The Other GCC Optimiser Issue / Bug [ formally gettimeofday seconds rollover problem ]

[Paul Whitfield]

Ian Caddy wrote:

> On another issue we are looking into a another Coldfire compiler bug 
> with gcc 3.2.3 which is listed here:
> 
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=8309
> 
> Basically in some of our code, uint16 are used, and if local it is 
> possible that they won't get incremented properly on Coldfire 
> processors, and possibly modifying other variables instead on any 
> optimisation level greater than 0.


Hi All,

This issue was raised in the original thread and seems to have
not been really followed up / lost in the discussions about
the optimiser reorganising code around interrupts.

While I have not seen this issue myself,
it is worrying me that it could be lurking in my code just
waiting to bite me!

Has anyone else observed this same bug?

If no one has a better solution I will probably apply the same
patch as Ian.

However, I am reluctant to move from the "official" gcc version.
(I am using RTEMS 4.6.5).

Regards

Paul



=========================================================================
The information in this e-mail is intended for the addressee only.
Unauthorised use, copying, disclosure or distribution by anyone else
is prohibited. Please let us know immediately if you receive this
e-mail in error. Thank you.
=========================================================================