[Bug c/26459] New: gcc-4.1.0 RC2 fails to build on certain PowerPC targets

[Bug c/26459] New: gcc-4.1.0 RC2 fails to build on certain PowerPC targets

[edmar at freescale dot com]

When configuring gcc with:
--target=powerpc-unknown-linux-gnuspe --host=powerpc-unknown-linux-gnuspe
--enable-e500_double
this file will be included:
gcc/config/rs6000/e500-double.h
which in turn uses the variable
rs6000_float_gprs_string
which is not defined anywhere. (So the build stops with undefined sysmbol
error)

This problem can be detected as far as 20060206.
Gcc 4.0.2 builds fine with the same configuration options.


-- 
           Summary: gcc-4.1.0 RC2 fails to build on certain PowerPC targets
           Product: gcc
           Version: 4.1.0
            Status: UNCONFIRMED
          Severity: critical
          Priority: P3
         Component: c
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: edmar at freescale dot com
 GCC build triplet: powerpc-unknown-linux-gnuspe
  GCC host triplet: powerpc-unknown-linux-gnuspe
GCC target triplet: powerpc-unknown-linux-gnuspe


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc-4.1.0 RC2 fails to build on powerpc e500-double targets

[pinskia at gcc dot gnu dot org]

------- Comment #1 from pinskia at gcc dot gnu dot org  2006-02-24 18:42 -------
It is obvious that it was caused by:
2005-05-05  Aldy Hernandez  <aldyh@redhat.com>
            David Edelsohn  <edelsohn@gnu.org>

        * config.gcc: Add .opt magic for the rs6000 ports.
.....


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|critical                    |normal
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|0                           |1
  GCC build triplet|powerpc-unknown-linux-gnuspe|
   GCC host triplet|powerpc-unknown-linux-gnuspe|
 GCC target triplet|powerpc-unknown-linux-gnuspe|powerpc-linux-gnuspe
   Last reconfirmed|0000-00-00 00:00:00         |2006-02-24 18:42:12
               date|                            |
            Summary|gcc-4.1.0 RC2 fails to build|[4.1/4.2 Regression] gcc-
                   |on certain PowerPC targets  |4.1.0 RC2 fails to build on
                   |                            |powerpc e500-double targets
   Target Milestone|---                         |4.1.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[pinskia at gcc dot gnu dot org]

------- Comment #2 from pinskia at gcc dot gnu dot org  2006-02-24 18:43 -------
It also tells us that testing powerpc-linux-gnuspe is not important as it has
been broken for over a half of a year.


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|[4.1/4.2 Regression] gcc-   |[4.1/4.2 Regression] gcc
                   |4.1.0 RC2 fails to build on |fails to build on powerpc
                   |powerpc e500-double targets |e500-double targets


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[edmar at freescale dot com]

------- Comment #3 from edmar at freescale dot com  2006-02-24 18:49 -------
(In reply to comment #2)
> It also tells us that testing powerpc-linux-gnuspe is not important as it has
> been broken for over a half of a year.
> 

We test all releases, the last one was 4.0.2 and it was OK.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[mmitchel at gcc dot gnu dot org]

------- Comment #4 from mmitchel at gcc dot gnu dot org  2006-02-24 19:04 -------
Edmar --

Does changing:

  if (rs6000_float_gprs_string == NULL) \
    rs6000_float_gprs = 2;

too:

  if (!rs6000_explicit_options.float_gprs) \
    rs6000_float_gprs = 2;

work?  Not just compile, but also generate the code you expect?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[edmar at freescale dot com]

------- Comment #5 from edmar at freescale dot com  2006-02-24 19:10 -------
(In reply to comment #4)
> Edmar --
> 
> Does changing:
> 
>   if (rs6000_float_gprs_string == NULL) \
>     rs6000_float_gprs = 2;
> 
> too:
> 
>   if (!rs6000_explicit_options.float_gprs) \
>     rs6000_float_gprs = 2;
> 
> work?  Not just compile, but also generate the code you expect?
> 

I started a build. This is a very slow target, it will take several hours...


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[mark at codesourcery dot com]

------- Comment #6 from mark at codesourcery dot com  2006-02-24 19:14 -------
Subject: Re:  [4.1/4.2 Regression] gcc fails to build on
 powerpc e500-double targets

edmar at freescale dot com wrote:

> I started a build. This is a very slow target, it will take several hours...

If you've got a fast workstation somewhere, you can set it up as a
cross-compiler; should take about five minutes to build a cross
compiler.  But, there's not going to be a release today, so a few hours
is not a big deal.

I would like to fix this by changing only a known-broken header file, if
possible, because that is a known-safe fix.  That would allow us to
support this target without risking any other platforms, where testing
has been performed all along.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[edmar at freescale dot com]

------- Comment #7 from edmar at freescale dot com  2006-02-24 20:45 -------
Created an attachment (id=10911)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=10911&action=view)
Input program that causes ICE on gcc-4.1.0 RC2

No deal. I got an ICE now:


[root:/local/edmar/obj_gcc/powerpc-unknown-linux-gnuspe/libiberty]
/local/edmar/obj_gcc/./gcc/xgcc -B/local/edmar/obj_gcc/./gcc/
-B/root/edmar/gcc-4.1-20060206/powerpc-unknown-linux-gnuspe/bin/
-B/root/edmar/gcc-4.1-20060206/powerpc-unknown-linux-gnuspe/lib/ -isystem
/root/edmar/gcc-4.1-20060206/powerpc-unknown-linux-gnuspe/include -isystem
/root/edmar/gcc-4.1-20060206/powerpc-unknown-linux-gnuspe/sys-include -c
-DHAVE_CONFIG_H -O2 -g -O2  -I.
-I../../../src/gcc4_sources/libiberty/../include  -W -Wall -pedantic
-Wwrite-strings -Wstrict-prototypes
../../../src/gcc4_sources/libiberty/floatformat.c -o floatformat.o -v
Reading specs from /local/edmar/obj_gcc/./gcc/specs
Target: powerpc-unknown-linux-gnuspe
Configured with: ../src/gcc4_sources/configure
--prefix=/root/edmar/gcc-4.1-20060206 --enable-languages=c,c++,f95
--with-as=/root/edmar/gcc-4.1-20060206/bin/as
--target=powerpc-unknown-linux-gnuspe --host=powerpc-unknown-linux-gnuspe
--enable-e500_double --disable-gdbtk --without-x
Thread model: posix
gcc version 4.1.0 20051026 (experimental)
 /local/edmar/obj_gcc/./gcc/cc1 -quiet -v -I.
-I../../../src/gcc4_sources/libiberty/../include -iprefix
/local/edmar/obj_gcc/gcc/../lib/gcc/powerpc-unknown-linux-gnuspe/4.1.0/
-isystem /local/edmar/obj_gcc/./gcc/include -D__unix__ -D__gnu_linux__
-D__linux__ -Dunix -D__unix -Dlinux -D__linux -Asystem=linux -Asystem=unix
-Asystem=posix -DHAVE_CONFIG_H -isystem
/root/edmar/gcc-4.1-20060206/powerpc-unknown-linux-gnuspe/include -isystem
/root/edmar/gcc-4.1-20060206/powerpc-unknown-linux-gnuspe/sys-include
../../../src/gcc4_sources/libiberty/floatformat.c -quiet -dumpbase
floatformat.c -auxbase-strip floatformat.o -g -O2 -O2 -W -Wall -pedantic
-Wwrite-strings -Wstrict-prototypes -version -o /tmp/ccl8DAd9.s
ignoring nonexistent directory
"/root/edmar/gcc-4.1-20060206/powerpc-unknown-linux-gnuspe/include"
ignoring nonexistent directory
"/root/edmar/gcc-4.1-20060206/powerpc-unknown-linux-gnuspe/sys-include"
ignoring nonexistent directory
"/local/edmar/obj_gcc/gcc/../lib/gcc/powerpc-unknown-linux-gnuspe/4.1.0/include"
ignoring nonexistent directory
"/local/edmar/obj_gcc/gcc/../lib/gcc/powerpc-unknown-linux-gnuspe/4.1.0/../../../../powerpc-unknown-linux-gnuspe/include"
ignoring nonexistent directory
"/root/edmar/gcc-4.1-20060206/lib/gcc/powerpc-unknown-linux-gnuspe/4.1.0/../../../../powerpc-unknown-linux-gnuspe/include"
#include "..." search starts here:
#include <...> search starts here:
 .
 ../../../src/gcc4_sources/libiberty/../include
 /local/edmar/obj_gcc/./gcc/include
 /usr/local/include
 /root/edmar/gcc-4.1-20060206/include

/root/edmar/gcc-4.1-20060206/lib/gcc/powerpc-unknown-linux-gnuspe/4.1.0/include
 /usr/include
End of search list.
GNU C version 4.1.0 20051026 (experimental) (powerpc-unknown-linux-gnuspe)
        compiled by GNU C version 2.95.4 20010319 (prerelease/franzo/20011204).
GGC heuristics: --param ggc-min-expand=30 --param ggc-min-heapsize=4096
Compiler executable checksum: a377e6d60b64947de4beadd578545a5b
../../../src/gcc4_sources/libiberty/floatformat.c: In function
'floatformat_to_double':
../../../src/gcc4_sources/libiberty/floatformat.c:382: error: unrecognizable
insn:
(insn 349 347 122 12 ../../../src/gcc4_sources/libiberty/floatformat.c:350 (set
(reg:DF 11 11)
        (mem/u/c/i:DF (lo_sum:SI (reg:SI 11 11)
                (symbol_ref/u:SI ("*.LC5") [flags 0x2])) [12 S8 A64])) -1 (nil)
    (nil))
../../../src/gcc4_sources/libiberty/floatformat.c:382: internal compiler error:
in extract_insn, at recog.c:2084
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://gcc.gnu.org/bugs.html> for instructions.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[mark at codesourcery dot com]

------- Comment #8 from mark at codesourcery dot com  2006-02-24 21:07 -------
Subject: Re:  [4.1/4.2 Regression] gcc fails to build on
 powerpc e500-double targets

edmar at freescale dot com wrote:
> ------- Comment #7 from edmar at freescale dot com  2006-02-24 20:45 -------
> Created an attachment (id=10911)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=10911&action=view)
>  --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=10911&action=view)
> Input program that causes ICE on gcc-4.1.0 RC2
> 
> No deal. I got an ICE now:
> (insn 349 347 122 12 ../../../src/gcc4_sources/libiberty/floatformat.c:350 (set
> (reg:DF 11 11)
>         (mem/u/c/i:DF (lo_sum:SI (reg:SI 11 11)
>                 (symbol_ref/u:SI ("*.LC5") [flags 0x2])) [12 S8 A64])) -1 (nil)
>     (nil))
> ../../../src/gcc4_sources/libiberty/floatformat.c:382: internal compiler error:
> in extract_insn, at recog.c:2084

Bummer.

Please attach preprocessed source and configuration options so that we
have a shot at reproducing.

However, I'm afraid that means you're going to be out of luck in GCC
4.1.0; hopefully the problem will be fixed before 4.1.1!

I'd consider  a patch for 4.1.0, but only if it's available very
shortly, and is obviously safe from the perspective of other targets; I
do not intend to do an RC3, so we cannot put other targets at risk.

Sorry,


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[edmar at freescale dot com]

------- Comment #9 from edmar at freescale dot com  2006-02-24 21:15 -------
(In reply to comment #8)
> Subject: Re:  [4.1/4.2 Regression] gcc fails to build on
>  powerpc e500-double targets
> 
> 
> 
> Bummer.
> 
> Please attach preprocessed source and configuration options so that we
> have a shot at reproducing.
> 
> However, I'm afraid that means you're going to be out of luck in GCC
> 4.1.0; hopefully the problem will be fixed before 4.1.1!
> 
> I'd consider  a patch for 4.1.0, but only if it's available very
> shortly, and is obviously safe from the perspective of other targets; I
> do not intend to do an RC3, so we cannot put other targets at risk.
> 
> Sorry,
> 

Mark,
On comment #7 I did attach floatformat.i, also the ICE output was done with -v.
The config options are visible there. Do you need more info ?




-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[mmitchel at gcc dot gnu dot org]

------- Comment #10 from mmitchel at gcc dot gnu dot org  2006-02-24 21:46 -------
Edmar --

Great, yes, that looks like the right information.  However, it's unlikely that
I'll be able to personally look at this before 4.1.0.

Thanks,

-- Mark


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[steven at gcc dot gnu dot org]

------- Comment #11 from steven at gcc dot gnu dot org  2006-02-27 21:42 -------
Re. comment #3, testing all releases is, I'm sorry to say, rather worthless.
Things usually break during development, not during the release process.

Obviously nobody can tell you what to do with your resources, but it shouldn't
be very hard for a company like FreeScale to set up an automatic testing system
and test the trunk more regularly, instead of testing it days before a release
and filing 11th hour bug reports.

The fact that Mark would accept a patch for GCC 4.1.0 for this problem really
surprises me. I'd use the opportunity and try to fix the problem asap ;-)


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[steven at gcc dot gnu dot org]

------- Comment #12 from steven at gcc dot gnu dot org  2006-02-27 22:36 -------
With "GNU C version 4.1.0 20060222 (prerelease) (powerpc-unknown-linux-gnuspe)"
I get a different ICE:

$ ./cc1 -O2 -fno-inline t.c
 foo
t.c: In function 'foo':
t.c:9: warning: incompatible implicit declaration of built-in function 'ldexp'

Analyzing compilation unitPerforming intraprocedural optimizations
Assembling functions:
 foo
t.c: In function 'foo':
t.c:12: error: unrecognizable insn:
(insn 77 60 32 2 (set (subreg:DF (reg:DI 3 3 [128]) 0)
        (mem/u/c/i:DF (reg/f:SI 9 9 [126]) [3 S8 A64])) -1 (nil)
    (expr_list:REG_DEAD (reg/f:SI 9 9 [126])
        (nil)))
t.c:12: internal compiler error: in extract_insn, at recog.c:2084
Please submit a full bug report,
with preprocessed source if appropriate.


$ cat t.c
void
foo (long exponent, double *to)
{
  double dto;

  dto = 0.0;

  if (!exponent)
    dto = ldexp (1.0, exponent);

  *to = dto;
}


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[steven at gcc dot gnu dot org]

------- Comment #13 from steven at gcc dot gnu dot org  2006-02-27 22:42 -------
The insn triggering my ICE appears in peephole2.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[steven at gcc dot gnu dot org]

------- Comment #14 from steven at gcc dot gnu dot org  2006-02-27 22:44 -------
Before peephole2:

(insn:HI 27 60 28 2 (set (reg:DF 0 0 [125])
        (mem/u/c/i:DF (reg/f:SI 9 9 [126]) [3 S8 A64])) 892
{*movdf_e500_double} (insn_list:REG_DEP_TRUE 26 (nil))
    (expr_list:REG_DEAD (reg/f:SI 9 9 [126])
        (expr_list:REG_EQUIV (const_double:DF 1.0e+0 [0x0.8p+1])
            (nil))))

(insn:HI 28 27 32 2 (set (subreg:DF (reg:DI 3 3 [128]) 0)
        (reg:DF 0 0 [125])) 889 {*frob_di_df_2} (insn_list:REG_DEP_TRUE 27
(nil))
    (expr_list:REG_DEAD (reg:DF 0 0 [125])
        (nil)))

After peephole2:
(insn 77 60 32 2 (set (subreg:DF (reg:DI 3 3 [128]) 0)
        (mem/u/c/i:DF (reg/f:SI 9 9 [126]) [3 S8 A64])) -1 (nil)
    (expr_list:REG_DEAD (reg/f:SI 9 9 [126])
        (nil)))


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[mmitchel at gcc dot gnu dot org]

-- 

mmitchel at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.1.0                       |4.1.1


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[amodra at bigpond dot net dot au]

------- Comment #15 from amodra at bigpond dot net dot au  2006-03-24 14:23 -------
Appears to be the peephole in rs6000.md immediately after this comment:
;; after inserting conditional returns we can sometimes have
;; unnecessary register moves.

(subreg:DF (reg:DI 3 3 [128]) 0) is a valid gpc_reg_operand.  Try this (totally
untested) patch.

Index: gcc/config/rs6000/rs6000.md
===================================================================
--- gcc/config/rs6000/rs6000.md (revision 112347)
+++ gcc/config/rs6000/rs6000.md (working copy)
@@ -9771,18 +9771,20 @@
   "operands[0] = widen_memory_access (operands[0], V2DFmode, 0);
    operands[1] = gen_rtx_REG (V2DFmode, REGNO (operands[1]));")

-;; after inserting conditional returns we can sometimes have
+;; After inserting conditional returns we can sometimes have
 ;; unnecessary register moves.  Unfortunately we cannot have a
 ;; modeless peephole here, because some single SImode sets have early
 ;; clobber outputs.  Although those sets expand to multi-ppc-insn
 ;; sequences, using get_attr_length here will smash the operands
 ;; array.  Neither is there an early_cobbler_p predicate.
+;; Disallow subregs for E500 so we don't munge frob_di_df_2.
 (define_peephole2
   [(set (match_operand:DF 0 "gpc_reg_operand" "")
        (match_operand:DF 1 "any_operand" ""))
    (set (match_operand:DF 2 "gpc_reg_operand" "")
        (match_dup 0))]
-  "peep2_reg_dead_p (2, operands[0])"
+  "!(TARGET_E500_DOUBLE && GET_CODE (operands[2]) == SUBREG)
+   && peep2_reg_dead_p (2, operands[0])"
   [(set (match_dup 2) (match_dup 1))])

 (define_peephole2


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[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|NEW                         |ASSIGNED
   Last reconfirmed|2006-02-24 18:42:12         |2006-03-25 02:59:46
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[amodra at bigpond dot net dot au]

------- Comment #16 from amodra at bigpond dot net dot au  2006-03-25 03:00 -------
Indeed, that patch cures the problem.


-- 

amodra at bigpond dot net dot au changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|amodra at bigpond dot net   |
                   |dot au                      |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[amodra at gcc dot gnu dot org]

------- Comment #17 from amodra at gcc dot gnu dot org  2006-03-26 22:48 -------
Subject: Bug 26459

Author: amodra
Date: Sun Mar 26 22:48:22 2006
New Revision: 112400

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=112400
Log:
        PR target/26459
        * config/rs6000/rs6000.md (DF reg move peephole): Prevent
        peephole removing spe frob_di_df_2.


Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/config/rs6000/rs6000.md


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[amodra at gcc dot gnu dot org]

------- Comment #18 from amodra at gcc dot gnu dot org  2006-03-26 22:49 -------
Subject: Bug 26459

Author: amodra
Date: Sun Mar 26 22:49:20 2006
New Revision: 112401

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=112401
Log:
        PR target/26459
        * config/rs6000/rs6000.md (DF reg move peephole): Prevent
        peephole removing spe frob_di_df_2.


Modified:
    branches/gcc-4_1-branch/gcc/ChangeLog
    branches/gcc-4_1-branch/gcc/config/rs6000/rs6000.md


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[amodra at bigpond dot net dot au]

------- Comment #19 from amodra at bigpond dot net dot au  2006-03-27 01:41 -------
After applying the patch, I decided I really ought to find where the regression
from 4.0 occurred.  It's a pity I didn't do that before developing the patch..
To my surprise I found that on the face of it, 4.0 ought to have the same
problem, because it has exactly the same peephole.  A debug session revealed
that 4.0 doesn't think that (subreg:DF (reg:DI 3 3 [128]) 0) is a valid
gpc_reg_operand for e500 double target, while 4.1 does.  After some
head-scratching, I discovered the reason for this is the TARGET_IEEEQUAD change
to rs6000.h:CANNOT_CHANGE_MODE_CLASS (well, really, it's the change to use IBM
extended double for powerpc-linux targets).

The first few lines of this macro are:

#define CANNOT_CHANGE_MODE_CLASS(FROM, TO, CLASS)                         \
  (!TARGET_IEEEQUAD                                                       \
   && GET_MODE_SIZE (FROM) >= 8 && GET_MODE_SIZE (TO) >= 8                \
   ? 0                                                                    \

This of course interacts badly with later TARGET_E500_DOUBLE tests for DFmode
and DImode, because now that TARGET_IEEEQUAD is false for linuxspe, we never
get to make the TARGET_E500_DOUBLE checks.  At the minimum, the
TARGET_E500_DOUBLE checks ought to be done before this test allows a zero
return.

These lines were added here
        PR target/11848
        * rs6000.h (CANNOT_CHANGE_MODE_CLASS): Allow change of mode
        in floating-point registers between TFmode and DImode.

Really, I think it would be better to check for TFmode<->DImode explicitly
rather than the rather obscure check that Geoff added.  We have quite a few
modes that have size greater than eight.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[patchapp at dberlin dot org]

------- Comment #20 from patchapp at dberlin dot org  2006-03-27 04:36 -------
Subject: Bug number PR26459

A patch for this bug has been added to the patch tracker.
The mailing list url for the patch is
http://gcc.gnu.org/ml/gcc-patches/2006-03/msg01519.html


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[amodra at gcc dot gnu dot org]

------- Comment #21 from amodra at gcc dot gnu dot org  2006-03-31 01:25 -------
Subject: Bug 26459

Author: amodra
Date: Fri Mar 31 01:25:35 2006
New Revision: 112561

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=112561
Log:
        PR target/26459
        * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Limit 2003-12-08
        change to FLOAT_REGS.


Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/config/rs6000/rs6000.h


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[amodra at gcc dot gnu dot org]

------- Comment #22 from amodra at gcc dot gnu dot org  2006-03-31 01:27 -------
Subject: Bug 26459

Author: amodra
Date: Fri Mar 31 01:27:44 2006
New Revision: 112562

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=112562
Log:
        PR target/26459
        * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Limit 2003-12-08
        change to FLOAT_REGS.


Modified:
    branches/gcc-4_1-branch/gcc/ChangeLog
    branches/gcc-4_1-branch/gcc/config/rs6000/rs6000.h


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[amodra at bigpond dot net dot au]

------- Comment #23 from amodra at bigpond dot net dot au  2006-03-31 01:54 -------
Fixed


-- 

amodra at bigpond dot net dot au changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                URL|                            |http://gcc.gnu.org/ml/gcc-
                   |                            |patches/2006-
                   |                            |03/msg01751.html
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[edmar at freescale dot com]

------- Comment #24 from edmar at freescale dot com  2006-04-10 15:42 -------
I am sorry, but the patches on comments 17, 18, 21, and 22 are no good without
the  
patch on comment 5, which seems, it was never commited into the repository...

Can you double check this. Thanks.

Edmar


-- 

edmar at freescale dot com changed:

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[amodra at gcc dot gnu dot org]

------- Comment #25 from amodra at gcc dot gnu dot org  2006-04-11 00:33 -------
Subject: Bug 26459

Author: amodra
Date: Tue Apr 11 00:33:29 2006
New Revision: 112843

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=112843
Log:
        PR target/26459
        * config/rs6000/e500-double.h (SUB3TARGET_OVERRIDE_OPTIONS): Test
        rs6000_explicit_options.float_gprs.


Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/config/rs6000/e500-double.h


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[amodra at gcc dot gnu dot org]

------- Comment #26 from amodra at gcc dot gnu dot org  2006-04-11 00:36 -------
Subject: Bug 26459

Author: amodra
Date: Tue Apr 11 00:36:50 2006
New Revision: 112844

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=112844
Log:
        PR target/26459
        * config/rs6000/e500-double.h (SUB3TARGET_OVERRIDE_OPTIONS): Test
        rs6000_explicit_options.float_gprs.


Modified:
    branches/gcc-4_1-branch/gcc/ChangeLog
    branches/gcc-4_1-branch/gcc/config/rs6000/e500-double.h


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[amodra at bigpond dot net dot au]

------- Comment #27 from amodra at bigpond dot net dot au  2006-04-11 00:37 -------
Sorry!  My apologies for closing this bug without checking that the original
problem had been fixed.  I was focusing on the code generation problems.


-- 

amodra at bigpond dot net dot au changed:

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[edmar at freescale dot com]

------- Comment #28 from edmar at freescale dot com  2006-04-12 15:36 -------
I double checked everything. The sources I checked out last night has the
patches of comments 17, 21, and 25.
This time I got a failure just like the on comment 7. The insn is the same....
The original attachment: floatformat.i, still good. I can reproduce the error
with it.



/local/gnu_toolchain/build_area/obj_gcc-trunk_e500v1/obj_gcc-trunk_e500v2/powerpc-unknown-linux-gnuspe/libiberty>
/local/gnu_toolchain/build_area/obj_gcc-trunk_e500v1/obj_gcc-trunk_e500v2/./gcc/xgcc
-B/local/gnu_toolchain/build_area/obj_gcc-trunk_e500v1/obj_gcc-trunk_e500v2/./gcc/
-B/local/gnu_toolchain/install_area/gcc-trunk-20060411-e500v2/powerpc-unknown-linux-gnuspe/bin/
-B/local/gnu_toolchain/install_area/gcc-trunk-20060411-e500v2/powerpc-unknown-linux-gnuspe/lib/
-isystem
/local/gnu_toolchain/install_area/gcc-trunk-20060411-e500v2/powerpc-unknown-linux-gnuspe/include
-isystem
/local/gnu_toolchain/install_area/gcc-trunk-20060411-e500v2/powerpc-unknown-linux-gnuspe/sys-include
-c -DHAVE_CONFIG_H -O2 -g -O2  -I. -I../../../gcc-trunk/libiberty/../include 
-W -Wall -pedantic -Wwrite-strings -Wstrict-prototypes -Wc++-compat
../../../gcc-trunk/libiberty/floatformat.c -o floatformat.o -v --save-temps
Reading specs from
/local/gnu_toolchain/build_area/obj_gcc-trunk_e500v1/obj_gcc-trunk_e500v2/./gcc/specs
Target: powerpc-unknown-linux-gnuspe
Configured with: ../gcc-trunk/configure
--prefix=/local/gnu_toolchain/install_area/gcc-trunk-20060411-e500v2
--with-local-prefix=/local/gnu_toolchain/install_area/gcc-trunk-20060411-e500v2
--enable-languages=c,c++ --enable-threads --target=powerpc-unknown-linux-gnuspe
--disable-shared --disable-multilib --disable-linux-futex --enable-e500_double
Thread model: posix
gcc version 4.2.0 20060411 (experimental)

/local/gnu_toolchain/build_area/obj_gcc-trunk_e500v1/obj_gcc-trunk_e500v2/./gcc/cc1
-E -quiet -v -I. -I../../../gcc-trunk/libiberty/../include -iprefix
/temp/gnu_toolchain/build_area/obj_gcc-trunk_e500v1/obj_gcc-trunk_e500v2/gcc/../lib/gcc/powerpc-unknown-linux-gnuspe/4.2.0/
-isystem
/local/gnu_toolchain/build_area/obj_gcc-trunk_e500v1/obj_gcc-trunk_e500v2/./gcc/include
-D__unix__ -D__gnu_linux__ -D__linux__ -Dunix -D__unix -Dlinux -D__linux
-Asystem=linux -Asystem=unix -Asystem=posix -DHAVE_CONFIG_H -isystem
/local/gnu_toolchain/install_area/gcc-trunk-20060411-e500v2/powerpc-unknown-linux-gnuspe/include
-isystem
/local/gnu_toolchain/install_area/gcc-trunk-20060411-e500v2/powerpc-unknown-linux-gnuspe/sys-include
../../../gcc-trunk/libiberty/floatformat.c -W -Wall -pedantic -Wwrite-strings
-Wstrict-prototypes -Wc++-compat -fworking-directory -O2 -O2 -fpch-preprocess
-o floatformat.i
ignoring nonexistent directory
"/local/gnu_toolchain/install_area/gcc-trunk-20060411-e500v2/powerpc-unknown-linux-gnuspe/include"
ignoring nonexistent directory
"/temp/gnu_toolchain/build_area/obj_gcc-trunk_e500v1/obj_gcc-trunk_e500v2/gcc/../lib/gcc/powerpc-unknown-linux-gnuspe/4.2.0/include"
ignoring nonexistent directory
"/temp/gnu_toolchain/build_area/obj_gcc-trunk_e500v1/obj_gcc-trunk_e500v2/gcc/../lib/gcc/powerpc-unknown-linux-gnuspe/4.2.0/../../../../powerpc-unknown-linux-gnuspe/sys-include"
ignoring nonexistent directory
"/temp/gnu_toolchain/build_area/obj_gcc-trunk_e500v1/obj_gcc-trunk_e500v2/gcc/../lib/gcc/powerpc-unknown-linux-gnuspe/4.2.0/../../../../powerpc-unknown-linux-gnuspe/include"
ignoring duplicate directory
"/local/gnu_toolchain/install_area/gcc-trunk-20060411-e500v2/lib/gcc/powerpc-unknown-linux-gnuspe/4.2.0/../../../../powerpc-unknown-linux-gnuspe/sys-include"
ignoring nonexistent directory
"/local/gnu_toolchain/install_area/gcc-trunk-20060411-e500v2/lib/gcc/powerpc-unknown-linux-gnuspe/4.2.0/../../../../powerpc-unknown-linux-gnuspe/include"
#include "..." search starts here:
#include <...> search starts here:
 .
 ../../../gcc-trunk/libiberty/../include

/local/gnu_toolchain/build_area/obj_gcc-trunk_e500v1/obj_gcc-trunk_e500v2/./gcc/include

/local/gnu_toolchain/install_area/gcc-trunk-20060411-e500v2/powerpc-unknown-linux-gnuspe/sys-include

/local/gnu_toolchain/install_area/gcc-trunk-20060411-e500v2/lib/gcc/powerpc-unknown-linux-gnuspe/4.2.0/include
End of search list.

/local/gnu_toolchain/build_area/obj_gcc-trunk_e500v1/obj_gcc-trunk_e500v2/./gcc/cc1
-fpreprocessed floatformat.i -quiet -dumpbase floatformat.c -auxbase-strip
floatformat.o -g -O2 -O2 -W -Wall -pedantic -Wwrite-strings -Wstrict-prototypes
-Wc++-compat -version -o floatformat.s
GNU C version 4.2.0 20060411 (experimental) (powerpc-unknown-linux-gnuspe)
        compiled by GNU C version 3.4.3.
GGC heuristics: --param ggc-min-expand=30 --param ggc-min-heapsize=4096
Compiler executable checksum: 46d139d3138e283288bcfc291691aee0
../../../gcc-trunk/libiberty/floatformat.c: In function
'floatformat_to_double':
../../../gcc-trunk/libiberty/floatformat.c:415: error: unrecognizable insn:
(insn 342 340 128 14 ../../../gcc-trunk/libiberty/floatformat.c:383 (set
(reg:DF 11 11)
        (mem/u/c/i:DF (lo_sum:SI (reg:SI 11 11)
                (symbol_ref/u:SI ("*.LC6") [flags 0x2])) [10 S8 A64])) -1 (nil)
    (nil))
../../../gcc-trunk/libiberty/floatformat.c:415: internal compiler error: in
extract_insn, at recog.c:2037
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://gcc.gnu.org/bugs.html> for instructions.


-- 

edmar at freescale dot com changed:

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[amodra at gcc dot gnu dot org]

------- Comment #29 from amodra at gcc dot gnu dot org  2006-04-13 05:10 -------
Subject: Bug 26459

Author: amodra
Date: Thu Apr 13 05:10:00 2006
New Revision: 112912

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=112912
Log:
        PR target/26459
        * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
        optimize to lo_sum for DFmode if !TARGET_FPRS.


Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/config/rs6000/rs6000.c


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[amodra at gcc dot gnu dot org]

------- Comment #30 from amodra at gcc dot gnu dot org  2006-04-13 05:46 -------
Subject: Bug 26459

Author: amodra
Date: Thu Apr 13 05:46:01 2006
New Revision: 112913

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=112913
Log:
        PR target/26459
        * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
        optimize to lo_sum for DFmode if !TARGET_FPRS.


Modified:
    branches/gcc-4_1-branch/gcc/ChangeLog
    branches/gcc-4_1-branch/gcc/config/rs6000/rs6000.c


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[mmitchel at gcc dot gnu dot org]

------- Comment #31 from mmitchel at gcc dot gnu dot org  2006-04-16 18:39 -------
Is there more to do?  The end of the audit trail contains patches from Alan, so
it's unclear to me if the issue should remain open.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[amodra at bigpond dot net dot au]

------- Comment #32 from amodra at bigpond dot net dot au  2006-04-17 00:09 -------
I've closed this bug incorrectly twice, the last time simply because I hadn't
committed a patch already in my tree.  Edmar, please resolve this bug as fixed
if e500 is back in bootstrap land.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[edmar at freescale dot com]

------- Comment #33 from edmar at freescale dot com  2006-04-17 15:22 -------
The latest round of patches cured the original problem. Unfortunately, the
target itself is still broken. This time because of a bug originaly reported
for the e500v1 target. (27075).
I am not sure what is the right administrative thing to do. This one was a P3,
27075 was changed to P2 recently.
Perhaps a comment on 27075 that it also affects e500v2 target ? I will leave to
you guys to decide...


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[mark at codesourcery dot com]

------- Comment #34 from mark at codesourcery dot com  2006-04-17 15:37 -------
Subject: Re:  [4.1/4.2 Regression] gcc fails to build on
 powerpc e500-double targets

edmar at freescale dot com wrote:
> ------- Comment #33 from edmar at freescale dot com  2006-04-17 15:22 -------
> The latest round of patches cured the original problem. Unfortunately, the
> target itself is still broken. This time because of a bug originaly reported
> for the e500v1 target. (27075).
> I am not sure what is the right administrative thing to do. This one was a P3,
> 27075 was changed to P2 recently.
> Perhaps a comment on 27075 that it also affects e500v2 target ? I will leave to
> you guys to decide...

If the problem reported in PR 26459 is resolved, then we should close PR
26459.  If the problem in PR 27075 also effects E500v2, then we should
not that there.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459

[Bug target/26459] [4.1/4.2 Regression] gcc fails to build on powerpc e500-double targets

[edmar at freescale dot com]

------- Comment #35 from edmar at freescale dot com  2006-04-17 15:44 -------
The original problem was solved. For futher bootstrap problems on this target,
see bug 27075.


-- 

edmar at freescale dot com changed:

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26459