[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <083fac943c8c422a9f667f6b0371153c@AcuMS.aculab.com>
Date: Mon, 7 Feb 2022 17:05:21 +0000
From: David Laight <David.Laight@...LAB.COM>
To: 'Willy Tarreau' <w@....eu>,
"Paul E . McKenney" <paulmck@...nel.org>
CC: Mark Brown <broonie@...nel.org>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: RE: [PATCH 11/42] tools/nolibc/types: move the FD_* functions to
macros in types.h
From: Willy Tarreau
> Sent: 07 February 2022 16:23
>
> FD_SET, FD_CLR, FD_ISSET, FD_ZERO are supposed to be macros and not
> functions.
Are you sure?
I'd have thought they could be either.
There are certainly systems where they are functions.
They can be implemented as an array of fd numbers rather than a bitmap.
> In addition we already have a file dedicated to such macros
> and types used by syscalls, it's types.h, so let's move them
> there and turn them to macros. FD_CLR() and FD_ISSET() were missing,
> so they were added. FD_ZERO() now deals with its own loop so that it
> doesn't rely on memset() that sets one byte at a time.
>
....
> +#define FD_CLR(fd, set) do { \
> + int __fd = (int)(fd); \
> + fd_set *__set = (fd_set *)(set); \
I'm not sure you really want either cast.
They are just likely to hide some horrid bugs.
+ if (__fd >= 0 && __fd < FD_SETSIZE) \
+ __set->fd32[__fd / 32] &= ~(1U << (__fd & 31)); \
+ } while (0)
+
Do you need the range check?
I don't think glibc has one.
Things just break in obscure ways when you use select on big fd.
David
-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)
Powered by blists - more mailing lists