You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Philip Martin <ph...@wandisco.com> on 2010/02/03 16:14:26 UTC
patch regression tests failures on trunk
I'm seeing patch regression tests failures on trunk:
CMD: svn patch /home/pm/sw/subversion/obj/subversion/tests/cmdline/svn-test-work/local_tmp/tmp3Grvg3 svn-test-work/working_copies/patch_tests-9 --dry-run --config-dir /home/pm/sw/subversion/obj/subversion/tests/cmdline/svn-test-work/local_tmp/config --password rayjandom --no-auth-cache --username jrandom
<TIME = 0.085810>
U svn-test-work/working_copies/patch_tests-9/A/mu
> applied hunk @@ -1,6 +1,7 @@ with fuzz 1
> applied hunk @@ -7,6 +8,7 @@ with fuzz 2
> applied hunk @@ -19,6 +20,7 @@ with offset 1 and fuzz 140733193388034
CMD: svn patch /home/pm/sw/subversion/obj/subversion/tests/cmdline/svn-test-work/local_tmp/tmp3Grvg3 svn-test-work/working_copies/patch_tests-9 --config-dir /home/pm/sw/subversion/obj/subversion/tests/cmdline/svn-test-work/local_tmp/config --password rayjandom --no-auth-cache --username jrandom
<TIME = 0.055382>
U svn-test-work/working_copies/patch_tests-9/A/mu
> applied hunk @@ -1,6 +1,7 @@ with fuzz 1
> applied hunk @@ -7,6 +8,7 @@ with fuzz 2
> applied hunk @@ -19,6 +20,7 @@ with offset 1 and fuzz 2
=============================================================
Outputs differ
'svn patch --dry-run' output:
U svn-test-work/working_copies/patch_tests-9/A/mu
> applied hunk @@ -1,6 +1,7 @@ with fuzz 1
> applied hunk @@ -7,6 +8,7 @@ with fuzz 2
> applied hunk @@ -19,6 +20,7 @@ with offset 1 and fuzz 140733193388034
'svn patch' output:
U svn-test-work/working_copies/patch_tests-9/A/mu
> applied hunk @@ -1,6 +1,7 @@ with fuzz 1
> applied hunk @@ -7,6 +8,7 @@ with fuzz 2
> applied hunk @@ -19,6 +20,7 @@ with offset 1 and fuzz 2
=============================================================
Note the silly fuzz values. Valgrind shows:
$ valgrind -q ../../../subversion/svn/.libs/lt-svn patch x.patch svn-test-work/working_copies/patch_tests-9
U svn-test-work/working_copies/patch_tests-9/A/mu
==391== Conditional jump or move depends on uninitialised value(s)
==391== at 0x5DFA863: conv_10_quad (in /usr/lib/libapr-1.so.0.2.12)
==391== by 0x5DFB90F: apr_vformatter (in /usr/lib/libapr-1.so.0.2.12)
==391== by 0x5E08316: apr_pvsprintf (in /usr/lib/libapr-1.so.0.2.12)
==391== by 0x5978D23: svn_cmdline_printf (cmdline.c:275)
==391== by 0x41706D: notify (notify.c:364)
==391== by 0x4E798A6: maybe_send_patch_notification (patch.c:1023)
==391== by 0x4E7A6DE: apply_one_patch (patch.c:1308)
==391== by 0x4E7A90C: apply_patches (patch.c:1372)
==391== by 0x50D67F9: svn_wc__call_with_write_lock (lock.c:1863)
==391== by 0x4E7AA5B: svn_client_patch (patch.c:1403)
==391== by 0x418B1E: svn_cl__patch (patch-cmd.c:79)
==391== by 0x414A11: main (main.c:2234)
> applied hunk @@ -1,6 +1,7 @@ with fuzz 1
> applied hunk @@ -7,6 +8,7 @@ with fuzz 2
> applied hunk @@ -19,6 +20,7 @@ with offset 1 and fuzz 2
With --dry-run there are also errors at notify.c:334:
==540== Conditional jump or move depends on uninitialised value(s)
==540== at 0x5DFA94A: conv_10_quad (in /usr/lib/libapr-1.so.0.2.12)
==540== by 0x5DFB90F: apr_vformatter (in /usr/lib/libapr-1.so.0.2.12)
==540== by 0x5E08316: apr_pvsprintf (in /usr/lib/libapr-1.so.0.2.12)
==540== by 0x5978D23: svn_cmdline_printf (cmdline.c:275)
==540== by 0x416F36: notify (notify.c:334)
==540== by 0x4E798A6: maybe_send_patch_notification (patch.c:1023)
==540== by 0x4E7A6DE: apply_one_patch (patch.c:1308)
==540== by 0x4E7A90C: apply_patches (patch.c:1372)
==540== by 0x50D67F9: svn_wc__call_with_write_lock (lock.c:1863)
==540== by 0x4E7AA5B: svn_client_patch (patch.c:1403)
==540== by 0x418B1E: svn_cl__patch (patch-cmd.c:79)
==540== by 0x414A11: main (main.c:2234)
--
Philip
Re: patch regression tests failures on trunk
Posted by Philip Martin <ph...@wandisco.com>.
Stefan Sperling <st...@elego.de> writes:
> Does this fix it?
>
> Index: subversion/svn/notify.c
> ===================================================================
> --- subversion/svn/notify.c (revision 906159)
> +++ subversion/svn/notify.c (working copy)
> @@ -334,7 +334,7 @@ notify(void *baton, const svn_wc_notify_t *n, apr_
> if ((err = svn_cmdline_printf(pool,
> apr_pstrcat(pool, s,
> "%"APR_UINT64_T_FMT
> - " and fuzz %lu\n",
> + " and fuzz %d\n",
> NULL),
> n->hunk_original_start,
> n->hunk_original_length,
> @@ -363,7 +363,7 @@ notify(void *baton, const svn_wc_notify_t *n, apr_
> {
> if ((err = svn_cmdline_printf(pool,
> _("> applied hunk @@ -%lu,%lu +%lu,%lu @@ "
> - "with fuzz %lu\n"),
> + "with fuzz %d\n"),
> n->hunk_original_start,
> n->hunk_original_length,
> n->hunk_modified_start,
Yes.
--
Philip
Re: patch regression tests failures on trunk
Posted by Stefan Sperling <st...@elego.de>.
On Wed, Feb 03, 2010 at 04:27:39PM +0000, Philip Martin wrote:
> Philip Martin <ph...@wandisco.com> writes:
>
> > Note the silly fuzz values. Valgrind shows:
>
> I just got r906107, the problem is still there but the call stack is
> slightly different:
>
> $ valgrind -q ../../../subversion/svn/.libs/lt-svn patch x.patch svn-test-work/working_copies/patch_tests-9
> U svn-test-work/working_copies/patch_tests-9/A/mu
> ==4268== Conditional jump or move depends on uninitialised value(s)
> ==4268== at 0x5E03BF3: conv_10_quad (apr_snprintf.c:395)
> ==4268== by 0x5E04C9F: apr_vformatter (apr_snprintf.c:862)
> ==4268== by 0x5E119CB: apr_pvsprintf (apr_pools.c:1682)
> ==4268== by 0x5978D23: svn_cmdline_printf (cmdline.c:275)
> ==4268== by 0x41706D: notify (notify.c:364)
> ==4268== by 0x4E79891: send_patch_notification (patch.c:1022)
> ==4268== by 0x4E7AA55: apply_patches (patch.c:1420)
> ==4268== by 0x50D67F9: svn_wc__call_with_write_lock (lock.c:1863)
> ==4268== by 0x4E7ABBA: svn_client_patch (patch.c:1449)
> ==4268== by 0x418B1E: svn_cl__patch (patch-cmd.c:79)
> ==4268== by 0x414A11: main (main.c:2234)
> > applied hunk @@ -1,6 +1,7 @@ with fuzz 1
> > applied hunk @@ -7,6 +8,7 @@ with fuzz 2
> > applied hunk @@ -19,6 +20,7 @@ with offset 1 and fuzz 2
Does this fix it?
Thanks,
Stefan
Index: subversion/svn/notify.c
===================================================================
--- subversion/svn/notify.c (revision 906159)
+++ subversion/svn/notify.c (working copy)
@@ -334,7 +334,7 @@ notify(void *baton, const svn_wc_notify_t *n, apr_
if ((err = svn_cmdline_printf(pool,
apr_pstrcat(pool, s,
"%"APR_UINT64_T_FMT
- " and fuzz %lu\n",
+ " and fuzz %d\n",
NULL),
n->hunk_original_start,
n->hunk_original_length,
@@ -363,7 +363,7 @@ notify(void *baton, const svn_wc_notify_t *n, apr_
{
if ((err = svn_cmdline_printf(pool,
_("> applied hunk @@ -%lu,%lu +%lu,%lu @@ "
- "with fuzz %lu\n"),
+ "with fuzz %d\n"),
n->hunk_original_start,
n->hunk_original_length,
n->hunk_modified_start,
Re: patch regression tests failures on trunk
Posted by Philip Martin <ph...@wandisco.com>.
Philip Martin <ph...@wandisco.com> writes:
> Note the silly fuzz values. Valgrind shows:
I just got r906107, the problem is still there but the call stack is
slightly different:
$ valgrind -q ../../../subversion/svn/.libs/lt-svn patch x.patch svn-test-work/working_copies/patch_tests-9
U svn-test-work/working_copies/patch_tests-9/A/mu
==4268== Conditional jump or move depends on uninitialised value(s)
==4268== at 0x5E03BF3: conv_10_quad (apr_snprintf.c:395)
==4268== by 0x5E04C9F: apr_vformatter (apr_snprintf.c:862)
==4268== by 0x5E119CB: apr_pvsprintf (apr_pools.c:1682)
==4268== by 0x5978D23: svn_cmdline_printf (cmdline.c:275)
==4268== by 0x41706D: notify (notify.c:364)
==4268== by 0x4E79891: send_patch_notification (patch.c:1022)
==4268== by 0x4E7AA55: apply_patches (patch.c:1420)
==4268== by 0x50D67F9: svn_wc__call_with_write_lock (lock.c:1863)
==4268== by 0x4E7ABBA: svn_client_patch (patch.c:1449)
==4268== by 0x418B1E: svn_cl__patch (patch-cmd.c:79)
==4268== by 0x414A11: main (main.c:2234)
> applied hunk @@ -1,6 +1,7 @@ with fuzz 1
> applied hunk @@ -7,6 +8,7 @@ with fuzz 2
> applied hunk @@ -19,6 +20,7 @@ with offset 1 and fuzz 2
--
Philip