Re: [rfa/testsuite] test script for pr gdb/1056, divide by zero in gdb

Re: [rfa/testsuite] test script for pr gdb/1056, divide by zero in gdb

[Michael Elizabeth Chastain]

Joel asks:
> What is our policy regarding the insertion of URLs pointing to GDB PRs,
> or URLs in general? I would prefer that we actually copy the relevant
> information from the URL rather than inserting the URL.

I'm not aware of an actual policy about this.

I like the URL because the PR database is the central repository
for information about bugs in gdb.  It's easy for anyone to add new
information to the PR database, but it requires an FSF copyright
assignment and maintainer approval to add information to a test case.

I think that this test case has enough information even if
the PR database disappears.  Specifically:

  # When SIGFPE happens, the operating system may restart the
  # offending instruction after the signal handler returns,
  # rather than proceeding to the next instruction.  This happens
  # on i686-pc-linux-gnu with a linux kernel.  If gdb has a naive
  # signal handler that just returns, then it will restart the
  # broken instruction and gdb gets an endless stream of SIGFPE's
  # and makes no progress.

If you want even more text in the test case, I'm open to patches.

Michael C

Re: [rfa/testsuite] test script for pr gdb/1056, divide by zero in gdb

[Andrew Cagney]

> Joel asks:
> 
>> What is our policy regarding the insertion of URLs pointing to GDB PRs,
>> or URLs in general? I would prefer that we actually copy the relevant
>> information from the URL rather than inserting the URL.
> 
> 
> I'm not aware of an actual policy about this.

I'm not either.

While ".../gdb/bugs/<PR>" will out last any switch to mozilla (re-number 
over my dead body :-), I don't think people will appreciate having 
"redhat" embedded in the file.

I'd just stick to quoting the GDB pr number and any relevant text.

> I like the URL because the PR database is the central repository
> for information about bugs in gdb.  It's easy for anyone to add new
> information to the PR database, but it requires an FSF copyright
> assignment and maintainer approval to add information to a test case.
> 
> I think that this test case has enough information even if
> the PR database disappears.  Specifically:
> 
>   # When SIGFPE happens, the operating system may restart the
>   # offending instruction after the signal handler returns,
>   # rather than proceeding to the next instruction.  This happens
>   # on i686-pc-linux-gnu with a linux kernel.  If gdb has a naive
>   # signal handler that just returns, then it will restart the
>   # broken instruction and gdb gets an endless stream of SIGFPE's
>   # and makes no progress.
> 
> If you want even more text in the test case, I'm open to patches.

Anyway, yes ok.

Andrew

Re: [rfa/testsuite] test script for pr gdb/1056, divide by zero in gdb

[Michael Elizabeth Chastain]

ac> While ".../gdb/bugs/<PR>" will out last any switch to mozilla (re-number 
ac> over my dead body :-), I don't think people will appreciate having 
ac> "redhat" embedded in the file.
ac> 
ac> I'd just stick to quoting the GDB pr number and any relevant text.

Err, okay.  I wish that I could say "http://gdb.gnu.org/bugs/1234".
But I can't.  So I'll delete this line.

  # See http://sources.redhat.com/gdb/bugs/1056

Modified patch below.  There are still plenty of places that tie it
to gdb pr/1056.

ac> Anyway, yes ok.

I'll take that as approval for the modified patch.  Wheee!

I'm committing this now.

Michael C

===

2003-10-20  Michael Chastain  <mec@shout.net>

	* gdb.base/gdb1056.exp: New test script.

# Copyright 2003 Free Software Foundation, Inc.

# This program is free software; you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation; either version 2 of the License, or
# (at your option) any later version.
# 
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License for more details.
# 
# You should have received a copy of the GNU General Public License
# along with this program; if not, write to the Free Software
# Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.  

# Test for PR gdb/1056.
# 2003-10-18  Michael Chastain <mec@shout.net>

if $tracelevel then {
    strace $tracelevel
}

# test SIGFPE (such as division by 0) inside gdb itself

set prms_id 0
set bug_id 0

gdb_start

# When SIGFPE happens, the operating system may restart the
# offending instruction after the signal handler returns,
# rather than proceeding to the next instruction.  This happens
# on i686-pc-linux-gnu with a linux kernel.  If gdb has a naive
# signal handler that just returns, then it will restart the
# broken instruction and gdb gets an endless stream of SIGFPE's
# and makes no progress.
#
# On a broken gdb this test will just time out.

gdb_test_multiple "print 1/0" "" {
    -re ".*$gdb_prompt $" {
	pass "print 1/0"
    }
    timeout {
	kfail "gdb/1056" "print 1/0"
    }
}

Re: [rfa/testsuite] test script for pr gdb/1056, divide by zero in gdb

[Joel Brobecker]

No real comment, just a general question:

> # See http://sources.redhat.com/gdb/bugs/1056

What is our policy regarding the insertion of URLs pointing to GDB PRs,
or URLs in general? I would prefer that we actually copy the relevant
information from the URL rather than inserting the URL.

One reason is that the URL may disappear in the future, maybe even
sooner than we think if we move to bugzilla. It also takes longer for
the reader to fetch the information referenced by this comment.

-- 
Joel

[rfa/testsuite] test script for pr gdb/1056, divide by zero in gdb

[Michael Elizabeth Chastain]

This is a new test script.  It tests for PR gdb/1056.

  http://sources.redhat.com/gdb/bugs/1056
  divide by 0 hangs gdb

The test script includes some comments which analyze the problem.

The test script accepts any output from gdb followed by $gdb_prompt
as okay.  A timeout results in a KFAIL.

Testing: I ran this script on native i686-pc-linux-gnu, red hat 8.0.
The result was a KFAIL.

Okay to commit?

Michael C

===

2003-10-18  Michael Chastain  <mec@shout.net>

	* gdb.base/gdb1056.exp: New test script.

===

# Copyright 2003 Free Software Foundation, Inc.

# This program is free software; you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation; either version 2 of the License, or
# (at your option) any later version.
# 
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License for more details.
# 
# You should have received a copy of the GNU General Public License
# along with this program; if not, write to the Free Software
# Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.  

# Test for PR gdb/1056.
# 2003-10-18  Michael Chastain <mec@shout.net>

if $tracelevel then {
    strace $tracelevel
}

# test SIGFPE (such as division by 0) inside gdb itself

set prms_id 0
set bug_id 0

gdb_start

# See http://sources.redhat.com/gdb/bugs/1056
#
# When SIGFPE happens, the operating system may restart the
# offending instruction after the signal handler returns,
# rather than proceeding to the next instruction.  This happens
# on i686-pc-linux-gnu with a linux kernel.  If gdb has a naive
# signal handler that just returns, then it will restart the
# broken instruction and gdb gets an endless stream of SIGFPE's
# and makes no progress.
#
# On a broken gdb this test will just time out.

gdb_test_multiple "print 1/0" "" {
    -re ".*$gdb_prompt $" {
	pass "print 1/0"
    }
    timeout {
	kfail "gdb/1056" "print 1/0"
    }
}