It makes a difference because m_RefreshPending was initialized to 0 by
the constructor, but the three bytes behind it were not, so the first
cmpxchg could fail.
Plus of course if someone decided to add another bool or a char array
or whatever at the end, that would get overwritten.
On 2015-11-03
The common computers never work with bits in memory. All the common
operations work at LEAST in bytes, and often in larger units, so
working with bits requires extra logic. Booleans are most commonly
stored as 1 byte because of that.
It is possible that come machines do use 1 bit, but they would
i thought it was 1 bit, not 1 byte... thanks for correcting my mistake :)
On Tue, Nov 3, 2015 at 10:00 AM, Thomas Faber
wrote:
> It makes a difference because m_RefreshPending was initialized to 0 by
> the constructor, but the three bytes behind it were not, so the
On 2015-11-03 22:02, Timo Kreuzer wrote:
> m_RefreshPending is a 32 bit BOOL and the constructor will initialize it
> completely with 0.
> If it was initializing only the 1st byte, the new code would still be wrong,
> since it makes no difference.
> InterlockedCompareExchange will exchange 32
m_RefreshPending is a 32 bit BOOL and the constructor will initialize it
completely with 0.
If it was initializing only the 1st byte, the new code would still be wrong,
since it makes no difference.
InterlockedCompareExchange will exchange 32 bit, no matter whether you pass a
bool or a BOOL.
Am 03.11.2015 um 22:09 schrieb Thomas Faber:
On 2015-11-03 22:02, Timo Kreuzer wrote:
m_RefreshPending is a 32 bit BOOL and the constructor will initialize it
completely with 0.
If it was initializing only the 1st byte, the new code would still be wrong,
since it makes no difference.
In fact in this case it wouldn't make a difference, since the bool would
be converted to a LONG. But using TRUE/FALSE seems to be appropriate here.
Am 22.10.2015 um 19:37 schrieb gedmur...@svn.reactos.org:
Author: gedmurphy
Date: Thu Oct 22 17:37:51 2015
New Revision: 69650
URL: