Bugs item #1274069, was opened at 2005-08-26 16:25
Message generated for change (Comment added) made by birkenfeld
You can respond by visiting:
https://sourceforge.net/tracker/?func=detail&atid=105470&aid=1274069&group_id=5470
Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: Extension Modules
Group: Python 2.5
Status: Open
Resolution: None
Priority: 5
Submitted By: Tim Peters (tim_one)
Assigned to: Nobody/Anonymous (nobody)
Summary: bz2module.c compiler warning
Initial Comment:
On Python HEAD, there's a new warning while
compiling bz2module.c (MSVC 7.1):
python\Modules\bz2module.c(1095) : warning
C4244: '=' :
conversion from 'Py_off_t' to 'int', possible loss of data
Looks like a legitimate complaint to me. The line in
question:
readsize = offset-bytesread;
Those have types int, Py_off_t, and int respectively. Is
it true that offset-bytesread _necessarily_ fits in an int?
If so, a comment should explain why, and a cast should
be added to squash the warning. Or If the difference
doesn't necessarily fit in an int, the code is wrong.
----------------------------------------------------------------------
>Comment By: Reinhold Birkenfeld (birkenfeld)
Date: 2005-08-27 16:36
Message:
Logged In: YES
user_id=1188172
Okay, next try. Someone with Windows should check this
before I check in.
----------------------------------------------------------------------
Comment By: Martin v. Löwis (loewis)
Date: 2005-08-27 14:55
Message:
Logged In: YES
user_id=21627
As Tim says: the cast deserves a comment, explaining why it
cannot overflow. Even when readsize is size_t, it might
still overflow, since Py_off_t might be wider than size_t.
Apart from that, the patch looks good - but I can't check
whether it silences the warning on Windows.
----------------------------------------------------------------------
Comment By: Reinhold Birkenfeld (birkenfeld)
Date: 2005-08-27 14:27
Message:
Logged In: YES
user_id=1188172
Attaching patch for f->pos and f->size as Py_off_t.
----------------------------------------------------------------------
Comment By: Martin v. Löwis (loewis)
Date: 2005-08-27 12:30
Message:
Logged In: YES
user_id=21627
Clearly, the difference is bound by buffersize (since, if it
is larger, readsize is set to buffersize). buffersize is
initialized to SMALLCHUNK, which is 8192, and never changed,
so yes, the difference fits to an int an all supported
platforms.
OTOH, the condition of the if statement is bogus:
if ((size_t)offset-bytesread > buffersize)
offset is of type Py_off_t, which is a 64-bit type on most
platforms. Casting it to size_t causes truncation on 32-bit
platforms.
Furthermore, I think self->pos should be of type Py_off_t,
as it will wrap around for files larger >2GB on 32-bit
platforms; the same for self->size.
----------------------------------------------------------------------
Comment By: Reinhold Birkenfeld (birkenfeld)
Date: 2005-08-26 17:05
Message:
Logged In: YES
user_id=1188172
I think declaring readsize as size_t should fix the problem.
----------------------------------------------------------------------
You can respond by visiting:
https://sourceforge.net/tracker/?func=detail&atid=105470&aid=1274069&group_id=5470
_______________________________________________
Python-bugs-list mailing list
Unsubscribe:
http://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com
