Re: Attempt to consolidate reading of XLOG page

Antonin Houska <ah@cybertec.at>

View thread View on postgr.es

From: Antonin Houska <ah@cybertec.at>

To: Alvaro Herrera <alvherre@2ndquadrant.com>

Cc: pgsql-hackers@postgresql.org

Date: 2019-05-02T16:17:36Z

Lists: pgsql-hackers

Attachments

v02_001_Unexport_XLogReaderInvalReadState.patch (text/x-diff) patch v2
v02_002_Remove_TLI_argument_from_XLR_callback.patch (text/x-diff) patch v2
v02_003_Introduce_XLogSegment.patch (text/x-diff) patch v2
v02_004_Use_only_one_implementation_of_XLogRead.patch (text/x-diff) patch v2
v02_005_Cleanup.patch (text/x-diff) patch v2

Antonin Houska <ah@cybertec.at> wrote:

> Alvaro Herrera <alvherre@2ndquadrant.com> wrote:
> 
> > Not sure about the walsize; maybe it can be a member in XLogReadPos, and
> > given to XLogReadInitPos()?  (Maybe rename XLogReadPos as
> > XLogReadContext or something like that, indicating it's not just the
> > read position.)
> 
> As pointed out by others, XLogReadPos is not necessary. So if XLogRead()
> receives XLogReaderState instead, it can get the segment size from there.

Eventually I found out that it's good to have a separate structure for the
read position because walsender calls the XLogRead() function directly, not
via the XLOG reader. Currently the structure name is XLogSegment (maybe
someone can propose better name) and it's a member of XLogReaderState. No
field of the new structure is duplicated now.

The next version of the patch is attached.

-- 
Antonin Houska
Web: https://www.cybertec-postgresql.com

Commits

Remove logical_read_local_xlog_page
- 5d0c2d5eba6b 13.0 landed
Refactor WAL file-reading code into WALRead()
- 0dc8ead46363 13.0 landed
Rework WAL-reading supporting structs
- 709d003fbd98 13.0 landed
Make XLogReaderInvalReadState static
- 25dcc9d35dfe 13.0 landed
Use pg_pread() and pg_pwrite() for data files and WAL.
- c24dcd0cfd94 12.0 cited