[Bug middle-end/55030] New: [4.8 Regression]: gcc.c-torture/execute/builtins/memcpy-chk.c execution, -Os (et al)

[Bug middle-end/55030] New: [4.8 Regression]: gcc.c-torture/execute/builtins/memcpy-chk.c execution, -Os (et al)

[hp at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55030

             Bug #: 55030
           Summary: [4.8 Regression]:
                    gcc.c-torture/execute/builtins/memcpy-chk.c execution,
                     -Os (et al)
    Classification: Unclassified
           Product: gcc
           Version: 4.8.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: middle-end
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: hp@gcc.gnu.org
            Target: i686-pc-linux-gnu


As noted in <http://gcc.gnu.org/ml/gcc-patches/2012-10/msg01968.html>, revision
r192676 introduced the following test-suite regressions for i686-pc-linux-gnu
or x86_64-pc-linux-gnu with -m32:

Running
/home/hp/gcctop/tmp/base/gcc/gcc/testsuite/gcc.c-torture/execute/builtins/builtins.exp
...
FAIL: gcc.c-torture/execute/builtins/memcpy-chk.c execution,  -Os
FAIL: gcc.c-torture/execute/builtins/memmove-chk.c execution,  -Os
FAIL: gcc.c-torture/execute/builtins/mempcpy-chk.c execution,  -Os
FAIL: gcc.c-torture/execute/builtins/memset-chk.c execution,  -Os
FAIL: gcc.c-torture/execute/builtins/snprintf-chk.c execution,  -Os
FAIL: gcc.c-torture/execute/builtins/sprintf-chk.c execution,  -Os
FAIL: gcc.c-torture/execute/builtins/stpcpy-chk.c execution,  -Os
FAIL: gcc.c-torture/execute/builtins/stpncpy-chk.c execution,  -Os
FAIL: gcc.c-torture/execute/builtins/strcat-chk.c execution,  -Os
FAIL: gcc.c-torture/execute/builtins/strcpy-chk.c execution,  -Os
FAIL: gcc.c-torture/execute/builtins/strncat-chk.c execution,  -Os
FAIL: gcc.c-torture/execute/builtins/strncpy-chk.c execution,  -Os
FAIL: gcc.c-torture/execute/builtins/vsnprintf-chk.c execution,  -Os
FAIL: gcc.c-torture/execute/builtins/vsprintf-chk.c execution,  -Os

[Bug middle-end/55030] [4.8 Regression]: gcc.c-torture/execute/builtins/memcpy-chk.c execution, -Os (et al)

[hp at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55030

--- Comment #1 from Hans-Peter Nilsson <hp at gcc dot gnu.org> 2012-10-23 01:05:29 UTC ---
Author: hp
Date: Tue Oct 23 01:05:25 2012
New Revision: 192701

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=192701
Log:
    PR middle-end/55030
    Revert:
    * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
    expand_builtin_setjmp_receiver.
    (expand_label): Adjust, call expand_builtin_setjmp_receiver
    with NULL for the label parameter.
    * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
    the frame-pointer.  Adjust comments.
    [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
    only if LABEL is non-NULL.

Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/builtins.c
    trunk/gcc/stmt.c

[Bug middle-end/55030] [4.8 Regression]: gcc.c-torture/execute/builtins/memcpy-chk.c execution, -Os (et al)

[hp at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55030

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2012-10-23
     Ever Confirmed|0                           |1

--- Comment #2 from Hans-Peter Nilsson <hp at gcc dot gnu.org> 2012-10-23 01:08:33 UTC ---
This PR is kept open until the underlying issue is analyzed and fixed.

[Bug middle-end/55030] [4.8 Regression]: gcc.c-torture/execute/builtins/memcpy-chk.c execution, -Os (et al)

[hp at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55030

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED
         AssignedTo|unassigned at gcc dot       |hp at gcc dot gnu.org
                   |gnu.org                     |

[Bug middle-end/55030] [4.8 Regression]: gcc.c-torture/execute/builtins/memcpy-chk.c execution, -Os (et al)

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55030

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

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

[Bug middle-end/55030] [4.8 Regression]: gcc.c-torture/execute/builtins/memcpy-chk.c execution, -Os (et al)

[hp at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55030

--- Comment #3 from Hans-Peter Nilsson <hp at gcc dot gnu.org> 2012-11-12 04:37:04 UTC ---
Created attachment 28664
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=28664
reduced test-case

Reduced test-case.  Do e.g. "cc1 -fpreprocessed pr55030-chk.i -quiet -dumpbase
pr55030-chk.c -m32 -mtune=generic -march=x86-64 -auxbase pr55030-chk -Os -w
-version -fno-diagnostics-show-caret -fno-tree-loop-distribute-patterns -o
pr55030-chk.s"

[Bug middle-end/55030] [4.8 Regression]: gcc.c-torture/execute/builtins/memcpy-chk.c execution, -Os (et al)

[hp at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55030

--- Comment #4 from Hans-Peter Nilsson <hp at gcc dot gnu.org> 2012-11-12 04:53:55 UTC ---
(In reply to comment #3)
> Created attachment 28664 [details]
> reduced test-case

Only intended for code inspection, not a self-contained test-case; that's
already covered as in the original description.

[Bug middle-end/55030] [4.8 Regression]: gcc.c-torture/execute/builtins/memcpy-chk.c execution, -Os (et al)

[hp at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55030

--- Comment #5 from Hans-Peter Nilsson <hp at gcc dot gnu.org> 2012-11-26 03:22:19 UTC ---
Author: hp
Date: Mon Nov 26 03:22:15 2012
New Revision: 193802

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=193802
Log:
    PR middle-end/55030
    * builtins.c (expand_builtin_setjmp_receiver): Update comment
    regarding purpose of blockage.
    * emit-rtl.c [!HAVE_blockage] (gen_blockage): Similarly for
    the head comment.
    * rtlanal.c (volatile_insn_p): Ditto.
    * doc/md.texi (blockage): Update similarly.  Change wording to
    require one of two forms, rather than implying a wider choice.
    * cse.c (cse_insn): Where checking for blocking insns, use
    volatile_insn_p instead of manual check for volatile ASM.
    * dse.c (scan_insn): Ditto.
    * cselib.c (cselib_process_insn): Ditto.

Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/builtins.c
    trunk/gcc/cse.c
    trunk/gcc/cselib.c
    trunk/gcc/doc/md.texi
    trunk/gcc/dse.c
    trunk/gcc/emit-rtl.c
    trunk/gcc/rtlanal.c

[Bug middle-end/55030] [4.8 Regression]: gcc.c-torture/execute/builtins/memcpy-chk.c execution, -Os (et al)

[hp at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55030

--- Comment #6 from Hans-Peter Nilsson <hp at gcc dot gnu.org> 2012-11-26 03:26:00 UTC ---
Author: hp
Date: Mon Nov 26 03:25:57 2012
New Revision: 193803

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=193803
Log:
    PR middle-end/55030
    * gcc.dg/guality/pr36728-1.c, gcc.dg/guality/pr36728-2.c (foo): Don't
    use volatile asms, use plain asms.   Where the output value for the
    asm is unused, write a global variable.

Modified:
    trunk/gcc/testsuite/ChangeLog
    trunk/gcc/testsuite/gcc.dg/guality/pr36728-1.c
    trunk/gcc/testsuite/gcc.dg/guality/pr36728-2.c

[Bug middle-end/55030] [4.8 Regression]: gcc.c-torture/execute/builtins/memcpy-chk.c execution, -Os (et al)

[krebbel at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55030

Andreas Krebbel <krebbel at gcc dot gnu.org> changed:

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

--- Comment #7 from Andreas Krebbel <krebbel at gcc dot gnu.org> 2012-11-27 07:40:41 UTC ---
(In reply to comment #5)
> Author: hp
> Date: Mon Nov 26 03:22:15 2012
> New Revision: 193802
> 
> URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=193802
> Log:
>     PR middle-end/55030
>     * builtins.c (expand_builtin_setjmp_receiver): Update comment
>     regarding purpose of blockage.
>     * emit-rtl.c [!HAVE_blockage] (gen_blockage): Similarly for
>     the head comment.
>     * rtlanal.c (volatile_insn_p): Ditto.
>     * doc/md.texi (blockage): Update similarly.  Change wording to
>     require one of two forms, rather than implying a wider choice.
>     * cse.c (cse_insn): Where checking for blocking insns, use
>     volatile_insn_p instead of manual check for volatile ASM.
>     * dse.c (scan_insn): Ditto.
>     * cselib.c (cselib_process_insn): Ditto.

This patch broke bootstrap on s390x.

/home/andreas/git/gcc-head/libgcc/libgcc2.c: In function ‘__powitf2’:
/home/andreas/git/gcc-head/libgcc/libgcc2.c:1779:1: internal compiler error: in
vt_expand_var_loc_chain, 
at var-tracking.c:8020
 }
 ^
0x80b9c219 vt_expand_var_loc_chain
        /home/andreas/git/gcc-head/gcc/var-tracking.c:8020
0x80b9cf23 vt_expand_loc_callback
        /home/andreas/git/gcc-head/gcc/var-tracking.c:8216
0x80305c7d cselib_expand_value_rtx_1
        /home/andreas/git/gcc-head/gcc/cselib.c:1683
0x80305e3b cselib_expand_value_rtx_1
        /home/andreas/git/gcc-head/gcc/cselib.c:1721
0x8030568d cselib_expand_value_rtx_cb(rtx_def*, bitmap_head_def*, int, rtx_def*
(*)(rtx_def*, bitmap_head
_def*, int, void*), void*)
        /home/andreas/git/gcc-head/gcc/cselib.c:1530
0x80b9c39b vt_expand_var_loc_chain
        /home/andreas/git/gcc-head/gcc/var-tracking.c:8054
0x80b9cf23 vt_expand_loc_callback
        /home/andreas/git/gcc-head/gcc/var-tracking.c:8216
0x80305c7d cselib_expand_value_rtx_1
        /home/andreas/git/gcc-head/gcc/cselib.c:1683
0x80305e3b cselib_expand_value_rtx_1
        /home/andreas/git/gcc-head/gcc/cselib.c:1721
0x8030568d cselib_expand_value_rtx_cb(rtx_def*, bitmap_head_def*, int, rtx_def*
(*)(rtx_def*, bitmap_head
_def*, int, void*), void*)
        /home/andreas/git/gcc-head/gcc/cselib.c:1530
0x80b9c39b vt_expand_var_loc_chain
        /home/andreas/git/gcc-head/gcc/var-tracking.c:8054
0x80b9cf23 vt_expand_loc_callback
        /home/andreas/git/gcc-head/gcc/var-tracking.c:8216
0x80305c7d cselib_expand_value_rtx_1
        /home/andreas/git/gcc-head/gcc/cselib.c:1683
0x8030568d cselib_expand_value_rtx_cb(rtx_def*, bitmap_head_def*, int, rtx_def*
(*)(rtx_def*, bitmap_head_def*, int, void*), void*)
        /home/andreas/git/gcc-head/gcc/cselib.c:1530
0x80b9c39b vt_expand_var_loc_chain
        /home/andreas/git/gcc-head/gcc/var-tracking.c:8054
0x80b9d67d vt_expand_1pvar
        /home/andreas/git/gcc-head/gcc/var-tracking.c:8331
0x80b9d9c7 emit_note_insn_var_location
        /home/andreas/git/gcc-head/gcc/var-tracking.c:8385
0x80f1a4e3 htab_traverse_noresize
        /home/andreas/git/gcc-head/libiberty/hashtab.c:784
0x80f1a5bf htab_traverse
        /home/andreas/git/gcc-head/libiberty/hashtab.c:800
0x80b9f613 emit_notes_for_changes
        /home/andreas/git/gcc-head/gcc/var-tracking.c:8749
Please submit a full bug report,
with preprocessed source if appropriate.
Please include the complete backtrace with any bug report.
See <http://gcc.gnu.org/bugs.html> for instructions.
make[5]: *** [_powitf2.o] Error 1
make[5]: *** Waiting for unfinished jobs....

[Bug middle-end/55030] [4.8 Regression]: gcc.c-torture/execute/builtins/memcpy-chk.c execution, -Os (et al)

[hp at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55030

--- Comment #8 from Hans-Peter Nilsson <hp at gcc dot gnu.org> 2012-11-27 11:54:47 UTC ---
(In reply to comment #7)
> This patch broke bootstrap on s390x.

Please open a new PR, assigned to me, and attach preprocessed code.
Make it depend on this PR.

[Bug middle-end/55030] [4.8 Regression]: gcc.c-torture/execute/builtins/memcpy-chk.c execution, -Os (et al)

[hp at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55030

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

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

--- Comment #9 from Hans-Peter Nilsson <hp at gcc dot gnu.org> 2012-11-27 23:16:27 UTC ---
This PR is fixed.  For fallouts like the one hinted in comment #7, please open
new PR's.

[Bug middle-end/55030] [4.8 Regression]: gcc.c-torture/execute/builtins/memcpy-chk.c execution, -Os (et al)

[ebotcazou at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55030

Eric Botcazou <ebotcazou at gcc dot gnu.org> changed:

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

--- Comment #10 from Eric Botcazou <ebotcazou at gcc dot gnu.org> 2013-02-16 09:37:22 UTC ---
I'm getting back to this because I think that we should reinstate the original
patch, now that the blockage patch has been installed.

I have run into the same issue as your original issue with a private port on
the 4.7 branch: the clobber causes the restoring of the frame pointer to be
deleted
  http://gcc.gnu.org/ml/gcc-patches/2012-10/msg01172.html
Later reload allocates a stack slot to a pseudo that is set before the setjmp
and used after, but the frame pointer doesn't have a consistent value...

Clearly the frame pointer needs to be restored so the clobber is wrong.  It was
there because the final blockage wasn't blocking enough, but the blockage patch
is supposed to have fixed that.

[Bug middle-end/55030] [4.8 Regression]: gcc.c-torture/execute/builtins/memcpy-chk.c execution, -Os (et al)

[ebotcazou at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55030

--- Comment #11 from Eric Botcazou <ebotcazou at gcc dot gnu.org> 2013-02-16 09:44:31 UTC ---
While we are at it, we could also revert the dse.c and cselib.c hunks of the
blockage patch, which weren't strictly necessary.  Jakub was really concerned
about their impact on volatile asms.

[Bug middle-end/55030] [4.8 Regression]: gcc.c-torture/execute/builtins/memcpy-chk.c execution, -Os (et al)

[hp at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55030

--- Comment #12 from Hans-Peter Nilsson <hp at gcc dot gnu.org> 2013-02-17 00:33:17 UTC ---
(In reply to comment #10)
> I'm getting back to this because I think that we should reinstate the original
> patch, now that the blockage patch has been installed.

*wake-up reactions*

> I have run into the same issue as your original issue with a private port on
> the 4.7 branch: the clobber causes the restoring of the frame pointer to be
> deleted
>   http://gcc.gnu.org/ml/gcc-patches/2012-10/msg01172.html
> Later reload allocates a stack slot to a pseudo that is set before the setjmp
> and used after, but the frame pointer doesn't have a consistent value...

Yup, this far I remember.

> Clearly the frame pointer needs to be restored so the clobber is wrong.  It was
> there because the final blockage wasn't blocking enough, but the blockage patch
> is supposed to have fixed that.

I've lost track. What was "the original patch", what do you mean by "the
blockage patch" (that has been installed) and I'm pretty sure there were
several follow-up patches, so I can't say I'm confident about reverting
something from just this subset.  (To wit: if it's something that causes
volatile asms to again be treated different from (other) blockages, then that's
wrong, as a volatile asm is the default blockage.)

Can you please a candidate (reverting?) patch gcc-patches@ *and CC me* (I'm far
behind on reading gcc lists).

[Bug middle-end/55030] [4.8 Regression]: gcc.c-torture/execute/builtins/memcpy-chk.c execution, -Os (et al)

[ebotcazou at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55030

--- Comment #13 from Eric Botcazou <ebotcazou at gcc dot gnu.org> 2013-02-18 10:48:18 UTC ---
> I've lost track. What was "the original patch", what do you mean by "the
> blockage patch" (that has been installed) and I'm pretty sure there were
> several follow-up patches, so I can't say I'm confident about reverting
> something from just this subset.  (To wit: if it's something that causes
> volatile asms to again be treated different from (other) blockages, then that's
> wrong, as a volatile asm is the default blockage.)

The original patch was the patch you reverted in comment #1 and the fix for the
original issue.  As far as I can see, you didn't put it back after installing
the blockage patch, which is the patch in comment #5.  The dse.c and cselib.c
hunks were not necessary and not in the submission, but I asked you to add them
for the sake of consistency; then Jakub pointed out that they pessimize
volatile asms.

[Bug middle-end/55030] [4.8 Regression]: gcc.c-torture/execute/builtins/memcpy-chk.c execution, -Os (et al)

[hp at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55030

--- Comment #14 from Hans-Peter Nilsson <hp at gcc dot gnu.org> 2013-02-18 22:21:28 UTC ---
(In reply to comment #13)
> The original patch was the patch you reverted in comment #1 and the fix for the
> original issue.  As far as I can see, you didn't put it back after installing
> the blockage patch, which is the patch in comment #5.  The dse.c and cselib.c
> hunks were not necessary and not in the submission, but I asked you to add them
> for the sake of consistency; then Jakub pointed out that they pessimize
> volatile asms.

Thanks for clarifying, I (think I) get it now.  Not sure when I can revisit,
maybe this month, maybe the next, so I won't feel bad if someone beats me to
it.

[Bug middle-end/55030] [4.8 Regression]: gcc.c-torture/execute/builtins/memcpy-chk.c execution, -Os (et al)

[ebotcazou at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55030

--- Comment #15 from Eric Botcazou <ebotcazou at gcc dot gnu.org> 2013-02-19 11:32:24 UTC ---
> Thanks for clarifying, I (think I) get it now.  Not sure when I can revisit,
> maybe this month, maybe the next, so I won't feel bad if someone beats me to
> it.

Sure, no urgency at all, thanks.

[Bug middle-end/55030] [4.8 Regression]: gcc.c-torture/execute/builtins/memcpy-chk.c execution, -Os (et al)

[mikpe at it dot uu.se]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55030

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

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

--- Comment #16 from Mikael Pettersson <mikpe at it dot uu.se> ---
FWIW, I've bootstrapped and regtested gcc 4.9 with r192676 reapplied and the
dse.c and cselib.c hunks of r193802 reverted on several hosts (x86_64, sparc64,
powerpc64, armv5tel, m68k) without regressions.

[Bug middle-end/55030] [4.8 Regression]: gcc.c-torture/execute/builtins/memcpy-chk.c execution, -Os (et al)

[hp at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55030

--- Comment #17 from Hans-Peter Nilsson <hp at gcc dot gnu.org> ---
(In reply to Mikael Pettersson from comment #16)
> FWIW, I've bootstrapped and regtested gcc 4.9 with r192676 reapplied and the
> dse.c and cselib.c hunks of r193802 reverted on several hosts (x86_64,
> sparc64, powerpc64, armv5tel, m68k) without regressions.

Wow, thanks.

So, what's needed is to just spell out the exact patch (a description as above
doesn't suffice unfortunately), send it to gcc-patches and CC Eric who I think
would ok it at a glance.  Here's me hoping someone will beat me to it, but it
just might happen *this* month...

[Bug middle-end/55030] [4.8 Regression]: gcc.c-torture/execute/builtins/memcpy-chk.c execution, -Os (et al)

[ebotcazou at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55030

--- Comment #18 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
> FWIW, I've bootstrapped and regtested gcc 4.9 with r192676 reapplied and the
> dse.c and cselib.c hunks of r193802 reverted on several hosts (x86_64,
> sparc64, powerpc64, armv5tel, m68k) without regressions.

Let's just reapply r192676 on the mainline.  We just need to make sure that the
failures on i686 don't reappear.

[Bug middle-end/55030] [4.8 Regression]: gcc.c-torture/execute/builtins/memcpy-chk.c execution, -Os (et al)

[hp at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55030

--- Comment #19 from Hans-Peter Nilsson <hp at gcc dot gnu.org> ---
(In reply to Eric Botcazou from comment #18)
> Let's just reapply r192676 on the mainline.  We just need to make sure that
> the failures on i686 don't reappear.

Done now less than half a year later!
(Just r192676, no dse.c or cselib.c changes, leaving that to interested
parties.)