On Mon, Aug 5, 2013 at 8:09 PM, Stefan Küng <tortoise...@gmail.com> wrote:
> On Sat, Jul 27, 2013 at 5:24 AM, Ben Reser <b...@reser.org> wrote: > >> On Fri, Jul 26, 2013 at 8:28 AM, Stefan Küng <tortoise...@gmail.com> >> wrote: >> > crash reports for 1.8.1 are coming in now. Top crash right now is this: >> > https://www.crash-server.com/Problem.aspx?ClientID=tsvn&ProblemID=31426 >> > >> >> libsvn_tsvn.dll!svn_path_join_internal(const char * >> >> base=0xfffffffffcfcc6f7, const char * component=0x0000000000000000, >> >> apr_pool_t * pool=0x0000000003033880) Line 115 C >> >> libsvn_tsvn.dll!checkout_dir(dir_context_t * dir=0x0000000003017e78, >> >> apr_pool_t * scratch_pool=0x0000000002fce380) Line 403 C >> >> libsvn_tsvn.dll!delete_entry(const char * path=0x000007fedc1f7708, long >> >> revision=50429560, void * parent_baton=0x0000000003033880, apr_pool_t * >> >> pool=0x000007fedc1f7708) Line 1546 C >> >> libsvn_tsvn.dll!do_item_commit(void * * dir_baton=0x0000000003f5ebb8, >> void >> >> * parent_baton=0x0000000003033880, void * >> callback_baton=0x0000000003f5ec70, >> >> const char * path=0x0000000002fcc3a8, apr_pool_t * >> pool=0x0000000003039838) >> >> Line 1590 C >> >> libsvn_tsvn.dll!svn_delta_path_driver2(const svn_delta_editor_t * >> >> editor=0x000000000301f7e0, void * edit_baton=0x0000000002fce6a0, const >> >> apr_array_header_t * paths=0x0000000003031870, int sort_paths=50438336, >> >> svn_error_t * (void * *, void *, void *, const char *, apr_pool_t *) * >> >> callback_func=0x000007fede5ebbd0, void * >> callback_baton=0x0000000003f5ec70, >> >> apr_pool_t * pool=0x000000000301f768) Line 263 C >> >> libsvn_tsvn.dll!svn_client__do_commit(const char * >> >> base_url=0x0000000002fcc310, const apr_array_header_t * >> >> commit_items=0x000000000302f7e8, const svn_delta_editor_t * >> >> editor=0x0000000002fce6a0, void * edit_baton=0x0000000002fce380, const >> char >> >> * notify_path_prefix=0x0000000003017960, apr_hash_t * * >> >> sha1_checksums=0x0000000003f5eda8, svn_client_ctx_t * >> >> ctx=0x000000000301a0c0, apr_pool_t * result_pool=0x00000000030176f8, >> >> apr_pool_t * scratch_pool=0x000000000301f768) Line 1808 C >> >> libsvn_tsvn.dll!svn_client_commit6(const apr_array_header_t * >> >> targets=0x0000000003017770, svn_depth_t depth=34, int keep_locks=0, int >> >> keep_changelists=0, int commit_as_operations=1, int >> >> include_file_externals=0, int include_dir_externals=0, const >> >> apr_array_header_t * changelists=0x0000000003017770, const apr_hash_t * >> >> revprop_table=0x0000000000000000, svn_error_t * (const >> svn_commit_info_t *, >> >> void *, apr_pool_t *) * commit_callback=0x000000013f3cc770, void * >> >> commit_baton=0x0000000000158de0, svn_client_ctx_t * >> ctx=0x000000000301a0c0, >> >> apr_pool_t * pool=0x00000000030176f8) Line 965 C >> > >> > >> > the dir->parent_dir->working_url in libsvn_ra_serf/commit.c, line 402 is >> > NULL which makes svn_path_join_internal access the null pointer. >> >> There clearly is a bug where dir->parent_dir->working_url is NULL, >> however the backtrace above seems to be showing that the component >> variable passed into svn_path_join_internal is NULL. checkout_dir() >> calls svn_path_url_add_component2() which uses uri_escape() on the >> component portion before calling svn_path_join_internal() with the >> result of uri_escape(). So I'm really not sure how the component is >> ending up NULL, but it appears to be a different issue from what you >> describe. >> > > > There's a small confusion about what the 'component' is I guess: > > > /* Implicitly checkout this dir now. */ > dir->working_url = svn_path_url_add_component2( > dir->parent_dir->working_url, > dir->name, dir->pool); > > It's actually dir->name that's NULL here, because: > > > const char * > svn_path_url_add_component2(const char *url, > const char *component, > apr_pool_t *pool) > { > /* = svn_path_uri_encode() but without always copying */ > > > component = uri_escape(component, svn_uri__char_validity, pool); > > return svn_path_join_internal(url, component, pool); > } > > > which means 'component' (the second param to svn_path_join_internal) is also > the second param in svn_path_url_add_component2. > > > But looking at the code of svn_path_url_add_component2(): > component = uri_escape(component, svn_uri__char_validity, pool); > > > how is that possible without the compiler complaining? 'component' is const, > so how is it possible to assign the return value of uri_escape() to it? > > > Could it be that this strange use of the component param made the compiler do > something wrong and that's why I get this crash? > > Since the debug view didn't show enough information, I used the disassembly view to gather more information: A few things to notice here: * the generated code is correct, even though uri_escape assigns a value to a const param * it's not 'component' that's NULL but 'base' (in svn_path_join_internal) which means the url and one level up dir->parent_dir->working_url, is NULL. That's why uri_escape works without a crash and the crash happens in svn_path_join_internal. Stefan -- ___ oo // \\ "De Chelonian Mobile" (_,\/ \_/ \ TortoiseSVN \ \_/_\_/> The coolest Interface to (Sub)Version Control /_/ \_\ http://tortoisesvn.net
