Peter Geoghegan <pg(at)bowt(dot)ie> writes:
> My nbtree patch [1] needs to call index_getprocinfo() with an
> exclusive buffer lock held during a leaf page split.
I think you should stop right there and ask why. Surely that info
can be obtained before starting the operation? Quite aside from the
deadlock hazard, I do not think holding an exclusive buffer lock
for long enough to go consult a system catalog will be acceptable
from a performance/concurrency standpoint.
regards, tom lane