[Bug target/47449] New: [32] can’t find a register in class ‘DIREG’ while reloading ‘asm’

[Bug target/47449] New: [32] can’t find a register in class ‘DIREG’ while reloading ‘asm’

[hjl.tools at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47449

           Summary: [32] can’t find a register in class ‘DIREG’ while
                    reloading ‘asm’
           Product: gcc
           Version: 4.6.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: target
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: hjl.tools@gmail.com
                CC: ubizjak@gmail.com


[hjl@gnu-6 ilp32-17]$ cat x.c
void bar (void *);
int
foo (void *trans)
{
    int ret;
    __asm __volatile ("foo %0\n\t" : "=D" (ret));
  bar (trans);
  return ret;
}
[hjl@gnu-6 ilp32-17]$ make x.s
/export/build/gnu/gcc-x32/build-x86_64-linux/gcc/xgcc
-B/export/build/gnu/gcc-x32/build-x86_64-linux/gcc/ -S -o x.s -mx32 -O2 -dp 
x.c
x.c: In function ‘foo’:
x.c:6:5: error: can’t find a register in class ‘DIREG’ while reloading ‘asm’
x.c:6:5: error: ‘asm’ operand has impossible constraints
make: *** [x.s] Error 1
[hjl@gnu-6 ilp32-17]$ 

Uros, any ideas? Thanks.

[Bug target/47449] [32] can’t find a register in class ‘DIREG’ while reloading ‘asm’

[hjl.tools at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47449

--- Comment #1 from H.J. Lu <hjl.tools at gmail dot com> 2011-01-24 23:01:31 UTC ---
RDI is used to pass the first argument.

[Bug target/47449] [32] can’t find a register in class ‘DIREG’ while reloading ‘asm’

[hjl.tools at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47449

--- Comment #2 from H.J. Lu <hjl.tools at gmail dot com> 2011-01-24 23:16:57 UTC ---
fwprop1 turns:

(insn 2 4 3 2 (set (reg/v/f:DI 61 [ trans ])
        (zero_extend:DI (reg:SI 5 di [ trans ]))) x.c:4 115
{*zero_extendsidi2_r
ex64}
     (nil))

(note 3 2 6 2 NOTE_INSN_FUNCTION_BEG)

(insn 6 3 7 2 (parallel [
            (set (reg:SI 62 [ ret ])
                (asm_operands:SI ("negl %0
    ") ("=D") 0 []
                     []
                     [] x.c:9))
            (clobber (reg:QI 18 fpsr))
            (clobber (reg:QI 17 flags))
        ]) x.c:6 -1
     (nil))

into

(insn 6 3 9 2 (parallel [
            (set (reg:SI 62 [ ret ])
                (asm_operands:SI ("negl %0
    ") ("=D") 0 []
                     []
                     [] x.c:9))
            (clobber (reg:QI 18 fpsr))
            (clobber (reg:QI 17 flags))
        ]) x.c:6 -1
     (expr_list:REG_UNUSED (reg:QI 18 fpsr)
        (expr_list:REG_UNUSED (reg:QI 17 flags)
            (nil))))

IRA can't deal with it.

[Bug target/47449] [32] can’t find a register in class ‘DIREG’ while reloading ‘asm’

[hjl.tools at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47449

H.J. Lu <hjl.tools at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |bonzini at gnu dot org

--- Comment #3 from H.J. Lu <hjl.tools at gmail dot com> 2011-01-24 23:18:59 UTC ---
fwprop can't handle subreg properly.

[Bug target/47449] [32] can’t find a register in class ‘DIREG’ while reloading ‘asm’

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47449

--- Comment #4 from Andrew Pinski <pinskia at gcc dot gnu.org> 2011-01-24 23:21:48 UTC ---
(In reply to comment #3)
> fwprop can't handle subreg properly.

Huh?  You don't show where the subreg of 61 is being used.  As far as I can
tell what fwprop is doing is correct.  I think we should just disable the code
which Alan added for doing fwprop for SMALL_REGISTERS(sp.) targets

[Bug target/47449] [32] can’t find a register in class ‘DIREG’ while reloading ‘asm’

[hjl.tools at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47449

--- Comment #5 from H.J. Lu <hjl.tools at gmail dot com> 2011-01-24 23:27:48 UTC ---
Before fwprop1:

---
;; Pred edge  ENTRY [100.0%]  (fallthru)
(note 4 0 2 2 [bb 2] NOTE_INSN_BASIC_BLOCK)

(insn 2 4 3 2 (set (reg/v/f:DI 61 [ trans ])
        (zero_extend:DI (reg:SI 5 di [ trans ]))) x.c:4 115
{*zero_extendsidi2_r
ex64}
     (nil))

(note 3 2 6 2 NOTE_INSN_FUNCTION_BEG)

(insn 6 3 7 2 (parallel [
            (set (reg:SI 62 [ ret ])
                (asm_operands:SI ("negl %0
    ") ("=D") 0 []
                     []
                     [] x.c:9))
            (clobber (reg:QI 18 fpsr))
            (clobber (reg:QI 17 flags))
        ]) x.c:6 -1
     (nil))

(insn 7 6 8 2 (set (reg/v:SI 59 [ ret ])
        (reg:SI 62 [ ret ])) x.c:6 64 {*movsi_internal}
     (nil))

(insn 8 7 9 2 (set (reg:SI 5 di)
        (subreg/s/u:SI (reg/v/f:DI 61 [ trans ]) 0)) x.c:7 64 {*movsi_internal}
     (nil))

(call_insn 9 8 10 2 (call (mem:QI (symbol_ref:DI ("bar") [flags 0x41] 
<function
_decl 0x7fa8a422ef00 bar>) [0 S1 A8])
        (const_int 0 [0])) x.c:7 623 {*call_0}
     (nil)
    (expr_list:REG_DEP_TRUE (use (reg:SI 5 di))
        (nil)))

(insn 10 9 14 2 (set (reg:SI 60 [ <retval> ])
        (reg:SI 62 [ ret ])) x.c:7 64 {*movsi_internal}
     (nil))

(insn 14 10 17 2 (set (reg/i:SI 0 ax)
        (reg:SI 62 [ ret ])) x.c:9 64 {*movsi_internal}
     (nil))

(insn 17 14 0 2 (use (reg/i:SI 0 ax)) x.c:9 -1
     (nil))
;; End of basic block 2 -> ( 1)
;; lr  out      0 [ax] 6 [bp] 7 [sp] 16 [argp] 20 [frame]


;; Succ edge  EXIT [100.0%]  (fallthru)
---

After fwprop1:

---
;; Pred edge  ENTRY [100.0%]  (fallthru)
(note 4 0 3 2 [bb 2] NOTE_INSN_BASIC_BLOCK)

(note 3 4 6 2 NOTE_INSN_FUNCTION_BEG)

(insn 6 3 9 2 (parallel [
            (set (reg:SI 62 [ ret ])
                (asm_operands:SI ("negl %0
    ") ("=D") 0 []
                     []
                     [] x.c:9))
            (clobber (reg:QI 18 fpsr))
            (clobber (reg:QI 17 flags))
        ]) x.c:6 -1
     (expr_list:REG_UNUSED (reg:QI 18 fpsr)
        (expr_list:REG_UNUSED (reg:QI 17 flags)
            (nil))))

(call_insn 9 6 14 2 (call (mem:QI (symbol_ref:DI ("bar") [flags 0x41] 
<function
_decl 0x7fa8a422ef00 bar>) [0 S1 A8])
        (const_int 0 [0])) x.c:7 623 {*call_0}
     (expr_list:REG_DEAD (reg:SI 5 di)
        (nil))
    (expr_list:REG_DEP_TRUE (use (reg:SI 5 di))
        (nil)))

(insn 14 9 17 2 (set (reg/i:SI 0 ax)
        (reg:SI 62 [ ret ])) x.c:9 64 {*movsi_internal}
     (expr_list:REG_DEAD (reg:SI 62 [ ret ])
        (nil)))

(insn 17 14 0 2 (use (reg/i:SI 0 ax)) x.c:9 -1
     (nil))
;; End of basic block 2 -> ( 1)
;; lr  out      0 [ax] 6 [bp] 7 [sp] 16 [argp] 20 [frame]
;; md  out     


;; Succ edge  EXIT [100.0%]  (fallthru)
---

For some reason, fwprop1 doesn't remove

(insn 2 4 3 2 (set (reg/v/f:DI 61 [ trans ])
        (reg:DI 5 di [ trans ])) x.c:4 62 {*movdi_internal_rex64}
     (nil))

for 64bit.

[Bug target/47449] [32] can’t find a register in class ‘DIREG’ while reloading ‘asm’

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47449

--- Comment #6 from Andrew Pinski <pinskia at gcc dot gnu.org> 2011-01-24 23:31:30 UTC ---
ok, first off the issue is proping over the asm that might clobber EDI.

[Bug target/47449] [32] can’t find a register in class ‘DIREG’ while reloading ‘asm’

[bonzini at gnu dot org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47449

--- Comment #7 from Paolo Bonzini <bonzini at gnu dot org> 2011-01-24 23:45:01 UTC ---
Is this a regression?

[Bug target/47449] [32] can’t find a register in class ‘DIREG’ while reloading ‘asm’

[bonzini at gnu dot org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47449

--- Comment #8 from Paolo Bonzini <bonzini at gnu dot org> 2011-01-24 23:47:10 UTC ---
Ah, [32] means 32-bit x86_64...

[Bug target/47449] [x32] can’t find a register in class ‘DIREG’ while reloading ‘asm’

[hjl.tools at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47449

--- Comment #9 from H.J. Lu <hjl.tools at gmail dot com> 2011-01-24 23:50:05 UTC ---
PR 47379 is another fwprop issue with subreg.

[Bug middle-end/47449] [x32] can’t find a register in class ‘DIREG’ while reloading ‘asm’

[hjl at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47449

--- Comment #10 from hjl at gcc dot gnu.org <hjl at gcc dot gnu.org> 2011-01-25 02:13:17 UTC ---
Author: hjl
Date: Tue Jan 25 02:13:14 2011
New Revision: 169191

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=169191
Log:
Don't propagate zero/sign extended hard register.

gcc/

2011-01-24  H.J. Lu  <hongjiu.lu@intel.com>

    PR middle-end/47449
    * fwprop.c (forward_propagate_subreg): Don't propagate zero/sign
    extended hard register.

gcc/testsuite/

2011-01-24  H.J. Lu  <hongjiu.lu@intel.com>

    PR middle-end/47449
    * gcc.target/i386/pr47449.c: New.

Added:
    branches/x32/gcc/testsuite/gcc.target/i386/pr47449.c
Modified:
    branches/x32/gcc/ChangeLog.x32
    branches/x32/gcc/fwprop.c
    branches/x32/gcc/testsuite/ChangeLog.x32

[Bug middle-end/47449] [x32] can’t find a register in class ‘DIREG’ while reloading ‘asm’

[hjl.tools at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47449

H.J. Lu <hjl.tools at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                URL|                            |http://gcc.gnu.org/ml/gcc-p
                   |                            |atches/2011-01/msg01701.htm
                   |                            |l

--- Comment #11 from H.J. Lu <hjl.tools at gmail dot com> 2011-01-25 02:17:29 UTC ---
A patch is posted at

http://gcc.gnu.org/ml/gcc-patches/2011-01/msg01701.html

[Bug middle-end/47449] [x32] can’t find a register in class ‘DIREG’ while reloading ‘asm’

[hjl at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47449

--- Comment #12 from hjl at gcc dot gnu.org <hjl at gcc dot gnu.org> 2011-01-25 15:14:54 UTC ---
Author: hjl
Date: Tue Jan 25 15:14:49 2011
New Revision: 169234

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=169234
Log:
Don't propagate hard register.

2011-01-25  H.J. Lu  <hongjiu.lu@intel.com>

    PR middle-end/47449
    * fwprop.c (forward_propagate_subreg): Don't propagate hard
    register.

Modified:
    branches/x32/gcc/ChangeLog.x32
    branches/x32/gcc/fwprop.c

[Bug rtl-optimization/47449] [x32] can’t find a register in class ‘DIREG’ while reloading ‘asm’

[hjl at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47449

--- Comment #13 from hjl at gcc dot gnu.org <hjl at gcc dot gnu.org> 2011-07-05 12:47:38 UTC ---
Author: hjl
Date: Tue Jul  5 12:47:35 2011
New Revision: 175849

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=175849
Log:
Don't propagate hard registers in forward_propagate_subreg.

2011-07-05  H.J. Lu  <hongjiu.lu@intel.com>

    PR rtl-optimization/47449
    * fwprop.c (forward_propagate_subreg): Don't propagate hard
    register nor zero/sign extended hard register.

Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/fwprop.c

[Bug rtl-optimization/47449] [x32] can’t find a register in class ‘DIREG’ while reloading ‘asm’

[hjl.tools at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47449

H.J. Lu <hjl.tools at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|                            |FIXED
   Target Milestone|---                         |4.7.0

--- Comment #14 from H.J. Lu <hjl.tools at gmail dot com> 2011-07-05 13:14:59 UTC ---
Fixed.

[Bug rtl-optimization/47449] [x32] can’t find a register in class ‘DIREG’ while reloading ‘asm’

[hjl at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47449

--- Comment #15 from hjl at gcc dot gnu.org <hjl at gcc dot gnu.org> 2011-08-06 14:36:55 UTC ---
Author: hjl
Date: Sat Aug  6 14:36:52 2011
New Revision: 177512

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=177512
Log:
Add testcases for PRs 47449/47446.

2011-08-06  H.J. Lu  <hongjiu.lu@intel.com>

    PR middle-end/47449
    * gcc.target/i386/pr47449.c: New.

    PR target/47446
    * gcc.dg/pr47446-1.c: New.
    * gcc.dg/pr47446-2.c: Likewise.

Added:
    trunk/gcc/testsuite/gcc.dg/pr47446-1.c
    trunk/gcc/testsuite/gcc.dg/pr47446-2.c
    trunk/gcc/testsuite/gcc.target/i386/pr47449.c
Modified:
    trunk/gcc/testsuite/ChangeLog