|From:||Ingo Molnar <mingo-AT-elte.hu>|
|To:||Al Viro <viro-AT-ftp.linux.org.uk>|
|Subject:||Re: [RFC] timers, pointers to functions and type safety|
|Date:||Mon, 4 Dec 2006 12:48:51 +0100|
|Cc:||Thomas Gleixner <tglx-AT-linutronix.de>, Matthew Wilcox <matthew-AT-wil.cx>, Linus Torvalds <torvalds-AT-osdl.org>, linux-arch-AT-vger.kernel.org, linux-kernel-AT-vger.kernel.org|
* Al Viro <firstname.lastname@example.org> wrote: > > This is going to make a lot of data structures smaller, when the > > timer_list is embedded in the structure itself and for the lot, > > which ignores the timer callback argument anyway. > > container_of => still lousy type safety. All over the sodding place. the question is: which is more important, the type safety of a container_of() [or type cast], which if we get it wrong produces a /very/ trivial crash that is trivial to fix - or embedded timers data structure size all around the kernel? I believe the latter is more important. and we could have a runtime debugging option to tie the type of the structure to the timer list entry. For example by using __builtin_classify_type(), sizeof() and offsetof() to fingerprint timer structs at init_timer time, and then checking for that at container_of() time - or something like that. In fact, gcc should really give us a better way to categorize types than __builtin_classify_type(). We could probably improve the situation by having some global registry of types known to the kernel, via a huge switch() around __builtin_types_compatible_p(). Ingo
Copyright © 2006, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds