On Tue, Jul 13, 2021 at 01:42:58AM +0200, Jeremie Courreges-Anglas wrote: > On Tue, Jul 13 2021, Matthieu Herrb <[email protected]> wrote: > > Hi, > > > > This patch adds riscv64 support to postgresql and postgresql-previous. > > copied from FreeBSD ports. > > > > ok ? > > Your patch *looks* fine from a quick glance, it's the same patch that > has been proposed upstream earlier. But at least Tom Lane (long time > postgres developer) wasn't keen on reusing gcc-style intrinsics. > > First report in 2016: > https://www.postgresql.org/message-id/flat/[email protected] > Same thread, patch similar to yours: > https://www.postgresql.org/message-id/20161120150905.GO11243%40redhat.com > > More recent thread, upstream doesn't seem to reach a conclusion: > > https://www.postgresql.org/message-id/20191019102234.s7utdy44npjqr7in%40alap3.anarazel.de > https://www.postgresql.org/message-id/16913.1571545387%40sss.pgh.pa.us > > https://www.postgresql.org/message-id/CA%2BhUKGKnXHpROE%2Bm-Qw006V77Yo9NfTKAW2B-xn7XmDiRwmUZQ%40mail.gmail.com > > Your patch is not sufficient since I already committed a workaround for > the lack of spinlock implementation earlier today. You'd want the > additional diff below (untested). I have no strong opinion regarding > the approach we should take. >
Thanks for the feedback. As far as I understand it's either your approach or mine (from FreeBSD). While I can understand that PostgreSQL developpers would prefer a better approach, as far as I can tell my patch was enough to allow me to build the packages and install them and then initialized a database and do some basic operations on it. I'have tested with just what you committed in the mean time. Anyways I don't have more time to work on this until next week. Dropping my patch until uptstreams adds proper riscv support is fine by me. -- Matthieu Herrb
