This is the mail archive of the
gdb-patches@sources.redhat.com
mailing list for the GDB project.
Re: [RFC]: Add test for verifying the fix to the SEGV error when printing F77 array element
- From: Daniel Jacobowitz <drow at false dot org>
- To: Wu Zhou <woodzltc at cn dot ibm dot com>
- Cc: gdb-patches at sources dot redhat dot com
- Date: Sun, 12 Jun 2005 22:55:51 -0400
- Subject: Re: [RFC]: Add test for verifying the fix to the SEGV error when printing F77 array element
- References: <Pine.LNX.4.63.0506101625290.13036@plinuxt18.cn.ibm.com>
On Fri, Jun 10, 2005 at 04:41:02PM +0800, Wu Zhou wrote:
> Daniel and all,
>
> This is the testcase associated with the following patch:
>
> http://sources.redhat.com/ml/gdb-patches/2005-06/msg00090.html
>
> Without the patch above, the last three tests fail. With the patch, all
> tests pass. It is tested on ppc64 with g77-3.3.3 and g77-3.4.3.
>
> It is my first attempt to add a new testcase into GDB testsuite. So if
> there is any errors or missings, please feel free to tell me. Thanks a
> lot!
Thanks. It does need some changes, but not big ones.
I'd appreciated it if you could provide new files as diffs. The normal
way to do this is by diffing them against /dev/null, with the
appropriate relative path to the new file. That way, I can save your
message and apply it as a patch.
Let's use hyphens instead of underscores in file names; there's already
a lot of hyphenated test cases. So array-element.exp.
> array_element.f
> ======================================
We're trying to add copyright notices to all new tests; please add one
here.
> dimension a(10)
> write(*,*)'This is a test.'
> call sub(a,10)
> write(*,*) a
> stop
> end
>
> subroutine sub(a,n)
> dimension a(n)
> do 100 i=1, n
> a(i)=i
> 100 continue
> return
> end
>
> array_element.exp
> =====================================
> if $tracelevel then {
> strace $tracelevel
> }
>
> #
> # test running programs
> #
> set prms_id 0
> set bug_id 0
This bit is not necessary.
> set testfile "array_element"
> set srcfile ${testfile}.f
> set binfile ${objdir}/${subdir}/${testfile}
>
> if { [gdb_compile "${srcdir}/${subdir}/${srcfile}" "${binfile}" executable {debug f77}] != "" } {
> gdb_suppress_entire_file "Testcase compile failed, so all tests in this file will automatically fail."
> }
Please don't use gdb_suppress_entire_file; use "untested" and then
"return". See the Java tests for an example. You can see what will
happen when no Fortran compiler is installed by running
"runtest array-element.exp F77_FOR_TARGET=false"; compare with
"runtest jprint.exp GCJ_UNDER_TEST=false".
You can get the right result by also adding "quiet" to the last
argument of gdb_compile, and using untested/return.
> if [get_compiler_info ${binfile} "f77"] {
> return -1
> }
What's this needed for? Also, get_compiler_info does not take a second
argument.
> gdb_exit
> gdb_start
> gdb_reinitialize_dir $srcdir/$subdir
> gdb_load ${binfile}
>
> if ![runto sub_] then {
> perror "couldn't run to breakpoint"
> continue
> }
>
> set bp_location [gdb_get_line_number "continue"]
> gdb_test "break $bp_location" \
> "Breakpoint.*at.* file .*$srcfile, line $bp_location\\." \
> "brealpoint at continue"
Typo there.
> gdb_test "continue" \
> "Continuing\\..*Breakpoint.*" \
> "continue to breakpoint"
> gdb_test "print a(1)" "= 1.*" "print the first element of array a"
>
> gdb_test "continue" \
> "Continuing\\..*Breakpoint.*" \
> "continue to breakpoint once again"
> gdb_test "print a(2)" "= 2.*" "print the first element of array a"
> gdb_exit
> return 0
This bit isn't actually necessary.
--
Daniel Jacobowitz
CodeSourcery, LLC