[Bug rtl-optimization/48331] New: [4.7 Regression] gcc.c-torture/execute/built-in-setjmp.c FAILs with -O -fira-algorithm=priority -fPIC

[Bug rtl-optimization/48331] New: [4.7 Regression] gcc.c-torture/execute/built-in-setjmp.c FAILs with -O -fira-algorithm=priority -fPIC

[zsojka at seznam dot cz]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48331

           Summary: [4.7 Regression]
                    gcc.c-torture/execute/built-in-setjmp.c FAILs with -O
                    -fira-algorithm=priority -fPIC
           Product: gcc
           Version: 4.7.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: rtl-optimization
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: zsojka@seznam.cz


Created attachment 23797
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=23797
reduced testcase

Output:
$ gcc -O -fira-algorithm=priority -fPIC testcase.c
$ valgrind -q ./a.out  
==1889== Use of uninitialised value of size 8
==1889==    at 0x400599: sub2 (testcase.c:6)
==1889==    by 0x4E4EB6C: (below main) (in /lib64/libc-2.11.3.so)
==1889== 
==1889== Jump to the invalid address stated on the next line
==1889==    at 0x0: ???
==1889==    by 0x4E4EB6C: (below main) (in /lib64/libc-2.11.3.so)
==1889==  Address 0x0 is not stack'd, malloc'd or (recently) free'd
==1889== 
==1889== 
==1889== Process terminating with default action of signal 11 (SIGSEGV)
==1889==  Bad permissions for mapped region at address 0x0
==1889==    at 0x0: ???
==1889==    by 0x4E4EB6C: (below main) (in /lib64/libc-2.11.3.so)
Segmentation fault


Tested revisions:
r171653 - fail
r171626 - OK

[Bug rtl-optimization/48331] [4.7 Regression] gcc.c-torture/execute/built-in-setjmp.c FAILs with -O -fira-algorithm=priority -fPIC

[zsojka at seznam dot cz]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48331

--- Comment #1 from Zdenek Sojka <zsojka at seznam dot cz> 2011-03-29 12:10:38 UTC ---
The "while (1)" part of the testcase is not actually needed.

The problem seems to be there (at the assembly level):
...
    mov    rax, QWORD PTR buf@GOTPCREL[rip]    #,
    mov    rax, QWORD PTR 8[rax]    #,
    mov    QWORD PTR -8[rbp], rax    # %sfp, # stores value
    mov    rax, QWORD PTR buf@GOTPCREL[rip]    #,
    mov    rbp, QWORD PTR [rax]    #, # rbp changes
    mov    rsp, QWORD PTR 16[rax]    #,
    jmp    [QWORD PTR -8[rbp]]    # %sfp # uses new value or rbp
...

Maybe this was just uncovered by recent changes - comparing asm output from
r171626 and r171653 shows the priority allocator now gives much worse results.

In r171626, the generated code is the same as with CB's allocator, but in
r171653 there is much higher stack usage, causing this problem to show up.

[Bug rtl-optimization/48331] [4.7 Regression] gcc.c-torture/execute/built-in-setjmp.c FAILs with -O -fira-algorithm=priority -fPIC

[zsojka at seznam dot cz]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48331

Zdenek Sojka <zsojka at seznam dot cz> changed:

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

--- Comment #2 from Zdenek Sojka <zsojka at seznam dot cz> 2011-03-29 14:35:34 UTC ---
It started with http://gcc.gnu.org/viewcvs?view=revision&revision=171649

I don't know what's the status of this allocator (how near is its end), nor if
there are any targets that have to use it as CB's allocator doesn't work for
them.

[Bug rtl-optimization/48331] [4.7 Regression] gcc.c-torture/execute/built-in-setjmp.c FAILs with -O -fira-algorithm=priority -fPIC

[vmakarov at redhat dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48331

Vladimir Makarov <vmakarov at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |vmakarov at redhat dot com

--- Comment #3 from Vladimir Makarov <vmakarov at redhat dot com> 2011-03-29 15:07:46 UTC ---
(In reply to comment #2)
> It started with http://gcc.gnu.org/viewcvs?view=revision&revision=171649
> 
> I don't know what's the status of this allocator (how near is its end), nor if
> there are any targets that have to use it as CB's allocator doesn't work for
> them.

Thanks for reporting.  The patch is to permit to use CB allocator for ports
which had to use the priority allocator.  The performance result of the
modified CB allocator is expected to be better than the usage of priority one
for the ports.

In perspective, priority coloring will be removed.  I'd recommend maintainers
of the ports using priority coloring to check CB coloring and plan to switch to
it by default.

The changes in IRA are big and complex and probably will result some port
problems for some time because RA is the most machine-dependent part of the
compiler.  Therefore the patch was committed to the trunk on the beginning of
stage1 to have more time to fix all the problems.

Meanwhile, I am going to work and try to fix this PR.

[Bug rtl-optimization/48331] [4.7 Regression] gcc.c-torture/execute/built-in-setjmp.c FAILs with -O -fira-algorithm=priority -fPIC

[vmakarov at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48331

--- Comment #4 from Vladimir Makarov <vmakarov at gcc dot gnu.org> 2011-03-29 20:51:18 UTC ---
Author: vmakarov
Date: Tue Mar 29 20:51:14 2011
New Revision: 171699

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=171699
Log:
2011-03-29  Vladimir Makarov  <vmakarov@redhat.com>

    PR rtl-optimization/48331
    PR rtl-optimization/48334
    * ira-color.c (color_allocnos): Call setup_profitable_hard_regs
    for any used algorithm.


Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/ira-color.c

[Bug rtl-optimization/48331] [4.7 Regression] gcc.c-torture/execute/built-in-setjmp.c FAILs with -O -fira-algorithm=priority -fPIC

[hp at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48331

Hans-Peter Nilsson <hp at gcc dot gnu.org> changed:

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

--- Comment #5 from Hans-Peter Nilsson <hp at gcc dot gnu.org> 2011-03-30 01:37:08 UTC ---
Should this be closed now?

[Bug rtl-optimization/48331] [4.7 Regression] gcc.c-torture/execute/built-in-setjmp.c FAILs with -O -fira-algorithm=priority -fPIC

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48331

Richard Guenther <rguenth at gcc dot gnu.org> changed:

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

--- Comment #6 from Richard Guenther <rguenth at gcc dot gnu.org> 2011-03-30 08:38:09 UTC ---
Assuming fixed.