What does the .align directive do?

Question

I'm learning x86, and I'm trying to figure out what the .align directive does and how it can be useful.

The Oracle Reference Manual says:

The .align directive causes the next data generated to be aligned modulo integer bytes. Integer must be a positive integer expression and must be a power of 2. If specified, pad is an integer byte value used for padding. The default value of pad for the text section is 0x90 (nop); for other sections, the default value of pad is zero (0).

But I'm not sure what it means. Would you refer to where I can read more about it or explain it briefly with an example?

Answer 1

The key to understanding what it does is to understand why it's there.

All computers have a natural boundary known as the wordsize. This boundary is typically 4-bytes or 8-bytes.

The CPU can load and store from memory faster and without wasting cache space if 4 and 8 byte values are located on these boundaries. Some types of CPU's cannot fetch misaligned values at all.

So, there must be a mechanism in the assembler that skips to the next boundary so that a label and a storage allocation directive can begin on a more efficient address.

For instructions, odd boundaries work on most computers, but they still have performance implications and waste cache space.

   .string "ab\0"
; this next address is 3

vs

    .string "ab\0"
    .align 4 # sometimes interpreted as 2**n, so, .align 2
; this next address is 4, and would still be 4 if the string was just "a"