[Bug c/11999] New: illegal instruction in libunwind-ia64.c

[Bug c/11999] New: illegal instruction in libunwind-ia64.c

[rscohnlkjh at netscape dot net]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11999

           Summary: illegal instruction in libunwind-ia64.c
           Product: gcc
           Version: 3.4
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: c
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: rscohnlkjh at netscape dot net
                CC: gcc-bugs at gcc dot gnu dot org
  GCC host triplet: ia64
GCC target triplet: ia64

There is an asm on line 2239 in gcc/config/ia64/libunwind-ia64.c that contains
the following instruction:
	"(p6) mov.i ar.fpsr = r30		\n\t"

The fpsr application register can only be referenced on the m unit and this
instruction attempts to reference it on the i unit. It would cause an illegal
instruction fault if it was executed. It should be:

"(p6) mov.m ar.fpsr = r30		\n\t"

See table 3.3 of the "Intel Itanium architecture software devloper's manual, vol 1"

I don't have a reproducer for this. The problem was found by a program that
inspects code, not by running a gcc compiled binary that executes the code.

Robert

[Bug target/11999] illegal instruction in libunwind-ia64.c

[pinskia at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11999


pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|c                           |target


------- Additional Comments From pinskia at gcc dot gnu dot org  2003-08-20 22:30 -------
Could you read <http://gcc.gnu.org/contribute.html> and submit a patch to gcc-
patches@gcc.gnu.org?

[Bug target/11999] illegal instruction in unwind-ia64.c

[pinskia at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11999


pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1
   Last reconfirmed|0000-00-00 00:00:00         |2003-08-21 14:35:52
               date|                            |
            Summary|illegal instruction in      |illegal instruction in
                   |libunwind-ia64.c            |unwind-ia64.c


------- Additional Comments From pinskia at gcc dot gnu dot org  2003-08-21 14:35 -------
The file is not called libunwind-ia64.c but unwind-ia64.c and it still exists in the mainline.

[Bug target/11999] illegal instruction in unwind-ia64.c

[rscohnlkjh at netscape dot net]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11999



------- Additional Comments From rscohnlkjh at netscape dot net  2003-08-22 18:59 -------
The function that contains the illegal asm, uw_install_context, is static and is
not called in the file. Should I just delete this unused and untested function?

[Bug target/11999] illegal instruction in unwind-ia64.c

[pinskia at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11999



------- Additional Comments From pinskia at gcc dot gnu dot org  2003-08-22 19:06 -------
It is not unused as it is used in the header unwind.inc which gets included at the end of the file.

[Bug target/11999] illegal instruction in unwind-ia64.c

[dhazeghi at yahoo dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11999


dhazeghi at yahoo dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|3.4                         |---

[Bug target/11999] illegal instruction in unwind-ia64.c

[dhazeghi at yahoo dot com]

------- Additional Comments From dhazeghi at yahoo dot com  2003-12-09 23:58 -------
Still present, I made a patch. Hopefully some ia64 expert can now okay it.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |patch


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11999

[Bug target/11999] illegal instruction in unwind-ia64.c

[dhazeghi at yahoo dot com]

------- Additional Comments From dhazeghi at yahoo dot com  2003-12-11 08:29 -------
Jim, would you mind commenting on this report and patch? Thanks.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |wilson at tuliptree dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11999

[Bug target/11999] illegal instruction in unwind-ia64.c

[wilson at tuliptree dot org]

------- Additional Comments From wilson at tuliptree dot org  2003-12-11 08:58 -------
Subject: Re:  illegal instruction in unwind-ia64.c

On Thu, 2003-12-11 at 00:29, dhazeghi at yahoo dot com wrote:
> Jim, would you mind commenting on this report and patch? Thanks.

I'm a few days behind on gcc-patches.  I was assuming that there was a
gcc-patches message that I would get to, but apparently I assumed wrong.

There is an assembler problem here that needs to be fixed in addition to
the compiler problem.  The assembler should not be accepting invalid
instructions.  I don't want the assembler problem forgotten.  Perhaps
this bug report can be kept open until the assembler problem is also
fixed?  If that isn't reasonable, I can handle this some other way.

Jakub Jelinek has contributed a larger patch that fixes this problem in
addition to several others.
	http://gcc.gnu.org/ml/gcc-patches/2003-12/msg00940.html
Since I need to review Jakub's patch anyways, and I see no reason to
force him to unnecessarily rewrite his patch, we probably won't need
this patch.

The fix in this patch is the same as one of the fixes in Jakub's patch.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11999

[Bug target/11999] illegal instruction in unwind-ia64.c

[dhazeghi at yahoo dot com]

------- Additional Comments From dhazeghi at yahoo dot com  2003-12-11 09:14 -------
This can certainly stay open until the assembler gets fixed (does binutils have a bug db?). Any 
patch that fixes it is fine by me (Jakub's or otherwise). Thanks for the feedback.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11999

[Bug target/11999] illegal instruction in unwind-ia64.c

[wilson at tuliptree dot org]

------- Additional Comments From wilson at tuliptree dot org  2003-12-11 18:40 -------
Subject: Re:  illegal instruction in unwind-ia64.c

On Thu, 2003-12-11 at 01:15, dhazeghi at yahoo dot com wrote:
> This can certainly stay open until the assembler gets fixed (does binutils have a bug db?). Any 
> patch that fixes it is fine by me (Jakub's or otherwise). Thanks for the feedback.

There is no binutils bug db.  I will mention this to HJ, maybe he will
be interested in fixing the assembler bug.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11999

[Bug target/11999] illegal instruction in unwind-ia64.c

[dhazeghi at yahoo dot com]

------- Additional Comments From dhazeghi at yahoo dot com  2004-01-19 20:54 -------
Can we close this now that Jakub has committed his patch? Thanks.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|patch                       |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11999

[Bug target/11999] illegal instruction in unwind-ia64.c

[wilson at specifixinc dot com]

------- Additional Comments From wilson at specifixinc dot com  2004-01-20 21:41 -------
Subject: Re:  illegal instruction in unwind-ia64.c

On Mon, 2004-01-19 at 12:54, dhazeghi at yahoo dot com wrote:
> Can we close this now that Jakub has committed his patch? Thanks.

There is still an assembler bug that needs to be fixed, and no binutils
bug database I can register it in, so there would be some use in keeping
the bug report open so that we don't forget about the assembler bug.  I
asked HJ if he could look at the assembler bug, and he put it on his
list of things to look at.  He hasn't gotten to it yet.

This shouldn't be holding up the gcc-3.4 release though.  The version
could be changed to 3.5 if that helps, and the priority could be
lowered.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11999

[Bug target/11999] [binutils bug] illegal instruction in unwind-ia64.c accepts by gas

[dhazeghi at yahoo dot com]

------- Additional Comments From dhazeghi at yahoo dot com  2004-01-21 01:00 -------
I just changed the summary accordingly. Thanks.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|normal                      |minor
            Summary|illegal instruction in      |[binutils bug] illegal
                   |unwind-ia64.c               |instruction in unwind-ia64.c
                   |                            |accepts by gas


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11999

[Bug target/11999] [binutils bug] illegal instruction in unwind-ia64.c accepts by gas

[wilson at gcc dot gnu dot org]

------- Additional Comments From wilson at gcc dot gnu dot org  2004-05-28 20:30 -------
HJ has written a patch to fix the binutils bug.  The original bogus assembler
code now gets an assembler error as it should.  The patch is here:
    http://sources.redhat.com/ml/binutils/2004-05/msg00565.html


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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11999