[Bug rtl-optimization/23650] Scheduler does not handle zero_extract correctly

[Bug rtl-optimization/23650] Scheduler does not handle zero_extract correctly

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23650

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |WAITING
   Last reconfirmed|                            |2012-01-11
     Ever Confirmed|0                           |1

--- Comment #10 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-01-11 14:23:13 UTC ---
Is this still an issue?

[Bug rtl-optimization/23650] Scheduler does not handle zero_extract correctly

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23650

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|WAITING                     |RESOLVED
         Resolution|                            |INVALID

--- Comment #11 from Andrew Pinski <pinskia at gcc dot gnu.org> 2013-03-20 00:48:42 UTC ---
No feedback in over an year so closing.

[Bug rtl-optimization/23650] Scheduler does not handle zero_extract correctly

[Erwin dot Unruh at fujitsu-siemens dot com]

------- Additional Comments From Erwin dot Unruh at fujitsu-siemens dot com  2005-09-20 07:22 -------
Created an attachment (id=9780)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=9780&action=view)
complete testcase (corrected)

My first try on the testcase depends on the endianess. S/390 is big endian,
intel is little endian. This revised testcase avoids this dependency. It runs
correctly on intel (-O0 or -O3).

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
Attachment #9773 is|0                           |1
           obsolete|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23650

[Bug rtl-optimization/23650] Scheduler does not handle zero_extract correctly

[rguenth at gcc dot gnu dot org]

------- Additional Comments From rguenth at gcc dot gnu dot org  2005-09-19 17:57 -------
Created an attachment (id=9774)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=9774&action=view)
complete testcase

Combining comment #1 and the testcase, declaring some functions and exchaning
printf for abort() I cannot reproduce the bug as the testcase aborts with
every optimization level, even -O0 -fno-strict-aliasing and shows this same
behavior even with the intel compiler.

Can you re-check, if the testcase is really working in some case?

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23650

[Bug rtl-optimization/23650] Scheduler does not handle zero_extract correctly

[Erwin dot Unruh at fujitsu-siemens dot com]

------- Additional Comments From Erwin dot Unruh at fujitsu-siemens dot com  2005-09-19 12:25 -------
Created an attachment (id=9773)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=9773&action=view)
testcase supplement

Here is some additional coding, which calls the function given in the bug
report. The exit code is 1 in case of error, zero otherwise.
I was able to verify my original analysis with this test.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23650

[Bug rtl-optimization/23650] Scheduler does not handle zero_extract correctly

[Erwin dot Unruh at fujitsu-siemens dot com]

------- Additional Comments From Erwin dot Unruh at fujitsu-siemens dot com  2005-09-02 09:08 -------
The zero_extract is generated by the combiner pass (x.c.17.combine). Note that 
the official documentation of zero_extract requires a mem:QI (see 
http://gcc.gnu.org/onlinedocs/gccint/Bit_002dFields.html).

I just tested the case with a 4.0.1 crosscompiler. The error is still present.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23650

[Bug rtl-optimization/23650] Scheduler does not handle zero_extract correctly

[rguenth at gcc dot gnu dot org]

------- Additional Comments From rguenth at gcc dot gnu dot org  2005-08-31 14:58 -------
Also note that gcc 3.3.2 is not exactly state of the art, but unmaintained,
and a fix would go at earliest into 3.4.5, which may be already fixed.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23650

[Bug rtl-optimization/23650] Scheduler does not handle zero_extract correctly

[bonzini at gcc dot gnu dot org]

------- Additional Comments From bonzini at gcc dot gnu dot org  2005-08-31 14:45 -------
Having a zero_extract that takes 16 bytes out of a QImode mem seems extremely
wrong.  Can you find out which pass generates that thing?  Just grep for

zero_extract:SI..mem[a-z/]:QI

in all the files dumped by -da.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23650

[Bug rtl-optimization/23650] Scheduler does not handle zero_extract correctly

[rguenth at gcc dot gnu dot org]

------- Additional Comments From rguenth at gcc dot gnu dot org  2005-08-31 14:39 -------
Can you please provide a testcase that can be executed and does abort() if the
outcome is wrong and has zero exit code if it is right?

Thanks,
Richard.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23650

[Bug rtl-optimization/23650] Scheduler does not handle zero_extract correctly

[Erwin dot Unruh at fujitsu-siemens dot com]

------- Additional Comments From Erwin dot Unruh at fujitsu-siemens dot com  2005-08-31 12:51 -------
With option -fno-strict-aliasing the generated code is still incorrect.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |UNCONFIRMED
         Resolution|DUPLICATE                   |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23650

[Bug rtl-optimization/23650] Scheduler does not handle zero_extract correctly

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-08-31 12:29 -------
You are violating C aliasing rules, either use an union (which is a GCC extension) or use -fno-strict-
alias.

(Char2 * ) &var

*** This bug has been marked as a duplicate of 21920 ***

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|                            |DUPLICATE


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23650