Re: [RFA/testsuite]: Stop dbx.exp insanity

[Fernando Nasser <fnasser@redhat.com>]

Daniel Jacobowitz wrote:
> 
> There were two issues with dbx.exp which could cause hard-to-identify
> testsuite failures:
> 
>   - It reimplemented gdb_file_cmd and left it changed.  This means that the
>     behavior of tests varied based on whether they ran before or after
>     dbx.exp.
>   - The reimplemented gdb_file_cmd was buggy in and of itself.
> 
> The attached patch has been verified to fix both things.  I restore the old
> gdb_file_cmd after dbx.exp is done; and for kicks I ran the testsuite
> without the line that restores it and verified that the other bugs were
> gone.  The problem was forgetting to call exec-file in some cases.
> 
> Fernando, is this OK to commit?
> 

Cool!  Yes, thank you.

Fernando


> Two examples in particular.  This is the one inexplicable failure caused by
> my stabs line-numbering patch:
> 
> Running ../../../src/gdb/testsuite/gdb.base/dbx.exp ...
> Running ../../../src/gdb/testsuite/gdb.base/reread.exp ...
> FAIL: gdb.base/reread.exp: second pass: breakpoint foo in first file
> 
> # of unexpected failures        1
> 
> vs.
> 
> Running ../../../src/gdb/testsuite/gdb.base/reread.exp ...
> 
> (no failures)
> 
> Here's one in dump.exp:
> 
> Running ../../../src/gdb/testsuite/gdb.base/dbx.exp ...
> Running ../../../src/gdb/testsuite/gdb.base/dump.exp ...
> FAIL: gdb.base/dump.exp: reload array as value, srec
> FAIL: gdb.base/dump.exp: reload struct as value, srec
> FAIL: gdb.base/dump.exp: reload array as memory, srec
> FAIL: gdb.base/dump.exp: reload struct as memory, srec
> FAIL: gdb.base/dump.exp: reload array as value, intel hex
> FAIL: gdb.base/dump.exp: reload struct as value, intel hex
> FAIL: gdb.base/dump.exp: reload array as memory, intel hex
> FAIL: gdb.base/dump.exp: reload struct as memory, intel hex
> FAIL: gdb.base/dump.exp: reload array as value, tekhex
> FAIL: gdb.base/dump.exp: reload struct as value, tekhex
> FAIL: gdb.base/dump.exp: reload array as memory, tekhex
> FAIL: gdb.base/dump.exp: reload struct as memory, tekhex
> 
> vs.
> 
> Running ../../../src/gdb/testsuite/gdb.base/dump.exp ...
> 
> (no failures)
> 
> --
> Daniel Jacobowitz                           Carnegie Mellon University
> MontaVista Software                         Debian GNU/Linux Developer
> 
> 2002-03-30  Daniel Jacobowitz  <drow@mvista.com>
> 
>         * gdb.base/dbx.exp: Restore old definition of gdb_file_cmd
>         when finished.  Make gdb_file_cmd send "exec-file" when
>         appropriate.
> 
> Index: dbx.exp
> ===================================================================
> RCS file: /cvs/src/src/gdb/testsuite/gdb.base/dbx.exp,v
> retrieving revision 1.3
> diff -u -r1.3 dbx.exp
> --- dbx.exp     2001/05/23 19:04:13     1.3
> +++ dbx.exp     2002/03/30 19:27:00
> @@ -168,6 +168,11 @@
>  # file into gdb for a dbx session. So why not just override gdb_file_cmd with the
>  # right sequence of events, allowing gdb_load to do its normal thing? This way
>  # remotes and simulators will work, too.
> +#
> +# [drow 2002-03-30]: We can restore the old gdb_file_cmd afterwards, though.
> +set old_gdb_file_cmd_args [info args gdb_file_cmd]
> +set old_gdb_file_cmd_body [info body gdb_file_cmd]
> +
>  proc gdb_file_cmd {arg} {
>      global verbose
>      global loadpath
> @@ -206,6 +211,11 @@
>              verbose "\t\tLoaded $arg into the $GDB"
>              send_gdb "exec-file $arg\n"
>              gdb_expect {
> +               -re "A program is being debugged already.*Kill it.*y or n. $" {
> +                   send_gdb "y\n"
> +                       verbose "\t\tKilling previous program being debugged"
> +                   exp_continue
> +               }
>                  -re ".*$gdb_prompt $" {
>                      verbose "\t\tLoaded $arg with new symbol table into $GDB"
>                      return 0
> @@ -221,23 +231,9 @@
>              perror "$arg wasn't compiled with \"-g\""
>              return -1
>          }
> -        -re "A program is being debugged already.*Kill it.*y or n. $" {
> -            send_gdb "y\n"
> -                verbose "\t\tKilling previous program being debugged"
> -            exp_continue
> -        }
>          -re "Load new symbol table from \".*\".*y or n. $" {
>              send_gdb "y\n"
> -            gdb_expect {
> -                -re "Reading symbols from.*done.*$gdb_prompt $" {
> -                    verbose "\t\tLoaded $arg with new symbol table into $GDB"
> -                    return 0
> -                }
> -                timeout {
> -                    perror "(timeout) Couldn't load $arg, other program already loaded."
> -                    return -1
> -                }
> -            }
> +           exp_continue
>          }
>          -re ".*No such file or directory.*$gdb_prompt $" {
>              perror "($arg) No such file or directory\n"
> @@ -339,4 +335,6 @@
>  gdb_exit
> 
>  set GDBFLAGS $saved_gdbflags
> +eval proc gdb_file_cmd {$old_gdb_file_cmd_args} {$old_gdb_file_cmd_body}
> +
>  return 0

-- 
Fernando Nasser
Red Hat Canada Ltd.                     E-Mail:  fnasser@redhat.com
2323 Yonge Street, Suite #300
Toronto, Ontario   M4P 2C9