Skip to main content

Further Reading

The Whole ISA​

If you want to search for an instruction, use this website. Each instruction has its own table with all possible operands and what they do. Note that imm8 means "8-bit immediate" (an 8-byte regular number), imm64 means a 64-bit immediate and so on. Similarly, reg32 means a 32-bit register and m16 for example means a 16-bit (2-byte) memory area. You'll see reg, imm and m combined with 8, 16, 32 and 64 depending on what each instruction does.

Caches​

Many programs access the same addresses repeatedly over a short period of time. Take a short 1000-step loop. It uses the same code 1000 times. It would be inefficient for the CPU to read the instructions directly from the RAM 1000 times. For this reason, there is an intermediary level of memory between the RAM and the registers, called the cache.

As their name implies, caches store the contents of some memory addresses that are frequently requested by the CPU. We say caches, in plural because they are laid out hierarchically, each lower level being faster and smaller than the ones below. Usually, CPUs have 3 levels of cache memory. You can query their sizes with the lscpu command:

root@kali:~$ lscpu
[...]
L1d cache:                       128 KiB
L1i cache:                       128 KiB
L2 cache:                        1 MiB
L3 cache:                        6 MiB
[...]

Notice the L1 (level 1) cache is split between a data cache (L1d) and an instruction cache L1i. The other caches do not store data and instructions separately.

Assembly Syntaxes​

This session we've used the Intel syntax for writing and displaying Assembly. We did so because it's more straightforward than its alternative: the AT&T syntax. You can find the differences on Wikipedia.

lea​

lea stands for "Load Effective Address". Its syntax is:

lea dest, [address]

It loads address into the dest register (it can only be a register). What's interesting about it is that it also uses the [...] syntax, but does not dereference the address. In the snippet below, 0xdeadbeef is simply copied to rax.

lea rax, [0xdeadbeef]

Its true power comes from the fact that it can also compute an address. For example, the code below will first compute the address given by rdi + rcx * 8 + 7 and then write this address into rax.

lea rax, [rdi + rcx * 8 + 7]