I noticed that pgbench calls FD_ISSET on a socket returned by
PQsocket() without first checking that it's not invalid. I don't think
there's a real problem here because the socket was already checked a few
lines above, but I think applying the same coding pattern to both places
is cleaner.
Any objections to changing it like this? I'd probably backpatch to 9.5,
but no further (even though this pattern actually appears all the way
back.)
*** a/src/bin/pgbench/pgbench.c
--- b/src/bin/pgbench/pgbench.c
***************
*** 3770,3780 **** threadRun(void *arg) Command **commands = sql_script[st->use_file].commands;
int prev_ecnt = st->ecnt;
! if (st->con && (FD_ISSET(PQsocket(st->con), &input_mask)
! || commands[st->state]->type == META_COMMAND)) {
! if (!doCustom(thread, st, &aggs))
! remains--; /* I've aborted */ } if (st->ecnt > prev_ecnt &&
commands[st->state]->type== META_COMMAND)
--- 3770,3790 ---- Command **commands = sql_script[st->use_file].commands; int
prev_ecnt= st->ecnt;
! if (st->con) {
! int sock = PQsocket(st->con);
!
! if (sock < 0)
! {
! fprintf(stderr, "bad socket: %s\n", strerror(errno));
! goto done;
! }
! if (FD_ISSET(sock, &input_mask) ||
! commands[st->state]->type == META_COMMAND)
! {
! if (!doCustom(thread, st, &aggs))
! remains--; /* I've aborted */
! } } if (st->ecnt > prev_ecnt && commands[st->state]->type == META_COMMAND)
--
Álvaro Herrera http://www.linkedin.com/in/alvherre
