[Bug 1193177] New: sqlite3_drv bad memory free and severe performance bug - with patch

bugzilla at redhat.com bugzilla at redhat.com
Mon Feb 16 18:45:16 UTC 2015


https://bugzilla.redhat.com/show_bug.cgi?id=1193177

            Bug ID: 1193177
           Summary: sqlite3_drv bad memory free and severe performance bug
                    - with patch
           Product: Fedora EPEL
           Version: el6
         Component: dspam
          Severity: medium
          Assignee: nathanael at gnat.ca
          Reporter: stuart at gathman.org
        QA Contact: extras-qa at fedoraproject.org
                CC: nathanael at gnat.ca, perl-devel at lists.fedoraproject.org



Created attachment 992325
  --> https://bugzilla.redhat.com/attachment.cgi?id=992325&action=edit
Patch for problems reported

Description of problem:
The dspam sqlite3_drv runs in auto_commit mode, resulting in 1000s of
transactions on a single email - rendering it unusably slow.  Further, it
crashes on sql errors (include expected failures such as with
_ds_verify_signature) because it uses the system free() on the error message
from sqlite3, instead of sqlite3_free().

Version-Release number of selected component (if applicable):
dspam-3.10.2-4.el6.x86_64

How reproducible:
always

Steps to Reproduce:
1. run dspam_train with sqlite3 configured to see performance bug.  
2. verify_signature error can be seen eventually in normal operation, but more
quickly with the pydspam-1.3 test suite, with Dspam.py modified to load the
sqlite3 driver (one line change).  Apply the performance bug fix first - or it
will be a while...
3.

Actual results:
Unusably slow, and crashes on verify_signature when signature does not exist.

Expected results:
Reasonable performance, and no crashes.

Additional info:

https://pypi.python.org/pypi?%3Aaction=pkg_edit&name=pydspam

Patch to fix the problems attached.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=kmUjakIn3F&a=cc_unsubscribe



More information about the perl-devel mailing list