Re: Test failure in libgcj on mainline

Re: Test failure in libgcj on mainline

[John David Anglin]

> Running /home/aph/gcc/gcc/libjava/testsuite/libjava.cni/cni.exp ...
> ERROR: tcl error sourcing /home/aph/gcc/gcc/libjava/testsuite/libjava.cni/cni.exp.
> ERROR: can't unset "env(SHLIB_PATH)": no such element in array

Sorry, this is caused by my recent LD_LIBRARY_PATH.  This didn't appear
in my testing.  There seems to be some difference in behavior of unsetenv
from one system to another.

The enclosed patch should fix the problem.  Tested on i686-pc-linux-gnu.
I will install as obvious.

Dave
-- 
J. David Anglin                                  dave.anglin@nrc-cnrc.gc.ca
National Research Council of Canada              (613) 990-0752 (FAX: 952-6602)

2004-11-26  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>

	* lib/target-libpath.exp (restore_ld_library_path_env_vars):
	Check existence of ld library path variables before unsetting.

Index: lib/target-libpath.exp
===================================================================
RCS file: /cvs/gcc/gcc/gcc/testsuite/lib/target-libpath.exp,v
retrieving revision 1.2
diff -u -3 -p -r1.2 target-libpath.exp
--- lib/target-libpath.exp	25 Nov 2004 04:27:13 -0000	1.2
+++ lib/target-libpath.exp	27 Nov 2004 04:38:08 -0000
@@ -196,42 +196,42 @@ proc restore_ld_library_path_env_vars { 
 
   if { $orig_ld_library_path_saved } {
     setenv LD_LIBRARY_PATH "$orig_ld_library_path"
-  } else {
+  } elseif [info exists env(LD_LIBRARY_PATH)] {
     unsetenv LD_LIBRARY_PATH
   }
   if { $orig_ld_run_path_saved } {
     setenv LD_RUN_PATH "$orig_ld_run_path"
-  } else {
+  } elseif [info exists env(LD_RUN_PATH)] {
     unsetenv LD_RUN_PATH
   }
   if { $orig_shlib_path_saved } {
     setenv SHLIB_PATH "$orig_shlib_path"
-  } else {
+  } elseif [info exists env(SHLIB_PATH)] {
     unsetenv SHLIB_PATH
   }
   if { $orig_ld_libraryn32_path_saved } {
     setenv LD_LIBRARYN32_PATH "$orig_ld_libraryn32_path"
-  } else {
+  } elseif [info exists env(LD_LIBRARYN32_PATH)] {
     unsetenv LD_LIBRARYN32_PATH
   }
   if { $orig_ld_library64_path_saved } {
     setenv LD_LIBRARY64_PATH "$orig_ld_library64_path"
-  } else {
+  } elseif [info exists env(LD_LIBRARY64_PATH)] {
     unsetenv LD_LIBRARY64_PATH
   }
   if { $orig_ld_library_path_32_saved } {
     setenv LD_LIBRARY_PATH_32 "$orig_ld_library_path_32"
-  } else {
+  } elseif [info exists env(LD_LIBRARY_PATH_32)] {
     unsetenv LD_LIBRARY_PATH_32
   }
   if { $orig_ld_library_path_64_saved } {
     setenv LD_LIBRARY_PATH_64 "$orig_ld_library_path_64"
-  } else {
+  } elseif [info exists env(LD_LIBRARY_PATH_64)] {
     unsetenv LD_LIBRARY_PATH_64
   }
   if { $orig_dyld_library_path_saved } {
     setenv DYLD_LIBRARY_PATH "$orig_dyld_library_path"
-  } else {
+  } elseif [info exists env(DYLD_LIBRARY_PATH)] {
     unsetenv DYLD_LIBRARY_PATH
   }
 }

Re: Test failure in libgcj on mainline

[Ranjit Mathew]

On Fri, 26 Nov 2004 23:55:05 -0500 (EST), John David Anglin
<dave@hiauly1.hia.nrc.ca> wrote:
> > Running /home/aph/gcc/gcc/libjava/testsuite/libjava.cni/cni.exp ...
> > ERROR: tcl error sourcing /home/aph/gcc/gcc/libjava/testsuite/libjava.cni/cni.exp.
> > ERROR: can't unset "env(SHLIB_PATH)": no such element in array
> 
> Sorry, this is caused by my recent LD_LIBRARY_PATH.  This didn't appear
> in my testing.  There seems to be some difference in behavior of unsetenv
> from one system to another.
> 
> The enclosed patch should fix the problem.  Tested on i686-pc-linux-gnu.
> I will install as obvious.

Thanks John. However, I now see:

Running /extra/src/gcc/gcc-20041127/libjava/testsuite/libjava.cni/cni.exp ...
ERROR: (DejaGnu) proc "set_ld_library_path_env_vars" does not exist.
The error code is NONE
The info on the error is:
close: spawn id exp6 not open
    while executing
"close -i exp6"
    invoked from within
"catch "close -i $spawn_id""

Ranjit

Re: Test failure in libgcj on mainline

[John David Anglin]

> Thanks John. However, I now see:
> 
> Running /extra/src/gcc/gcc-20041127/libjava/testsuite/libjava.cni/cni.exp
> ...
> ERROR: (DejaGnu) proc "set_ld_library_path_env_vars" does not exist.

I'm puzzled by this one.  I didn't see it in my last build on i686-linux
or hppa-linux:

  <http://gcc.gnu.org/ml/gcc-testresults/2004-11/msg01204.html>
  <http://gcc.gnu.org/ml/gcc-testresults/2004-11/msg01205.html>.

The proc "set_ld_library_path_env_vars" is in
gcc/gcc/testsuite/lib/target-libpath.exp.  This file is loaded by
libjava.exp.  This is done using the same proc as other libraries
use to load files in the gcc testsuite directory.  This depends
on the setting of srcdir in the libjava testsuite Makefile.

Dave
-- 
J. David Anglin                                  dave.anglin@nrc-cnrc.gc.ca
National Research Council of Canada              (613) 990-0752 (FAX: 952-6602)

Re: Test failure in libgcj on mainline

[Ranjit Mathew]

On Sat, 27 Nov 2004 13:41:23 -0500 (EST), John David Anglin
<dave@hiauly1.hia.nrc.ca> wrote:
> > Thanks John. However, I now see:
> >
> > Running /extra/src/gcc/gcc-20041127/libjava/testsuite/libjava.cni/cni.exp
> > ...
> > ERROR: (DejaGnu) proc "set_ld_library_path_env_vars" does not exist.
> 
> I'm puzzled by this one.  I didn't see it in my last build on i686-linux
> or hppa-linux:
> 
>   <http://gcc.gnu.org/ml/gcc-testresults/2004-11/msg01204.html>
>   <http://gcc.gnu.org/ml/gcc-testresults/2004-11/msg01205.html>.
> 
> The proc "set_ld_library_path_env_vars" is in
> gcc/gcc/testsuite/lib/target-libpath.exp.  This file is loaded by
> libjava.exp.  This is done using the same proc as other libraries
> use to load files in the gcc testsuite directory.  This depends
> on the setting of srcdir in the libjava testsuite Makefile.

Today (2004-11-29, IST) I tried this on another machine
(i686-pc-linux-gnu) and did not see this problem.  So I
guess it was a temporary disturbance in the force. :-/

Thanks,
Ranjit.

-- 
Ranjit Mathew          Email: rmathew AT gmail DOT com

Bangalore, INDIA.      Web: http://ranjitmathew.tripod.com/

Re: Test failure in libgcj on mainline

[Ranjit Mathew]

On Sat, 27 Nov 2004 13:41:23 -0500 (EST), John David Anglin
<dave@hiauly1.hia.nrc.ca> wrote:
> > Thanks John. However, I now see:
> >
> > Running /extra/src/gcc/gcc-20041127/libjava/testsuite/libjava.cni/cni.exp
> > ...
> > ERROR: (DejaGnu) proc "set_ld_library_path_env_vars" does not exist.
> 
> I'm puzzled by this one.  I didn't see it in my last build on i686-linux
> or hppa-linux:
> 
>   <http://gcc.gnu.org/ml/gcc-testresults/2004-11/msg01204.html>
>   <http://gcc.gnu.org/ml/gcc-testresults/2004-11/msg01205.html>.
> 
> The proc "set_ld_library_path_env_vars" is in
> gcc/gcc/testsuite/lib/target-libpath.exp.  This file is loaded by
> libjava.exp.  This is done using the same proc as other libraries
> use to load files in the gcc testsuite directory.  This depends
> on the setting of srcdir in the libjava testsuite Makefile.

This turned out to be my bad - due to disc space constraints on my home 
PC, I normally create a snapshot of the GCC CVS sources containing
just enough files to help me compile c,c++,java on i686-pc-linux-gnu and
run the Java testsuite.

Till now, I used to get away with not copying "gcc/testsuite" as I was
only interested in the libjava testsuite. I have now modified the script
to copy over the "lib" and "config" folders of this folder and everything
is back on track.

Sorry for the noise,
Ranjit.

Test failure in libgcj on mainline

[Andrew Haley]

This seems to be because SHLIB_PATH isn't being set.  I can trivially
fix this with the patch at the end of his mail, but I don't know that
is the right fix.


Running /home/aph/gcc/gcc/libjava/testsuite/libjava.cni/cni.exp ...
ERROR: tcl error sourcing /home/aph/gcc/gcc/libjava/testsuite/libjava.cni/cni.exp.
ERROR: can't unset "env(SHLIB_PATH)": no such element in array
    while executing
"unset env($var)"
    (procedure "unsetenv" line 3)
    invoked from within
"unsetenv SHLIB_PATH"
    (procedure "restore_ld_library_path_env_vars" line 37)
    invoked from within
"restore_ld_library_path_env_vars"
    (procedure "gcj_invoke" line 20)
    invoked from within
"gcj_invoke $main [file rootname $file].out {}"
    (procedure "gcj_cni_test_one" line 35)
    invoked from within
"gcj_cni_test_one $x"
    (procedure "gcj_cni_run" line 10)
    invoked from within
"gcj_cni_run
"
    (file "/home/aph/gcc/gcc/libjava/testsuite/libjava.cni/cni.exp" line 119)
    invoked from within
"source /home/aph/gcc/gcc/libjava/testsuite/libjava.cni/cni.exp"
    ("uplevel" body line 1)
    invoked from within
"uplevel #0 source /home/aph/gcc/gcc/libjava/testsuite/libjava.cni/cni.exp"
    invoked from within
"catch "uplevel #0 source $test_file_name""


Andrew.



Index: target-libpath.exp
===================================================================
RCS file: /cvs/gcc/gcc/gcc/testsuite/lib/target-libpath.exp,v
retrieving revision 1.2
diff -u -r1.2 target-libpath.exp
--- target-libpath.exp	25 Nov 2004 04:27:13 -0000	1.2
+++ target-libpath.exp	25 Nov 2004 12:39:30 -0000
@@ -207,7 +207,7 @@
   if { $orig_shlib_path_saved } {
     setenv SHLIB_PATH "$orig_shlib_path"
   } else {
-    unsetenv SHLIB_PATH
+      catch { unsetenv SHLIB_PATH }
   }
   if { $orig_ld_libraryn32_path_saved } {
     setenv LD_LIBRARYN32_PATH "$orig_ld_libraryn32_path"

Re: Test failure in libgcj on mainline

[Ranjit Mathew]

Andrew Haley wrote:
> This seems to be because SHLIB_PATH isn't being set.  I can trivially
> fix this with the patch at the end of his mail, but I don't know that
> is the right fix.
> 
> 
> Running /home/aph/gcc/gcc/libjava/testsuite/libjava.cni/cni.exp ...
> ERROR: tcl error sourcing /home/aph/gcc/gcc/libjava/testsuite/libjava.cni/cni.exp.
> ERROR: can't unset "env(SHLIB_PATH)": no such element in array

FWIW, I get a similar failure for LD_LIBRARY_PATH (i686-pc-linux-gnu).

With included Jacks, I get the following from today's build:

                  === libjava Summary ===

  # of expected passes            7785
  # of unexpected failures        2
  # of unexpected successes       9
  # of expected failures          730
  # of untested testcases         12

Details:

  XPASS: 15.28-string-13
  FAIL: 15.21-equal-3
  XPASS: 15.15-precedence-5
  FAIL: 9.2-implicit-4
  XPASS: non-jls-argument-expansion-tokens-2
  XPASS: non-jls-argument-expansion-tokens-4
  XPASS: non-jls-argument-expansion-tokens-5
  XPASS: 8.4.6.4-multiple-runtime-1
  XPASS: 15.18.1.1-runtime-5
  XPASS: pr13107_3 compilation from bytecode
  XPASS: pr13107_3 -O3 compilation from bytecode

Ranjit.

-- 
Ranjit Mathew          Email: rmathew AT gmail DOT com

Bangalore, INDIA.      Web: http://ranjitmathew.tripod.com/