GCC 3.0 Status Report

[John David Anglin]

Forwarded message:
>From dave Tue Apr 17 19:11:04 EDT 2001
Subject: Re: GCC 3.0 Status Report
To: gcc@gcc.gn.org, law@redhat.com
Date: Tue, 17 Apr 2001 19:11:04 -0400 (EDT)
From: "John David Anglin" <dave@hiauly1>
X-Mailer: ELM [version 2.4 PL25]
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Content-Length: 2640      

> Last Week
> =========

> Jeff Law reported some initial success building GCC 3.0 on HPUX.  The
> compiler is now bootstrapping, but libstdc++ V3 is still not working.
> If you can help out with the debugging task (either because you're a
> PA expert or a V3 expert or a general GCC expert), please do so.  The
> best way to help is probably to contact Jeff Law (law@redhat.com)
> directly.

I am currently doing a build with a hack to avoid the std::errno
problem with respect to libstdc++-v3.  However, the number of failures
is still about the same:

                === g++ Summary for unix/-fPIC ===

# of expected passes            5531
# of unexpected failures        868
# of unexpected successes       2
# of expected failures          93
# of untested testcases         20

I looked at one of them briefly.  The test builds and runs successfully
if linked with `-static'.  When linked with the shared libstdc++-v3,
the test fails almost immediately with a segmentation fault (I think
running __cxx_new_vec.  When I try to single step with gdb, it dumps
core.  It helps to link with immediate binding.  Adb works better than
gdb.  Maybe this gives a clue as to the problem:

>adb g++-bugs-900220_03-C.x
:r
g++-bugs-900220_03-C.x: running (process 4061)
segmentation violation
stopped at      0xC132DA80:     LDW             1980(r20),r20
0xC132DA70,8?ia
0xC132DA70:     LDW             -260(r30),r19
0xC132DA74:     STW             r28,-256(r30)
0xC132DA78:     ADDIL           L%0xffff7000,r19
0xC132DA7C:     OR              r1,r0,r20
0xC132DA80:     LDW             1980(r20),r20
0xC132DA84:     OR              r20,r0,r22
0xC132DA88:     LDWS            0(r20),r20
0xC132DA8C:     LDO             1(r20),r20
0xC132DA90:
$r
pcoqh 0xC132DA83
pcoqt 0xC132DA87
rp    0xC132DA73

arg0  7B006C90      arg1  0             arg2  0             arg3  7AFF289C
sp    7B03AE60      ret0  7B006BB0      ret1  10C8C00       dp    40001308
r1    0xFFFF7000    r3    7AFFA89C      r4    1             r5    7B03A730
r6    7B03A738      r7    7B03A860      r8    7B03A860      r9    4003A2C0
r10   4003AAC0      r11   400372C0      r12   4003AB04      r13   3F
r14   3F            r15   3D            r16   3D            r17   3A
r18   40036822      r19   0             r20   0xFFFF7000    r21   0xC0087EC0
r22   7B006BB0      r31   7B006C90      sar   8             sr0   0
sr1   1138          sr2   5A94          sr3   0             sr4   62DE

The address in r20 is clearly wacko.

Dave
-- 
J. David Anglin                                  dave.anglin@nrc.ca
National Research Council of Canada              (613) 990-0752 (FAX: 952-6605)


-- 
J. David Anglin                                  dave.anglin@nrc.ca
National Research Council of Canada              (613) 990-0752 (FAX: 952-6605)