showdown demnächts im anmarsch!!
msi weigert sich allem anschein nach die frage der lösbarkeit bzgl. der smbus und co probleme zu beseitigen! 3x nachgefragt sie ignorieren das und antworten dann nur bzgl. der expresscard!
heute wieder aw bekommen das sie nix machen können und audiointerface hersteller/ expresscard hersteller und microsoft schuld sind und sie natürlich nicht!!
ich kontaktier jetzt mal div medien und werde euch und auch die medien die ansprechpartner dafür rausrücken auf cc be meiner nächsten antwort die so lauten wird:
Thanks for checking into this but regarding the PCI to PCI bridge for WinXP this is not correct.
Windows and also the PCI Driver is not able to handle this as windows XP is not PCIE aware OS and has no resourcerebalancing built in.
So please also show the information inside this mail to your software engineer, as it looks like he did a mistake.
for sure he thought about Vista which is pcie aware and has resourcerebalancing built in.
I`ve also put the relevant Informations (from microsoft) regarding XP and PCI to PCI bridge configuration in genereal and when hotplugged inside this mail now.
There is the exact information about this.
As Windows XP has no Resourcerebalancing built in and is not able to configure devices which are behind hotplugged pci to pci bridges.
Please see pages 9-13 in attached documentation from microsoft:
Also you are able to allocate ressources & configure devices if inserted on Boot via Bios. You also allready do that when system boots.
Also the ones which are behind Rootport #1 are controlable via bios on boot.
Thats how PCIE and PCI Devices are in general configured @ Post. (Devices Main initialization)
If this wouldn`t be possible Computerhardware in general won`t work.
Furthermore again: it works with every other notebook and pc and also mac without problems also when hotplugged. (so additional to normal function when inserted before boot, where bios configures whole system and also all found pci to pci bridges, other vendors also found a fix via bios regarding whats written in the microsoft document, where microsoft states that xp can`t configure hotplugged pci to pci bridges while windows is running.)
so for xp both has to be configured from bios / bios has to provide config settings.
Please don`t tell me that it`s microsoft fault / audio / expresscard problem. We (also johannes > he tested with 4 different notebooks) have tested this with different notebooks.
If you also have possibility to get notebook from eg. sony with xp in your hands > insert fw expresscard and check for cacheline size on the firewire expresscards bridge, with pci register tool.
you will see that it`s also not 0 there.
I`m also in IT and Telco since many years and easiest way is always to say we are not, all the others are problem source.
* Every one in IT knows that allready
* It`s working with different nb`s without any problems.
I`ve seen this so often while working in and with big it companys in the last years, as i told you in pdf #2. but there we have penaltys for such things
I`ve provided all the necessary informations which are needed to track down the problem and to check how to built in the needed function.
For me this allways seems that msi don`t enough trust their customers, there are not only noobs without technical background out there which buy msi products.
With 3.48 the System configured the Expresscard with cacheline size 8. (Test Scenario exactly described in PDF#1 & #2. The test where it worked from itself.)
How do you think i have found that the problem is cacheline size?
>>> I compared snapshots from pci register tool with snapshots where it did not worked and then while it was not working configured the pci registers (from the working one snapshot) in realtime until it worked. and this was when i`ve set cacheline size to 8.
Please, why don`t you answer the questions i now send for the 4th time regarding all the other things like smbus and co.?
This are things which are wrong and faulty out of the box! So MSI sold Nb`S which are not 100% working because there are Errors reported by windows (devmgr) and also some shown in pci status registers.
Otherwise we users can`t see them. Look for SMBus in your Pr200device manger. Oh where is it`it`s? not there but as described in pdf #2 how to make it viewable > it shows that it has problems.
also acpi / motherboard resources show out of the box resource conflicts in there resource tabs >> Is MSI able to fix this or is it not? Please provide an answer. Yes or No?
And regarding Vendor of expresscard. Again: The expresscards have no problems! Have you read my last big mail?
It was allready also tested with SIIG 2 Port Fw 400 adapter. (certified from expresscard.org with the rabbit!)
This was done From two different Persons with two different Audio Interfaces.
+ different other expresscards and the problem is allways that expresscards got wrong config.
Also as written in pdf#1 where it got 8 in cachelinesize was only possbile with bios 3.48 > unistalled x3100 drivers and all that stuff.
but with 1.49 / 1.50 this did not work anymore.
There are no drivers for FW Expresscards pci to pci bridge / rom chip. Microsoft is provider for all FW-Card 400 drivers and also for PCI & PCI to PCI bridge drivers.
And as it works with every other notebook! There is no Problem with this.
This is also why you wrote me you don`t have seen such things regarding this in your last mail from 31.3.09. because there are no other notebook vendors with this problem.
otherwise the whole internet would be full with people complaining this stuff.
So again this is bios/pci bios issue. pci bios is responsible for device configuration. also additional it`s possible that pr200 acpi bios has problem.
(As written above this is all well documented in the attached .doc from microsoft. pages 9-13)
Resource Allocation in Windows XP
Windows XP, Windows Server 2003, and Windows 2000 are PCI hot plug aware. However, none of these operating systems natively handles the insertion or removal event request of PCI or PCIe hot plug. These operating systems support the PCI Hot-Plug Controller model of hot-plugging PCI add-in devices, and they also support PCI resource assignment when notified by a bus scan request from the ACPI hot-plug insertion event.
Problems with resource allocation occur when downstream PCIe and PCI bridges are discovered within the ExpressCard module after it is hot plugged.
To prevent this, the hot-plug module must be inserted before the system is powered on. As firmware enumerates the PCI or PCIe tree, resources are assigned to all functions that are present, thus enabling the previously nonfunctional devices.
If a PCI or PCIe hot-plug module is removed after firmware has enumerated and assigned resources to the hot-plugged bridge configuration registers, Windows XP does not alter the resources that were assigned to the system-board’s downstream hot-plug port. But if the same module is removed and re-inserted during the same power-up session, it demonstrates the same starvation of resources as described earlier.
Summary
Firmware manufacturers should follow these recommendations to support ExpressCard in firmware on systems that run Windows operating systems:
For systems that run Windows Vista, implement an _OSC method in firmware, as described in this paper.
For systems that run Windows XP and Windows 2000:
Define a unique table in the ACPI namespace for each PCIe port.
Define eject dependencies through an SB.EXCx method in the ACPI namespace.
Supply a platform-specific INF that declares a user-friendly name for each _HID entry in a per-manufacturer [Models] section.
To ensure adequate resource allocation for PCI bridges, firmware manufacturers should:
Design firmware to program the bridge base registers that support PCI and PCIe hot-plug insertion as part of the POST process, before the operating system loads.
please have a look to attached pdf: PCI-PCI_Bridges_When_Designing.pdf
there is exact info about cacheline size & pci to pci bridges and all about that:
The PCI BIOS is required to traverse the PCI bus hierarchy, and
identify and configure all devices connected to it. Instead of
assuming there is only a certain number of bus segments or a
particular bus structure (i.e. daisy-chained versus parallel or a
combination thereof), the BIOS should be generic, not written
for a special case.
During configuration, the BIOS should program the bridge
device’s Latency Timer registers and Cacheline Size register to a
non-zero value. Otherwise, system performance could be
severely hampered. For example, leaving the Latency Timer
register programmed to 0 could cause the bridge to disconnect
bursts after only one data phase. Although bridge devices do not
have a MIN_GNT register since they do not know what device
will be behind it, the Latency Timer register still needs to be
programmed.
So as you know it`s also not working when expresscard inserted before system startup and that is exact the case where bios is responsible for correct pci to pci bridge configuration.
Also from Microsoft:
For Windows 2000, Windows XP, and Windows Server 2003, hardware manufacturers should not assume that the operating system will reconfigure bridges at runtime to address specific memory requirements and I/O windows needed by child devices.
Hardware manufacturers should:
Understand resource requirements of devices and the hierarchical nature of resource allocation in Windows.
Design devices to use memory space rather than I/O space as much as possible, because memory space is much less constrained.
Understand how resources are allocated to their devices in cold boot and hot-plug scenarios.
Make sure the BIOS configures all bridges in the system as required for all child devices to receive sufficient resources.
Remember that PCI-to-PCI bridges do not have subsystem and subsystem vendor IDs, which prevents the operating system from applying workarounds based on the presence of a particular bridge in a particular subsystem or on a particular add-in card.
So again for XP it`s the bios responsibility to program the pci registers of expresscards pci to pci bridges. (the only well known expresscards with pci to pci bridges are firewire expresscards and expresscard which use the pcie part of the expresscard slot to cardbus adapters.)
if you have expresscard to cardbus adapter in your office (which uses the pcie part of the expresscard slot not the usb part!) i bet it won`t work with pr200 for 100% because it also has pcie to pci bridge inside.
and time ago i and also other people have allready contacted, motu/alesis, expresscard vendors, intel ( and also regrading the vga card bios update mentionend in pdf#2 > response from intel > only possbile via oem bios update).
Please give me an answer regarding the other questions. can msi fix the out of the box issues? > smbus/ acpi resource conflicts and all this stuff < yes or no)
thanks