Re: XID formatting and SLRU refactorings (was: Add 64-bit XIDs into PostgreSQL 15)

Hi Noah,

> This yielded commit 4ed8f09 "Index SLRUs by 64-bit integers rather than by
> 32-bit integers" and left some expressions coercing SLRU page numbers to int.
> Two sources:
>
> grep -i 'int\b.*page' $(git grep -l SimpleLruInit)
> make && touch $(git grep -l SimpleLruInit) && make PROFILE=-Wconversion 2>&1 | less -p '.int. from .int64. may alter its value'
>
> (Not every match needs to change.)

I examined the new warnings introduced by 4ed8f09. Most of them seem
to be harmless, for instance:

```
slru.c:657:43: warning: conversion from ‘int64’ {aka ‘long int’} to
‘int’ may change value [-Wconversion]
657 | int rpageno = pageno %
SLRU_PAGES_PER_SEGMENT;
```

```
slru.c: In function ‘SlruReportIOError’:
slru.c:962:43: warning: conversion from ‘int64’ {aka ‘long int’} to
‘int’ may change value [-Wconversion]
962 | int rpageno = pageno %
SLRU_PAGES_PER_SEGMENT;
```

Interestingly the patch decreased the overall number of warnings.

I prepared the patch for clog.c. The rest of the warnings don't strike
me as something we should immediately act on unless we have a bug
report. Or perhaps there is a particular warning that worries you?

> > @@ -127,7 +127,15 @@ typedef struct SlruCtlData
> > * the behavior of this callback has no functional implications.) Use
> > * SlruPagePrecedesUnitTests() in SLRUs meeting its criteria.
> > */
> > - bool (*PagePrecedes) (int, int);
> > + bool (*PagePrecedes) (int64, int64);
> > +
> > + /*
> > + * If true, use long segment filenames formed from lower 48 bits of the
> > + * segment number, e.g. pg_xact/000000001234. Otherwise, use short
> > + * filenames formed from lower 16 bits of the segment number e.g.
> > + * pg_xact/1234.
> > + */
> > + bool long_segment_names;
>
> SlruFileName() makes 15-character (60-bit) file names. Where does the 48-bit
> limit arise? How does the SlruFileName() comment about a 24-bit limit for
> short names relate this comment's 16-bit limit?

Yes, this comment is wrong. Here is a fix.

[1]: https://www.postgresql.org/message-id/CAJ7c6TNMuKWUuMfh5KWgJJBoJGqPHYdZeN4t%2BLB6WdRLbDfVTw%40mail.gmail.com

--
Best regards,
Aleksander Alekseev