Problem with XP scheduler?
in [Win32 Kernel]
|
From: Leo Davidson on 23 Dec 2009 12:09 On Dec 23, 12:55 pm, ta...(a)mongo.net (tanix) wrote: > And that means that win is NOT a fully multitasking system. That's a silly thing to say. It's like concluding that a car isn't a true "moving-forwards machine" because it stops working if fitted with bad tyres which blow out and cause it to crash into a tree. I'm sure there are cases where Linux locks up as well. Maybe there's someone out there who has never experienced your Firefox/Windows lock- up but has experienced a lock-up on Linux, saying that "Linux is NOT a fully multitasking system, unlike Windows." That'd be a silly thing to say as well.
From: Boba on 22 Dec 2009 23:45 "Don Burn" <burn(a)stopspam.windrvr.com> wrote in message news:euMp3ywgKHA.6096(a)TK2MSFTNGP02.phx.gbl... > ... > Tim Roberts is a highly experienced windows developer, absolutely agree; that's why i couldn't hold it. > please show your data, since the only published data I know of contradicts > you. would this particular news thread serve the one? wouldn't it? "published data" you're talking about is exactly a sign of system behaving properly (at least responsive to some extend); what my concern is those unfortunate situations when you can't even get any dump from completely frozen box caused by powerpnt, winword, excel to name a few.
From: Leo Davidson on 23 Dec 2009 02:49 On Dec 23, 4:45 am, "Boba" <B...(a)somewhere.net> wrote: > "Don Burn" <b...(a)stopspam.windrvr.com> wrote in message > > get any dump from completely frozen box caused by > powerpnt, winword, excel to name a few. You're just assuming, with no evidence, that it is powerpnt, winword, excel or firefox that are responsible for the lock-ups you've seen when they may just be the trigger. In fact, I'd replace "may" with "must" because even if a user-level app does something deeply wrong, the drivers (or OS itself) are still at fault if they allow that to lock-up the entire system. At this point people usually reply asking why, if it's the drivers (or OS) at fault, does the problem only occur when particular software is used. The obvious answer is that different software does different things -- even when performing the same high-level tasks there are many ways to accomplish the same things -- and sometimes only the things that certain apps do trigger bugs in drivers. As a simple example, you might have a program which does lots of concurrent file access on multiple threads which triggers a problem in a filesystem driver which isn't protecting its shared data properly. A weird, and completely rule (happened to several people), example I know if is where people copying data with one program found their files were corrupted. People blamed the program at first but on investigation it turned out to be the system bus drivers. The reason one program resulted in corrupt data but others didn't is because that program had a progress dialog which had a lot more information in it and the extra GDI calls to update the dialog were causing enough bus traffic that it triggered a bug where the bus dropped data destined for the IDE controller. People upgraded their motherboard drivers and the problem went away.
From: Boba on 23 Dec 2009 06:03 "Leo Davidson" <leonudeldavidson(a)googlemail.com> wrote in message news:da4e6a99-fc5b-432d-bb23-81329d14289f(a)v30g2000yqm.googlegroups.com... >You're just assuming, with no evidence, that it is powerpnt, winword, >excel or firefox that are responsible for the lock-ups you've seen >when they may just be the trigger. this was not an assumption. that is a solid fact. i don't understand why you are trying to separate user applications from the rest of the system: if the box is dead - it is dead; for a user computer's heart is not beating any more and all he or she can do is to press that most popular button - the reset button. there's no need in trying to explain the user that it was just some part of the software and not the other: there are no parts in that dead body... yes - one may say - look, the fans are spinning - it is still alive :-) >In fact, I'd replace "may" with "must" because even if a user-level >app does something deeply wrong, the drivers (or OS itself) are still >at fault if they allow that to lock-up the entire system. no, and no again. the trigger was that dude who pushed the power button on that box to start it up and allowed the Windows to be loaded. >At this point people usually reply asking why, if it's the drivers (or >OS) at fault, does the problem only occur when particular software is >used. The obvious answer is that different software does different >things -- even when performing the same high-level tasks there are >many ways to accomplish the same things -- and sometimes only the >things that certain apps do trigger bugs in drivers. and my point is there are applications from MS that do that, and that there were not a single version of Windows yet that is free of that. (well, i have not touched W7 yet). the whole idea of having UM was/is to avoid the situations like this by keeping the system alive no matter what. >As a simple example, you might have a program which does lots of >concurrent file access on multiple threads which triggers a problem in >a filesystem driver which isn't protecting its shared data properly. which reminds me that explorer and iexplore are on the list too. >People upgraded their motherboard drivers and the problem went away. isn't it true that any user level problem could be solved by replacing either the core or the user himself?
From: Pavel A. on 23 Dec 2009 06:38
"Boba" <Boba(a)somewhere.net> wrote in message news:OulXT#7gKHA.5528(a)TK2MSFTNGP05.phx.gbl... > "Leo Davidson" <leonudeldavidson(a)googlemail.com> wrote in message > news:da4e6a99-fc5b-432d-bb23-81329d14289f(a)v30g2000yqm.googlegroups.com... > >>You're just assuming, with no evidence, that it is powerpnt, winword, >>excel or firefox that are responsible for the lock-ups you've seen >>when they may just be the trigger. > > this was not an assumption. that is a solid fact. > i don't understand why you are trying to separate user applications > from the rest of the system: if the box is dead - it is dead; Then let me help you understand. "Rest of the system" includes 3rd party drivers (video is especially troublesome, but others as well), for which you can not blame MS. Yes, they've architected OS that makes possible for 3rd party drivers to crash or hang the entire system, but this architecture was state of the art when it was created. The OP has been adviced to get the dump and help finding the exact reason of the problem, but he was not enough motivated to do this. So the whole issue is moot. Another Windoze crash, yeah. > isn't it true that any user level problem could be solved by replacing > either the core or the user himself? Replace XP with win7. Same h/w, same user, same application - the problem is gone. Now, has MS & industry made some progress, or not? --pa |