<kenaan>
arigo sandbox-2 906b820ecdac /rpython/: in-progress: enforce a review on functions manipulating directly low-level pointers or doing any other strange th...
<arigato>
may take a while and another thread may run
<arigato>
now what if the buffer is shrunk by another thread?
nunatak has joined #pypy
<antocuni>
segfault?
<arigato>
likely
<arigato>
found during the review I'm doing now for sandboxing
<mattip>
isn't any modification of a shared array/bytestring dangerous?
<mattip>
or for that matter does python protect lists and dictionaries?
<mattip>
s/btyestring/bytearray
nunatak has quit [Quit: Leaving]
<arigato>
you're not supposed to obtain a segfault using only built-in types or 'array'
<arigato>
or 'file' or '_io'
jcea has quit [Remote host closed the connection]
<kenaan>
arigo default 215ea348ccb4 /pypy/module/: Check for a rare case of someone shrinking the buffer from another thread while using it in a read() variant
<arigato>
for example, ArrayBuffer keeps a reference to the app-level array object, and not to the raw data, so if it moves it's fine
<mattip>
cpython seems to be careful around keeping the references fresh
<arigato>
yes, because it would be easier to crash
<arigato>
in our case I could fix the problem just by checking again the length after the read() system call
<kenaan>
arigo py3.6 6bb742c50a9f /pypy/config/pypyoption.py: With either the sandbox or the reverse-debugger, the hash used in pypy3 needs to be the old 'fnv' hash instead of 'si...
<kenaan>
arigo default 56e99531241f /rpython/rlib/rsiphash.py: Forbid calls to enable_siphash24() in sandbox or revdb mode
<kenaan>
arigo default a6c1ae01424e /pypy/module/posix/interp_posix.py: CPython raises NotImplementedError if /dev/urandom cannot be found. To maximize compatibility, we should also raise...