Hi!
5-Янв-2005 12:07 [EMAIL PROTECTED] (Pat Villani) wrote to
[email protected]:
>> :) We can't protect from such hardware failures (when executed random
>>pieces of code). :(
PV> Actually, you can.
No - because hardware failures (on which code works) may be _very_
different and all of them is unpredictable.
Hm. After rereading your (skipped) paragraph, I suggest that you mean
"for handling external event was called code, which earlier wasn't tested".
In this case, we discuss not "hardware failure", but "untested code".
PV> Changing something like this is the difference
PV> between a stable and unstable kernel. BTW -- that wasn't random code
PV> execution. It was an untested piece of code that didn't get executed
PV> because that hardware failure never occurred.
But in given case (`break' after `return_user()') there are _no_
untested code, and return_user() doesn't returns back to switch in
inthndlr.c unconditionally. So, `break' there is _really_ not need, but its
missing (or missing comment about it) may mislead programmer and somewhat
affect optimizer.
PV> How much of the kernel
PV> can any of us say has actually been executed? Are there any hardware
PV> error cases that are not caught in the driver or has just not been
PV> exercised because the error almost never happens?
-------------------------------------------------------
The SF.Net email is sponsored by: Beat the post-holiday blues
Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt
_______________________________________________
Freedos-kernel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/freedos-kernel
