[Bug c/18180] New: Disable special handling of functions with --freestanding

[Bug c/18180] New: Disable special handling of functions with --freestanding

[alex dot neyman at auriga dot ru]

The documentation on GCC states that in freestanding environment, no 
assumptions are made: 
 
======== 
To make it act as a conforming freestanding 
implementation for a freestanding environment, use the option 
`-ffreestanding'; it will then define `__STDC_HOSTED__' to `0' and not 
make assumptions about the meanings of function names from the standard 
library 
======== 
 
However, a simple testcase shows that 
- different code is generated for function issuing the setjmp() call and any 
other function call in freestanding env (actually, the code for a function 
using setjmp() is the same both in hosted and freestanding env). 
 
- The 'variable might be clobbered' warning is emitted for setjmp() call even 
in freestanding env 
 
Try compiling the test case with the following flags: 
*1) gcc -DSETJMP=setjmp -ffreestanding -s ff.c 
*2) gcc -DSETJMP=setjmp -fhosted -s ff.c 
*3) gcc -DSETJMP=mysetjmp -ffreestanding -s ff.c 
 
and compare the output. *1 and *2 give the same output, which differs from *3.

-- 
           Summary: Disable special handling of functions with --
                    freestanding
           Product: gcc
           Version: 4.0.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: c
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: alex dot neyman at auriga dot ru
                CC: gcc-bugs at gcc dot gnu dot org
 GCC build triplet: i686-pc-linux-gnu
  GCC host triplet: i686-pc-linux-gnu
GCC target triplet: i686-pc-linux-gnu


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18180

[Bug c/18180] Disable special handling of functions with --freestanding

[alex dot neyman at auriga dot ru]

------- Additional Comments From alex dot neyman at auriga dot ru  2004-10-27 14:38 -------
Created an attachment (id=7416)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=7416&action=view)
The test case


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18180

[Bug c/18180] Disable special handling of functions with --freestanding

[alex dot neyman at auriga dot ru]

------- Additional Comments From alex dot neyman at auriga dot ru  2004-10-27 14:55 -------
Created an attachment (id=7418)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=7418&action=view)
Proposed patch

The patch, however, leads to a build failure:
optabs.h and c-common.h have a clash with CTI_MAX enumeration value.
Patch was tested by rename with CTI_MAX in optabs.[ch] renamed to CTI_max.
(not regression-tested, though). The resulting assembly is the same in *2 and
*3, and differs from *1. No "might be clobbered" warnings are generated
in -freestanding mode.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18180

[Bug c/18180] Disable special handling of functions with --freestanding

[alex dot neyman at auriga dot ru]

------- Additional Comments From alex dot neyman at auriga dot ru  2004-10-27 14:56 -------
Just to make it clear: this test case never produces "variable might be 
clobbered" warning, another test case was used for that. 

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18180

[Bug c/18180] Disable special handling of functions with --freestanding

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-10-27 15:00 -------
Try with -fno-builtins.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18180

[Bug c/18180] Disable special handling of functions with --freestanding

[alex dot neyman at auriga dot ru]

------- Additional Comments From alex dot neyman at auriga dot ru  2004-10-27 15:09 -------
Tried, the result of "-ffreestanding -fno-builtin" is the same as just 
"-ffreestanding". Also, IIRC -ffreestanding implies -fno-builtin, doesn't it? 

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18180

[Bug c/18180] Disable special handling of functions with --freestanding

[jsm at polyomino dot org dot uk]

------- Additional Comments From jsm at polyomino dot org dot uk  2004-10-27 15:39 -------
Subject: Re:  New: Disable special handling of functions with
 --freestanding

On Wed, 27 Oct 2004, alex dot neyman at auriga dot ru wrote:

> - different code is generated for function issuing the setjmp() call and any 
> other function call in freestanding env (actually, the code for a function 
> using setjmp() is the same both in hosted and freestanding env). 

The bug would be that special_function_p checks function names at all.

We have at least three different methods of applying special properties to 
functions based on their names: built-in functions with attributes, 
special_function_f and cp/cfns.gperf.  We should subsume the latter two in 
the first.  See projects/beginner.html.  The special handling may be 
needed to generate correct code in some cases such as setjmp, so disabling 
the special handling without first having an attribute allowing it to be 
reenabled for particular declarations is not a good solution.

I would advise following the usual procedures for contributing (see 
contribute.html) to discuss the issues on gcc-patches and work through 
getting a patch for bug 14411 into GCC.  That is necessary to solve the 
issues for bug 14411.  It would be best to do a series of patches, one for 
each attribute (discussed and revised as necessary) followed by one 
replacing the special_function_p handling with use of that attribute 
(including fixincludes work as necessary).



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18180

[Bug c/18180] Disable special handling of functions with --freestanding

[alex dot neyman at auriga dot ru]

------- Additional Comments From alex dot neyman at auriga dot ru  2004-10-27 16:01 -------
The bug 14411 was submitted by me as well, though from a different email 
address. I updated the patch in it to apply to 4.0. 
 
 

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18180

[Bug c/18180] Disable special handling of functions with --freestanding

[jsm at polyomino dot org dot uk]

------- Additional Comments From jsm at polyomino dot org dot uk  2004-10-27 17:05 -------
Subject: Re:  Disable special handling of functions with
 --freestanding

On Wed, 27 Oct 2004, alex dot neyman at auriga dot ru wrote:

> The bug 14411 was submitted by me as well, though from a different email 
> address. I updated the patch in it to apply to 4.0. 

You will need to *send the patch to gcc-patches* for reviewers to look at 
it, and iterate as necessary until a version is approved.  See the full 
instructions at <http://gcc.gnu.org/contribute.html> first.  You will need 
to include testcases in the patch, and I suggest naming the setjmp 
attribute something like returns_twice - make it general so it can handle 
vfork as well.  If you don't have the copyright assignment paperwork on 
file yet, start getting it in now.  I don't know whether the present patch 
will require it, but a series of successive patches along those lines 
will.

++ @code{unused}, @code{deprecated}, @code{weak}, @code{malloc}, @code{setjmp},
++ @code{longjmp}, @code{alias}, and @code{nonnull}.  Several other attributes
++ are defined for functions on particular target systems.  Other attributes,
++ including @code{section} are supported for variables declarations

in your patch looks wrong; why add the lines beginning with an extra "+"?

You need two spaces after a "." at the end of a sentence in Texinfo.

K&R-style function definitions are no longer used in GCC.

When adding flags to tree_decl you need to update the comment lower down 
about how many spare bits there are (before another 32-bit word would be 
used).  Note the flags are listed grouped into bytes in the declaration.

`%s' quoting is no longer used in GCC; you should use %qs instead.



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18180

[Bug c/18180] move special handling out functions into attributes

[pinskia at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|normal                      |enhancement
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1
           Keywords|                            |diagnostic, wrong-code
   Last reconfirmed|0000-00-00 00:00:00         |2004-10-29 17:38:20
               date|                            |
            Summary|Disable special handling of |move special handling out
                   |functions with --           |functions into attributes
                   |freestanding                |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18180

[Bug c/18180] move special handling out functions into attributes

[pinskia at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|2004-10-29 17:38:20         |2005-01-28 06:28:44
               date|                            |
   Target Milestone|---                         |4.1.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18180

[Bug c/18180] move special handling out functions into attributes

[egallager at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=18180

Eric Gallager <egallager at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |FIXED
             Status|NEW                         |RESOLVED
                 CC|                            |egallager at gcc dot gnu.org

--- Comment #10 from Eric Gallager <egallager at gcc dot gnu.org> ---
(In reply to Andrew Pinski from comment #9)
> I think this is all fixed with the patch for PR 71876.

ok, closing as FIXED then

[Bug c/18180] move special handling out functions into attributes

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=18180

--- Comment #9 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
I think this is all fixed with the patch for PR 71876.

[Bug c/18180] move special handling out functions into attributes

[pinskia at gcc dot gnu dot org]

-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.1.0                       |---


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18180