Recently I read somewhere that calloc() does lazy allocation, in that it doesn't allocate any real memory, but just virtual memory. Real memory pages are only allocated as writes to it are performed, and any reads performed on uninitialized memory returns 0, just like a sparse file. This made a lot of sense to me, because it seemed strange that ANSI would come up w/a whole new function to do the same thing that malloc() followed by bzero() could do, and besides, I've never found a use for calloc() in my entire career, so far. So I thought that this would be great, because if I have a method that writes a blob to a db, and if the blob column in that db can be up to 1Gb, then my method can lazily allocate a whole Gb, and even if my app only ever handles 10-byte blobs, the most memory that would be wasted is the better part of a page.
Since then I've tried to read up on this, and what I've read is that malloc() and calloc() allocate memory in the same way, and calloc() does actively initialize the memory. Furthermore, that lazy allocation I'm talking about is system-dependent, and if calloc() on a particular system does it, then so does malloc(). Can anyone confirm any of this for me? I'm mostly interested in the Linux kernel, since that seems to be all I program in these days.
