Remove support for HPPA (a/k/a PA-RISC) architecture.
authorTom Lane <tgl@sss.pgh.pa.us>
Mon, 1 Jul 2024 17:55:52 +0000 (13:55 -0400)
committerTom Lane <tgl@sss.pgh.pa.us>
Mon, 1 Jul 2024 17:55:52 +0000 (13:55 -0400)
commitedadeb0710e838f6ce1f95973fae683a56fdba06
tree05faefc07fcff0da2ed3e23c80af65159ac4d652
parent7967d10c5b49ccb82f67a0b80678a1a932bccdee
Remove support for HPPA (a/k/a PA-RISC) architecture.

This old CPU architecture hasn't been produced in decades, and
whatever instances might still survive are surely too underpowered
for anyone to consider running Postgres on in production.  We'd
nonetheless continued to carry code support for it (largely at my
insistence), because its unique implementation of spinlocks seemed
like a good edge case for our spinlock infrastructure.  However,
our last buildfarm animal of this type was retired last year, and
it seems quite unlikely that another will emerge.  Without the ability
to run tests, the argument that this is useful test code fails to
hold water.  Furthermore, carrying code support for an untestable
architecture has costs not to be ignored.  So, remove HPPA-specific
code, in the same vein as commits 718aa43a4 and 92d70b77e.

Discussion: https://postgr.es/m/3351991.1697728588@sss.pgh.pa.us
contrib/pgcrypto/crypt-blowfish.c
doc/src/sgml/installation.sgml
src/backend/storage/lmgr/s_lock.c
src/include/port/atomics.h
src/include/port/atomics/arch-hppa.h [deleted file]
src/include/port/atomics/fallback.h
src/include/storage/s_lock.h
src/tools/pginclude/headerscheck