Undefined behavior in glibc

[Joseph Myers]

On Fri, 5 Feb 2016, Alexander Cherepanov wrote:

> Then let's look at strlen (&co.) from another angle. It starts accessing its
> parameter as chars and then continues accessing it as longs, like this:
> 
>   ...
>   char *p = ...;
>   *p++;
>   *(long *)p;
> 
> Is there a situation where this is valid in GNU C? For ISO C the question

If the underlying object has effective type an array of longs it seems 
valid to me.

> BTW is there a formal method to tell test files from important files? I see
> .../tst-* and .../test-* files but io/pwd.c looks like a test too.

You need to look at what files are listed in the relevant makefile 
variables.

-- 
Joseph S. Myers
joseph@codesourcery.com