[Bug c/13873] New: gcc/longlong.h uses test for PPC

[Bug c/13873] New: gcc/longlong.h uses test for PPC

[sean at mcneil dot com]

PPC is much too generic a define for testing to see if it is a PowerPC target. 
This is found to complicate things with a vxWorks target other than PowerPC. For
vxWorks, PPC is defined (along with many other CPUs) all the time - regardless
of the target CPU type.

There are two tests for PPC.  They should be modified or removed.

-- 
           Summary: gcc/longlong.h uses test for PPC
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: c
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: sean at mcneil dot com
                CC: gcc-bugs at gcc dot gnu dot org
  GCC host triplet: i386-unknown-freebsd5.2
GCC target triplet: mips-wrs-vxworks


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13873

[Bug other/13873] gcc/longlong.h uses test for PPC

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-01-27 05:53 -------
Two things, longlong.c comes from GMP, report it to them, we can take a patch iff you sent it 
upstream also.
Please just send a patch to gcc-patches@gcc.gnu.org.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
          Component|c                           |other
     Ever Confirmed|                            |1
           Keywords|                            |build
   Last reconfirmed|0000-00-00 00:00:00         |2004-01-27 05:53:49
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13873

[Bug other/13873] gcc/longlong.h uses test for PPC

[sean at mcneil dot com]

------- Additional Comments From sean at mcneil dot com  2004-01-27 14:06 -------
Subject: Re:  gcc/longlong.h uses test for PPC

This is a restatement of the bug in longlong.h for the sake of reporting
to GMP.  I've also included a simple patch to remove to test of PPC. 
PPC is too generic of a name to test for the target type.  In vxWorks,
there are numerous definitions of CPU values and PPC is one of them. 
Regardless of the target type.  Tests like these should always be
preceded with at least one underscore.  The most proper tests should be
for __cputype__ and these are already accounted for.  The comment
specifies it to be for GNU/Linux, SysV.  Please use another test if
required.  __powerpc__ should have been good enough, however.

*** gcc/longlong.h      22 Jan 2004 00:15:25 -0000      1.41
--- gcc/longlong.h      27 Jan 2004 13:59:56 -0000
*************** UDItype __umulsidi3 (USItype, USItype);
*** 639,645 ****
       || defined (__powerpc__) /* gcc */                               \
       || defined (__POWERPC__) /* BEOS */                              \
       || defined (__ppc__)     /* Darwin */                            \
-      || defined (PPC)         /* GNU/Linux, SysV */                   \
       ) && W_TYPE_SIZE == 32
  #define add_ssaaaa(sh, sl, ah, al, bh, bl) \
    do {                                                                        \
--- 639,644 ----
*************** UDItype __umulsidi3 (USItype, USItype);
*** 677,683 ****
    __asm__ ("{cntlz|cntlzw} %0,%1" : "=r" (count) : "r" (x))
  #define COUNT_LEADING_ZEROS_0 32
  #if defined (_ARCH_PPC) || defined (__powerpc__) || defined (__POWERPC__) \
!   || defined (__ppc__) || defined (PPC)
  #define umul_ppmm(ph, pl, m0, m1) \
    do {                                                                        \
      USItype __m0 = (m0), __m1 = (m1);                                 \
--- 676,682 ----
    __asm__ ("{cntlz|cntlzw} %0,%1" : "=r" (count) : "r" (x))
  #define COUNT_LEADING_ZEROS_0 32
  #if defined (_ARCH_PPC) || defined (__powerpc__) || defined (__POWERPC__) \
!   || defined (__ppc__)
  #define umul_ppmm(ph, pl, m0, m1) \
    do {                                                                        \
      USItype __m0 = (m0), __m1 = (m1);                                 \

On Mon, 2004-01-26 at 21:53, pinskia at gcc dot gnu dot org wrote:
> ------- Additional Comments From pinskia at gcc dot gnu dot org  2004-01-27 05:53 -------
> Two things, longlong.c comes from GMP, report it to them, we can take a patch iff you sent it 
> upstream also.
> Please just send a patch to gcc-patches@gcc.gnu.org.



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13873

[Bug other/13873] gcc/longlong.h uses test for PPC

[rsandifo at gcc dot gnu dot org]

------- Additional Comments From rsandifo at gcc dot gnu dot org  2004-01-27 22:20 -------
Changing target... this is a powerpc rather than a mips PR.


-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
 GCC target triplet|mips-wrs-vxworks            |powerpc-wrs-vxworks


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13873

[Bug other/13873] gcc/longlong.h uses test for PPC

[sean at mcneil dot com]

------- Additional Comments From sean at mcneil dot com  2004-01-27 22:29 -------
Subject: Re:  gcc/longlong.h uses test for PPC

Actually, It is a problem building for mips as the fact that PPC is
defined, we end up with a powerpc instruction inserted into the prologue
of a function.

Please change back to a mips target bug.

Thanks,
Sean

On Tue, 2004-01-27 at 14:20, rsandifo at gcc dot gnu dot org wrote:
> ------- Additional Comments From rsandifo at gcc dot gnu dot org  2004-01-27 22:20 -------
> Changing target... this is a powerpc rather than a mips PR.
> 



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13873

[Bug other/13873] gcc/longlong.h uses test for PPC

[user42 at zip dot com dot au]

------- Additional Comments From user42 at zip dot com dot au  2004-02-09 23:58 -------
Subject: Re:  gcc/longlong.h uses test for PPC

Sean McNeil <sean@mcneil.com> writes:
>
> This is a restatement of the bug in longlong.h for the sake of reporting
> to GMP.

Thanks.

> PPC is too generic of a name to test for the target type.

I think I'd seen it somewhere under config/rs6000 in a past version of
gcc, though I see it's gone now.  (I don't think I just made it up :-)

> Please use another test if required.  __powerpc__ should have been
> good enough, however.

In gmp we make a HAVE_HOST_CPU_FAMILY_powerpc, driven from the --host
tuple.  I think we might use that, to remove any doubt.  (It's already
used in other places.)


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13873

[Bug other/13873] gcc/longlong.h uses test for PPC

[sean at mcneil dot com]

------- Additional Comments From sean at mcneil dot com  2004-02-10 00:04 -------
Subject: Re:  gcc/longlong.h uses test for PPC

That would be great.  Do you want to close the bug reported to gcc, or
would you like me to do that?

Thanks,
Sean

On Mon, 2004-02-09 at 15:58, Kevin Ryde wrote:
> Sean McNeil <sean@mcneil.com> writes:
> >
> > This is a restatement of the bug in longlong.h for the sake of reporting
> > to GMP.
> 
> Thanks.
> 
> > PPC is too generic of a name to test for the target type.
> 
> I think I'd seen it somewhere under config/rs6000 in a past version of
> gcc, though I see it's gone now.  (I don't think I just made it up :-)
> 
> > Please use another test if required.  __powerpc__ should have been
> > good enough, however.
> 
> In gmp we make a HAVE_HOST_CPU_FAMILY_powerpc, driven from the --host
> tuple.  I think we might use that, to remove any doubt.  (It's already
> used in other places.)
> 



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13873

[Bug other/13873] gcc/longlong.h uses test for PPC

[user42 at zip dot com dot au]

------- Additional Comments From user42 at zip dot com dot au  2004-02-10 00:12 -------
Subject: Re:  gcc/longlong.h uses test for PPC

Sean McNeil <sean@mcneil.com> writes:
>
> That would be great.  Do you want to close the bug reported to gcc, or
> would you like me to do that?

I'm looking only at the gmp side.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13873

[Bug other/13873] gcc/longlong.h uses test for PPC

[sean at mcneil dot com]

------- Additional Comments From sean at mcneil dot com  2004-02-10 04:20 -------
Awaiting GMP fix and import into gcc.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |SUSPENDED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13873

[Bug other/13873] gcc/longlong.h uses test for PPC

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-03-05 00:04 -------
*** Bug 14437 has been marked as a duplicate of this bug. ***

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |mfhaigh at acm dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13873

[Bug other/13873] gcc/longlong.h uses test for PPC

[sean at mcneil dot com]

------- Additional Comments From sean at mcneil dot com  2004-03-05 04:45 -------
Subject: Re:  gcc/longlong.h uses test for PPC

Sorry, I take that back.  I see that TARGET_OS_CPP_BUILTINS defines
__PPC.  It looks like the author of longlong.h didn't realize that
builtin_define_std prepends __ to the string and thought it would be PPC
instead of __PPC.  And for vxWorks, a check for __ppc should be added.




-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13873

[Bug other/13873] gcc/longlong.h uses test for PPC

[mfhaigh at acm dot org]

------- Additional Comments From mfhaigh at acm dot org  2004-03-05 05:28 -------
(In reply to comment #13)
> Subject: Re:  gcc/longlong.h uses test for PPC
> 
> Sorry, I take that back.  I see that TARGET_OS_CPP_BUILTINS defines
> __PPC.  It looks like the author of longlong.h didn't realize that
> builtin_define_std prepends __ to the string and thought it would be PPC
> instead of __PPC.  And for vxWorks, a check for __ppc should be added.
> 

Upon closer inspection, builtin_define_std will actually surround the macro with
leading and following __'s, if the macro is in the user namespace:

  /* If it was in user's namespace...  */
  if (p != buff + 2)
    {
      /* Define the macro with leading and following __.  */

The attached patch is against the 3.4 branch, and cleans up a couple of
testcases as well.  It does what you suggested-- use __ppc for vxWorks, and
change PPC to __PPC__.


Mark F. Haigh

2004-03-04 Mark F. Haigh <mfhaigh@acm.org>

   * longlong.h: PPC macro cleanup (fixes vxWorks targets)
   * gcc/testsuite/gcc.dg/20020103-1.c: PPC macro usage changed to __PPC__
   * gcc/testsuite/gcc.dg/20020919-1.c: Same as 20020103-1.c




-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13873

[Bug other/13873] gcc/longlong.h uses test for PPC

[amodra at bigpond dot net dot au]

-- 

amodra at bigpond dot net dot au changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |3.4.6


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13873

[Bug other/13873] gcc/longlong.h uses test for PPC

[amodra at bigpond dot net dot au]

------- Comment #20 from amodra at bigpond dot net dot au  2005-12-06 10:06 -------
Fixed.


-- 

amodra at bigpond dot net dot au changed:

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13873

[Bug other/13873] gcc/longlong.h uses test for PPC

[amodra at gcc dot gnu dot org]

------- Comment #19 from amodra at gcc dot gnu dot org  2005-12-06 10:05 -------
Subject: Bug 13873

Author: amodra
Date: Tue Dec  6 10:05:47 2005
New Revision: 108113

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=108113
Log:
        PR other/13873
        * longlong.h: Merge PPC tests from upstream.


Modified:
    branches/gcc-3_4-branch/gcc/ChangeLog
    branches/gcc-3_4-branch/gcc/longlong.h


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13873

[Bug other/13873] gcc/longlong.h uses test for PPC

[amodra at gcc dot gnu dot org]

------- Comment #18 from amodra at gcc dot gnu dot org  2005-12-06 10:04 -------
Subject: Bug 13873

Author: amodra
Date: Tue Dec  6 10:04:04 2005
New Revision: 108112

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=108112
Log:
        PR other/13873
        * longlong.h: Merge PPC tests from upstream.


Modified:
    branches/gcc-4_0-branch/gcc/ChangeLog
    branches/gcc-4_0-branch/gcc/longlong.h


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13873

[Bug other/13873] gcc/longlong.h uses test for PPC

[amodra at gcc dot gnu dot org]

------- Comment #17 from amodra at gcc dot gnu dot org  2005-12-06 10:03 -------
Subject: Bug 13873

Author: amodra
Date: Tue Dec  6 10:02:57 2005
New Revision: 108111

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=108111
Log:
        PR other/13873
        * longlong.h: Merge PPC tests from upstream.


Modified:
    branches/gcc-4_1-branch/gcc/ChangeLog
    branches/gcc-4_1-branch/gcc/longlong.h


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13873

[Bug other/13873] gcc/longlong.h uses test for PPC

[amodra at gcc dot gnu dot org]

------- Comment #16 from amodra at gcc dot gnu dot org  2005-12-06 09:55 -------
Subject: Bug 13873

Author: amodra
Date: Tue Dec  6 09:55:04 2005
New Revision: 108110

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=108110
Log:
        PR other/13873
        * longlong.h: Merge PPC tests from upstream.


Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/longlong.h


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13873

[Bug other/13873] gcc/longlong.h uses test for PPC

[amodra at bigpond dot net dot au]

-- 

amodra at bigpond dot net dot au changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |amodra at bigpond dot net
                   |dot org                     |dot au
             Status|SUSPENDED                   |ASSIGNED
   Last reconfirmed|2005-10-02 20:55:10         |2005-12-02 12:42:31
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13873