public inbox for [email protected]
help / color / mirror / Atom feedFrom: Fujii Masao <[email protected]>
To: Amit Kapila <[email protected]>
Cc: Zhijie Hou (Fujitsu) <[email protected]>
Cc: Xuneng Zhou <[email protected]>
Cc: Srinath Reddy Sadipiralla <[email protected]>
Cc: PostgreSQL Hackers <[email protected]>
Subject: Re: Fix race in ReplicationSlotRelease for ephemeral slots
Date: Thu, 11 Jun 2026 22:19:33 +0900
Message-ID: <CAHGQGwG_3ff4HciHtTZ_uMvbJgSDWsz4Yawj_zQpDG6Yj=Mjng@mail.gmail.com> (raw)
In-Reply-To: <CAA4eK1LkRdbm5XA=qa82Rp_y4rnyJh8pypMWVqOezOZpzy=Oaw@mail.gmail.com>
References: <TY4PR01MB177184FF9EE916F577E1F554194082@TY4PR01MB17718.jpnprd01.prod.outlook.com>
<CAFC+b6o-hD5VxVLZQovmHSYykF8Qzq3eiuBU-U1F_yR9-y6P_w@mail.gmail.com>
<TY4PR01MB177180A7CE60BCDF286B1C6F594172@TY4PR01MB17718.jpnprd01.prod.outlook.com>
<CABPTF7VyH1-W2xnDspECDEzFGQj=WTFpZBCqKfM11OAZa6gQHQ@mail.gmail.com>
<CAHGQGwE+2WSqiAYgNJRkf_twdB+uRGozjjGhUn76vUKZ8dzbSA@mail.gmail.com>
<CABPTF7VeA8szPv7LYDVY9_7LftV-HM8NFVQR2natPKmr73JW+A@mail.gmail.com>
<TY4PR01MB1771887D33612C5A45F7E9CDF941E2@TY4PR01MB17718.jpnprd01.prod.outlook.com>
<CAA4eK1LqFBKCkX2eoX3iQPxJJnzWTaCpdh9zNotxuoG8BgjdtA@mail.gmail.com>
<CAA4eK1LkRdbm5XA=qa82Rp_y4rnyJh8pypMWVqOezOZpzy=Oaw@mail.gmail.com>
On Thu, Jun 11, 2026 at 8:18 PM Amit Kapila <[email protected]> wrote:
> 1. Stale name read in local_sync_slot_required(): The reused cell
> holds a different name. local_sync_slot_required() might return false
> (drop needed). But then the in_use && synced spinlock check sees
> synced = false and skips the actual drop. The wrong decision is
> caught.
Yes, we could skip the actual drop. But then wouldn't we still emit
the log message "dropped replication slot ..." even though no slot was
actually dropped?
> 2. Wrong database OID read at line 551: The reused cell holds OID_B
> from the new slot. We lock OID_B, then at lines 563–565 we see synced
> = false, skip the drop, and unlock OID_B at line 579. Since no drop
> occurred, the cell is still the same non-synced slot, so the lock and
> unlock see the same OID_B. Symmetric — no lock leak.
What happens if the slot for OID_B is dropped after we lock
OID_B, and then a new slot for OID_C reuses the same array entry? In
that case, wouldn't the later unlock read OID_C from
local_slot->data.database even though the lock was originally taken on
OID_B?
Regards,
--
Fujii Masao
view thread (27+ messages) latest in thread
reply
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Reply to all the recipients using the --to and --cc options:
reply via email
To: [email protected]
Cc: [email protected], [email protected], [email protected], [email protected], [email protected], [email protected]
Subject: Re: Fix race in ReplicationSlotRelease for ephemeral slots
In-Reply-To: <CAHGQGwG_3ff4HciHtTZ_uMvbJgSDWsz4Yawj_zQpDG6Yj=Mjng@mail.gmail.com>
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
This inbox is served by agora; see mirroring instructions
for how to clone and mirror all data and code used for this inbox