(4-bytes)  (Py_ssize_t) references count
(4-bytes)  (pointer)    object type
(4-bytes)  (pointer)    memory address of the buffer location
(10-bytes) (array)      preview
(4-bytes)  (Py_ssize_t) buffer size
(4-bytes)  (Py_ssize_t) string size

Binary IO with structs

32-bit gives 4-billions addresses (that's enough addressing for
max. 4G of memory)

address arithmetic?

overflow is when there isn't any space left to put a number -- i.e,
fixed size.

pmap $PYTHON_PID -> process map (check the heap)
/proc/$PYTHON_PID/maps

string needs to be NULL terminated
strncpy as a for-loop

Prototypes are for error-checking. You can't know how many
arguments a function from an external object takes.

Structures align the memory.

Big/little endian

4-bytes chunks, MAXSIZE=10 has a size of 12

dynamic loading (dload.so)


>>> import spam
>>> s = spam.Spam()
>>> s.put_stuff("Hello, I am the king of the world!")
34
>>> s.dump()

% hd dump-file
00000000  02 00 00 00 80 41 f4 b7  60 dc 1c 08 48 65 6c 6c  |.....A..`...Hell|
00000010  6f 2c 20 49 20 61 6d 20  00 01 00 00 22 00 00 00  |o, I am ...."...|
00000020

>>> def get_addr(bytes):
...    return "0x" + "".join(reversed(bytes.split()))
... 
>>> mem = open("/proc/%i/mem" % os.getpid())
>>> get_addr("60 dc 1c 08")
'0x081cdc60'
>>> mem.seek(0x081cdc60)
>>> f.read(34)
'Hello, I am the king of the world!'

 % cat /proc/21289/maps 
08048000-0813f000 r-xp 00000000 08:01 1832494    /usr/bin/python2.5
0813f000-08164000 rw-p 000f6000 08:01 1832494    /usr/bin/python2.5
08164000-081dd000 rw-p 08164000 00:00 0          [heap]
...
b7f43000-b7f44000 r-xp 00000000 08:02 1425420    /home/alex/code/python/ext/build/lib.linux-i686-2.5/spam.so
b7f44000-b7f45000 rw-p 00001000 08:02 1425420    /home/alex/code/python/ext/build/lib.linux-i686-2.5/spam.so
b7f45000-b7f48000 rw-p b7f45000 00:00 0 
b7f48000-b7f61000 r-xp 00000000 08:01 2649039    /lib/ld-2.5.so
b7f61000-b7f63000 rw-p 00019000 08:01 2649039    /lib/ld-2.5.so
bfacf000-bfae5000 rw-p bfacf000 00:00 0          [stack]
ffffe000-fffff000 r-xp 00000000 00:00 0          [vdso]

>>> f = open("dump-file")
>>> data = f.read()
>>> import struct
>>> struct.unpack("<12slll", data[12:])
('Hello World!', 256, 12, 3)
>>> s = struct.unpack("<l12slll", data[8:])
>>> s
(136109904, 'Hello World!', 256, 12, 3)
>>> "%.8x" % s[0]
'081cdf50'