Checked build problem

Hi!

Does anyone have problem with installing WinXP checked build? Because I
have. Computer just reboots and reboots during driver load (blue screen
based at startup). I have downloaded image (En_WinXP_Pro_ISO.IMG ->
WXPCHK_EN) two times and tried to burn it with many different computers. I
have tried to install XP from CD (boot) and also tried to update from w2k to
XP. Nothing just seems to help.

Any ideas?

Jussi Rytilahti

Everybody got your question the first time; most are waking up right
about now…

Jussi Rytilahti wrote:

Hi!

Does anyone have problem with installing WinXP checked build? Because I
have. Computer just reboots and reboots during driver load (blue screen
based at startup). I have downloaded image (En_WinXP_Pro_ISO.IMG ->
WXPCHK_EN) two times and tried to burn it with many different computers. I
have tried to install XP from CD (boot) and also tried to update from w2k to
XP. Nothing just seems to help.

Any ideas?

Jussi Rytilahti


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

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

First check carefully, is the crash loading drivers or when the blue screen
says “Starting Windows”. It is extremely rare to crash loading drivers, but
once you start Windows this is more likely. If it is the latter, take a
look at my article http://www.osronline.com/article.cfm?id=264 on Driver
Installation at install. There is a good discussion of how to get WinDBG to
run with the blue screen windows, fire up the debugger and see where the
problem is.


Don Burn (MVP, Windows DDK)
Windows 2k/XP/2k3 Filesystem and Driver Consulting
Remove StopSpam from the email to reply

“Jussi Rytilahti” wrote in message
news:xxxxx@ntdev…
> Hi!
>
> Does anyone have problem with installing WinXP checked build? Because I
> have. Computer just reboots and reboots during driver load (blue screen
> based at startup). I have downloaded image (En_WinXP_Pro_ISO.IMG ->
> WXPCHK_EN) two times and tried to burn it with many different computers. I
> have tried to install XP from CD (boot) and also tried to update from w2k
> to
> XP. Nothing just seems to help.
>
> Any ideas?
>
> Jussi Rytilahti
>
>
>

Thanks Don,

Installation reboot first time after “Windows Setup” (blue screen based,
“Setup is loading file…”). After reboot installation asks: “Start Windows
2000… Or Windows XP setup” after that selecting “Windows XP setup”
installation goes to “Starting Windows…” and reboots.

I changed TXTSETUP.SIF ( from “OsLoadOptions = “/fastdetect /noguiboot
/nodebug”” to
“OsLoadOptions = “/fastdetect /noguiboot /baudrate=115200 /break””) I
modified image by using methods of http://www.nu2.nu/bootcd/.

Result : Nothing…No debug data to my debug-machine (running WinDbg at
“Kernel Debug”. Connected via COM ports (Target COM1) like it is running
normally)

Is there still something what I do not notice?

/Jussi

“Don Burn” kirjoitti viestissä:xxxxx@ntdev…
> First check carefully, is the crash loading drivers or when the blue
> screen says “Starting Windows”. It is extremely rare to crash loading
> drivers, but once you start Windows this is more likely. If it is the
> latter, take a look at my article
> http://www.osronline.com/article.cfm?id=264 on Driver Installation at
> install. There is a good discussion of how to get WinDBG to run with the
> blue screen windows, fire up the debugger and see where the problem is.
>
> –
> Don Burn (MVP, Windows DDK)
> Windows 2k/XP/2k3 Filesystem and Driver Consulting
> Remove StopSpam from the email to reply
>
>
>
>
> “Jussi Rytilahti” wrote in message
> news:xxxxx@ntdev…
>> Hi!
>>
>> Does anyone have problem with installing WinXP checked build? Because I
>> have. Computer just reboots and reboots during driver load (blue screen
>> based at startup). I have downloaded image (En_WinXP_Pro_ISO.IMG ->
>> WXPCHK_EN) two times and tried to burn it with many different computers.
>> I
>> have tried to install XP from CD (boot) and also tried to update from w2k
>> to
>> XP. Nothing just seems to help.
>>
>> Any ideas?
>>
>> Jussi Rytilahti
>>
>>
>>
>
>
>

Jussi,

This does sound like there is a problem in the actual startup. I know
of no way to debug this, perhaps the folks from Redmond on the list can
help.


Don Burn (MVP, Windows DDK)
Windows 2k/XP/2k3 Filesystem and Driver Consulting
Remove StopSpam from the email to reply

“Jussi Rytilahti” wrote in message
news:xxxxx@ntdev…
> Thanks Don,
>
> Installation reboot first time after “Windows Setup” (blue screen based,
> “Setup is loading file…”). After reboot installation asks: “Start
> Windows 2000… Or Windows XP setup” after that selecting “Windows XP
> setup” installation goes to “Starting Windows…” and reboots.
>
> I changed TXTSETUP.SIF ( from “OsLoadOptions = “/fastdetect /noguiboot
> /nodebug”” to
> “OsLoadOptions = “/fastdetect /noguiboot /baudrate=115200 /break””) I
> modified image by using methods of http://www.nu2.nu/bootcd/.
>
> Result : Nothing…No debug data to my debug-machine (running WinDbg at
> “Kernel Debug”. Connected via COM ports (Target COM1) like it is running
> normally)
>
> Is there still something what I do not notice?
>
> /Jussi
>
> “Don Burn” kirjoitti viestissä:xxxxx@ntdev…
>> First check carefully, is the crash loading drivers or when the blue
>> screen says “Starting Windows”. It is extremely rare to crash loading
>> drivers, but once you start Windows this is more likely. If it is the
>> latter, take a look at my article
>> http://www.osronline.com/article.cfm?id=264 on Driver Installation at
>> install. There is a good discussion of how to get WinDBG to run with the
>> blue screen windows, fire up the debugger and see where the problem is.
>>
>> –
>> Don Burn (MVP, Windows DDK)
>> Windows 2k/XP/2k3 Filesystem and Driver Consulting
>> Remove StopSpam from the email to reply
>>
>>
>>
>>
>> “Jussi Rytilahti” wrote in message
>> news:xxxxx@ntdev…
>>> Hi!
>>>
>>> Does anyone have problem with installing WinXP checked build? Because I
>>> have. Computer just reboots and reboots during driver load (blue screen
>>> based at startup). I have downloaded image (En_WinXP_Pro_ISO.IMG ->
>>> WXPCHK_EN) two times and tried to burn it with many different computers.
>>> I
>>> have tried to install XP from CD (boot) and also tried to update from
>>> w2k to
>>> XP. Nothing just seems to help.
>>>
>>> Any ideas?
>>>
>>> Jussi Rytilahti
>>>
>>>
>>>
>>
>>
>>
>
>
>

Jussi,

I use to have a certain box where every time I reloaded it, it would do
this same thing regardless of the OS version. Seemed to be some sort of
hardware conflict, removing all unnecessary peripherals seemed to do the
trick. The systems’ MB seemed a little unusual in design and some of the
peripherals were not on the HCL, seems as if the right (wrong) type of
hardware combination caused this in my case.

Try removing what you don’t need during install, once the OS is loaded
(if it will), add the peripherals back…

Jussi Rytilahti wrote:

Thanks Don,

Installation reboot first time after “Windows Setup” (blue screen based,
“Setup is loading file…”). After reboot installation asks: “Start Windows
2000… Or Windows XP setup” after that selecting “Windows XP setup”
installation goes to “Starting Windows…” and reboots.

I changed TXTSETUP.SIF ( from “OsLoadOptions = “/fastdetect /noguiboot
/nodebug”” to
“OsLoadOptions = “/fastdetect /noguiboot /baudrate=115200 /break””) I
modified image by using methods of http://www.nu2.nu/bootcd/.

Result : Nothing…No debug data to my debug-machine (running WinDbg at
“Kernel Debug”. Connected via COM ports (Target COM1) like it is running
normally)

Is there still something what I do not notice?

/Jussi

“Don Burn” kirjoitti viestissä:xxxxx@ntdev…
>
>
>>First check carefully, is the crash loading drivers or when the blue
>>screen says “Starting Windows”. It is extremely rare to crash loading
>>drivers, but once you start Windows this is more likely. If it is the
>>latter, take a look at my article
>>http://www.osronline.com/article.cfm?id=264 on Driver Installation at
>>install. There is a good discussion of how to get WinDBG to run with the
>>blue screen windows, fire up the debugger and see where the problem is.
>>
>>–
>>Don Burn (MVP, Windows DDK)
>>Windows 2k/XP/2k3 Filesystem and Driver Consulting
>>Remove StopSpam from the email to reply
>>
>>
>>
>>
>>“Jussi Rytilahti” wrote in message
>>news:xxxxx@ntdev…
>>
>>
>>>Hi!
>>>
>>>Does anyone have problem with installing WinXP checked build? Because I
>>>have. Computer just reboots and reboots during driver load (blue screen
>>>based at startup). I have downloaded image (En_WinXP_Pro_ISO.IMG ->
>>>WXPCHK_EN) two times and tried to burn it with many different computers.
>>>I
>>>have tried to install XP from CD (boot) and also tried to update from w2k
>>>to
>>>XP. Nothing just seems to help.
>>>
>>>Any ideas?
>>>
>>>Jussi Rytilahti
>>>
>>>
>>>
>>>
>>>
>>
>>
>>
>
>
>
>—
>Questions? First check the Kernel Driver FAQ at http://www.osronline.com/article.cfm?id=256
>
>You are currently subscribed to ntdev as: xxxxx@comcast.net
>To unsubscribe send a blank email to xxxxx@lists.osr.com
>
>
>

Hi,

I removed everything else than CD-drive.

Result: No effect to the problem.

This happends only with Checked XP build…

/Jussi

“MM” kirjoitti viestissä:xxxxx@ntdev…
> Jussi,
>
> I use to have a certain box where every time I reloaded it, it would do
> this same thing regardless of the OS version. Seemed to be some sort of
> hardware conflict, removing all unnecessary peripherals seemed to do the
> trick. The systems’ MB seemed a little unusual in design and some of the
> peripherals were not on the HCL, seems as if the right (wrong) type of
> hardware combination caused this in my case.
>
> Try removing what you don’t need during install, once the OS is loaded (if
> it will), add the peripherals back…
>
> Jussi Rytilahti wrote:
>
>>Thanks Don,
>>
>>Installation reboot first time after “Windows Setup” (blue screen based,
>>“Setup is loading file…”). After reboot installation asks: “Start
>>Windows 2000… Or Windows XP setup” after that selecting “Windows XP
>>setup” installation goes to “Starting Windows…” and reboots.
>>
>>I changed TXTSETUP.SIF ( from “OsLoadOptions = “/fastdetect /noguiboot
>>/nodebug”” to
>>“OsLoadOptions = “/fastdetect /noguiboot /baudrate=115200 /break””) I
>>modified image by using methods of http://www.nu2.nu/bootcd/.
>>
>>Result : Nothing…No debug data to my debug-machine (running WinDbg at
>>“Kernel Debug”. Connected via COM ports (Target COM1) like it is running
>>normally)
>>
>>Is there still something what I do not notice?
>>
>>/Jussi
>>
>>“Don Burn” kirjoitti viestissä:xxxxx@ntdev…
>>
>>>First check carefully, is the crash loading drivers or when the blue
>>>screen says “Starting Windows”. It is extremely rare to crash loading
>>>drivers, but once you start Windows this is more likely. If it is the
>>>latter, take a look at my article
>>>http://www.osronline.com/article.cfm?id=264 on Driver Installation at
>>>install. There is a good discussion of how to get WinDBG to run with the
>>>blue screen windows, fire up the debugger and see where the problem is.
>>>
>>>–
>>>Don Burn (MVP, Windows DDK)
>>>Windows 2k/XP/2k3 Filesystem and Driver Consulting
>>>Remove StopSpam from the email to reply
>>>
>>>
>>>
>>>
>>>“Jussi Rytilahti” wrote in message
>>>news:xxxxx@ntdev…
>>>
>>>>Hi!
>>>>
>>>>Does anyone have problem with installing WinXP checked build? Because I
>>>>have. Computer just reboots and reboots during driver load (blue screen
>>>>based at startup). I have downloaded image (En_WinXP_Pro_ISO.IMG ->
>>>>WXPCHK_EN) two times and tried to burn it with many different computers.
>>>>I
>>>>have tried to install XP from CD (boot) and also tried to update from
>>>>w2k to
>>>>XP. Nothing just seems to help.
>>>>
>>>>Any ideas?
>>>>
>>>>Jussi Rytilahti
>>>>
>>>>
>>>>
>>>>
>>>
>>>
>>
>>
>>
>>—
>>Questions? First check the Kernel Driver FAQ at
>>http://www.osronline.com/article.cfm?id=256
>>
>>You are currently subscribed to ntdev as: xxxxx@comcast.net
>>To unsubscribe send a blank email to xxxxx@lists.osr.com
>>
>>
>
>

uhh, that fixed it for me when I had the problem… If you can’t get
the full checked installed, remember you can always do a partial checked
install, the instructions are on the OSR website. A partial checked
install is probably all you need any how; it’s faster and pretty easy to
set up.

Jussi Rytilahti wrote:

Hi,

I removed everything else than CD-drive.

Result: No effect to the problem.

This happends only with Checked XP build…

/Jussi

“MM” kirjoitti viestissä:xxxxx@ntdev…
>
>
>>Jussi,
>>
>>I use to have a certain box where every time I reloaded it, it would do
>>this same thing regardless of the OS version. Seemed to be some sort of
>>hardware conflict, removing all unnecessary peripherals seemed to do the
>>trick. The systems’ MB seemed a little unusual in design and some of the
>>peripherals were not on the HCL, seems as if the right (wrong) type of
>>hardware combination caused this in my case.
>>
>>Try removing what you don’t need during install, once the OS is loaded (if
>>it will), add the peripherals back…
>>
>>Jussi Rytilahti wrote:
>>
>>
>>
>>>Thanks Don,
>>>
>>>Installation reboot first time after “Windows Setup” (blue screen based,
>>>“Setup is loading file…”). After reboot installation asks: “Start
>>>Windows 2000… Or Windows XP setup” after that selecting “Windows XP
>>>setup” installation goes to “Starting Windows…” and reboots.
>>>
>>>I changed TXTSETUP.SIF ( from “OsLoadOptions = “/fastdetect /noguiboot
>>>/nodebug”” to
>>>“OsLoadOptions = “/fastdetect /noguiboot /baudrate=115200 /break””) I
>>>modified image by using methods of http://www.nu2.nu/bootcd/.
>>>
>>>Result : Nothing…No debug data to my debug-machine (running WinDbg at
>>>“Kernel Debug”. Connected via COM ports (Target COM1) like it is running
>>>normally)
>>>
>>>Is there still something what I do not notice?
>>>
>>>/Jussi
>>>
>>>“Don Burn” kirjoitti viestissä:xxxxx@ntdev…
>>>
>>>
>>>
>>>>First check carefully, is the crash loading drivers or when the blue
>>>>screen says “Starting Windows”. It is extremely rare to crash loading
>>>>drivers, but once you start Windows this is more likely. If it is the
>>>>latter, take a look at my article
>>>>http://www.osronline.com/article.cfm?id=264 on Driver Installation at
>>>>install. There is a good discussion of how to get WinDBG to run with the
>>>>blue screen windows, fire up the debugger and see where the problem is.
>>>>
>>>>–
>>>>Don Burn (MVP, Windows DDK)
>>>>Windows 2k/XP/2k3 Filesystem and Driver Consulting
>>>>Remove StopSpam from the email to reply
>>>>
>>>>
>>>>
>>>>
>>>>“Jussi Rytilahti” wrote in message
>>>>news:xxxxx@ntdev…
>>>>
>>>>
>>>>
>>>>>Hi!
>>>>>
>>>>>Does anyone have problem with installing WinXP checked build? Because I
>>>>>have. Computer just reboots and reboots during driver load (blue screen
>>>>>based at startup). I have downloaded image (En_WinXP_Pro_ISO.IMG ->
>>>>>WXPCHK_EN) two times and tried to burn it with many different computers.
>>>>>I
>>>>>have tried to install XP from CD (boot) and also tried to update from
>>>>>w2k to
>>>>>XP. Nothing just seems to help.
>>>>>
>>>>>Any ideas?
>>>>>
>>>>>Jussi Rytilahti
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>
>>>>
>>>
>>>—
>>>Questions? First check the Kernel Driver FAQ at
>>>http://www.osronline.com/article.cfm?id=256
>>>
>>>You are currently subscribed to ntdev as: xxxxx@comcast.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@comcast.net
>To unsubscribe send a blank email to xxxxx@lists.osr.com
>
>
>

“MM” wrote in message news:xxxxx@ntdev…
> uhh, that fixed it for me when I had the problem… If you can’t get the
> full checked installed, remember you can always do a partial checked
> install, the instructions are on the OSR website. A partial checked
> install is probably all you need any how; it’s faster and pretty easy to
> set up.
>
Matt,

I really have to take exception with the partial build claims. Unless
you are doing a software only device driver there are capabilities that you
can only get with more than the partial. Yes, in theory one can start
stuffing in checked drivers to replace things such as ndis.sys, but when do
you stop. Interestingly, I have had drivers and software packages known for
their quality fail on a full checked build, a little investigation and a
real bug is found. Personally, I don’t consider a product stable till it
runs on the full checked build.


Don Burn (MVP, Windows DDK)
Windows 2k/XP/2k3 Filesystem and Driver Consulting
Remove StopSpam from the email to reply

Understood…

“have to take exception with the partial build claims” - don’t really
think I went too terribly overboard… The system does run faster,
takes less time to setup, and it is easy to do…

However, since your way more experienced than I, I do have to ask this.
I’ve been debugging small projects with partial checked
versions of the OS - all symbols are correct. Everything looks fine, are
you saying there could be some nasty bugs unseen in my
driver because I’m not using the FULL checked version? Is there some
voodoo I’m not aware of? (no sarcasm intended)

Matt

Don Burn wrote:

“MM” wrote in message news:xxxxx@ntdev…
>
>
>>uhh, that fixed it for me when I had the problem… If you can’t get the
>>full checked installed, remember you can always do a partial checked
>>install, the instructions are on the OSR website. A partial checked
>>install is probably all you need any how; it’s faster and pretty easy to
>>set up.
>>
>>
>>
>Matt,
>
> I really have to take exception with the partial build claims. Unless
>you are doing a software only device driver there are capabilities that you
>can only get with more than the partial. Yes, in theory one can start
>stuffing in checked drivers to replace things such as ndis.sys, but when do
>you stop. Interestingly, I have had drivers and software packages known for
>their quality fail on a full checked build, a little investigation and a
>real bug is found. Personally, I don’t consider a product stable till it
>runs on the full checked build.
>
>
>
>

A full checked build has ASSERT’s in almost all the drivers as well as
kernel and HAL. If this is something like an NDIS driver, this can be
critical since NDIS will sometimes assert when the kernel will not. Now
think of all the drivers that come into play in a system (use DeviceTree if
needed), don’t you want those driver’s ASSERT’s to be present.

The other thing to remember is all the diagnostic data the checked build
has. Microsoft only documents some of the logging you can get in knowledge
base articles. You can find some more of it in NTDDK.h, but even this is
not all of it. Right now I have a friend with a nasty power bug, it turns
out there is some tracing for power in the kernel, but it is not documented
anywhere.

Yes I use the partial checked build myself, I also mix and match adding
additional checked drivers at times. You are right that this is fast and
fairly easy. But I would not ship a driver to a customer if it did run
clean on a full checked build.


Don Burn (MVP, Windows DDK)
Windows 2k/XP/2k3 Filesystem and Driver Consulting
Remove StopSpam from the email to reply

“MM” wrote in message news:xxxxx@ntdev…
> Understood…
>
> “have to take exception with the partial build claims” - don’t really
> think I went too terribly overboard… The system does run faster,
> takes less time to setup, and it is easy to do…
>
> However, since your way more experienced than I, I do have to ask this.
> I’ve been debugging small projects with partial checked
> versions of the OS - all symbols are correct. Everything looks fine, are
> you saying there could be some nasty bugs unseen in my
> driver because I’m not using the FULL checked version? Is there some
> voodoo I’m not aware of? (no sarcasm intended)
>
> Matt
>
>
>
>
>
>
> Don Burn wrote:
>
>>“MM” wrote in message news:xxxxx@ntdev…
>>
>>>uhh, that fixed it for me when I had the problem… If you can’t get the
>>>full checked installed, remember you can always do a partial checked
>>>install, the instructions are on the OSR website. A partial checked
>>>install is probably all you need any how; it’s faster and pretty easy to
>>>set up.
>>>
>>>
>>Matt,
>>
>> I really have to take exception with the partial build claims.
>> Unless you are doing a software only device driver there are capabilities
>> that you can only get with more than the partial. Yes, in theory one can
>> start stuffing in checked drivers to replace things such as ndis.sys, but
>> when do you stop. Interestingly, I have had drivers and software
>> packages known for their quality fail on a full checked build, a little
>> investigation and a real bug is found. Personally, I don’t consider a
>> product stable till it runs on the full checked build.
>>
>>
>>
>
>

Maybe if the boys and girls at SMBios had followed that policy they would
not have a released driver in the field that throws an assert and fails in
someone else’s HCT test run.

“Compuware and SMBios … a new cash cow for a new year.” :slight_smile:

Gary G. Little

-----Original Message-----
From: xxxxx@lists.osr.com
[mailto:xxxxx@lists.osr.com] On Behalf Of Don Burn
Sent: Thursday, December 22, 2005 8:45 AM
To: Windows System Software Devs Interest List
Subject: Re:[ntdev] Checked build problem

A full checked build has ASSERT’s in almost all the drivers as well as
kernel and HAL. If this is something like an NDIS driver, this can be
critical since NDIS will sometimes assert when the kernel will not. Now
think of all the drivers that come into play in a system (use DeviceTree
if
needed), don’t you want those driver’s ASSERT’s to be present.

The other thing to remember is all the diagnostic data the checked build
has. Microsoft only documents some of the logging you can get in
knowledge
base articles. You can find some more of it in NTDDK.h, but even this is
not all of it. Right now I have a friend with a nasty power bug, it turns

out there is some tracing for power in the kernel, but it is not
documented
anywhere.

Yes I use the partial checked build myself, I also mix and match adding
additional checked drivers at times. You are right that this is fast and
fairly easy. But I would not ship a driver to a customer if it did run
clean on a full checked build.


Don Burn (MVP, Windows DDK)
Windows 2k/XP/2k3 Filesystem and Driver Consulting
Remove StopSpam from the email to reply

“MM” wrote in message news:xxxxx@ntdev…
> Understood…
>
> “have to take exception with the partial build claims” - don’t really
> think I went too terribly overboard… The system does run faster,
> takes less time to setup, and it is easy to do…
>
> However, since your way more experienced than I, I do have to ask this.
> I’ve been debugging small projects with partial checked
> versions of the OS - all symbols are correct. Everything looks fine, are

> you saying there could be some nasty bugs unseen in my
> driver because I’m not using the FULL checked version? Is there some
> voodoo I’m not aware of? (no sarcasm intended)
>
> Matt
>
>
>
>
>
>
> Don Burn wrote:
>
>>“MM” wrote in message news:xxxxx@ntdev…
>>
>>>uhh, that fixed it for me when I had the problem… If you can’t get
the
>>>full checked installed, remember you can always do a partial checked
>>>install, the instructions are on the OSR website. A partial checked
>>>install is probably all you need any how; it’s faster and pretty easy
to
>>>set up.
>>>
>>>
>>Matt,
>>
>> I really have to take exception with the partial build claims.
>> Unless you are doing a software only device driver there are
capabilities
>> that you can only get with more than the partial. Yes, in theory one
can
>> start stuffing in checked drivers to replace things such as ndis.sys,
but
>> when do you stop. Interestingly, I have had drivers and software
>> packages known for their quality fail on a full checked build, a little

>> investigation and a real bug is found. Personally, I don’t consider a
>> product stable till it runs on the full checked build.
>>
>>
>>
>
>


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

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

That you Mr. Burns…

“Right now I have a friend with a nasty power bug” - The Compuware bug?
I definatly see how a full checked version would help there…

I don’t mean to go way off subject (lying), but regarding power
managerment, is the KMDF all that it is cracked up to be; or have you not
yet had the chance to play with it?

Don Burn wrote:

A full checked build has ASSERT’s in almost all the drivers as well as
kernel and HAL. If this is something like an NDIS driver, this can be
critical since NDIS will sometimes assert when the kernel will not. Now
think of all the drivers that come into play in a system (use DeviceTree if
needed), don’t you want those driver’s ASSERT’s to be present.

The other thing to remember is all the diagnostic data the checked build
has. Microsoft only documents some of the logging you can get in knowledge
base articles. You can find some more of it in NTDDK.h, but even this is
not all of it. Right now I have a friend with a nasty power bug, it turns
out there is some tracing for power in the kernel, but it is not documented
anywhere.

Yes I use the partial checked build myself, I also mix and match adding
additional checked drivers at times. You are right that this is fast and
fairly easy. But I would not ship a driver to a customer if it did run
clean on a full checked build.

KMDF is a huge improvement in the entire area of PnP/Power Mangement as it
actually explicitly combines these two state machines into a single unified
consistent state machine that has a real good chance of being correct. Not
using KMDF for new PnP drivers (that, grrrrr, don’t need to get run on W2K
platforms) is silly.

=====================
Mark Roddy DDK MVP
Windows 2003/XP/2000 Consulting
Hollis Technology Solutions 603-321-1032
www.hollistech.com

-----Original Message-----
From: xxxxx@lists.osr.com
[mailto:xxxxx@lists.osr.com] On Behalf Of MM
Sent: Thursday, December 22, 2005 10:15 AM
To: Windows System Software Devs Interest List
Subject: Re: [ntdev] Checked build problem

That you Mr. Burns…

“Right now I have a friend with a nasty power bug” - The
Compuware bug?
I definatly see how a full checked version would help there…

I don’t mean to go way off subject (lying), but regarding
power managerment, is the KMDF all that it is cracked up to
be; or have you not yet had the chance to play with it?

Don Burn wrote:

>A full checked build has ASSERT’s in almost all the drivers
as well as
>kernel and HAL. If this is something like an NDIS driver,
this can be
>critical since NDIS will sometimes assert when the kernel
will not. Now
>think of all the drivers that come into play in a system
(use DeviceTree if
>needed), don’t you want those driver’s ASSERT’s to be present.
>
>The other thing to remember is all the diagnostic data the
checked build
>has. Microsoft only documents some of the logging you can
get in knowledge
>base articles. You can find some more of it in NTDDK.h, but
even this is
>not all of it. Right now I have a friend with a nasty power
bug, it turns
>out there is some tracing for power in the kernel, but it is
not documented
>anywhere.
>
>Yes I use the partial checked build myself, I also mix and
match adding
>additional checked drivers at times. You are right that
this is fast and
>fairly easy. But I would not ship a driver to a customer if
it did run
>clean on a full checked build.
>
>
>
>


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

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

We’re working on the win2k deployment story as we speak. Like I posted
previously, the community feedback is much appreciated in getting this
issue resolved.

d

-----Original Message-----
From: xxxxx@lists.osr.com
[mailto:xxxxx@lists.osr.com] On Behalf Of Mark Roddy
Sent: Thursday, December 22, 2005 7:38 AM
To: Windows System Software Devs Interest List
Subject: RE: [ntdev] Checked build problem

KMDF is a huge improvement in the entire area of PnP/Power Mangement as
it
actually explicitly combines these two state machines into a single
unified
consistent state machine that has a real good chance of being correct.
Not
using KMDF for new PnP drivers (that, grrrrr, don’t need to get run on
W2K
platforms) is silly.

=====================
Mark Roddy DDK MVP
Windows 2003/XP/2000 Consulting
Hollis Technology Solutions 603-321-1032
www.hollistech.com

-----Original Message-----
From: xxxxx@lists.osr.com
[mailto:xxxxx@lists.osr.com] On Behalf Of MM
Sent: Thursday, December 22, 2005 10:15 AM
To: Windows System Software Devs Interest List
Subject: Re: [ntdev] Checked build problem

That you Mr. Burns…

“Right now I have a friend with a nasty power bug” - The
Compuware bug?
I definatly see how a full checked version would help there…

I don’t mean to go way off subject (lying), but regarding
power managerment, is the KMDF all that it is cracked up to
be; or have you not yet had the chance to play with it?

Don Burn wrote:

>A full checked build has ASSERT’s in almost all the drivers
as well as
>kernel and HAL. If this is something like an NDIS driver,
this can be
>critical since NDIS will sometimes assert when the kernel
will not. Now
>think of all the drivers that come into play in a system
(use DeviceTree if
>needed), don’t you want those driver’s ASSERT’s to be present.
>
>The other thing to remember is all the diagnostic data the
checked build
>has. Microsoft only documents some of the logging you can
get in knowledge
>base articles. You can find some more of it in NTDDK.h, but
even this is
>not all of it. Right now I have a friend with a nasty power
bug, it turns
>out there is some tracing for power in the kernel, but it is
not documented
>anywhere.
>
>Yes I use the partial checked build myself, I also mix and
match adding
>additional checked drivers at times. You are right that
this is fast and
>fairly easy. But I would not ship a driver to a customer if
it did run
>clean on a full checked build.
>
>
>
>


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

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

Lack of w2k support is bad. Having promised but
revoking it later is evil.

I used to worked for company that was forced to form a
team to support NT4 and 9x for its latest h/w
technology which is not meant for legacy OSs, even
though support for these OS has been offically dropped
by MSFT for years. The fact is once some big accounts
ask for it, you have to do it.

If I was to develop a new PNP driver for mass
production devices, I will not use WDF unless w2k is
supported.

Calvin Guan (Windows DDK MVP)
NetXtreme Longhorn Miniport Prime
Broadcom Corp. www.broadcom.com

— Doron Holan wrote:

> We’re working on the win2k deployment story as we
> speak. Like I posted
> previously, the community feedback is much
> appreciated in getting this
> issue resolved.
>
> d
>
> -----Original Message-----
> From: xxxxx@lists.osr.com
> [mailto:xxxxx@lists.osr.com] On Behalf
> Of Mark Roddy
> Sent: Thursday, December 22, 2005 7:38 AM
> To: Windows System Software Devs Interest List
> Subject: RE: [ntdev] Checked build problem
>
> KMDF is a huge improvement in the entire area of
> PnP/Power Mangement as
> it
> actually explicitly combines these two state
> machines into a single
> unified
> consistent state machine that has a real good chance
> of being correct.
> Not
> using KMDF for new PnP drivers (that, grrrrr, don’t
> need to get run on
> W2K
> platforms) is silly.
>
> =====================
> Mark Roddy DDK MVP
> Windows 2003/XP/2000 Consulting
> Hollis Technology Solutions 603-321-1032
> www.hollistech.com
>
> > -----Original Message-----
> > From: xxxxx@lists.osr.com
> > [mailto:xxxxx@lists.osr.com] On
> Behalf Of MM
> > Sent: Thursday, December 22, 2005 10:15 AM
> > To: Windows System Software Devs Interest List
> > Subject: Re: [ntdev] Checked build problem
> >
> > That you Mr. Burns…
> >
> > “Right now I have a friend with a nasty power bug”
> - The
> > Compuware bug?
> > I definatly see how a full checked version would
> help there…
> >
> > I don’t mean to go way off subject (lying), but
> regarding
> > power managerment, is the KMDF all that it is
> cracked up to
> > be; or have you not yet had the chance to play
> with it?
> >
> >
> > Don Burn wrote:
> >
> > >A full checked build has ASSERT’s in almost all
> the drivers
> > as well as
> > >kernel and HAL. If this is something like an
> NDIS driver,
> > this can be
> > >critical since NDIS will sometimes assert when
> the kernel
> > will not. Now
> > >think of all the drivers that come into play in a
> system
> > (use DeviceTree if
> > >needed), don’t you want those driver’s ASSERT’s
> to be present.
> > >
> > >The other thing to remember is all the diagnostic
> data the
> > checked build
> > >has. Microsoft only documents some of the
> logging you can
> > get in knowledge
> > >base articles. You can find some more of it in
> NTDDK.h, but
> > even this is
> > >not all of it. Right now I have a friend with a
> nasty power
> > bug, it turns
> > >out there is some tracing for power in the
> kernel, but it is
> > not documented
> > >anywhere.
> > >
> > >Yes I use the partial checked build myself, I
> also mix and
> > match adding
> > >additional checked drivers at times. You are
> right that
> > this is fast and
> > >fairly easy. But I would not ship a driver to a
> customer if
> > it did run
> > >clean on a full checked build.
> > >
> > >
> > >
> > >
> >
> >
> > —
> > Questions? First check the Kernel Driver FAQ at
> > http://www.osronline.com/article.cfm?id=256
> >
> > You are currently subscribed to ntdev as:
> xxxxx@hollistech.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@microsoft.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: unknown
> lmsubst tag argument: ‘’
> To unsubscribe send a blank email to
> xxxxx@lists.osr.com
>

__________________________________________________________
Find your next car at http://autos.yahoo.ca

> ----------

From: xxxxx@lists.osr.com[SMTP:xxxxx@lists.osr.com] on behalf of Mark Roddy[SMTP:xxxxx@hollistech.com]
Reply To: Windows System Software Devs Interest List
Sent: Thursday, December 22, 2005 4:37 PM
To: Windows System Software Devs Interest List
Subject: RE: [ntdev] Checked build problem

KMDF is a huge improvement in the entire area of PnP/Power Mangement as it
actually explicitly combines these two state machines into a single unified
consistent state machine that has a real good chance of being correct.

Unfortunately, I have a very simple USB driver written using KMDF which can lock up on surprise removal. It doesn’t do anything special and it is probably OS USB drivers bug. On the other hand, without KMDF I’m able to make workaround which decreases probability of this scenario at least in order of magnitude. Customers don’t care about state machine correctness but they do care if their machine locks up during suspend or hibernate just because they detached USB device some time before.

Not using KMDF for new PnP drivers (that, grrrrr, don’t need to get run on W2K
platforms) is silly.

Are KMDF sources available? I didn’t have time to check it, yet. If not, I’d say using framework without sources is silly (see Gary’s mail about Compuware).

Best regards,

Michal Vodicka
UPEK, Inc.
[xxxxx@upek.com, http://www.upek.com]

Calvin Guan wrote:

I used to worked for company that was forced to form a
team to support NT4 and 9x for its latest h/w
technology which is not meant for legacy OSs, even
though support for these OS has been offically dropped
by MSFT for years. The fact is once some big accounts
ask for it, you have to do it.

Amen. As late as 2001, S3 was still doing NT 3.51 drivers for their new
graphics products. Apparently, there is a large base of ATM software
written for the OS/2 subsystem in NT, certified only on NT 3.51. The
client wanted to make absolutely sure that it’s vendors were still
committed to the platform.


Tim Roberts, xxxxx@probo.com
Providenza & Boekelheide, Inc.

IMHO, this is all exactly right. We all know for a fact that MS code has
bugs. Unless they had some programming superhero write KMDF code, it also
has bugs. It’s impossible for me to introduce bugs that can’t be fixed into
my code, even if I have a cool PnP state machine as a compensation. Plus, of
course, this W2K problem on its own is serious enough to preclude using it.
Publishing KMDF code would also take care of the W2K problem - people would
figure out how to fix it to support W2K.

BTW, AFAIK, Compuware supplies source code for their framework, so Gary can
fix it.

“Michal Vodicka” wrote in message
news:xxxxx@ntdev…
> ----------
> From:
> xxxxx@lists.osr.com[SMTP:xxxxx@lists.osr.com]
> on behalf of Mark Roddy[SMTP:xxxxx@hollistech.com]
> Reply To: Windows System Software Devs Interest List
> Sent: Thursday, December 22, 2005 4:37 PM
> To: Windows System Software Devs Interest List
> Subject: RE: [ntdev] Checked build problem
>
> KMDF is a huge improvement in the entire area of PnP/Power Mangement as it
> actually explicitly combines these two state machines into a single
> unified
> consistent state machine that has a real good chance of being correct.
>
Unfortunately, I have a very simple USB driver written using KMDF which can
lock up on surprise removal. It doesn’t do anything special and it is
probably OS USB drivers bug. On the other hand, without KMDF I’m able to
make workaround which decreases probability of this scenario at least in
order of magnitude. Customers don’t care about state machine correctness but
they do care if their machine locks up during suspend or hibernate just
because they detached USB device some time before.

> Not using KMDF for new PnP drivers (that, grrrrr, don’t need to get run on
> W2K
> platforms) is silly.
>
Are KMDF sources available? I didn’t have time to check it, yet. If not, I’d
say using framework without sources is silly (see Gary’s mail about
Compuware).

Best regards,

Michal Vodicka
UPEK, Inc.
[xxxxx@upek.com, http://www.upek.com]

“Ivan Bublikov” wrote in message
news:xxxxx@ntdev…
> IMHO, this is all exactly right. We all know for a fact that MS code has
> bugs. Unless they had some programming superhero write KMDF code, it also
> has bugs. It’s impossible for me to introduce bugs that can’t be fixed
> into my code, even if I have a cool PnP state machine as a compensation.
> Plus, of course, this W2K problem on its own is serious enough to preclude
> using it. Publishing KMDF code would also take care of the W2K problem -
> people would figure out how to fix it to support W2K.
>
> BTW, AFAIK, Compuware supplies source code for their framework, so Gary
> can fix it.
>
Well, you still have the problem, you don’t have source code to the things
that Microsoft supplies in the current environment, i.e. the kernel, the
support drivers, etc. Lets remember that the folks from Redmond who post
here have been fighting for us. It is just the fact that like all large
companies there are a certain number of “pointy haired” bosses, who get in
the way of things such as Windows 2000 support and a release of the source
code. Note: I don’t think they ever said that the source code we might get
would be buildable, just a reference folks.

You are right about Compuware supplying source code, but you are wrong about
Gary fixing it. The last time I did a static code analysis of the crap all
of the tools failed with “too many errors”, and a lot of the complaints were
not the nitpicking, they were things like using a pointer without
initializing it!


Don Burn (MVP, Windows DDK)
Windows 2k/XP/2k3 Filesystem and Driver Consulting
Remove StopSpam from the email to reply