[tahoe-lafs-trac-stream] [tahoe-lafs] #1418: "cannot convert float NaN to integer" in next_power_of_k, during upload via helper

[tahoe-lafs]

#1418: "cannot convert float NaN to integer" in next_power_of_k, during upload via
helper
------------------------+---------------------------
     Reporter:  rycee   |      Owner:  rycee
         Type:  defect  |     Status:  new
     Priority:  major   |  Milestone:  undecided
    Component:  code    |    Version:  1.8.2
   Resolution:          |   Keywords:  helper upload
Launchpad Bug:          |
------------------------+---------------------------

Comment (by zooko):

 Hm, judging from http://docs.python.org/library/fpectl.html , any Python
 code (in the CPython interpreter or in an extension module) which does
 floating point math is supposed to wrap the calls to floating point
 functions in macros named {{{PyFPE_START_PROTECT}}} and
 {{{PyFPE_END_PROTECT}}} in order to trap the exception and raise a nice
 Python exception. Apparently this hasn't happened!

 Hopefully you can get a run under valgrind without all the noise from
 false positives. Maybe whatever problem ends in the floating point
 exception starts with a memory mis-usage that valgrind can report.

 Jason Riedy on identi.ca wrote: "If the 30L somehow is pushed on a stack
 as an integer, the resulting corruption likely would produce a NaN." --
 http://identi.ca/notice/77048991

-- 
Ticket URL: <http://tahoe-lafs.org/trac/tahoe-lafs/ticket/1418#comment:19>
tahoe-lafs <http://tahoe-lafs.org>
secure decentralized storage