Horton

Re: LogwrtResult contended spinlock

Alvaro Herrera <alvherre@alvh.no-ip.org>

View thread View on postgr.es
From: Alvaro Herrera <alvherre@alvh.no-ip.org>
To: Anastasia Lubennikova <a.lubennikova@postgrespro.ru>
Cc: Andres Freund <andres@anarazel.de>, pgsql-hackers@lists.postgresql.org, Jaime Casanova <jaime.casanova@2ndquadrant.com>
Date: 2020-11-24T15:01:53Z
Lists: pgsql-hackers 
On 2020-Nov-24, Anastasia Lubennikova wrote:

> On 04.09.2020 20:13, Andres Freund wrote:

> > Re general routine: On second thought, it might actually be worth having
> > it. Even just for LSNs - there's plenty places where it's useful to
> > ensure a variable is at least a certain size.  I think I would be in
> > favor of a general helper function.
> Do you mean by general helper function something like this?
> 
> void
> swap_lsn(XLogRecPtr old_value, XLogRecPtr new_value, bool to_largest)

Something like that, yeah, though maybe name it "pg_atomic_increase_lsn"
or some similar name that makes it clear that 

1. it is supposed to use atomics
2. it can only be used to *advance* a value rather than a generic swap.

(I'm not 100% clear that that's the exact API we need.)

> This CF entry was inactive for a while. Alvaro, are you going to continue
> working on it?

Yes, please move it forward.  I'll post an update sometime before the
next CF.

Commits

  1. Remove bogus assertion in pg_atomic_monotonic_advance_u64

  2. Add XLogCtl->logInsertResult

  3. Operate XLogCtl->log{Write,Flush}Result with atomics

  4. Split XLogCtl->LogwrtResult into separate struct members

  5. Introduce atomic read/write functions with full barrier semantics.

  6. Reduce the number of GetFlushRecPtr() calls done by walsenders.

  7. Remove most volatile qualifiers from xlog.c