[Python-Dev] [Python-checkins] cpython: Issue #14716: Change integer overflow check in unicode_writer_prepare() (original) (raw)

Mark Dickinson dickinsm at gmail.com
Mon May 7 13:35:27 CEST 2012

On Mon, May 7, 2012 at 12:08 PM, victor.stinner <python-checkins at python.org> wrote:

http://hg.python.org/cpython/rev/ab500b297900 changeset:   76821:ab500b297900 user:        Victor Stinner <victor.stinner at gmail.com> date:        Mon May 07 13:02:44 2012 +0200 summary:  Issue #14716: Change integer overflow check in unicodewriterprepare() to compute the limit at compile time instead of runtime. Patch writen by Serhiy Storchaka.

if (newlen > PyUnicodeGETLENGTH(writer->buffer)) { -        /* overallocate 25% to limit the number of resize */ -        if (newlen <= (PYSSIZETMAX - newlen / 4)) +        /* Overallocate 25% to limit the number of resize. +           Check for integer overflow: +           (newlen + newlen / 4) <= PYSSIZETMAX */ +        if (newlen <= (PYSSIZETMAX - PYSSIZETMAX / 5)) newlen += newlen / 4;

Hmm. Very clever, but it's not obvious that that overflow check is mathematically sound. As it turns out, the maths works provided that PY_SSIZE_T_MAX isn't congruent to 4 modulo 5; since PY_SSIZE_T_MAX will almost always be one less than a power of 2 and powers of 2 are always congruent to 1, 2 or 4 modulo 5, we're safe.

Is the gain from this kind of micro-optimization really worth the cost of replacing obviously correct code with code whose correctness needs several minutes of thought?

Mark

More information about the Python-Dev mailing list