments From mkermani at latticesemi dot com  2003-10-15 19:33 -------
Subject: This is in regards to Bug 10719 and 12627

The issue is what should happen when the code,
 
   short x = va_arg(ap, short)
   
is executed. On NT and Solaris and any other compiler
this code works fine.  On GCC 3.3.1 it seg-faults.

Now the code needs to be conditionally compiled for GCC 
because it is not dealt with like all other compilers.
You can point to an ANSI or ISO documents which may
say this is undefined behavior, but if your are swimming 
up stream you are not doing it the right way.  It appears 
to me GCC should deal with this in the right way like 
all other compilers do. To say the behavior is undefined
it does not mean GCC should deal with it the worst possible 
way.  It could also mean hey we deal with it.  Why not 
choose the second.

Masoud.

--
Masoud Fadai Kermani, 
masoud.kermani@latticesemi.com
The man who views the world at 50 the same as he did at 20 has 
wasted 30 years of his life. (Muhammad Ali)

[Bug c/10719] invalid code generated (x86, "int $5") with __builtin_va_arg(va, char);

[pinskia at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10719



------- Additional Comments From pinskia at gcc dot gnu dot org  2003-10-15 19:44 -------
Here is the quote from the C standard, from a search of "va_arg undefined behavior promotion":
<http://lists.debian.org/debian-powerpc/2000/debian-powerpc-200010/msg00384.html>:
Here is the relevant part of ISO C Working Draft, 1997-11-21, WG14/N794 J11/97-158:

     7.12.1.2 The va_arg macro

     Synopsis

1               #include <stdarg.h>
                type va_arg(va_list ap, type);

     Description

2 The va_arg macro expands to an expression that has the type and value of the next
     argument in the call.  The parameter ap shall be the same as the va_list ap
     initialized by va_start.  Each invocation of va_arg modifies ap so that the values
     of successive arguments are returned in turn.  The parameter type is a type name
     specified such that the type of a pointer to an object that has the specified type can be
     obtained simply by postfixing a * to type.  If there is no actual next argument, or if
     type is not compatible with the type of the actual next argument (as promoted
     according to the default argument promotions), the behavior is undefined.

[Bug c/10719] invalid code generated (x86, "int $5") with __builtin_va_arg(va, char);

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-06-05 07:29 -------
Reopening to ...

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|FIXED                       |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10719

[Bug c/10719] invalid code generated (x86, "int $5") with __builtin_va_arg(va, char);

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-06-05 07:29 -------
Mark as invalid.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|REOPENED                    |RESOLVED
         Resolution|                            |INVALID


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10719

[Bug c/10719] invalid code generated (x86, "int $5") with __builtin_va_arg(va, char);

[appfault at hotmail dot com]

------- Additional Comments From appfault at hotmail dot com  2005-09-11 08:04 -------
Invalid?  So whenever behavior is undefined by the C standard, would it be ok 
to delete the user's harddrive as well?  This is ridiculous - fix the bug.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10719

[Bug c/10719] invalid code generated (x86, "int $5") with __builtin_va_arg(va, char);

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-09-11 14:34 -------
(In reply to comment #12)
> Invalid?  So whenever behavior is undefined by the C standard, would it be ok 
> to delete the user's harddrive as well?  This is ridiculous - fix the bug.

It is undefined which means anything can happen.  The comment about delete the user's harddrive is 
mainly a joke about the C standard says about this.

There is no bug here except in your code as the variable is passed as an int and not as a char.
To fix your code, replace char with int and it works just fine.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10719

[Bug c/10719] invalid code generated (x86, "int $5") with __builtin_va_arg(va, char);

[appfault at hotmail dot com]

------- Additional Comments From appfault at hotmail dot com  2005-09-12 06:17 -------
I agree that gcc apparently complies with the standard as currently 
implemented.  So this zilla is not a defect, but an enhancement request.

Why not reopen this to add a -Wundefined-behavior, so that at least bugs like 
that could be caught up front when using -Werror?

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10719

[Bug c/10719] invalid code generated (x86, "int $5") with __builtin_va_arg(va, char);

[falk at debian dot org]

------- Additional Comments From falk at debian dot org  2005-09-12 19:19 -------
(In reply to comment #14)

> Why not reopen this to add a -Wundefined-behavior, so that at least bugs like 
> that could be caught up front when using -Werror?

There is already an unconditional warning, so what would be the point?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10719

[Bug c/10719] invalid code generated (x86, "int $5") with __builtin_va_arg(va, char);

[appfault at hotmail dot com]

------- Additional Comments From appfault at hotmail dot com  2005-09-12 23:34 -------
Oh?  I had -Werror on, and was not getting any warning at all from my code that 
was generating 'int $0x5' with gcc 3.4.1.  It's perhaps a slightly different 
case than comment 0, because I was casting an int to a char* in the varargs?

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10719

[Bug c/10719] invalid code generated (x86, "int $5") with __builtin_va_arg(va, char);

[falk at debian dot org]

------- Additional Comments From falk at debian dot org  2005-09-13 06:19 -------
(In reply to comment #16)
> Oh?  I had -Werror on, and was not getting any warning at all from my code that 
> was generating 'int $0x5' with gcc 3.4.1.  It's perhaps a slightly different 
> case than comment 0, because I was casting an int to a char* in the varargs?

Please attach a test case.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10719

[Bug c/10719] invalid code generated (x86, "int $5") with __builtin_va_arg(va, char);

[appfault at hotmail dot com]

------- Additional Comments From appfault at hotmail dot com  2005-09-14 00:16 -------
Ok, disregard comment 16, the issue I saw was the same as comment 0.  
Unfortunately, there was a '-w' sneakily in a 3rd-party makefile which hid the 
warning.  Maybe I should open another zilla for warning you that '-w -Werror' 
is nonsensical.

I still say generating code that is not executable is a ridiculous way to 
handle this ambiguity in the standard...

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10719

[Bug c/10719] invalid code generated (x86, "int $5") with __builtin_va_arg(va, char);

[ebotcazou at gcc dot gnu dot org]

------- Additional Comments From ebotcazou at gcc dot gnu dot org  2005-09-14 06:15 -------
> I still say generating code that is not executable is a ridiculous way to 
> handle this ambiguity in the standard...

You still don't get the point.  Read again comment #2, the bottom line is that
it's the best code the compiler can generate portably.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10719

[Bug c/10719] invalid code generated (x86, "int $5") with __builtin_va_arg(va, char);

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-09-14 06:25 -------
"as promoted according to the default argument promotions" is what makes this undefined by the way.  
char is promoted to int.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10719

[Bug c/10719] invalid code generated (x86, "int $5") with __builtin_va_arg(va, char);

[appfault at hotmail dot com]

------- Additional Comments From appfault at hotmail dot com  2005-09-14 16:09 -------
Ok, so that's the best code it can generate, fine.  So if instant segfault is 
the best possible generated code, I think NOT generating any code is far more 
helpful to the user.  If not generating any code doesn't conform to the 
standard, then it seems like it should be at least optional.  Who knows how 
many of these "int $5" ticking timebombs are lurking out there in mission 
critical software.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10719

[Bug c/10719] invalid code generated (x86, "int $5") with __builtin_va_arg(va, char);

[ebotcazou at gcc dot gnu dot org]

------- Additional Comments From ebotcazou at gcc dot gnu dot org  2005-09-14 16:23 -------
> Ok, so that's the best code it can generate, fine.  So if instant segfault is 
> the best possible generated code, I think NOT generating any code is far more 
> helpful to the user.  If not generating any code doesn't conform to the 
> standard, then it seems like it should be at least optional.

-Werror precisely exists for that purpose.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10719

[Bug c/10719] invalid code generated (x86, "int $5") with __builtin_va_arg(va, char);

[appfault at hotmail dot com]

------- Additional Comments From appfault at hotmail dot com  2005-09-15 04:22 -------
Yes well I don't think you should have to go out of your way to ask the 
compiler to not generate invalid code.  Not generating invalid code should be 
the default behavior.

We're talking about the difference between software that's POSSIBLE to use, and 
software that's EASY to use.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10719

[Bug c/10719] invalid code generated (x86, "int $5") with __builtin_va_arg(va, char);

[ebotcazou at gcc dot gnu dot org]

------- Additional Comments From ebotcazou at gcc dot gnu dot org  2005-09-15 04:52 -------
> Yes well I don't think you should have to go out of your way to ask the 
> compiler to not generate invalid code.  Not generating invalid code should be 
> the default behavior.

Again the compiler violates the ISO C standard if it stops.  I agree that this
is counter-intuitive, but you have an unconditional warning.  Simply ban -w.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10719

[Bug c/10719] invalid code generated (x86, "int $5") with __builtin_va_arg(va, char);

[appfault at hotmail dot com]

------- Comment #25 from appfault at hotmail dot com  2005-11-06 07:26 -------
How does stopping violate the standard?  If the standard says behavior is
undefined, then you can do anything you want, including stopping.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10719

[Bug c/10719] invalid code generated (x86, "int $5") with __builtin_va_arg(va, char);

[ebotcazou at gcc dot gnu dot org]

------- Comment #26 from ebotcazou at gcc dot gnu dot org  2005-11-06 08:42 -------
> How does stopping violate the standard?  If the standard says behavior is
> undefined, then you can do anything you want, including stopping.

You're confusing compile time and run time.  Please read the whole audit trail,
in particular comment #3.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10719