[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: Berkeley 3.2.9 bug [#4164]

From: <cmpilato_at_collab.net>
Date: 2001-06-25 06:50:01 CEST

David,

Thank you on behalf of all the Subversion developers for your timely
and informative response to the little buglet we found, and for the
O-fficial patch which clears it up.

Mike Pilato
cmpilato@collab.net

PS -- So, uh, when can we expect to see those beautiful Berkeley
sources available via anonymous CVS? No, wait--I bet you're just
waiting for Subversion to reach 1.0, right? :-) Thanks again, David.

David Krinsky <krinsky@sleepycat.com> writes:

> > The second set of these reads (after the append of more data) loops
> > forever because the .size field in the result DBT is always set to
> > 100. Eventually, we get to the point where our offset is off the end
> > of the record, and *still* .size comes back 100.
>
> Hi, Michael--
>
> My name is David Krinsky, and I'm an engineer with Sleepycat Software;
> I'll own your Support Request #4164.
>
> This is a bug we recently found in Berkeley DB; the problem is that
> we were inconsistent about whether a partial get with a "doff"
> extending past the end of the record would return only the bytes
> available, or return a full dlen bytes and fill any space past the end
> of the record with padding. Your analysis is exactly correct; the
> mainline Btree code in 3.2.9 exhibits the former behavior (which we
> decided was correct), but records that are large enough to be put on
> an overflow page use a different code base and do the latter instead.
>
> The patch you have is correct, but we have committed to the DB sources
> a slightly different version, which allocates only as much memory
> as is needed. I have attached it below; it will also be included
> in the next release of Berkeley DB.
>
> Thanks for your detailed report; please let us know if you have
> any further questions or problems.
>
> Regards,
> Dave.
>
> --
> David Krinsky krinsky@sleepycat.com
> Sleepycat Software http://www.sleepycat.com
>
> Index: db/db_overflow.c
> ===================================================================
> RCS file: /b/CVSROOT/db/db/db_overflow.c,v
> retrieving revision 11.32
> diff -c -r11.32 db_overflow.c
> *** db/db_overflow.c 2001/05/17 17:23:49 11.32
> --- db/db_overflow.c 2001/06/14 18:24:01
> ***************
> *** 99,105 ****
> */
> if (F_ISSET(dbt, DB_DBT_PARTIAL)) {
> start = dbt->doff;
> ! needed = dbt->dlen;
> } else {
> start = 0;
> needed = tlen;
> --- 99,110 ----
> */
> if (F_ISSET(dbt, DB_DBT_PARTIAL)) {
> start = dbt->doff;
> ! if (start > tlen)
> ! needed = 0;
> ! else if (dbt->dlen > tlen - start)
> ! needed = tlen - start;
> ! else
> ! needed = dbt->dlen;
> } else {
> start = 0;
> needed = tlen;

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Sat Oct 21 14:36:32 2006

This is an archived mail posted to the Subversion Dev mailing list.

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.