Date: Fri, 3 Nov 2006 09:31:13 +0000 From: MQ <antinvidia@gmail.com> To: "Max Laier" <max@love2party.net> Cc: freebsd-net@freebsd.org Subject: Re: Reentrant problem with inet_ntoa in the kernel Message-ID: <be0088ce0611030131s482a9082rd51871a68894a2c8@mail.gmail.com> In-Reply-To: <200611021045.09774.max@love2party.net> References: <be0088ce0611020026y4fe07749pd5a984f8744769b@mail.gmail.com> <200611021045.09774.max@love2party.net>
next in thread | previous in thread | raw e-mail | index | archive | help
2006/11/2, Max Laier <max@love2party.net>: > > On Thursday 02 November 2006 09:26, . wrote: > > Hi, > > > > I am confused by the use of inet_ntoa function in the kernel. > > > > The function inet_ntoa in the /sys/libkern/inet_ntoa.c uses a static > > array static char buf[4 * sizeof "123"]; > > to store the result. And it returns the address of the array to the > > caller. > > > > I think this inet_ntoa is not reentrant, though there are several > > functions calling it. If two functions call it simultaneously, the > > result will be corrupted. Though I haven't really encountered this > > situation, it may occur someday, especially when using > > multi-processors. > > > > There is another reentrant version of inet_ntoa called inet_ntoa_r in > > the same file. It has been there for several years, but just used by > > ipfw2 for about four times in 7-CURRENT. In my patch, I replaced all > > the calls to inet_ntoa with calls to inet_ntoa_r. > > > > By the way, some of the original calls is written in this style: > > strcpy(buf, inet_ntoa(ip)) > > The modified code is written in this style > > inet_ntoa_r(ip, buf) > > This change avoids a call to strcpy, and can save a little time. > > > > Here is the patch. > > http://people.freebsd.org/~delphij/misc/patch-itoa-by-nodummy-at-yeah-n > >et > > > > I've already sent to PR(kern/104738), but got no reply, maybe it should > > be discussed here first? > > In general, correct IPs in logs and debugging messages are a good thing. > I'm not sure, however, it is a good thing to put 17 bytes of buffer space > on every function stack that might want to print an IP address. I think > it's less intrusive and equally good to have a hand full of static > buffers available which are given out in a round-robin fashion - as > attempted in ip6_sprintf. Obviously the buffer rotation needs to be > atomic, though. If a caller needs the result for more than logging - or > cares strongly - it can still allocate a private buffer and use the _r > version. A general replacement of all applications of inet_ntoa just > seems bloat. > > -- > /"\ Best regards, | mlaier@freebsd.org > \ / Max Laier | ICQ #67774661 > X http://pf4freebsd.love2party.net/ | mlaier@EFnet > / \ ASCII Ribbon Campaign | Against HTML Mail and News Yes, not all the occurrence of inet_ntoa should be replaced. I've replaced several inet_ntoa in /sys/boot/ just to be consistent with other files. MAYBE they will never be corrupted by *simultaneous*ly calling inet_ntoa. But the calls in the network stack can really break the results. If that happens, the users will be confused by the strange outputs and/or logs. So I think adding 16 bytes of memory consumption on the stack is worthy.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?be0088ce0611030131s482a9082rd51871a68894a2c8>