hivex: Test failure for Perl, OCaml, Python bindings on sparc
On a Debian/unstable chroot environment on smetana.debian.org ("Linux smetana 2.6.32-5-sparc64-smp #1 SMP Tue Mar 8 01:00:11 UTC 2011 sparc64 GNU/Linux") node_set_value and node_values fail with a "Bad address" error, see below. I am going to make an attempt at debugging this tonight, any advice would be appreciated. Cheers, -Hilko ,---- | (sid)bengen@smetana:~/hivex.git/perl$ make check | make check-TESTS | make[1]: Entering directory `/home/bengen/hivex.git/perl' | make -f Makefile-pl | make[2]: Entering directory `/home/bengen/hivex.git/perl' | make[2]: Leaving directory `/home/bengen/hivex.git/perl' | make[2]: Entering directory `/home/bengen/hivex.git/perl' | PERL_DL_NONLAZY=1 /usr/bin/perl "-MExtUtils::Command::MM" "-e" "test_harness(0, 'blib/lib', 'blib/arch')" t/*.t | t/005-pod.t .............. skipped: Test::Pod 1.00 required for testing POD | t/006-pod-coverage.t ..... skipped: Test::Pod::Coverage 1.00 required for testing POD | t/010-load.t ............. ok | t/020-open.t ............. ok | t/021-close.t ............ ok | t/200-write.t ............ ok | t/210-setvalue.t ......... 1/11 node_set_value: Bad address at t/210-setvalue.t line 46. | # Looks like you planned 11 tests but ran 5. | # Looks like your test exited with 14 just after 5. | t/210-setvalue.t ......... Dubious, test returned 14 (wstat 3584, 0xe00) | Failed 6/11 subtests | t/510-regedit-load.t ..... ok | t/550-regedit-export.t ... 1/8 node_values: Bad address at /home/bengen/hivex.git/perl/blib/lib/Win/Hivex/Regedit.pm line 518. | # Looks like you planned 8 tests but ran 6. | # Looks like your test exited with 14 just after 6. | t/550-regedit-export.t ... Dubious, test returned 14 (wstat 3584, 0xe00) | Failed 2/8 subtests | t/560-regedit-import.t ... 1/16 node_values: Bad address at /home/bengen/hivex.git/perl/blib/lib/Win/Hivex/Regedit.pm line 518. | # Looks like you planned 16 tests but ran 2. | # Looks like your test exited with 14 just after 2. | t/560-regedit-import.t ... Dubious, test returned 14 (wstat 3584, 0xe00) | Failed 14/16 subtests | t/570-regedit-import2.t .. ok | | Test Summary Report | ------------------- | t/210-setvalue.t (Wstat: 3584 Tests: 5 Failed: 0) | Non-zero exit status: 14 | Parse errors: Bad plan. You planned 11 tests but ran 5. | t/550-regedit-export.t (Wstat: 3584 Tests: 6 Failed: 0) | Non-zero exit status: 14 | Parse errors: Bad plan. You planned 8 tests but ran 6. | t/560-regedit-import.t (Wstat: 3584 Tests: 2 Failed: 0) | Non-zero exit status: 14 | Parse errors: Bad plan. You planned 16 tests but ran 2. | Files=11, Tests=30, 1 wallclock secs ( 0.18 usr 0.04 sys + 1.44 cusr 0.16 csys = 1.82 CPU) | Result: FAIL | Failed 3/11 test programs. 0/30 subtests failed. | make[2]: *** [test_dynamic] Error 255 | make[2]: Leaving directory `/home/bengen/hivex.git/perl' | FAIL: run-perl-tests | ================== | 1 of 1 test failed | ================== | make[1]: *** [check-TESTS] Error 1 | make[1]: Leaving directory `/home/bengen/hivex.git/perl' | make: *** [check-am] Error 2 `---- ,---- | (sid)bengen@smetana:~/hivex.git/python$ make check | make check-TESTS | make[1]: Entering directory `/home/bengen/hivex.git/python' | 010-import.py | 020-open.py | 021-close.py | 200-write.py | 210-setvalue.py | Traceback (most recent call last): | File "t/210-setvalue.py", line 44, in <module> | h.node_set_value (b, value1) | File "/home/bengen/hivex.git/python/hivex.py", line 130, in node_set_value | return libhivexmod.node_set_value (self._o, node, val) | RuntimeError: Bad address | FAIL: run-python-tests | ================== | 1 of 1 test failed | ================== | make[1]: *** [check-TESTS] Error 1 | make[1]: Leaving directory `/home/bengen/hivex.git/python' | make: *** [check-am] Error 2 `---- ,---- | (sid)bengen@smetana:~/hivex.git/ocaml$ make check | make check-TESTS | make[1]: Entering directory `/home/bengen/hivex.git/ocaml' | PASS: t/hivex_005_load | PASS: t/hivex_010_open | PASS: t/hivex_020_root | 01 non-existent file | 02 closed handle | 03 write to read-only file | 04 node_get_child node not found | PASS: t/hivex_100_errors | PASS: t/hivex_110_gc_handle | PASS: t/hivex_200_write | Fatal error: exception Hivex.Error("node_values", 9, "Bad address") | FAIL: t/hivex_300_fold | =================== | 1 of 7 tests failed | =================== | make[1]: *** [check-TESTS] Error 1 | make[1]: Leaving directory `/home/bengen/hivex.git/ocaml' | make: *** [check-am] Error 2 `----
On Wed, May 11, 2011 at 09:37:45AM +0200, Hilko Bengen wrote:
On a Debian/unstable chroot environment on smetana.debian.org ("Linux smetana 2.6.32-5-sparc64-smp #1 SMP Tue Mar 8 01:00:11 UTC 2011 sparc64 GNU/Linux") node_set_value and node_values fail with a "Bad address" error, see below.
I am going to make an attempt at debugging this tonight, any advice would be appreciated.
What is the C integer model of sparc64? I notice that we're storing the size_t (hive_value_t) values in Perl integers using newSViv. That's going to be a problem if Perl int is 32 bit, because those are 64 bit pointers. If that is the problem, the fix would be to replace newSViv with my_newSVll. Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones libguestfs lets you edit virtual machines. Supports shell scripting, bindings from many languages. http://libguestfs.org
* Richard W.M. Jones:
On Wed, May 11, 2011 at 09:37:45AM +0200, Hilko Bengen wrote:
I am going to make an attempt at debugging this tonight, any advice would be appreciated.
What is the C integer model of sparc64?
Apparently, only the kernel is 64bit, so posting the uname string was misleading. The generated binaries are 32bit binaries, this is also what gcc generates by default. (-m64 is supported by gcc.) ,---- | (sid)bengen@smetana:~/hivex.git$ file lib/.libs/libhivex.so.0.0.0 sh/.libs/hivexsh | lib/.libs/libhivex.so.0.0.0: ELF 32-bit MSB shared object, SPARC32PLUS, V8+ Required, version 1 (SYSV), dynamically linked, not stripped | sh/.libs/hivexsh: ELF 32-bit MSB executable, SPARC32PLUS, V8+ Required, version 1 (SYSV), dynamically linked (uses shared libs), for GNU/Linux 2.6.26, not stripped `----
I notice that we're storing the size_t (hive_value_t) values in Perl integers using newSViv.
Both size_t and int are 32 bit values. An endianess issue, maybe? Cheers, -Hilko
On Wed, May 11, 2011 at 05:41:38PM +0200, Hilko Bengen wrote:
* Richard W.M. Jones:
On Wed, May 11, 2011 at 09:37:45AM +0200, Hilko Bengen wrote:
I am going to make an attempt at debugging this tonight, any advice would be appreciated.
What is the C integer model of sparc64?
Apparently, only the kernel is 64bit, so posting the uname string was misleading. The generated binaries are 32bit binaries, this is also what gcc generates by default. (-m64 is supported by gcc.)
,---- | (sid)bengen@smetana:~/hivex.git$ file lib/.libs/libhivex.so.0.0.0 sh/.libs/hivexsh | lib/.libs/libhivex.so.0.0.0: ELF 32-bit MSB shared object, SPARC32PLUS, V8+ Required, version 1 (SYSV), dynamically linked, not stripped | sh/.libs/hivexsh: ELF 32-bit MSB executable, SPARC32PLUS, V8+ Required, version 1 (SYSV), dynamically linked (uses shared libs), for GNU/Linux 2.6.26, not stripped `----
I notice that we're storing the size_t (hive_value_t) values in Perl integers using newSViv.
Both size_t and int are 32 bit values. An endianess issue, maybe?
I guess it might be. We're supposed to be doing le32toh / be32toh everywhere as appropriate, but we might be missing one. The code is mainly tested on little endian arches. Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones virt-top is 'top' for virtual machines. Tiny program with many powerful monitoring features, net stats, disk stats, logging, etc. http://et.redhat.com/~rjones/virt-top
participants (2)
-
Hilko Bengen -
Richard W.M. Jones