Thu, 12 Apr 2007
Shell-less piping in PerlIn my previous article, I said:
Unfortunately, there is no easy way to avoid the shell when running a command that is attached to the parent process via a pipe. Perl provides open "| command arg arg arg...", which is what I used, and which is analogous to [system STRING], involving the shell. But it provides nothing analogous to [system ARGLIST], which avoids the shell. If it did, then I probably would have used it, writing something like this:Several people wrote to point out that, as of Perl 5.8.0, Perl does provide this, with a syntax almost identical to what I proposed:
open M, "|-", $MAILER, "-fnobody\@plover.com", $addre;Why didn't I use this? The program was written in late 2002, and Perl 5.8.0 was released in July 2002, so I expect it's just that I wasn't familiar with the new feature yet. Why didn't I mention it in the original article? As I said, I just got back from Asia, and I am still terribly jetlagged.
(Jet lag when travelling with a toddler is worse than normal jet lag, because nobody else can get over the jet lag until the toddler does.)
Jeff Weisberg also pointed out that even prior to 5.8.0, you can write:
open(F, "|-") || exec("program", "arg", "arg", "arg");Why didn't I use this construction? I have run out of excuses. Perhaps I was jetlagged in 2002 also.
RFC 822John Berthels wrote to point out that my proposed fix, which rejects all inputs containing spaces, also rejects some RFC822-valid addresses. Someone whose address was actually something like "Mark Dominus"@example.com would be unable to use the web form to subscribe to the mailing list.
Quite so. Such addresses are extremely rare, and people who use them are expected to figure out how to subscribe by email, rather than using the web form.
qmailNobody has expressed confusion on this point, but I want to expliticly state that, in my opinion, the security problem I described was entirely my fault, and was not due to any deficiency in the qmail mail system, or in its qmail-inject or qmail-queue components.
Moreover, since I have previously been paid to give classes at large conferences on how to avoid exactly this sort of problem, I deserve whatever scorn and ridicule comes my way because of this.
Thanks to everyone who wrote in.