| Re: [AD] Bug in MMX linear_clear_to_color8 with asynchronous usage |
[ Thread Index |
Date Index
| More lists.liballeg.org/allegro-developers Archives
]
> I think we ought not patch the IRQ stub. The problem has existed with
> normal FP code from the very beginning, before the MMX code was added, and
> I guess that the decision not to save the FPU state was made knowingly by
> Shawn.
Here's what I propose to add to the docs.
--
Eric Botcazou
--- /home/eric/cvs/allegro/docs/src/allegro._tx Fri Feb 7 14:23:40 2003
+++ allegro/docs/src/allegro._tx Fri Feb 14 11:33:39 2003
@@ -1816,7 +1816,9 @@
Under DOS it will constantly reprogram the clock to make sure they are all
called at the correct times. Because they alter the low level timer chip
settings, these routines should not be used together with other DOS timer
-functions like the djgpp uclock() routine.
+functions like the djgpp uclock() routine. Moreover, the FPU state is not
+preserved across Allegro interrupts so you ought not to use floating point
+or MMX code inside timer interrupt handlers.
Under other platforms, they are usually implemented using threads, which run
parallel to the main thread. Therefore timer callbacks on such platforms