Re: Re: Give me more details of some bits in infomask!!

From: Tomas Vondra
Date: 2023-02-26 23:23
To: jacktby(at)gmail(dot)com; pgsql-hackers
Subject: Re: Give me more details of some bits in infomask!!
On 2/26/23 15:30, jacktby(at)gmail(dot)com wrote:
> here are the source codes from src/include/access/htup_details.h.
> /*
> * information stored in t_infomask:
> */
> #define HEAP_HASNULL0x0001/* has null attribute(s) */
> #define HEAP_HASVARWIDTH0x0002/* has variable-width attribute(s) */
> #define HEAP_HASEXTERNAL0x0004/* has external stored attribute(s) */
> #define HEAP_HASOID_OLD0x0008/* has an object-id field */
> #define HEAP_XMAX_KEYSHR_LOCK0x0010/* xmax is a key-shared locker */
> #define HEAP_COMBOCID0x0020/* t_cid is a combo CID */
> #define HEAP_XMAX_EXCL_LOCK0x0040/* xmax is exclusive locker */
> #define HEAP_XMAX_LOCK_ONLY0x0080/* xmax, if valid, is only a locker */
>
> And I can't understand these attrs:

I suggest you try something like 'git grep HEAP_HASEXTERNAL' which shows
you where the flag is used, which should tell you what it means. These
short descriptions generally assume you know enough about the internals.

> 1. external stored attribute(s), what is this? can you give a create
> statement to show me?

external = value stored in a TOAST table

> 2. xmax is a key-shared locker/exclusive locker/only a locker, so how
> you use this? can you give me a scenario?
> let me try to explain it:
> if there is a txn is trying to read this heaptuple,
> the HEAP_XMAX_KEYSHR_LOCK bit will be set to 1.
> if there is a txn is trying to delete/update this heaptuple,
> the HEAP_XMAX_EXCL_LOCK bit will be set to 1.
> but for HEAP_XMAX_LOCK_ONLY, I can't understand.
> And another thought is that these three bit can have only one to be set
> 1 at most.

I believe HEAP_XMAX_LOCK_ONLY means the xmax transaction only locked the
tuple, without deleting/updating it.

> 3. t_cid is a combo CID? what's a CID? give me an example please.

CID means "command ID" i.e. sequential ID assigned to commands in a
single session (for visibility checks, so that a query doesn't see data
deleted by earlier commands in the same session). See
src/backend/utils/time/combocid.c for basic explanation of what "combo
CID" is.


regards

--
Tomas Vondra
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company

> I believe HEAP_XMAX_LOCK_ONLY means the xmax transaction only locked the
> tuple, without deleting/updating it.
if so, you mean when I read this tuple, this bit will be set 1, but I think this is duplicat with HEAP_XMAX_KEYSHR_LOCK.

> CID means "command ID" i.e. sequential ID assigned to commands in a
> single session (for visibility checks, so that a query doesn't see data
> deleted by earlier commands in the same session). See
> src/backend/utils/time/combocid.c for basic explanation of what "combo
> CID" is.
I think if cid is used for visibility checks in one session, that's meaingless, beacause we can use the t_xmin and t_xmax to
get this goal. Is tis