testcase help

testcase help

[Manoj Iyer]

I am writing a testcase for gdb in which I want to make GDB continue to
the next breakpoint, so I used gdb_continue "FUNC" to get to the
next breakpoint.

Here is the problem that I have ...

============ testcase code ================
gdb_continue "tf"
gdb_expect {
    -re "Breakpoint .* at .*\r\n$gdb_prompt $" {
        pass "continue to tf";
    }
    -re ".*$gdb_prompt $" {
        fail "continue to tf";
        gdb_suppress_tests;
    }
    timeout {
        fail "continue to tf (timeout)";
        gdb_suppress_tests;
    }
}

=========== Out put ==============
continue
30          for (n = 0; n < N; ++n)
(gdb) FAIL: gdb.threads/thread_break.exp: continue to tf
continue
Continuing.
[New Thread 32769 (LWP 29005)]
[New Thread 16386 (LWP 29006)]
[Switching to Thread 16386 (LWP 29006)]

Breakpoint 2, tf (arg=0x0) at gdb.threads/tbug.c:14
14          int n = (int) (long int) arg;
(gdb) PASS: gdb.threads/thread_break.exp: continue to tf

but a gdb_expect with correct parameters catches this continue as success.
I cannot get it to continue correctly with send_gdb "continue" that is why
I used gdb_continue.

What am I doing wrong?

Thanks
----- ----
Manoj Iyer
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+ Cognito ergo sum                                                          +
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

Re: testcase help

[Michael Chastain]

I don't even use gdb_continue ... how about just plain:

  gdb_test_multiple "continue" "continue to foo ..." {
    ...
  }

Or maybe I'm not getting the problem.

Michael C

Re: testcase help

[Manoj Iyer]

Michael,

I tried ur suggestion but it still did not work, so I used send_gdb and
gdb_expect{} and got it to work. Thanks for the help.

Thanks
----- ----
Manoj Iyer
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+ Cognito ergo sum                                                          +
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

On Wed, 4 Aug 2004, Michael Chastain wrote:

> I don't even use gdb_continue ... how about just plain:
>
>   gdb_test_multiple "continue" "continue to foo ..." {
>     ...
>   }
>
> Or maybe I'm not getting the problem.
>
> Michael C
>

thread testcase compile issue (question)

[Manoj Iyer]

When I execute tests under gdb.threads/ directory for instance killed.exp,
I see 2 complies of the same testcase, the first one gets a linker error
/usr/bin/ld: cannot find -lpthreads but the second time it works with no
errors. Anyone has any idea why???

Here is the output on x86 for you to compare:

                === gdb tests ===

Schedule of variations:
    unix

Running target unix
Using /home/manjo/projects/gdb/june/src/dejagnu/baseboards/unix.exp as
board des
cription file for target.
Using /home/manjo/projects/gdb/june/src/dejagnu/config/unix.exp as generic
inter
face file for target.
Using ./config/unix.exp as tool-and-target-specific interface file.
Running ./gdb.threads/killed.exp ...
Executing on host: gcc ./gdb.threads/killed.c
-I/home/manjo/projects/gdb/june/s
rc/gdb/testsuite -g  -lpthreads -lm   -o
/home/manjo/projects/gdb/june/src/gdb/t
estsuite/gdb.threads/killed    (timeout = 300)
/usr/bin/ld: cannot find -lpthreads
collect2: ld returned 1 exit status
compiler exited with status 1
output is:
/usr/bin/ld: cannot find -lpthreads
collect2: ld returned 1 exit status

Executing on host: gcc ./gdb.threads/killed.c
-I/home/manjo/projects/gdb/june/s
rc/gdb/testsuite -g  -lpthread -lm   -o
/home/manjo/projects/gdb/june/src/gdb/te
stsuite/gdb.threads/killed    (timeout = 300)
PASS: gdb.threads/killed.exp: successfully compiled posix threads test
case
GNU gdb 2004-06-30-cvs


Thanks
----- ----
Manoj Iyer
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+ Cognito ergo sum                                                          +
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

On Thu, 5 Aug 2004, Manoj Iyer wrote:

>
> Michael,
>
> I tried ur suggestion but it still did not work, so I used send_gdb and
> gdb_expect{} and got it to work. Thanks for the help.
>
> Thanks
> ----- ----
> Manoj Iyer
> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> + Cognito ergo sum                                                          +
> +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
>
> On Wed, 4 Aug 2004, Michael Chastain wrote:
>
> > I don't even use gdb_continue ... how about just plain:
> >
> >   gdb_test_multiple "continue" "continue to foo ..." {
> >     ...
> >   }
> >
> > Or maybe I'm not getting the problem.
> >
> > Michael C
> >
>

Re: thread testcase compile issue (question)

[Michael Chastain]

Manoj Iyer <manjo@austin.ibm.com> wrote:
> When I execute tests under gdb.threads/ directory for instance killed.exp,
> I see 2 complies of the same testcase, the first one gets a linker error
> /usr/bin/ld: cannot find -lpthreads but the second time it works with no
> errors. Anyone has any idea why???

This is normal.

Different systems keep the thread library in different places.
So the test suite tries three different "-l" options until it finds
one that works.

The specific code is in: testsuite/lib/gdb.exp, proc gdb_compile_pthreads.

Michael C