Hyper Threading Technology

NTDEV
1

Hi All,
What all things I should keep in mind while designing a SCSI miniport driver on a Intel P-4 processor with Hyper- Threading Technology? . What are the differences between processor with hyper threading support and multi processor? Whether my SCSI miniport driver that works perfectly fine on a multi-processor machine will work properly on a hyper-threaded processor?

Thanks

Sarun.M.N.

2

There should be effect at all in the functioning of your Scsi miniport
driver. First, scsiport serializes all miniport operations, so there is no
MP issue, and secondly hyperthreading just looks like MP to programs. You
should of course test on MP platforms, just to be safe.

=====================
Mark Roddy
Hollis Technology Solutions
www.hollistech.com
xxxxx@hollistech.com

-----Original Message-----
From: sarun [mailto:xxxxx@integramicro.com]
Sent: Monday, August 04, 2003 12:04 PM
To: Windows System Software Developers Interest List
Subject: [ntdev] Hyper Threading Technology

Hi All,

What all things I should keep in mind while designing a SCSI
miniport driver on a Intel P-4 processor with Hyper- Threading Technology? .
What are the differences between processor with hyper threading support and
multi processor? Whether my SCSI miniport driver that works perfectly fine
on a multi-processor machine will work properly on a hyper-threaded
processor?

Thanks

Sarun.M.N.

Questions? First check the Kernel Driver FAQ at
http://www.osronline.com/article.cfm?id=256

You are currently subscribed to ntdev as: xxxxx@stratus.com
To unsubscribe send a blank email to xxxxx@lists.osr.com

3

If you driver works multi-processor it will work fine on a hyper-threaded
system, Microsoft treats the hyper-threaded processors as 2 processors.

Don Burn (MVP, Windows DDK)
Windows 2k/XP/2k3 Filesystem and Driver Consulting

----- Original Message -----
From: sarun
To: Windows System Software Developers Interest List
Sent: Monday, August 04, 2003 12:04 PM
Subject: [ntdev] Hyper Threading Technology

Hi All,
What all things I should keep in mind while designing a SCSI
miniport driver on a Intel P-4 processor with Hyper- Threading Technology? .
What are the differences between processor with hyper threading support and
multi processor? Whether my SCSI miniport driver that works perfectly fine
on a multi-processor machine will work properly on a hyper-threaded
processor?
Thanks
Sarun.M.N.

Questions? First check the Kernel Driver FAQ at
http://www.osronline.com/article.cfm?id=256

You are currently subscribed to ntdev as: xxxxx@acm.org
To unsubscribe send a blank email to xxxxx@lists.osr.com

4

No difference between hyper-threading and SMP from the kernel mode driver point of view.

----- Original Message -----
From: sarun
To: Windows System Software Developers Interest List
Sent: Monday, August 04, 2003 8:04 PM
Subject: [ntdev] Hyper Threading Technology

Hi All,
What all things I should keep in mind while designing a SCSI miniport driver on a Intel P-4 processor with Hyper- Threading Technology? . What are the differences between processor with hyper threading support and multi processor? Whether my SCSI miniport driver that works perfectly fine on a multi-processor machine will work properly on a hyper-threaded processor?

Thanks

Sarun.M.N.

Questions? First check the Kernel Driver FAQ at http://www.osronline.com/article.cfm?id=256

You are currently subscribed to ntdev as: xxxxx@storagecraft.com
To unsubscribe send a blank email to xxxxx@lists.osr.com

5

This is going to be interesting. The latest P4s have hyperthreading, and there are a lot of drivers that have not been tested against a dual processor. We better have the method to disable hyperthreading handy.

–
Gary G. Little
Seagate Technologies, LLC
xxxxx@seagate.com
“sarun” wrote in message news:xxxxx@ntdev…
Hi All,
What all things I should keep in mind while designing a SCSI miniport driver on a Intel P-4 processor with Hyper- Threading Technology? . What are the differences between processor with hyper threading support and multi processor? Whether my SCSI miniport driver that works perfectly fine on a multi-processor machine will work properly on a hyper-threaded processor?

Thanks

Sarun.M.N.

6

Change the HAL to the uniprocessor one.

Maxim Shatskih, Windows DDK MVP
StorageCraft Corporation
xxxxx@storagecraft.com
http://www.storagecraft.com

----- Original Message -----
From: Gary G. Little
Newsgroups: ntdev
To: Windows System Software Developers Interest List
Sent: Wednesday, August 06, 2003 7:53 PM
Subject: [ntdev] Re: Hyper Threading Technology

This is going to be interesting. The latest P4s have hyperthreading, and there are a lot of drivers that have not been tested against a dual processor. We better have the method to disable hyperthreading handy.

–
Gary G. Little
Seagate Technologies, LLC
xxxxx@seagate.com
“sarun” wrote in message news:xxxxx@ntdev…
Hi All,
What all things I should keep in mind while designing a SCSI miniport driver on a Intel P-4 processor with Hyper- Threading Technology? . What are the differences between processor with hyper threading support and multi processor? Whether my SCSI miniport driver that works perfectly fine on a multi-processor machine will work properly on a hyper-threaded processor?

Thanks

Sarun.M.N.

—
Questions? First check the Kernel Driver FAQ at http://www.osronline.com/article.cfm?id=256

You are currently subscribed to ntdev as: xxxxx@storagecraft.com
To unsubscribe send a blank email to xxxxx@lists.osr.com

7

Try /numproc=1 on the boot.ini line.
“Maxim S. Shatskih” wrote in message news:xxxxx@ntdev…

Change the HAL to the uniprocessor one.

Maxim Shatskih, Windows DDK MVP
StorageCraft Corporation
xxxxx@storagecraft.com
http://www.storagecraft.com

----- Original Message -----
From: Gary G. Little
Newsgroups: ntdev
To: Windows System Software Developers Interest List
Sent: Wednesday, August 06, 2003 7:53 PM
Subject: [ntdev] Re: Hyper Threading Technology

This is going to be interesting. The latest P4s have hyperthreading, and there are a lot of drivers that have not been tested against a dual processor. We better have the method to disable hyperthreading handy.

–
Gary G. Little
Seagate Technologies, LLC
xxxxx@seagate.com
“sarun” wrote in message news:xxxxx@ntdev…
Hi All,
What all things I should keep in mind while designing a SCSI miniport driver on a Intel P-4 processor with Hyper- Threading Technology? . What are the differences between processor with hyper threading support and multi processor? Whether my SCSI miniport driver that works perfectly fine on a multi-processor machine will work properly on a hyper-threaded processor?

Thanks

Sarun.M.N.

—
Questions? First check the Kernel Driver FAQ at http://www.osronline.com/article.cfm?id=256

You are currently subscribed to ntdev as: xxxxx@storagecraft.com
To unsubscribe send a blank email to xxxxx@lists.osr.com

8

/PROCESSORS=1 if I recall correctly.
----- Original Message -----
From: Maxim S. Shatskih
To: Windows System Software Developers Interest List
Sent: Saturday, August 09, 2003 12:51 AM
Subject: [ntdev] Re: Hyper Threading Technology

Change the HAL to the uniprocessor one.

Maxim Shatskih, Windows DDK MVP
StorageCraft Corporation
xxxxx@storagecraft.com
http://www.storagecraft.com

----- Original Message -----
From: Gary G. Little
Newsgroups: ntdev
To: Windows System Software Developers Interest List
Sent: Wednesday, August 06, 2003 7:53 PM
Subject: [ntdev] Re: Hyper Threading Technology

This is going to be interesting. The latest P4s have hyperthreading, and there are a lot of drivers that have not been tested against a dual processor. We better have the method to disable hyperthreading handy.

–
Gary G. Little
Seagate Technologies, LLC
xxxxx@seagate.com
“sarun” wrote in message news:xxxxx@ntdev…
Hi All,
What all things I should keep in mind while designing a SCSI miniport driver on a Intel P-4 processor with Hyper- Threading Technology? . What are the differences between processor with hyper threading support and multi processor? Whether my SCSI miniport driver that works perfectly fine on a multi-processor machine will work properly on a hyper-threaded processor?

Thanks

Sarun.M.N.

—
Questions? First check the Kernel Driver FAQ at http://www.osronline.com/article.cfm?id=256

You are currently subscribed to ntdev as: xxxxx@storagecraft.com
To unsubscribe send a blank email to xxxxx@lists.osr.com
—
Questions? First check the Kernel Driver FAQ at http://www.osronline.com/article.cfm?id=256

You are currently subscribed to ntdev as: xxxxx@earthlink.net
To unsubscribe send a blank email to xxxxx@lists.osr.com

9

Nope, /numproc=1 or /onecpu work on boot.ini.

Don Burn (MVP, Windows DDK)
Windows 2k/XP/2k3 Filesystem and Driver Consulting

----- Original Message -----
From: Loren Wilton
To: Windows System Software Developers Interest List
Sent: Saturday, August 09, 2003 12:46 PM
Subject: [ntdev] Re: Hyper Threading Technology

/PROCESSORS=1 if I recall correctly.
----- Original Message -----
From: Maxim S. Shatskih
To: Windows System Software Developers Interest List
Sent: Saturday, August 09, 2003 12:51 AM
Subject: [ntdev] Re: Hyper Threading Technology

Change the HAL to the uniprocessor one.

Maxim Shatskih, Windows DDK MVP
StorageCraft Corporation
xxxxx@storagecraft.com
http://www.storagecraft.com

----- Original Message -----
From: Gary G. Little
Newsgroups: ntdev
To: Windows System Software Developers Interest List
Sent: Wednesday, August 06, 2003 7:53 PM
Subject: [ntdev] Re: Hyper Threading Technology

This is going to be interesting. The latest P4s have hyperthreading, and there are a lot of drivers that have not been tested against a dual processor. We better have the method to disable hyperthreading handy.

–
Gary G. Little
Seagate Technologies, LLC
xxxxx@seagate.com
“sarun” wrote in message news:xxxxx@ntdev…
Hi All,
What all things I should keep in mind while designing a SCSI miniport driver on a Intel P-4 processor with Hyper- Threading Technology? . What are the differences between processor with hyper threading support and multi processor? Whether my SCSI miniport driver that works perfectly fine on a multi-processor machine will work properly on a hyper-threaded processor?

Thanks

Sarun.M.N.

—
Questions? First check the Kernel Driver FAQ at http://www.osronline.com/article.cfm?id=256

You are currently subscribed to ntdev as: xxxxx@storagecraft.com
To unsubscribe send a blank email to xxxxx@lists.osr.com
—
Questions? First check the Kernel Driver FAQ at http://www.osronline.com/article.cfm?id=256

You are currently subscribed to ntdev as: xxxxx@earthlink.net
To unsubscribe send a blank email to xxxxx@lists.osr.com
—
Questions? First check the Kernel Driver FAQ at http://www.osronline.com/article.cfm?id=256

You are currently subscribed to ntdev as: xxxxx@acm.org
To unsubscribe send a blank email to xxxxx@lists.osr.com

10

This is all well and good boys and girls, but the real point of issue here is that are an unknown number of drivers out there that are going to fail because they were not tested on SMP boxes. As soon as hyper-threading is enabled on the systems with those drivers we will start hearing the same old songs “Well it worked on DOS” or “98 works fine” or “The old machine didn’t do that”; etc. etc. etc.

and who in the hell is still using HTML on this list!?!?!?

–
Gary G. Little
Seagate Technologies, LLC
xxxxx@seagate.com
“Don Burn” wrote in message news:xxxxx@ntdev…
Nope, /numproc=1 or /onecpu work on boot.ini.

Don Burn (MVP, Windows DDK)
Windows 2k/XP/2k3 Filesystem and Driver Consulting

----- Original Message -----
From: Loren Wilton
To: Windows System Software Developers Interest List
Sent: Saturday, August 09, 2003 12:46 PM
Subject: [ntdev] Re: Hyper Threading Technology

/PROCESSORS=1 if I recall correctly.
----- Original Message -----
From: Maxim S. Shatskih
To: Windows System Software Developers Interest List
Sent: Saturday, August 09, 2003 12:51 AM
Subject: [ntdev] Re: Hyper Threading Technology

Change the HAL to the uniprocessor one.

Maxim Shatskih, Windows DDK MVP
StorageCraft Corporation
xxxxx@storagecraft.com
http://www.storagecraft.com

----- Original Message -----
From: Gary G. Little
Newsgroups: ntdev
To: Windows System Software Developers Interest List
Sent: Wednesday, August 06, 2003 7:53 PM
Subject: [ntdev] Re: Hyper Threading Technology

This is going to be interesting. The latest P4s have hyperthreading, and there are a lot of drivers that have not been tested against a dual processor. We better have the method to disable hyperthreading handy.

–
Gary G. Little
Seagate Technologies, LLC
xxxxx@seagate.com
“sarun” wrote in message news:xxxxx@ntdev…
Hi All,
What all things I should keep in mind while designing a SCSI miniport driver on a Intel P-4 processor with Hyper- Threading Technology? . What are the differences between processor with hyper threading support and multi processor? Whether my SCSI miniport driver that works perfectly fine on a multi-processor machine will work properly on a hyper-threaded processor?

Thanks

Sarun.M.N.

—
Questions? First check the Kernel Driver FAQ at http://www.osronline.com/article.cfm?id=256

You are currently subscribed to ntdev as: xxxxx@storagecraft.com
To unsubscribe send a blank email to xxxxx@lists.osr.com
—
Questions? First check the Kernel Driver FAQ at http://www.osronline.com/article.cfm?id=256

You are currently subscribed to ntdev as: xxxxx@earthlink.net
To unsubscribe send a blank email to xxxxx@lists.osr.com
—
Questions? First check the Kernel Driver FAQ at http://www.osronline.com/article.cfm?id=256

You are currently subscribed to ntdev as: xxxxx@acm.org
To unsubscribe send a blank email to xxxxx@lists.osr.com

11

Way too many people still use HTML. I bet those driver writers didn’t mark
their driver as being not loadable on a SMP system. We did that with one
company until we had the hardware to test SMP. Still a dual CPU system,
while a 8x is better to validate most of the less likely conditions. I
think the driver just worked when we stopped marking it for single cpu
systems.

“Gary G. Little” wrote in message
news:xxxxx@ntdev…
This is all well and good boys and girls, but the real point of issue here
is that are an unknown number of drivers out there that are going to fail
because they were not tested on SMP boxes. As soon as hyper-threading is
enabled on the systems with those drivers we will start hearing the same old
songs “Well it worked on DOS” or “98 works fine” or “The old machine didn’t
do that”; etc. etc. etc.

and who in the hell is still using HTML on this list!?!?!?

–
Gary G. Little
Seagate Technologies, LLC
xxxxx@seagate.com
“Don Burn” wrote in message news:xxxxx@ntdev…
Nope, /numproc=1 or /onecpu work on boot.ini.

Don Burn (MVP, Windows DDK)
Windows 2k/XP/2k3 Filesystem and Driver Consulting

----- Original Message -----
From: Loren Wilton
To: Windows System Software Developers Interest List
Sent: Saturday, August 09, 2003 12:46 PM
Subject: [ntdev] Re: Hyper Threading Technology

/PROCESSORS=1 if I recall correctly.
----- Original Message -----
From: Maxim S. Shatskih
To: Windows System Software Developers Interest List
Sent: Saturday, August 09, 2003 12:51 AM
Subject: [ntdev] Re: Hyper Threading Technology

Change the HAL to the uniprocessor one.

Maxim Shatskih, Windows DDK MVP
StorageCraft Corporation
xxxxx@storagecraft.com
http://www.storagecraft.com

----- Original Message -----
From: Gary G. Little
Newsgroups: ntdev
To: Windows System Software Developers Interest List
Sent: Wednesday, August 06, 2003 7:53 PM
Subject: [ntdev] Re: Hyper Threading Technology

This is going to be interesting. The latest P4s have hyperthreading, and
there are a lot of drivers that have not been tested against a dual
processor. We better have the method to disable hyperthreading handy.

–
Gary G. Little
Seagate Technologies, LLC
xxxxx@seagate.com
“sarun” wrote in message news:xxxxx@ntdev…
Hi All,
What all things I should keep in mind while designing a SCSI
miniport driver on a Intel P-4 processor with Hyper- Threading Technology? .
What are the differences between processor with hyper threading support and
multi processor? Whether my SCSI miniport driver that works perfectly fine
on a multi-processor machine will work properly on a hyper-threaded
processor?
Thanks
Sarun.M.N.

—
Questions? First check the Kernel Driver FAQ at
http://www.osronline.com/article.cfm?id=256

You are currently subscribed to ntdev as: xxxxx@storagecraft.com
To unsubscribe send a blank email to xxxxx@lists.osr.com
—
Questions? First check the Kernel Driver FAQ at
http://www.osronline.com/article.cfm?id=256

You are currently subscribed to ntdev as: xxxxx@earthlink.net
To unsubscribe send a blank email to xxxxx@lists.osr.com
—
Questions? First check the Kernel Driver FAQ at
http://www.osronline.com/article.cfm?id=256

You are currently subscribed to ntdev as: xxxxx@acm.org
To unsubscribe send a blank email to xxxxx@lists.osr.com

12

EVERY driver should/MUST be designed with MP in mind. Have a dual proc test
machine sitting right beside your primary development machine. I you do not,
you have no business distributing your code to anyone!!!

Jamey

-----Original Message-----
From: xxxxx@lists.osr.com [mailto:xxxxx@lists.osr.com]
On Behalf Of Gary G. Little
Sent: Saturday, August 09, 2003 10:46 AM
To: Windows System Software Developers Interest List
Subject: [ntdev] Re: Hyper Threading Technology

This is all well and good boys and girls, but the real point of issue here
is that are an unknown number of drivers out there that are going to fail
because they were not tested on SMP boxes. As soon as hyper-threading is
enabled on the systems with those drivers we will start hearing the same old
songs “Well it worked on DOS” or “98 works fine” or “The old machine didn’t
do that”; etc. etc. etc.

and who in the hell is still using HTML on this list!?!?!?

–
Gary G. Little
Seagate Technologies, LLC
xxxxx@seagate.com

“Don Burn” wrote in message news:xxxxx@ntdev…

Nope, /numproc=1 or /onecpu work on boot.ini.

Don Burn (MVP, Windows DDK)
Windows 2k/XP/2k3 Filesystem and Driver Consulting

----- Original Message -----

From: Loren Wilton mailto:xxxxx

To: Windows System mailto:xxxxx Software Developers
Interest List

Sent: Saturday, August 09, 2003 12:46 PM

Subject: [ntdev] Re: Hyper Threading Technology

/PROCESSORS=1 if I recall correctly.

----- Original Message -----

From: Maxim S. mailto:xxxxx Shatskih

To: Windows System mailto:xxxxx Software Developers
Interest List

Sent: Saturday, August 09, 2003 12:51 AM

Subject: [ntdev] Re: Hyper Threading Technology

Change the HAL to the uniprocessor one.

Maxim Shatskih, Windows DDK MVP
StorageCraft Corporation
xxxxx@storagecraft.com
http://www.storagecraft.com

----- Original Message -----

From: Gary mailto:xxxxx G. Little

Newsgroups: ntdev

To: Windows System mailto:xxxxx Software Developers
Interest List

Sent: Wednesday, August 06, 2003 7:53 PM

Subject: [ntdev] Re: Hyper Threading Technology

This is going to be interesting. The latest P4s have hyperthreading, and
there are a lot of drivers that have not been tested against a dual
processor. We better have the method to disable hyperthreading handy.

–
Gary G. Little
Seagate Technologies, LLC
xxxxx@seagate.com

“sarun” wrote in message news:xxxxx@ntdev…

Hi All,

What all things I should keep in mind while designing a SCSI
miniport driver on a Intel P-4 processor with Hyper- Threading Technology? .
What are the differences between processor with hyper threading support and
multi processor? Whether my SCSI miniport driver that works perfectly fine
on a multi-processor machine will work properly on a hyper-threaded
processor?

Thanks

Sarun.M.N.

—
Questions? First check the Kernel Driver FAQ at
http://www.osronline.com/article.cfm?id=256

You are currently subscribed to ntdev as: xxxxx@storagecraft.com
To unsubscribe send a blank email to xxxxx@lists.osr.com

—
Questions? First check the Kernel Driver FAQ at
http://www.osronline.com/article.cfm?id=256

You are currently subscribed to ntdev as: xxxxx@earthlink.net
To unsubscribe send a blank email to xxxxx@lists.osr.com

—
Questions? First check the Kernel Driver FAQ at
http://www.osronline.com/article.cfm?id=256

You are currently subscribed to ntdev as: xxxxx@acm.org
To unsubscribe send a blank email to xxxxx@lists.osr.com

—
Questions? First check the Kernel Driver FAQ at
http://www.osronline.com/article.cfm?id=256

You are currently subscribed to ntdev as: xxxxx@storagecraft.com
To unsubscribe send a blank email to xxxxx@lists.osr.com</mailto:xxxxx></mailto:xxxxx></mailto:xxxxx></mailto:xxxxx></mailto:xxxxx></mailto:xxxxx>

13

From some the questions I have seen on this and other newsgroups, I suspect
that I would want some of those drivers to be linked with the “don’t load on
SMP” switch set. If they can’t even complete an IRP correctly, how can they
get SMP issues correct? I know programmers that can’t get reentrancy on a
single processor to work. Maybe they are really coders and bad ones at
that. They do make better managers, sometimes.

“Jamey Kirby” wrote in message news:xxxxx@ntdev…
EVERY driver should/MUST be designed with MP in mind. Have a dual proc test
machine sitting right beside your primary development machine. I you do not,
you have no business distributing your code to anyone!!!

Jamey

-----Original Message-----
From: xxxxx@lists.osr.com [mailto:xxxxx@lists.osr.com]
On Behalf Of Gary G. Little
Sent: Saturday, August 09, 2003 10:46 AM
To: Windows System Software Developers Interest List
Subject: [ntdev] Re: Hyper Threading Technology

This is all well and good boys and girls, but the real point of issue here
is that are an unknown number of drivers out there that are going to fail
because they were not tested on SMP boxes. As soon as hyper-threading is
enabled on the systems with those drivers we will start hearing the same old
songs “Well it worked on DOS” or “98 works fine” or “The old machine didn’t
do that”; etc. etc. etc.

and who in the hell is still using HTML on this list!?!?!?

–
Gary G. Little
Seagate Technologies, LLC
xxxxx@seagate.com
“Don Burn” wrote in message news:xxxxx@ntdev…
Nope, /numproc=1 or /onecpu work on boot.ini.

Don Burn (MVP, Windows DDK)
Windows 2k/XP/2k3 Filesystem and Driver Consulting
----- Original Message -----
From: Loren Wilton
To: Windows System Software Developers Interest List
Sent: Saturday, August 09, 2003 12:46 PM
Subject: [ntdev] Re: Hyper Threading Technology

/PROCESSORS=1 if I recall correctly.
----- Original Message -----
From: Maxim S. Shatskih
To: Windows System Software Developers Interest List
Sent: Saturday, August 09, 2003 12:51 AM
Subject: [ntdev] Re: Hyper Threading Technology

Change the HAL to the uniprocessor one.

Maxim Shatskih, Windows DDK MVP
StorageCraft Corporation
xxxxx@storagecraft.com
http://www.storagecraft.com
----- Original Message -----
From: Gary G. Little
Newsgroups: ntdev
To: Windows System Software Developers Interest List
Sent: Wednesday, August 06, 2003 7:53 PM
Subject: [ntdev] Re: Hyper Threading Technology

This is going to be interesting. The latest P4s have hyperthreading, and
there are a lot of drivers that have not been tested against a dual
processor. We better have the method to disable hyperthreading handy.

–
Gary G. Little
Seagate Technologies, LLC
xxxxx@seagate.com
“sarun” wrote in message news:xxxxx@ntdev…
Hi All,
What all things I should keep in mind while designing a SCSI
miniport driver on a Intel P-4 processor with Hyper- Threading Technology? .
What are the differences between processor with hyper threading support and
multi processor? Whether my SCSI miniport driver that works perfectly fine
on a multi-processor machine will work properly on a hyper-threaded
processor?
Thanks
Sarun.M.N.

—
Questions? First check the Kernel Driver FAQ at
http://www.osronline.com/article.cfm?id=256

You are currently subscribed to ntdev as: xxxxx@storagecraft.com
To unsubscribe send a blank email to xxxxx@lists.osr.com
—
Questions? First check the Kernel Driver FAQ at
http://www.osronline.com/article.cfm?id=256

You are currently subscribed to ntdev as: xxxxx@earthlink.net
To unsubscribe send a blank email to xxxxx@lists.osr.com
—
Questions? First check the Kernel Driver FAQ at
http://www.osronline.com/article.cfm?id=256

You are currently subscribed to ntdev as: xxxxx@acm.org
To unsubscribe send a blank email to xxxxx@lists.osr.com
—
Questions? First check the Kernel Driver FAQ at
http://www.osronline.com/article.cfm?id=256

You are currently subscribed to ntdev as: xxxxx@storagecraft.com
To unsubscribe send a blank email to xxxxx@lists.osr.com

14

(1) Note only do drivers need to run on SMP machines but they need to be
stressed - of course the processors the better for some stress conditions.
All types of IRPs processing in a stress condition to expose “windows of
opportunity” where the probability of code coverage due to who gets the
Spin Locks needs to be done and done in an unnatural very high stress
environments. This is testing as I see it now.
(2) Also I think code for drivers should be submitted along with getting a
driver digitally signed. Maybe there can be programs along with multiple
processors that can test drivers and test them even it they would run say
on 100000000 processors i.e making them mathematically sound. Techniques
for analysis of code running and looking the permutations that can occur
but indeed also look at the unnatural ones as well. Getting all the bugs
out is the goal.
Perhaps these techniques are more for mathematics and / or probability.
Given the success of verifier as a static analysis tool maybe we can
extend this endeavor into code coverage - perfmonance - and bug
elimination.
(3) These techniques could also be useful in application programming as
well.

15

Here we have a problem. Can you test a driver until it has no errors
remaining? The answer is NO, unless you have infinite test machines and
infinite time. So, that means NO. The design by developers who understand
the concepts of handling requests in a SMP environment is the only solution.
The hack writers generally fail early, but some keep testing and hacking at
the code until they can get it past the testers. Then, of course the
support calls start and as word spreads the sales fall and the supports
calls finally drop.

Design and having fellow developers who are competent helping with code and
design reviews are the only way to get it right. No matter how much
experience one person has, they will write purely “dumb” code some of the
time. Some people I know call that a “brain fart”. Even the authors of
most computer books have to issue errata and fixes to examples. Walter Oney
is one who issues them and keeps them available years after the book is a
collector’s item.

I don’t think it will ever be possible until we develop Asimov’s robots for
computers to check code, especially SMP type code as exists in drivers. Add
ISR, dispatch routines, DPCs, completion routines, PnP, and power (just to
name a few) to the mix and how can a computer find all the correct flow and
data access conflicts. It might be possible to force each activity to a
single thread process such as the floppy driver uses, but it would kill
performance, which is not an issue for the floppies. Firewire and USB need
drivers that can keep the pipes busy. I have discovered that while the
floppy chips have a possible 500kbps speed, the actual speed could only be
from 300-350kbps.

“William Michael Jones” wrote in message
news:xxxxx@ntdev…
>
> (1) Note only do drivers need to run on SMP machines but they need to be
> stressed - of course the processors the better for some stress conditions.
> All types of IRPs processing in a stress condition to expose “windows of
> opportunity” where the probability of code coverage due to who gets the
> Spin Locks needs to be done and done in an unnatural very high stress
> environments. This is testing as I see it now.
> (2) Also I think code for drivers should be submitted along with getting a
> driver digitally signed. Maybe there can be programs along with multiple
> processors that can test drivers and test them even it they would run say
> on 100000000 processors i.e making them mathematically sound. Techniques
> for analysis of code running and looking the permutations that can occur
> but indeed also look at the unnatural ones as well. Getting all the bugs
> out is the goal.
> Perhaps these techniques are more for mathematics and / or probability.
> Given the success of verifier as a static analysis tool maybe we can
> extend this endeavor into code coverage - perfmonance - and bug
> elimination.
> (3) These techniques could also be useful in application programming as
> well.
>
>

16

With Dell and others selling hyperthreaded 2x server boxes at entry level
prices of less than US$1,000, there is simply no excuse for not doing SMP
testing of drivers.

=====================
Mark Roddy
Hollis Technology Solutions
www.hollistech.com
xxxxx@hollistech.com

17

Correct me if I’m wrong, but isn’t proper multi-threaded design (using
sync objects) also (and automatically) enables SMP-safety? After all
(AFAIK) all the sync objects (not just spinlocks) provide SMP-sync as
well as MT-sync. Is Hyper Threading changing anything in this picture?

-----Original Message-----
From: Roddy, Mark [mailto:xxxxx@stratus.com]
Sent: Monday, August 11, 2003 7:35 AM
To: Windows System Software Developers Interest List
Subject: [ntdev] Re: Hyper Threading Technology

With Dell and others selling hyperthreaded 2x server boxes at entry
level
prices of less than US$1,000, there is simply no excuse for not doing
SMP
testing of drivers.

=====================
Mark Roddy
Hollis Technology Solutions
www.hollistech.com
xxxxx@hollistech.com

Questions? First check the Kernel Driver FAQ at
http://www.osronline.com/article.cfm?id=256

You are currently subscribed to ntdev as:
xxxxx@borland.com
To unsubscribe send a blank email to xxxxx@lists.osr.com

18

Sort of. It is more of a testing and code coverage issue. There are code
paths that are not ‘multi-threaded’ reentrant, but are ‘multi-processor’
reentrant, as in your driver lower edge interrupt driven code paths.
Finally, spinlocks are essentially NOPs on uniprocessors, so you are running
somewhat different code on these systems than on MP systems. You are very
unlikely to see a spinlock-order deadlock on a UP system.

=====================
Mark Roddy
Hollis Technology Solutions
www.hollistech.com
xxxxx@hollistech.com

-----Original Message-----
From: Vladimir Chtchetkine [mailto:xxxxx@borland.com]
Sent: Monday, August 11, 2003 11:11 AM
To: Windows System Software Developers Interest List
Subject: [ntdev] Re: Hyper Threading Technology

Correct me if I’m wrong, but isn’t proper multi-threaded design (using sync
objects) also (and automatically) enables SMP-safety? After all
(AFAIK) all the sync objects (not just spinlocks) provide SMP-sync as well
as MT-sync. Is Hyper Threading changing anything in this picture?

-----Original Message-----
From: Roddy, Mark [mailto:xxxxx@stratus.com]
Sent: Monday, August 11, 2003 7:35 AM
To: Windows System Software Developers Interest List
Subject: [ntdev] Re: Hyper Threading Technology

With Dell and others selling hyperthreaded 2x server boxes at entry level
prices of less than US$1,000, there is simply no excuse for not doing SMP
testing of drivers.

=====================
Mark Roddy
Hollis Technology Solutions
www.hollistech.com
xxxxx@hollistech.com

Questions? First check the Kernel Driver FAQ at
http://www.osronline.com/article.cfm?id=256

You are currently subscribed to ntdev as: xxxxx@borland.com
To unsubscribe send a blank email to xxxxx@lists.osr.com

Questions? First check the Kernel Driver FAQ at
http://www.osronline.com/article.cfm?id=256

You are currently subscribed to ntdev as: xxxxx@stratus.com To
unsubscribe send a blank email to xxxxx@lists.osr.com

19

You are right. But some kinds of programming errors don’t become operative
on single processor machines because code running at the same elevated IRQL
cannot interrupt each other there.
Now Hypertreading puts quasi multi processor machines into every nursery.
But the good news: work won’t be less :slight_smile:

-----Original Message-----
From: Vladimir Chtchetkine [mailto:xxxxx@borland.com]
Sent: Monday, August 11, 2003 5:11 PM
To: Windows System Software Developers Interest List
Subject: [ntdev] Re: Hyper Threading Technology

Correct me if I’m wrong, but isn’t proper multi-threaded design (using
sync objects) also (and automatically) enables SMP-safety? After all
(AFAIK) all the sync objects (not just spinlocks) provide SMP-sync as
well as MT-sync. Is Hyper Threading changing anything in this picture?

-----Original Message-----
From: Roddy, Mark [mailto:xxxxx@stratus.com]
Sent: Monday, August 11, 2003 7:35 AM
To: Windows System Software Developers Interest List
Subject: [ntdev] Re: Hyper Threading Technology

With Dell and others selling hyperthreaded 2x server boxes at entry
level
prices of less than US$1,000, there is simply no excuse for not doing
SMP
testing of drivers.

=====================
Mark Roddy
Hollis Technology Solutions
www.hollistech.com
xxxxx@hollistech.com

Questions? First check the Kernel Driver FAQ at
http://www.osronline.com/article.cfm?id=256

You are currently subscribed to ntdev as:
xxxxx@borland.com
To unsubscribe send a blank email to xxxxx@lists.osr.com

Questions? First check the Kernel Driver FAQ at
http://www.osronline.com/article.cfm?id=256

You are currently subscribed to ntdev as: xxxxx@baslerweb.com
To unsubscribe send a blank email to xxxxx@lists.osr.com

20

Then the second Q naturally arises :slight_smile: Can testing on HT boxes be
considered the same as testing on SMP boxes? After all, budget is
everything these times :slight_smile:

-----Original Message-----
From: Moebius, V. [mailto:xxxxx@baslerweb.com]
Sent: Monday, August 11, 2003 8:32 AM
To: Windows System Software Developers Interest List
Subject: [ntdev] Re: Hyper Threading Technology

You are right. But some kinds of programming errors don’t become
operative
on single processor machines because code running at the same elevated
IRQL
cannot interrupt each other there.
Now Hypertreading puts quasi multi processor machines into every
nursery.
But the good news: work won’t be less :slight_smile:

-----Original Message-----
From: Vladimir Chtchetkine [mailto:xxxxx@borland.com]
Sent: Monday, August 11, 2003 5:11 PM
To: Windows System Software Developers Interest List
Subject: [ntdev] Re: Hyper Threading Technology

Correct me if I’m wrong, but isn’t proper multi-threaded design (using
sync objects) also (and automatically) enables SMP-safety? After all
(AFAIK) all the sync objects (not just spinlocks) provide SMP-sync as
well as MT-sync. Is Hyper Threading changing anything in this picture?

-----Original Message-----
From: Roddy, Mark [mailto:xxxxx@stratus.com]
Sent: Monday, August 11, 2003 7:35 AM
To: Windows System Software Developers Interest List
Subject: [ntdev] Re: Hyper Threading Technology

With Dell and others selling hyperthreaded 2x server boxes at entry
level
prices of less than US$1,000, there is simply no excuse for not doing
SMP
testing of drivers.

=====================
Mark Roddy
Hollis Technology Solutions
www.hollistech.com
xxxxx@hollistech.com

Questions? First check the Kernel Driver FAQ at
http://www.osronline.com/article.cfm?id=256

You are currently subscribed to ntdev as:
xxxxx@borland.com
To unsubscribe send a blank email to xxxxx@lists.osr.com

Questions? First check the Kernel Driver FAQ at
http://www.osronline.com/article.cfm?id=256

You are currently subscribed to ntdev as: xxxxx@baslerweb.com
To unsubscribe send a blank email to xxxxx@lists.osr.com

Questions? First check the Kernel Driver FAQ at
http://www.osronline.com/article.cfm?id=256

You are currently subscribed to ntdev as:
xxxxx@borland.com
To unsubscribe send a blank email to xxxxx@lists.osr.com