On Tue, Oct 11, 2022 at 3:19 PM Richard Guo <guofenglinux@gmail.com> wrote: > On Tue, Oct 11, 2022 at 1:44 PM Kyotaro Horiguchi <horikyota.ntt@gmail.com> wrote: >> At Mon, 10 Oct 2022 08:53:55 +0530, Bharath Rupireddy <bharath.rupireddyforpostgres@gmail.com> wrote in >> > It looks like we have an unnecessary XLogSegNoOffsetToRecPtr() in >> > XLogReaderValidatePageHeader(). We pass the start LSN of the WAL page >> > and check if it matches with the LSN that was stored in the WAL page >> > header (xlp_pageaddr). We find segno, offset and LSN again using >> > XLogSegNoOffsetToRecPtr(). This happens to be the same as the passed >> > in LSN 'recptr'. >> >> Yeah, that's obviously useless. It looks like a thinko in pg93 when >> recptr became to be directly passed from the caller instead of >> calculating from static variables for file, segment and in-segment >> offset. > > > +1. This should be introduced in 7fcbf6a4 as a thinko. A grep search > shows other callers of XLogSegNoOffsetToRecPtr have no such issue.
Thanks for reviewing. It's a pretty-old code that exists in 9.5 or earlier [1], definitely not introduced by 7fcbf6a4.
