On Tue, Nov 05, 2019 at 06:39:25PM +0000, Ben Dooks wrote: > > I was wondering if we could do something like export the > check_assignment_types and maybe also add check_pointer_asn() in > evaluate.c? The code in check_assignment_types is quite long and > I am not sure I really want to try and recreate it. Sure, it's out of the question to recreate it. But better to create a wrapper for it (and later to adapt it to more specific needs) than export it. I'm not sure what you mean exactly by check_pointer_asn() but I've no principles against doing this. > I'm still not quite sure over the rules over what is and isn't allowed > to printf, it may be that we need to be able to get the sizes of the > types and verify those, which is very architecture specific. Well, I would say that the main rules are: * be useful * be essentially compatible with gcc and current kernel practice. Or said differently, what's is important it's not what is allowed or not but what devs needs to be warned about. The sizes are not what need to be checked, only the types matter. And yes, I think that checking these types is much desired but this can be done in a later step. For the architcture specific aspect, there is no need to worry much about it because sparse, by necessity, have already the correct types/sizes (for most archs and certainly the most important ones). > I would like to see this merged soon, as it looks like I will be doing > an amount of travelling for work, which can be time consuming and tiring. I understand and I'll be glad to merge it once the last batch of changes here are done. The reason I prefer that you move the code in a separate is not technical but so that you can be properly credited for it (when using git blame and such). Best regards, -- Luc
