Re: AIX support

Thomas Munro <thomas.munro@gmail.com>

From: Thomas Munro <thomas.munro@gmail.com>

To: Heikki Linnakangas <hlinnaka@iki.fi>

Cc: Srirama Kucherlapati <sriram.rk@in.ibm.com>, "pgsql-hackers@postgresql.org" <pgsql-hackers@postgresql.org>, Robert Haas <robertmhaas@gmail.com>, Bruce Momjian <bruce@momjian.us>, Peter Eisentraut <peter@eisentraut.org>, Alvaro Herrera <alvherre@alvh.no-ip.org>, Laurenz Albe <laurenz.albe@cybertec.at>, Noah Misch <noah@leadboat.com>, Michael Paquier <michael@paquier.xyz>, Andres Freund <andres@anarazel.de>, "tvk1271@gmail.com" <tvk1271@gmail.com>, Tom Lane <tgl@sss.pgh.pa.us>, Tristan Partin <tristan@neon.tech>, wenhui qiu <qiuwenhuifx@gmail.com>, "postgres-ibm-aix@wwpdl.vnet.ibm.com" <postgres-ibm-aix@wwpdl.vnet.ibm.com>

Date: 2025-04-25T01:31:46Z

Lists: pgsql-hackers

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

Restore AIX support.
- 4a1b05caa55d 19 (unreleased) landed
pg_createsubscriber: Improve error messages.
- 898c131b58a0 18.0 cited
Use <stdint.h> and <inttypes.h> for c.h integers.
- 962da900ac8f 18.0 cited
Stabilize jsonb_path_query test case.
- af2115226831 18.0 cited
Fix C23 compiler warning
- d2b4b4c2259e 18.0 cited
pg_stat_statements: Add tests for nested queries with level tracking
- 45e0ba30fc40 18.0 cited
Add missing newline at the end of index_including.sql
- 54b69f1bd730 17.0 cited
Remove AIX support
- 0b16bb8776bb 17.0 cited
Fix s_lock.h PPC assembly code to be compatible with native AIX assembler.
- c41a1215f049 9.6.0 cited
Use a non-locking initial test in TAS_SPIN on PPC.
- bc2a050d4097 9.2.0 cited
Use LWSYNC in place of SYNC/ISYNC in PPC spinlocks, where possible.
- 631beeac3598 9.2.0 cited
Use mutex hint bit in PPC LWARX instructions, where possible.
- 5cfa8dd3007d 9.2.0 cited
Adjust TAS assembly as per recent discussions: use "+m"(*lock) everywhere
- 109867748259 8.0.0 cited
Apple's assembler likes the inlined TAS syntax too, so no reason to
- f9ba0a7fe563 7.4.1 cited
Tighten up register usage for inline PPC version of tas().
- eb5e4c58d137 7.4.1 cited
Put the isync where it's supposed to be.
- cd35d601b859 7.4.1 cited
> > I'll re-check that with the ppc architecture guy here.
- ceb4f5ea9c2c 7.4.1 cited
Fix PPC s_lock operations to work correctly on multi-CPU machines.
- 7233aae50bea 7.3.1 cited
I tried to build PostgreSQL with the following step to see backends hung
- 50938576d482 7.3.1 cited
Complete merge of all old man page information.
- f2f43efbe1d5 7.1.1 cited
s_lock aix patch.
- e3b06a871b63 7.1.1 cited

On Mon, Apr 7, 2025 at 10:04 PM Heikki Linnakangas <hlinnaka@iki.fi> wrote:
> I'm surprised how big the difference is, because I actually expected the
> compiler to detect the memory-zeroing loop and replace it with some
> fancy vector instructions (does powerpc have any?).

It certainly does, and we've played with explicit AltiVec
vectorisation before with the idea that our abstraction over x86 and
ARM instructions could be extended to POWER, but that was just trying
stuff and learning while wondering if those abstractions are general
enough and how well all the instructions match up.  No one with modern
hardware and a vested interest has seriously investigated it.  That's
independent of automatic vectorisation done by the compiler or libc,
and I guess it would not really be AIX-specific since it would also
apply to Linux on POWER.  One of those patches is linked from this
small list of "ideas for a future PostgreSQL/AIX maintainer":

https://wiki.postgresql.org/wiki/AIX