Re: [PATCH] vsprintf: drop comment claiming %n is ignored

[Al Viro <viro@ZenIV.linux.org.uk>]

On Fri, Sep 13, 2013 at 04:03:25PM -0700, Kees Cook wrote:

> Maybe I missed this somewhere in the thread, but I'm not sure I
> understand the move to "void". Here's what I see, please correct me:
> 
> 1- seq_printf currently returns success/failure
> 2- some callers of seq_printf (correctly) use the return value as
> success/failure indication

Not all uses as success/failure are right, at that - you should *NOT*
return non-zero from ->show() on overflow.  Ever.  It should only
happen when you have a hard error and do *not* want the operation
retried.  On success ->show() returns zero - not the number of characters
written or anything like that.

> 3- some callers of seq_printf (incorrectly) use the return value as a
> length indication

> 4- both success/failure and length are important outputs from seq_printf

Not really.  Success/failure is used if you want to optimize ->show() a bit;
usually you don't.  Again, failure == overflow and it's both rare *and*
will cause all subsequent seq_...() to be no-ops until the retry.  Which
retry will follow.  Shortly.

> 5- we need a way to access the length written during the call

_Very_ rarely.  And I'd seriously suggest the use of %n in such cases.

> 6- want to minimize impact on the code base

Majority of the code base simply calls seq_printf() without caring what
it returns.  Very small minority is broken and needs to be fixed.

> Due to 1 and 2, it seems like there's no sense in changing the return
> value to void. Success/failure is already returned, and there are
> users of it. No sense changing them.

... most of them incorrect.

> The normal way to handle multiple return values (4 and 5) is to add a
> pointer argument. For example: seq_printf(s, &len, fmt, args...) where
> len can be NULL. But this runs against 6.
> 
> Due to 6, to solve 4 and 5, usually macro or inline tricks are used,
> for example:
> 
> __printf(3, 4) int seq_printf_len(struct seq_file *, size_t *len, ...);
> #define seq_printf(s, fmt, ...) seq_printf_len(s, NULL, fmt, ##__VA_ARGS__)
> 
> With this, solving 3 becomes possible (your void patch has already
> detected all the users of the return value, so we can sort out which
> expect length and which expect success/failure), and lets us actually
> remove the %n uses trivially too.

Consider that NAKed.  Reason: fucking ugly and pointless at the same time.

I don't believe that seq_printf() itself needs to be changed at all (or that
%n should be removed, actually).  Callers should be audited and fixed, of
course.  With dire warnings added to seq_file.[ch].  And no, it shouldn't
be returning void - the current calling conventions are actually right.