GnuCash
Contact   Instructions
Bug 797285 - Free of invalid ptr in webkit1 load_to_stream.
Summary: Free of invalid ptr in webkit1 load_to_stream.
Status: RESOLVED INCOMPLETE
Alias: None
Product: GnuCash
Classification: Unclassified
Component: Import - QIF (show other bugs)
Version: 3.5
Hardware: PC Mac OS
: Normal normal
Target Milestone: ---
Assignee: import
QA Contact: import
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2019-06-21 16:27 EDT by edwardfloehr
Modified: 2021-09-23 17:41 EDT (History)
4 users (show)

See Also:


Attachments
OSX crash info dump (79.29 KB, text/plain)
2019-06-21 16:27 EDT, edwardfloehr
no flags Details

Description edwardfloehr 2019-06-21 16:27:25 EDT
Created attachment 373303 [details]
OSX crash info dump

QIF imports failing to show any imported transactions (this has worked for me for years).  Then gnucash crashes.  See attached dump from MacOSX:
Comment 1 edwardfloehr 2019-06-21 16:30:29 EDT
This might be a possible lead from the dump:

Application Specific Information:
abort() called
*** error for object 0x1f000000000000: pointer being freed was not allocated
Comment 2 John Ralls 2019-06-21 17:13:32 EDT
The stack in the crash report is for displaying a report and it looks sane. Does importing the same QIF reliably crash the same way? Does importing a different QIF work correctly?
Comment 3 edwardfloehr 2019-06-21 19:31:53 EDT
I can't reproduce the crash on demand.  The failed import was operator error, but a bit user-unfriendly - I was missing a space in my account name, so the import silently went into "Unspecified" account.  I'd prefer it alerted me that I was trying to import into a non-existent account.  Thanks.
Comment 4 John Ralls 2019-06-21 20:46:17 EDT
OK. Did the crash actually happen after the import failed or did you then try to run a report and GnuCash crashed then?
Comment 5 John Ralls 2019-11-04 14:27:08 EST
Comment on attachment 373303 [details]
OSX crash info dump

I've worked out from the assembly that the crash is at line 507 in gnucash/html/webkit1.c.
  g_free(priv->html_string);
Unfortunately I don't see any obvious way that priv->html_string could be corrupted. It's initialized to NULL in gnc_html_webkit_init (part of GObject construction, so it must have run) and re-nulled the only other place it's freed. Here it's immediately reassigned to a new g_strdup which would return a validly allocated ptr or NULL.
Comment 6 John Ralls 2021-09-23 17:41:22 EDT
AFAIK this problem has never resurfaced and without being able to reproduce the problem I unable to determine the cause.

Note You need to log in before you can comment on or make changes to this bug.