On Wed, Nov 16, 2011 at 12:57 PM, Stefan Küng <tortoise...@gmail.com> wrote:
> On 16.11.2011 18:56, Paul Burba wrote:
>>
>> On Wed, Nov 16, 2011 at 12:24 PM, Stefan Küng<tortoise...@gmail.com>
>> wrote:
>>>
>>> Hi,
>>>
>>> A few crash reports for TSVN show a problem in libsvn_client\merge.c,
>>> function merge_props_changed:
>>>
>>> if (strcmp(prop->name, SVN_PROP_MERGEINFO) == 0)
>>> {
>>> /* Does LOCAL_ABSPATH have any pristine mergeinfo? */
>>> svn_boolean_t has_pristine_mergeinfo = FALSE;
>>> apr_hash_t *pristine_props;
>>>
>>> SVN_ERR(svn_wc_get_pristine_props(&pristine_props,
>>> ctx->wc_ctx,
>>> local_abspath,
>>> scratch_pool,
>>> scratch_pool));
>>>
>>> if (apr_hash_get(pristine_props, SVN_PROP_MERGEINFO,
>>> APR_HASH_KEY_STRING))
>>> has_pristine_mergeinfo = TRUE;
>>>
>>> The svn_wc_get_pristine_props() sets pristine_props to NULL, which then
>>> results in a segfault in apr_hash_get().
>>>
>>> svn_wc_get_pristine_props() can set pristine_props to NULL and still
>>> return
>>> SVN_NO_ERROR if status is either incomplete, excluded, server_excluded or
>>> not_present. So it _can_ happen.
>>>
>>> Checked the source of the svn trunk: the issue is still there. Haven't
>>> found
>>> an entry in the issue tracker so I'm reporting this here.
>>
>> Thanks Stefan,
>>
>> Fixed in r1202807and nominated for backport to 1.7.2.
>
> That was fast!
It was just so obviously wrong (and I happen to be working on merge.c
already). The API is pretty clear:
/** Return the set of "pristine" properties for @a local_abspath.
*
* There are node states where properties do not make sense. For these
* cases, NULL will be returned in @a *props.
^^^^^^^^^
...
svn_error_t *
svn_wc_get_pristine_props(apr_hash_t **props,
svn_wc_context_t *wc_ctx,
const char *local_abspath,
apr_pool_t *result_pool,
apr_pool_t *scratch_pool);
Paul
> Thanks!
>
> Stefan
>
> --
> ___
> oo // \\ "De Chelonian Mobile"
> (_,\/ \_/ \ TortoiseSVN
> \ \_/_\_/> The coolest Interface to (Sub)Version Control
> /_/ \_\ http://tortoisesvn.net
>
