I use pickle over sockets to make 2 processes communicate and I experience hangs when the resulting pickle message is exactly 4097 bytes.

Attached file is a small script reproducing the issue, trying a few values around 4096. It runs on both Python 2.7 & 3.3:

- Python 2.7.3: ok (script exits)
- Python 3.3.2: last iteration hangs

With strace, you can easily see that after reading the 4097 bytes:

[pid 17100] read(3,  <unfinished ...>
[pid 17100] <... read resumed> "\200\2X\367\17\0\0                         "..., 4096) = 4096
[pid 17100] read(3, ".", 4096)          = 1

it reads again without reason:

[pid 17100] read(3, 

No issue with the Python implementation, that you can try with little change:

- Python 2.7.3: ok   (from pickle import Unpickler)
- Python 3.3.2: ok   (from pickle import _Unpickler as Unpickler)

What I see in gdb is that that read is occurring due to the unpickling thread reading peeking ahead in the stream due to buffered io being used.

#0  0x00007ffff7bcbcbd in read () from /lib/x86_64-linux-gnu/libpthread.so.0
#1  0x000000000061f521 in fileio_readinto (self=0x7ffff6491528, args=0x7ffff66dc190) at ./Modules/_io/fileio.c:541
#2  0x00000000004b96f0 in PyCFunction_Call (func=0x7ffff601d598, arg=0x7ffff66dc190, kw=0x0) at Objects/methodobject.c:90
#3  0x000000000045a0c2 in PyObject_Call (func=0x7ffff601d598, arg=0x7ffff66dc190, kw=0x0) at Objects/abstract.c:2105
#4  0x000000000045ad3e in PyObject_CallMethodObjArgs (callable=0x7ffff601d598, name=0x7ffff7f3ccf0) at Objects/abstract.c:2341
#5  0x0000000000626e65 in _bufferedreader_raw_read (self=0x7ffff6777e58, start=0xa89210 ".\002X\367\017", len=4096) at ./Modules/_io/bufferedio.c:1438
#6  0x00000000006270ed in _bufferedreader_fill_buffer (self=0x7ffff6777e59) at ./Modules/_io/bufferedio.c:1470
#7  0x0000000000628474 in _bufferedreader_peek_unlocked (self=0x7ffff6777e58) at ./Modules/_io/bufferedio.c:1714
#8  0x0000000000624b1d in buffered_peek (self=0x7ffff6777e58, args=0x7ffff66d4948) at ./Modules/_io/bufferedio.c:880

As a possible workaround if you disable buffering when opening your pipe for input it seems to prevent the problem (though I am not sure about performance implications).

22c22
< load = Unpickler(os.fdopen(r, 'rb')).load
---
> load = Unpickler(os.fdopen(r, 'rb', 0)).load

For the record, this is one of the things that the framing proposal in pickle protocol 4 (pep 3154 - http://www.python.org/dev/peps/pep-3154/#framing) is able to address.

Seems as the patch for issue17897 fixes this case.

If issue17897 fixes this bug (I'm not absolutely sure), perhaps it should be backported to 3.3.