Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss
Segfaults/gpf using DBI 1.622 with threads
6 views
Skip to first unread message
Matthew Musgrove
Oct 26, 2012, 4:24:46 PM10/26/12
to dbi-...@perl.org
So once again we're seeing segmentation and general protection faults when
using DBI with threads. I've tested it with Perl 5.12.3, 5.14.3, and
5.16.1. The backtraces keep pointing to DBI.xs line 3265:
#0 0x00002aaaaaaddd8f in XS_DBI_dispatch (my_perl=0x2ed7310, cv=0x3448528)
at DBI.xs:3265
3265 if ((i = DBIc_DEBUGIV(imp_xxh))) { /* merge handle into global
*/
Has anyone else seen this?
I don't even know how to begin to recreate the failure outside of this one
daemon process.
I'll try reverting to DBI 1.61 (I can't go further back that than or I'll
hit bug 77137 again) and see it is still an issue.
Matt
using DBI with threads. I've tested it with Perl 5.12.3, 5.14.3, and
5.16.1. The backtraces keep pointing to DBI.xs line 3265:
#0 0x00002aaaaaaddd8f in XS_DBI_dispatch (my_perl=0x2ed7310, cv=0x3448528)
at DBI.xs:3265
3265 if ((i = DBIc_DEBUGIV(imp_xxh))) { /* merge handle into global
*/
Has anyone else seen this?
I don't even know how to begin to recreate the failure outside of this one
daemon process.
I'll try reverting to DBI 1.61 (I can't go further back that than or I'll
hit bug 77137 again) and see it is still an issue.
Matt
Matthew Musgrove
Oct 26, 2012, 4:30:40 PM10/26/12
to dbi-...@perl.org
Matthew Musgrove
Nov 8, 2012, 5:10:14 PM11/8/12
to dbi-...@perl.org
This is more for future reference in case anyone encounters this problem.
I reported a bug <https://rt.cpan.org/Ticket/Display.html?id=80474> on
10/29. Tim Bunce has updated the bug report today with a possible fix that
looks good to me (no segfaults on 5.12.3 or 5.16.2).
Matt
Matt
On Fri, Oct 26, 2012 at 3:44 PM, Matthew Musgrove <mr.mu...@gmail.com>wrote:
> I didn't think that it would matter but it still happened after
> downgrading DBI to 1.621.
>
> Here's a full backtrace.
I reported a bug <https://rt.cpan.org/Ticket/Display.html?id=80474> on
10/29. Tim Bunce has updated the bug report today with a possible fix that
looks good to me (no segfaults on 5.12.3 or 5.16.2).
Matt
Matt
On Fri, Oct 26, 2012 at 3:44 PM, Matthew Musgrove <mr.mu...@gmail.com>wrote:
> I didn't think that it would matter but it still happened after
> downgrading DBI to 1.621.
>
> Here's a full backtrace.
0 new messages