| From: | Alexander Korotkov <aekorotkov(at)gmail(dot)com> |
|---|---|
| To: | Heikki Linnakangas <hlinnakangas(at)vmware(dot)com> |
| Cc: | Tomas Vondra <tv(at)fuzzy(dot)cz>, pgsql-hackers <pgsql-hackers(at)postgresql(dot)org> |
| Subject: | Re: GIN improvements part 1: additional information |
| Date: | 2014-01-24 08:53:26 |
| Message-ID: | CAPpHfdugQ=dWDNf_=DFd5sj6yg=vajtY-xg9AXeJXY=02DsT=A@mail.gmail.com |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
On Fri, Jan 24, 2014 at 12:50 PM, Heikki Linnakangas <
hlinnakangas(at)vmware(dot)com> wrote:
> On 01/24/2014 10:03 AM, Alexander Korotkov wrote:
>
>> ITSM I found this bug. ginVacuumPostingTreeLeaf re-encodes only some
>> segments. Others are not even re-palloced. They are moved left
>> in dataPlaceToPageLeafRecompress by memcpy. But it's incorrect to with
>> memcpy, proper solution is memmove. Using memcpy platform dependently can
>> lead to page corruption. Another solution is to re-palloc segments in
>> ginVacuumPostingTreeLeaf.
>>
>
> Good catch. Thanks, committed, changing memcpy to memmove. Will have to
> switch to pallocing everything in the future, if we make leafRepackItems
> smarter, so that it doesn't rewrite all the segments after the first
> modified one.
OK. What about previous fix in assert?
------
With best regards,
Alexander Korotkov.
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Dimitri Fontaine | 2014-01-24 08:57:49 | Re: extension_control_path |
| Previous Message | Heikki Linnakangas | 2014-01-24 08:50:48 | Re: GIN improvements part 1: additional information |