ash: fix 'read' built-in performance regression

Commits 8e6991733 and b2901ce8e fixed problems with the 'read'
shell built-in when the '-t' option was used.  However, they
result in a performance penalty.  This pipeline:

   seq -w 0 999999 | while read line; do :; done

takes 10 times longer than prior to the changes.

If no timeout is specified don't call poll(2).

Costs 16 bytes in a 32-bit build; 0 in 64-bit.

1 files changed, 4 insertions, 0 deletions

diff --git a/shell/shell_common.c b/shell/shell_common.c
index da157ea0e..7fb5f8c58 100644
--- a/shell/shell_common.c
+++ b/shell/shell_common.c
@@ -221,6 +221,10 @@ shell_builtin_read(struct builtin_read_params *params)
                 errno = 0;
                 pfd[0].events = POLLIN;
 //TODO race with a signal arriving just before the poll!
+#if ENABLE_PLATFORM_MINGW32
+                /* Don't poll if timeout is -1, it hurts performance. */
+                if (timeout >= 0)
+#endif
                 if (poll(pfd, 1, timeout) <= 0) {
                         /* timed out, or EINTR */
                         err = errno;