Re: the s_lock_stuck on perform_spin_delay - Mailing list pgsql-hackers

On Mon, Jan 8, 2024 at 9:40 PM Andy Fan <zhihuifan1213@163.com> wrote:
> The singler handler I was refering to is 'CHECK_FOR_INTERRUPTS', Based
> on this, spin_lock and lwlock are acted pretty differently.

CHECK_FOR_INTERRUPTS() is not a signal handler, and it's OK to acquire
and release spin locks or lwlocks there. We have had (and I think
still do have) cases where signal handlers do non-trivial work,
resulting in serious problems in some cases. A bunch of that stuff has
been rewritten to just set a flag and then let the calling code sort
it out, but not everything has been rewritten that way (I think) and
there's always a danger of future hackers introducing new problem
cases.

--
Robert Haas
EDB: http://www.enterprisedb.com