UUvF+Ii/P6zqJmew";
er.org"; posting-account="U2FsdGVkX1/f0dySh9V0Bn+BQHtG37Y7"
Gecko/20110616 Thunderbird/3.1.11
From: Peter Flass
On 7/13/2011 5:07 AM, Ruediger Ihle wrote:
> On Tue, 12 Jul 2011 04:29:19 UTC, Steve Wendt wrote:
>
>> As I understand it, if Mozilla is using TIMER0.SYS, other apps can't.
>> I don't know all the details, but I don't think this is a Mozilla
>> problem: http://www.os2site.com/sw/drivers/sound/misc/hrtx-v1.0r03.txt
>
> Actually Mozilla isn't using TIMER0.SYS, but the DosTmrXXX APIs. The
> problem is, that these APIs cease to work when an(other) application
> in the system uses TIMER0.SYS in it's standard version. AFAIK, this
> behavour is not documented anywhere.
It's too bad Timur Tabi doesn't hang out here. Wasn't he responsibility
for TIMER0.SYS?
>
>
>> I'm also not certain what it actually uses the timer for, but I
>> suspect any issues with it are exacerbated with SMP.
>
> ISTR, that a long time ago a developer wanted a better timer accuracy
> in order to perform benchmarking. He added the DosTmrXXX code (which is
> similar to the Windows implementation) without knowing/caring about the
> TIMER0 side effect. In result of my complaint, Mike Kaply introduced
> the environment variable. At that time it was thought that it would be
> a good idea to use the newer implentation by default. However, over the
> years it turned out that it would have been better to do the opposite.
It's always a good idea to look at what windoze does and do the opposite.
It's too bad that LINK386 doesn't [afaik] operate like mainframe linkage
editors and let you relink already linked executables. It would be
possible to replace individual kernel functions with rewritten ones
without needing the complete source, which it doesn't look like we're
going to ever get.
--- Internet Rex 2.31
* Origin: A noiseless patient Spider (1:261/20.999)
|
