On Tue, 24 Oct 2006, Mandeep Sandhu wrote:
I read some time ago (can't remember where!), that 0 is not always an invalid address. Though I think the C compiler is supposed to generate one for any such assignments (as shown above). Is this a C standard? Are all compilers supposed to do it? Shouldn't NULL be a better choice so that we can #define it to an invalid value, where this #define is specific to an architechture?? CMIIW.
This is a confusing issue which crops up now and then. There's a very lucid explanation of 0 vs. NULL vs. illegal values in the C FAQ, see for instance http://c-faq.com/null/index.html .
In short, when a compiler sees a 0 in a pointer context, it converts into the internal representation for a null pointer for the architecture in question. It doesn't matter if it's used in assigment ( p = 0 ) or as part of an expression ( if (p == 0) ), or even implicitly in an expression ( if (!p) ).
It is also true that most processors actually use the value 0 for the null pointer.
Semantically NULL is identical to 0, so if you write p = 0 or p = NULL there is absolutely no difference as far as the compiler is concerned. NULL is usually #defined to 0, and 0 will cause the compiler to create a null pointer in a pointer context. In many cases NULL is #defined to (void *) 0 in order to catch type conversion bugs.
At any rate, many people feel that NULL is better than 0 when used with pointers as it indicates that the expression is a pointer type. But it makes no difference to the compiler or to the resulting code.
/Ricard -- Ricard Wolf Wanderlöf ricardw(at)axis.com Axis Communications AB, Lund, Sweden www.axis.com Phone +46 46 272 2016 Fax +46 46 13 61 30 -- Kernelnewbies: Help each other learn about the Linux kernel. Archive: http://mail.nl.linux.org/kernelnewbies/ FAQ: http://kernelnewbies.org/faq/
