On Sat, Aug 7, 2021 at 11:40 AM Andres Freund <andres@anarazel.de> wrote:
Hi,
On 2021-07-12 17:28:15 +0530, Ashutosh Bapat wrote: > On Mon, Jul 12, 2021 at 8:39 AM Amit Kapila <amit.kapila16@gmail.com> wrote: > > > On Mon, Jul 5, 2021 at 12:54 PM Masahiko Sawada <sawada.mshk@gmail.com> > > Today, again looking at this patch, I don't think doing elog inside > > spinlock is a good idea. We can either release spinlock before it or > > use some variable to remember that we need to write such an elog and > > do it after releasing the lock. What do you think? > > > The elog will be effective only under DEBUG1, otherwise it will be almost a > NOOP. I am wondering whether it's worth adding a bool assignment and move > the elog out only for DEBUG1. Anyway, will defer it to you.
It's *definitely* not ok to do an elog() while holding a spinlock. Consider what happens if the elog tries to format the message and runs out of memory. Or if elog detects the stack depth is too deep. An ERROR would be thrown, and we'd loose track of the held spinlock.