How to disable resource rebalancing

Hi
I test non commercial (only Lab) host platform with internal lab USB device (with specific USB driver) and win10 on test mode.
Unfortunately , after my first write operation ( only first time after the boot), the specific USB device driver disappear from device manager and back after few seconds.
I’m pretty sure, it is because of resource re-balancing. because, on another lab host, disabling some PCI devices and USB devices , solves the issue, but on this host still happens.

Notes:

  1. I can’t change the driver (don;t have sources)
  2. re-balancing happens only once, after boot and then never again until next boot

Is there any way to disable the re-balancing for just this abnormal test ?
Or any ideas ?
Thank you

A resource rebalance should not make the device disappear. Capture a usb etw log and see what is happening at the bus layer. Sounds like your device is not responding properly on the wire.

Bent from my phone


From: xxxxx@lists.osr.com on behalf of xxxxx@hotmail.com
Sent: Tuesday, March 13, 2018 2:16:25 PM
To: Windows System Software Devs Interest List
Subject: [ntdev] How to disable resource rebalancing

Hi
I test non commercial (only Lab) host platform with internal lab USB device (with specific USB driver) and win10 on test mode.
Unfortunately , after my first write operation ( only first time after the boot), the specific USB device driver disappear from device manager and back after few seconds.
I’m pretty sure, it is because of resource re-balancing. because, on another lab host, disabling some PCI devices and USB devices , solves the issue, but on this host still happens.

Notes:
1. I can’t change the driver (don;t have sources)
2. re-balancing happens only once, after boot and then never again until next boot

Is there any way to disable the re-balancing for just this abnormal test ?
Or any ideas ?
Thank you


NTDEV is sponsored by OSR

Visit the list online at: https:

MONTHLY seminars on crash dump analysis, WDF, Windows internals and software drivers!
Details at https:

To unsubscribe, visit the List Server section of OSR Online at https:</https:></https:></https:>

On Mar 13, 2018, at 2:16 PM, xxxxx@hotmail.com wrote:
>
> I test non commercial (only Lab) host platform with internal lab USB device (with specific USB driver) and win10 on test mode.
> Unfortunately , after my first write operation ( only first time after the boot), the specific USB device driver disappear from device manager and back after few seconds.
> I’m pretty sure, it is because of resource re-balancing.

USB devices don’t need rebalancing.

> because, on another lab host, disabling some PCI devices and USB devices , solves the issue, but on this host still happens.

How many USB devices do you have? There is an extremely serious and little-known limitation on Intel host controllers that limits them to no more than 63 configured endpoints on the entire bus, and that includes endpoint 0 for all of the devices and hubs. (I’m not sure how they get away with that, since it violates the USB spec.) In a “rich” environment, it’s not difficult to bump up against that ceiling.

What kind of a device is it? Does your write operation trigger a bunch of activity? Is it possible you are violating the port’s power supply? I have certainly seen that cause devices to drop out.

Are you using high-quality USB cables? I’ve seen cables with slightly out-of-spec connectors cause this kind of headache.

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

thanks,
in fact my first thinking was that the device isn’t react properly to something, but if so, why this USB device is working well on other systems / platforms with all same configuration and devices?

The USB device just opening a thunderbolt port such that thunderbolt device that already connected to this port gets link with the platform . - at this moment the usb device disappear from the device manager and back again in few seconds.
Is this possible that once thunderbolt device make link , it somehow affect on USB device?

WHAT!? Seriously? Citation, please?

I’m not QUESTIONING this revelation… rather, I’m *shocked* to hear this, AND having just spent a few days with a client that has NINETY something USB devices on their system, this might be relevant to their “issues”… So I want to follow-up on this and understand the specifics of the limitation.

Thanks!

Peter
OSR
@OSRDrivers

xxxxx@osr.com wrote:

WHAT!? Seriously? Citation, please?

I’m not QUESTIONING this revelation… rather, I’m *shocked* to hear this, AND having just spent a few days with a client that has NINETY something USB devices on their system, this might be relevant to their “issues”… So I want to follow-up on this and understand the specifics of the limitation.

I, also, was shocked when I actually encountered this in the real world,
on a manufacturing test I was writing.  When Intel designed their EHCI
controllers a hundred years ago, they required that the BIOS allocate a
chunk of DOS memory for communicating between BIOS services and
operating system services.  That chunk of DOS memory has 4kB of state
info per endpoint, and at a fixed size of 512kB, allowed for 128
endpoints.  It was believed that would be plenty.

When they created their XHCI controllers, that “backwards compatibility”
bugaboo required that the memory chunk remain the same size, but now
each endpoint requires 8kB, meaning they can only hold 64.  It IS
possible to encounter this limitation in real world systems.

This is why Intel motherboards usually include at least two XHCI
controllers.

https://embedded.communities.intel.com/thread/7500


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

xxxxx@hotmail.com wrote:

in fact my first thinking was that the device isn’t react properly to something, but if so, why this USB device is working well on other systems / platforms with all same configuration and devices?

Electrical and timing problems can be tricky beasts, even on supposedly
identical hardware.  For example, say that one of your systems is at the
very edge of the voltage tolerances for the USB Vbus signal, and your
activity causes it to draw too much current.

The USB device just opening a thunderbolt port such that thunderbolt device that already connected to this port gets link with the platform . - at this moment the usb device disappear from the device manager and back again in few seconds.
Is this possible that once thunderbolt device make link , it somehow affect on USB device?

How are you “opening” the Thunderbolt port?  Could it be pulling current
from USB to do that?  Thunderbolt runs at pretty high speed, so board
layout is critical.  Have you taken steps to ensure you don’t have any
capacitive coupling between the Thunderbolt lines and the USB lines?

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

/aside

https://embedded.communities.intel.com/thread/7500

That is THE scariest thread I think that I’ve ever read. The inanity runs *really* deep. Not only do the Intel people *constantly* confuse/conflate the concepts of devices and endpoints, but the thread ends with “this information needs a non-disclosure agreement”…

An NDA. To be able to tell you WTF is going on… and that some Intel USB controllers only apparently support 64 endpoints each, or 128 endpoints, or… something?

I guess this isn’t a Big Deal because most computer have just a few USB devices. But I’m shocked… shocked. Shit. I have no idea what to tell my customer (who has happily had 91 USB devices working on a Windows XP system for ages). They were trying to tell me that they had trouble buying systems, and had to order some VERY SPECIFIC hardware for their system to work properly… I assumed they were insane. Now… Maybe not so much.

THANK YOU Mr. Roberts.

Peter
OSR
@OSRDrivers

Apologies to the OP for thread drift – If it can happen in the thread *I* started, it can happen in your thread too… I guess :wink:

Tim,
You speak about a long time ago .
today there’s only XHCI controller. even two controllers.
I think it is still 128, but I’ll check it, soon…

Regarding my question , can you help please guys ?
I do not understand why I have the USB driver disappear then back few sec later as a react to thunderbolt port activation ?
there are no too much devices on the system , mouse , keyboard , external hub to which my device connected …

Any suggestion where and how to start investigate this issue in order to get some info ?

As others already told you - it seems you have a hardware problem, not a driver-related one. So - connect USB bus analyzer to understand what is happening there.

Regards,
Alex Krol

-----Original Message-----
From: xxxxx@lists.osr.com [mailto:xxxxx@lists.osr.com] On Behalf Of xxxxx@hotmail.com
Sent: Monday, March 19, 2018 4:11 PM
To: Windows System Software Devs Interest List
Subject: RE:[ntdev] How to disable resource rebalancing

CAUTION: This email originated from outside of Kodak. Do not click links or open attachments unless you recognize the sender and know the content is safe.

Tim,
You speak about a long time ago .
today there’s only XHCI controller. even two controllers.
I think it is still 128, but I’ll check it, soon…

Regarding my question , can you help please guys ?
I do not understand why I have the USB driver disappear then back few sec later as a react to thunderbolt port activation ?
there are no too much devices on the system , mouse , keyboard , external hub to which my device connected …

Any suggestion where and how to start investigate this issue in order to get some info ?


NTDEV is sponsored by OSR

Visit the list online at: http:

MONTHLY seminars on crash dump analysis, WDF, Windows internals and software drivers!
Details at http:

To unsubscribe, visit the List Server section of OSR Online at http:</http:></http:></http:>

>thunderbolt port activation

Hmmmm… Thunderbolt support is, ah, complicated and often problematic.

Just google “Windows Thunderbolt” and read what comes up.

If you already know this, and your job is to do what you can to smooth out these differences, you likely have a complex task ahead of you. As Mr. Krol suggested, I would *start* with a USB Bus Analyzer.

Peter
OSR
@OSRDrivers

xxxxx@hotmail.com wrote:

You speak about a long time ago .
today there’s only XHCI controller. even two controllers.
I think it is still 128, but I’ll check it, soon…

I encountered this problem in state-of-the-art motherboards in 2016. 
It’s not a long time ago.  But it’s unlikely to be related to your problem.
 

Regarding my question , can you help please guys ?
I do not understand why I have the USB driver disappear then back few sec later as a react to thunderbolt port activation ?
there are no too much devices on the system , mouse , keyboard , external hub to which my device connected …

Any suggestion where and how to start investigate this issue in order to get some info ?

My first guess is voltage drop due to current draw.  Does the
Thunderbolt get powered up using bus power?


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