[Bug rtl-optimization/53352] New: Incorrect CSE optimization on RTL expressions with a paradoxical subreg

[Bug rtl-optimization/53352] New: Incorrect CSE optimization on RTL expressions with a paradoxical subreg

[meadori at codesourcery dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53352

             Bug #: 53352
           Summary: Incorrect CSE optimization on RTL expressions with a
                    paradoxical subreg
    Classification: Unclassified
           Product: gcc
           Version: 4.8.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: rtl-optimization
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: meadori@codesourcery.com
              Host: GNU/Linux
            Target: arm-none-linux-gnueabi


The attached reproduction case exhibits what I believe to be a bug
in the CSE pass.  The problem can be reproduced from the mainline
at r187470.  The compiler was built for arm-none-linux-gnueabi
and the reproduction case was compiled with -O3.

Out of the gate the expanded RTL of interest looks like:

(insn 12 11 13 3 (set (reg:SI 140)
        (lshiftrt:SI (reg/v:SI 135 [ tmp1 ])
            (const_int 16 [0x10])))
     (nil))

(insn 13 12 14 3 (set (reg:QI 141)
        (subreg:QI (reg:SI 140) 0))
     (nil))

(insn 14 13 15 3 (set (reg:SI 142)
        (subreg:SI (reg:QI 141) 0))
     (nil))

(insn 15 14 16 3 (set (reg:SI 134 [ tmp1$2 ])
        (and:SI (reg:SI 142)
            (const_int 255 [0xff])))
     (nil))

...

(insn 29 28 30 3 (set (reg:SI 0 r0)
        (const_int 0 [0]))
     (nil))

after "cse1" things look like:

(insn 12 11 13 2 (set (reg:SI 140)
        (const_int 65280 [0xff00]))
     (nil))

(insn 13 12 14 2 (set (reg:QI 141)
        (subreg:QI (reg:SI 140) 0))
     (expr_list:REG_EQUAL (const_int 0 [0])
        (nil)))

;; This is *not* equal to zero because the upper
;; two bytes are undefined.
(insn 14 13 15 2 (set (reg:SI 142)
        (subreg:SI (reg:QI 141) 0))
     (expr_list:REG_EQUAL (const_int 0 [0])
        (nil)))

(insn 15 14 16 2 (set (reg:SI 134 [ tmp1$2 ])
        (reg:SI 142))
     (expr_list:REG_EQUAL (const_int 0 [0])
        (nil)))

...

(insn 29 28 30 2 (set (reg:SI 0 r0)
        (reg:SI 142))
     (expr_list:REG_EQUAL (const_int 0 [0])
        (nil)))

I believe the REG_EQUAL note on the set involving a paradoxical subreg
is incorrect.  It eventually causes 0xFF00 to be passed to the function
'foo'.

I doesn't look like 'equiv_constant' handles the paradoxical subreg case
correctly.  I fixed it as follows and am testing the patch now:

Index: gcc/cse.c
===================================================================
--- gcc/cse.c   (revision 187470)
+++ gcc/cse.c   (working copy)
@@ -3786,8 +3786,11 @@
            }
        }

-      /* Otherwise see if we already have a constant for the inner REG.  */
+      /* Otherwise see if we already have a constant for the inner REG.
+        Don't bother with paradoxical subregs because we have no way
+        of knowing what the upper bytes are.  */
       if (REG_P (SUBREG_REG (x))
+         && (GET_MODE_SIZE (mode) <= GET_MODE_SIZE (imode))
          && (new_rtx = equiv_constant (SUBREG_REG (x))) != 0)
         return simplify_subreg (mode, new_rtx, imode, SUBREG_BYTE (x));

[Bug rtl-optimization/53352] Incorrect CSE optimization on RTL expressions with a paradoxical subreg

[meadori at codesourcery dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53352

--- Comment #1 from Meador Inge <meadori at codesourcery dot com> 2012-05-14 22:52:41 UTC ---
Created attachment 27405
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=27405
Reproduction case

[Bug rtl-optimization/53352] Incorrect CSE optimization on RTL expressions with a paradoxical subreg

[mikpe at it dot uu.se]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53352

Mikael Pettersson <mikpe at it dot uu.se> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |mikpe at it dot uu.se

--- Comment #2 from Mikael Pettersson <mikpe at it dot uu.se> 2012-05-17 16:36:38 UTC ---
Can you make the test case standalone?  Currently it compiles but doesn't link,
and there's nothing in it to suggest where the error could be observed.

[Bug rtl-optimization/53352] Incorrect CSE optimization on RTL expressions with a paradoxical subreg

[meadori at codesourcery dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53352

--- Comment #3 from Meador Inge <meadori at codesourcery dot com> 2012-05-17 16:42:04 UTC ---
It can easily be seen my inspection of the assembly code.  As I mentioned
before, 0xFF00 gets passed to 'foo'.  Anyway, I posted an executable case along
with my patch here: http://gcc.gnu.org/ml/gcc-patches/2012-05/msg01042.html.

[Bug rtl-optimization/53352] Incorrect CSE optimization on RTL expressions with a paradoxical subreg

[steven at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53352

Steven Bosscher <steven at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |steven at gcc dot gnu.org

--- Comment #4 from Steven Bosscher <steven at gcc dot gnu.org> 2012-05-17 18:29:27 UTC ---
(In reply to comment #3)

That test case isn't particularly good, either. If something like IPA-CP or
inlining becomes the default at -O1, your whole test cases optimizes to a
no-op. And, at least AFAICT, your test case doesn't even specifically test for
the incorrect transformation.

[Bug rtl-optimization/53352] Incorrect CSE optimization on RTL expressions with a paradoxical subreg

[mikpe at it dot uu.se]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53352

--- Comment #5 from Mikael Pettersson <mikpe at it dot uu.se> 2012-05-17 18:35:00 UTC ---
There is no call to 'foo' in the test case here.  I guess you meant 'bar', but
even so executable test cases are strongly preferred over manual inspection of
assembly code.

FWIW, using the test case in the gcc-patches submission I've reproduced the bug
with both gcc-4.6.3 and gcc-4.5.3 on armv5tel-linux-gnueabi.

[Bug rtl-optimization/53352] Incorrect CSE optimization on RTL expressions with a paradoxical subreg

[steven at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53352

--- Comment #6 from Steven Bosscher <steven at gcc dot gnu.org> 2012-05-17 19:09:14 UTC ---
Mikael, is it possible for you to test if this is caused by this patch, please?
 http://gcc.gnu.org/viewcvs?view=revision&revision=142443

[Bug rtl-optimization/53352] Incorrect CSE optimization on RTL expressions with a paradoxical subreg

[meadori at codesourcery dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53352

--- Comment #7 from Meador Inge <meadori at codesourcery dot com> 2012-05-17 20:31:28 UTC ---
(In reply to comment #5)
> There is no call to 'foo' in the test case here.  I guess you meant 'bar', but
> even so executable test cases are strongly preferred over manual inspection of
> assembly code.

Yup, 'bar'.  Sorry.

> FWIW, using the test case in the gcc-patches submission I've reproduced the bug
> with both gcc-4.6.3 and gcc-4.5.3 on armv5tel-linux-gnueabi.

Thanks for checking that out.

[Bug rtl-optimization/53352] Incorrect CSE optimization on RTL expressions with a paradoxical subreg

[mikpe at it dot uu.se]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53352

--- Comment #8 from Mikael Pettersson <mikpe at it dot uu.se> 2012-05-17 21:07:20 UTC ---
(In reply to comment #6)
> Mikael, is it possible for you to test if this is caused by this patch, please?
>  http://gcc.gnu.org/viewcvs?view=revision&revision=142443

The error appears to be older than that, I can reproduce it with both r142442
and r142443.

[Bug rtl-optimization/53352] Incorrect CSE optimization on RTL expressions with a paradoxical subreg

[rsandifo at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53352

--- Comment #9 from rsandifo at gcc dot gnu.org <rsandifo at gcc dot gnu.org> 2012-05-18 09:04:47 UTC ---
Author: rsandifo
Date: Fri May 18 09:04:38 2012
New Revision: 187648

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=187648
Log:
gcc/
2012-05-15  Meador Inge  <meadori@codesourcery.com>

    PR rtl-optimization/53352
    * cse.c (equiv_constant): Ignore paradoxical subregs.

gcc/testsuite/
2012-05-15  Meador Inge  <meadori@codesourcery.com>

    PR rtl-optimization/53352
    * gcc.dg/pr53352.c: New test.

Added:
    trunk/gcc/testsuite/gcc.dg/pr53352.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/cse.c
    trunk/gcc/testsuite/ChangeLog

[Bug rtl-optimization/53352] Incorrect CSE optimization on RTL expressions with a paradoxical subreg

[hjl.tools at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53352

--- Comment #10 from H.J. Lu <hjl.tools at gmail dot com> 2012-05-18 14:16:46 UTC ---
The test fails on Linux/x86 and Linux/x86-64.

[Bug rtl-optimization/53352] Incorrect CSE optimization on RTL expressions with a paradoxical subreg

[ramana at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53352

Ramana Radhakrishnan <ramana at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |ramana at gcc dot gnu.org

--- Comment #11 from Ramana Radhakrishnan <ramana at gcc dot gnu.org> 2012-07-13 16:24:34 UTC ---
(In reply to comment #10)
> The test fails on Linux/x86 and Linux/x86-64.

http://gcc.gnu.org/ml/gcc-testresults/2012-07/msg00958.html doesn't suggest
this test fails anymore ? 

HJ, should this now be marked as FIXED ?


ramana

[Bug rtl-optimization/53352] Incorrect CSE optimization on RTL expressions with a paradoxical subreg

[meadori at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53352

meadori at gcc dot gnu.org changed:

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

--- Comment #12 from meadori at gcc dot gnu.org 2012-07-13 16:34:14 UTC ---
The test case was fixed:
http://gcc.gnu.org/viewcvs?view=revision&revision=187654.

[Bug rtl-optimization/53352] Incorrect CSE optimization on RTL expressions with a paradoxical subreg

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53352

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.8.0

[Bug rtl-optimization/53352] Incorrect CSE optimization on RTL expressions with a paradoxical subreg

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53352

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |acn1 at cam dot ac.uk

--- Comment #13 from Andrew Pinski <pinskia at gcc dot gnu.org> 2013-02-04 00:05:48 UTC ---
*** Bug 55108 has been marked as a duplicate of this bug. ***