enabling tcp segmentation offload leads to poorer perfomance

TheMSsForum.com: The Microsoft Software Forum

The MSS Forum ‹ Drivers
- Archive
  - Biz
  - MCSE
  - CRM
  - Drivers
  - Framework
  - ADO
  - ASP
  - Compact
  - Forms
  - Dotnet
  - C#
  - VB
  - FontpageGen
  - Excel
  - WorkSheet
  - Exchange
  - Setup
  - Fox
  - Fontpage
  - ASP
  - IIS
  - Entourage
  - Money
  - Messanger
  - PocketPC
  - Powerpoint
  - Project
  - Publisher
  - Excel
  - VB
  - Security
  - Portal
  - Services
  - SQLServerDev
  - SVCS
  - SQLServer
  - VB
  - VC
  - MFC
  - ExcelGen

- Previous
  - 1
    - build error with KdPrintEx Hi, I am using KdPrintEx in my driver but i keeep getting the 'identifier not found' error. I have included ntddk.h where it is supposed to be defined. I am building teh driver using 'windows checked build environment for XP'. I have looked into teh sample DDK code 'cdrom' which uses KdPrintEx and I see no difference in the way its used in sample and my driver. The cdrom sample builds without errors. Please let me know what mistake I might be making. Regds. Nitin Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75639
  - 2
    - IoIsWdmVersionAvailable Hi, I would like to use IoIsWdmVersionAvailable(IN UCHAR MajorVersion, IN UCHAR Minor Version) call to check whether the driver is running in Vista. Is the MajorVersion and MinorVersion defined for Vista? Thanks KS Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75636
  - 3
    - multi-function USB drivers with IAD Hi.. I'm trying to get a multi-function USB device to talk with windows properly. One of the functions is RNDIS, which has 2 interfaces, along with a custom function with 1 interface. In my config, RNDIS is using interfaces 0 and 1, while the custom function uses interface 2. In the config descriptor, I use this IAD descriptor before the RNDIS interface (but not one before the custom interface): 08 0B 00 02 02 02 FF 09 The problem is that when I plug in the device, windows (XP SP2) doesn't detect it as a composite device and add the MI_xx field onto the id that it compares to in my .inf file. I set up a dummy config with 2 functions that had only 1 interface apiece and the composite driver loaded correctly, so my guess would be that the IAD is not being set up properly. Is IAD support in XP SP2? Are there debug logs I can look at to see what is going on? Do I have to set up my .inf file to load the composite driver somehow? thanks for your help, Ben Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75628
  - 4
    - How to get ACPI notifications I am writing an ACPI device driver. I have search/study for a long time to implement the WDM of getting the ACPI event which is trigged by the BIOS ASL code notify (, 80) and notify (, 81). Which approach I should try in order to get these ACPI notifications? Do you get any sample code, opinion or comment? Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75627
  - 5
    - What is the latest available DDK? Hi driver experts, What is the latest version of DDK? Is the DDK for future MS OS (Long horn?) available for download? Thanks. Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75624
  - 6
    - Driver installation signing issues This is a multi-part message in MIME format. ------=_NextPart_000_0007_01C62C07.E00931D0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Hello, I have a driver that I am attempting to create an installer for using DIFx. The driver has passed WHQL certification but I have since updated it. My inf file references a cat file named the same as my driver. This originally was the cat file supplied from WHQL that I just resigned using my Authenticode cert. I then tried to create a new cat file using makecat but I get the same results. When trying to install I get an error on the inf stating "Error code 0x57: The parameter is incorrect.) Searching MSDN and Google doesn't come up with anything useful. Anyone know what this is or what I might be doing wrong? Any help would be greatly appreciated. Thanks, -- Sean McGinnis ------=_NextPart_000_0007_01C62C07.E00931D0 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN"> <HTML><HEAD> <META http-equiv=3DContent-Type content=3D"text/html; = charset=3Diso-8859-1"> <META content=3D"MSHTML 6.00.2900.2802" name=3DGENERATOR> <STYLE></STYLE> </HEAD> <BODY bgColor=3D#ffffff> <DIV><FONT face=3DArial>Hello,</FONT></DIV> <DIV><FONT face=3DArial></FONT> </DIV> <DIV><FONT face=3DArial>I have a driver that I am attempting to create = an=20 installer for using DIFx. The driver has passed WHQL certification but I = have=20 since updated it. My inf file references a cat file named the same as my = driver.=20 This originally was the cat file supplied from WHQL that I just resigned = using=20 my Authenticode cert. I then tried to create a new cat file using = makecat but I=20 get the same results. When trying to install I get an error on the inf = stating=20 "Error code 0x57: The parameter is incorrect.) Searching MSDN and Google = doesn't=20 come up with anything useful. </FONT></DIV> <DIV><FONT face=3DArial></FONT> </DIV> <DIV><FONT face=3DArial>Anyone know what this is or what I might be = doing wrong?=20 Any help would be greatly appreciated.</FONT></DIV> <DIV><FONT face=3DArial></FONT> </DIV> <DIV><FONT face=3DArial>Thanks,<BR>-- <BR>Sean = McGinnis</FONT></DIV></BODY></HTML> ------=_NextPart_000_0007_01C62C07.E00931D0-- Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75623
  - 7
    - Re-using a user mode DLL for implementing a HID joystick? Hello, I'm just playing around with an USB interface card ( http://www.velleman.be/ot/en/product/view/?id=351980 ). This interface card is controlled via calls to an (user mode) DLL (K8055D.DLL) for setting and getting the analogue and digital output/input. My idea is to use this card as a base for a custom build flight sim yoke. As far as I'm understanding the DDK docs, I will have to write an HID minidriver to get DirectInput to "see" the interface card as joystick device. Problem: The minidriver is a kernel mode driver. The provided DLL is user mode. So, I assume using the high-level (and comfortable) DLL calls out of the minidriver is not possible, isn't it? Is there a other solution than reverse-engineer and re-implement the low-level DLL to interface card USB communication to pretend the card to be a joystick for windows? Kind Regards, Horst. Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75611
  - 8
    - ACPI and Thermal zones Hi I try to get the data of the ACPI thermal zones. Currently I don't know how. As far as I believe I must access the acpi.sys driver "in a way" to get the data. Unfortunately I dont know how. When I searched the web for references I found a program named "mobilemeter". This programm accesses the ACPI in a way and displays the thermal zone. It looks like this programm accesses the ACPI driver since no driver is shipped with this program... Hope anybody has a idea or example. I use VC++ for development... Thnaks in advance P. Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75607
  - 9
    - netvmini sample bug? Hey, I just experienced a bugcheck in my NDIS driver in my MiniportInitialize() function. I had been inspired by the netvmini sample's use of the two macros MP_INC_REF() and MP_DEC_REF(). The latter calls NdisSetEvent() whenever the reference count has reached zero. However, this results in a problem in the MiniportInitialize() function. In the sample, one of the last things done in the function is to call NdisInitializeEvent(). But if initialization fails at an earlier point, NdisInitializeEvent() is never executed and when MP_DEC_REF() is called to clean-up memory a bugcheck occurs (as the event hasn't been initialized). What's up with that? Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75602
  - 10
    - query regarding printer driver hi i am a software engineer & i am working on a UNIDRV printer driver using Microsoft plug-ins IPRINTOEMUI & IPRINTOEMUNI . I have some queries: 1. how can i reduce the size of TIFF image? 2. how can i get the name of Document from which i give a print command. ex: suppose i give a print command from Microsoft Word document name tmp.doc so i want to get the name (tmp.doc & Microsoft word) using functions & structures given with above mentioned plug-ins? 3. how i can implement TIFFSetField(tiff *, TIFFTAG_COMPRESSION,7). when i am trying this i am not able to build the TIFF file.? 4. how i can implement TIFFSetField(tiff *, TIFFTAG_COLORMAP,???).. what is the value of ?? in TIFFSetField function/ 5.how can i get the typeface of all the FONTS used in a document from which i give print command. ex: suppose i give a print command from a document named tmp.doc & i am using Times New Roman, Arial, Gothica fonts in that document so i want to know the name of these fonts using functions & structures given with above mentioned plug-ins? regards Himanshu himanshu.agar...@netcreativemind.com Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75601
  - 11
    - PnPAddDevice Return Value Hello Everyone, I am trying to return a error value from PnPAddDevice routine of my PCI Device driver for the system to not load the driver for that particular device. I am trying with STATUS_NOT_SUPPORTED. I see that the AddDevice keeps getting called. The background for this trial is as follows: My driver is a PnP Driver and there is a legacy driver available for the same PCI device with different functionality. The system has 10 similar PCI devices and I want to load my driver only on two of them. I am using the PCI three-tuples BUS-DEVICE-FUNCTION for choosing my devices. But, for automatic installation we have to use the PCI VendorID and Device ID which is same for all the 10 PCI devices. If I try automatic installation, the system tries to load my driver for all the 10 PCI devices. So, I check for specific BUS-DEVICE-FUNCTION in my AddDevice and return STATUS_NOT_SUPPORTED. I expect the system to ignore my driver and allow the legacy driver to use the device. But, it does not seem to happen and my AddDevice keeps getting called. Is there any specific STATUS_* value I have to return from AddDevice to achieve what I want? Please let me know the value or point me to the place where I can get it from. Thanks in advance for the big help from all of you Best Regards, Raj Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75600
  - 12
    - Question about IoAllocateIrp and IoInitializeIrp Hi all, the DDk document say that "Do not use IoInitializeIrp to initialize the IRP before its first use". If I misuse IoInitializeIrp next to IoAllocateIrp will cause any problem ? Is it possible to cause system crash ? Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75594
  - 13
    - Pre-loading USB Device Drivers I have a client who has created a USB device. The chipset vendor of the USB chipset provides a set of driver files. The drivers install correctly and everything is working so far. The customer would like me to have the primary software installation program pre-load the USB drivers so that if the user plugs in the device, Windows will automatically install the drivers without prompting the user for the CDROM, etc. The installation files that I have for the USB chipset is a series of INF files, dlls, and device drivers. Can anyone give me an idea of how I can pre-load these drivers on the user system to meet the customer's goal? Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75586
  - 14
    - CSQ and IoMarkIrpPending Hi, I had an impression that I don't need to call IoMarkIrpPending before IoCsqInsertIrp, as the latter internally marks the inserted IRP pending anyway. Is this correct? Thanks -- Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75584
  - 15
    - determine x64 Hi, I need to determine which version of my driver to load depending of if it is a 32-bit or 64 bit OS. What's a reliable way to determine if I am running on some x64 version of windows. I could always check the OS version with GetVersionEx but that doesn't seem reliable. Thanks Hua-Ying Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75579
  - 16
    - SetupDiCallClassInstaller getting error 0xE000023B on DIF_REMOVE My code uninstalls an NDIS driver with SetupDiCallClassInstaller(DIF_REMOVE,..), much as in the DDK example. On Windows 2003 Server, I get an error 0xE000023B, and can't figure out what this means. The nearest error I could find was STATUS_CONNECTION_ACTIVE (0xC000023B), but that doesn't seem to make sense-- I tried disabling my driver and still get the same error. The same code works fine on Windows 2000 and XP. Anyone know what this error means or what might be causing it? Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75577
  - 17
    - Device Co Installer and Microsft In Box Drivers I have developed a Device coinstaller that utilizes in box MS drivers based off the DDK toaster sample. Since I cannot redistribute in box drivers how to I have my co installer point to the in box driver without the Files needed box showing up? I noticed if I copy the driver to my distribution folder and tell the installer the location I do not recieve the Files needed box anymore. But leaving the driver in the system32 drivers directory and pointing the driverpath to that directory I get the box Any suggestions? Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75576
  - 18
    - Windows VISTA and SCSI drivers Does anyone possess or know where to find Adaptec SCSI drivers for Vista? -- Dennis @ H-P Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75575
  - 19
    - Sending Irp from SCSI miniport driver to bus driver I have a bus driver similar to the toaster bus driver from the DDK. I create a PDO (by using an application to send IOCTL_BUSENUM_PLUGIN_HARDWARE), and then I load my SCSI miniport driver onto that PDO. My bus driver exports an interface that the SCSI miniport driver uses. In the SCSI miniport code, I want to send IRP_MN_QUERY_INTERFACE to the child PDO that corresponds to the PDO on which the miniport driver is installed. In the miniport code, I have been sending the IRP by using the code below, where targetObject is the DEVICE_OBJECT field of the DRIVER_OBJECT parameter passed to the miniport's DriverEntry function. I believe that targetObject would then correspond to the FDO (not the child PDO that I want), but, nevertheless, this was working with no problems at all. However, I recently switched to another test PC, and now I get Bug Check 0x35 (NO_MORE_IRP_STACK_LOCATIONS) on the call to IoCallDriver. 1) Any idea why I get this Bug Check? 2) Is there a way for the miniport driver to get the PDO pointer that I need? KeInitializeEvent(&event, NotificationEvent, FALSE); irp = IoBuildSynchronousFsdRequest(IRP_MJ_PNP, targetObject, NULL, 0, NULL, &event, &ioStatusBlock); if(NULL == irp) { status = STATUS_INSUFFICIENT_RESOURCES; goto End; } irpStack = IoGetNextIrpStackLocation(irp); irpStack->MinorFunction = IRP_MN_QUERY_INTERFACE; irpStack->Parameters.QueryInterface.InterfaceType = (LPGUID)&ChannelBusProtocolGuid; irpStack->Parameters.QueryInterface.Size = sizeof(CHANNEL_BUS); irpStack->Parameters.QueryInterface.Version = 1; irpStack->Parameters.QueryInterface.Interface = (PINTERFACE)pChannelBus; irpStack->Parameters.QueryInterface.InterfaceSpecificData = NULL; irp->IoStatus.Status = STATUS_NOT_SUPPORTED; status = IoCallDriver(targetObject, irp); if(STATUS_PENDING == status) { KeWaitForSingleObject(&event, Executive, KernelMode, FALSE, NULL); status = ioStatusBlock.Status; } Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75574
  - 20
    - Howto extract info from an HTASK Hi, I am conscient that it might not be the place to ask that question here. I am trying to identify which task is presently active for the user. I am trying to filter WM_ACTIVATEAPP: But the param is an htask How do I get information such as which application name is? Or perhaps convert htask in ProcessID so I can extract the module name. Thanks Denis Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75570
  - 21
    - training with focus on device driver installation Is anyone offering training with focus on Windows device driver installation Thanks Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75569
  - 22
    - Non-paged pool leak in HID Class driver Hi, I used pooltag and found that the non-paged pool bytes used in HID Class driver is increasing about 200 KB per hour and the number of allocations minus the frees is also increasing (1 per second). The problem occurs in Windows 2000 but does not occur in XP. We have a program and a driver that interacts with the Hid Class driver. What in our program/driver can cause a leak in Windows 2000 and not in XP? Thanks, Daniel Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75567
  - 23
    - why do the encrypted files become plaintext when i reboot computer I develop the encyrpting/decrypting file system based on sfilter.c (IFS kit) . In fat32 file /NTFS file format , all are rigth. But in NTFS compressed file format , when i reboot computer , the file is in plaintext. I dispose the IRP_READ & IRP_WRITE, only disposed ' if((Irp->Flags &(IRP_NOCACHE|IRP_PAGING_IO|IRP_SYNCHRONOUS_PAGING_IO)))'. in IRP_WRITE dispatch function , encrypt file , in IRP_READ dispath function , decrypt file . Why? Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75565
  - 24
    - SP_DEVICE_INTERFACE_DETAIL_DATA Are the following facts trivial? Maybe not trivial enough. In the C and C++ languages, if the name of an array is used in an expression then it usually converts to a pointer type, but not always. Also array types in function parameters usually convert. However, if an array is being defined and storage is being reserved for the elements, then the name of the array is the name of the array itself, and does not convert to a pointer. Let's hope those facts are understood by most device driver writers, so that they don't cause many blue screens. Let's hope. Let's hope those facts are understood by big software vendors who have been known to write device drivers. Oops. http://msdn.microsoft.com/library/default.asp?url=/library/en-us/devio/base/sp_device_interface_detail_data_str.asp says: > DevicePath > Pointer to a NULL-terminated string that specifies the device path. Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75563
  - 25
    - Libc.lib is missing in WDK. Hi all, I am new to WDK development and hope that I can get from helps from the experts. My device driver built fine on DDK but I cannot get build success on WDK. The error messages: BUILDMSG: error BLD1004 : MSVCRT lib listed explicitly - use USE_MSVCRT=1 (c:\s b2\smbslave\exe\server) BUILDMSG: error BLD1006 : LIBC lib listed explicitly (c:\smb2\smbslave\exe\server) NMAKE : U1073: don't know how to make 'C:\WinDDK\5270\lib\wxp\i386\libc.lib' BUILD: nmake.exe failed - rc = 2 Does anyone know how I can resolve the above error message? Thanks. HS Phuah Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75560
- Next
  - 1
    - "Driver" or "Interface Object" - yes or no? I am having the following debate: I am trying to convince a vendor that the proper way to provide an interface to their embedded system (that hooks up to a PC using TCPIP) is to provide a software component on the PC (like a driver). For some reason, they are of the oppinion that just providing the syntax for the messages that go back and forth should be enough (i.e. the actual data that corresponeds to a particular message in an agreed format). I believe that it is better software engineering to provide an encapsulation and buffer layer in shape of a software component (such as a DLL etc.). Am I right? Wrong? How do I convince them? Any litterature to refer to? Thanks Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75552
  - 2
    - BroadcastSystemMessage and system-level device drivers according to the sdk "The BroadcastSystemMessage function sends a message to the specified recipients. The recipients can be applications, installable drivers, network drivers, system-level device drivers, or any combination of these system components" Note: system-level device drivers Is this documentation correct? If so, how does the WDM-type system-level device drivers get this message, possibbly via IoRegisterPlugPlayNotification? which parameters to pass to this call? Or maybe another registration function? And also, if so what parameters to pass in BroadcastSystemMessage( DWORD dwFlags, LPDWORD lpdwRecipients, UINT uiMessage, WPARAM wParam, LPARAM lParam ) ??? such that it can be seen by a wdm driver thanks tom Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75548
  - 3
    - PS/2 Mouse Packets and USB Mouse Reports under Windows Vista Hello Everybody, Back in 2004, Microsoft said that Vista was going to support the new WM_MOUSEHWHEEL message for horizontal scrolling (tilt wheel), which implies that tilt wheel data will be part of the standard data packet sent by a mouse to the OS. Does anyone know if any information has been published about the standard PS/2 packet structure for a mouse with a tilt wheel? Will there be any other information contained in the new packet format, such as the 4th and 5th buttons? If there are more buttons, what functionality is expected from them? How should a tilt-wheel PS/2 mouse report itself to the system to inform that it supports the new packet structure? Has any information been published about HID usage pages and usages to be used to transfer tilt wheel values to the OS over a USB connection? Any leads will be greatly appreciated. Thank you! Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75542
  - 4
    - UniDriver plugin -- "custom papersize" does not appear in my DEVMODE UniDriver plugin -- "custom papersize" does not appear in my DEVMODE Why is it that when I enter a "custom papersize", in an application like MS Word, that I don't see the same values (conversions accounted for) in my public DEVMODE (dmPaperSize, dmPaperWidth, dmPaperLength) in the driver, especially when printing? What do I need to do? Any ideas? Rudy Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75539
  - 5
    - pci expansion support I have finally found some time to rewrite a filter driver for a pci expansion system. Given the product I have a requirment to increase the resource windows of buses that are upstream of my product, because the system may be hot plugged, some BIOSes don't support us, and some of the devices downstream (third party) may require more then the default 64MB/64MB/4K. My driver acts as an upper filter to PCI.SYS. I would like to modify the PCI config space of the upstream buses before PCI.SYS reads them. The information I am looking for is at what point in the process does PCI.sys read the config space before resource arbitration to decide what resources to allocate? I will figure it out through expirimentation, but if someone out there has some street knowledge, or can point me to some relevant documentaion in this area that could save me some time I would appreciate it. The driver I am replacing makes modifications in the IRP_MN_FILTER_RESOURCE_REQUIREMENTS pnp IRP, this is a direction I could go as I have a mostly working example, it just seems like modifying the config space and letting PCI.sys do its job is an cleaner route to take. My worry is that I have somewhat of a chicken and egg dilema as I need pci.sys to make modifications to the config space, but I don't want to make the modifications too late. I want to work within the accepted practices of the OS as I want the driver to support Vista and 64 bit versions as well without too many modifications. Thanks, Brian Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75538
  - 6
    - !obtrace Hi All, How to set the "Object Reference Tracing" in the gflags to enable the !obtrace to work for _DEVICE_OBJECT objects? TIA Andrew Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75535
  - 7
    - IoCallDriver() returns STATUS_INVALID_DEVICE_REQUEST 0xc0000010 Hi All, I have a problem formatting and sending an Irp with Urb. I have an usb upper function filter driver and inside my custom IOCTL handler I call UsbBuildSelectConfigurationRequest to format an Urb and then I call SendAwaitUrb to sumbit the Urb. However, IoCallDriver inside SendAwaitUrb is returning STATUS_INVALID_DEVICE_REQUEST 0xc0000010. I must be wrong somewere in the code below, because as we all know the SendWaitUrb and the other code from W.Oney's book is a great source to learn and practicise. I have checked the chain Fido->Fdo->Pdo, it's ok, I believe UsbBuildSelectConfigurationRequest with NULL formats a proper setup packet... but still I am stuck and could not go ahead. Please take a look at the code below and let me know what do you think. Best regards, Dim switch (irpStack->MajorFunction) { case IRP_MJ_CREATE: DebugPrint(("Create \n")); break; case IRP_MJ_CLOSE: DebugPrint(("Close \n")); break; case IRP_MJ_CLEANUP: DebugPrint(("Cleanup \n")); break; case IRP_MJ_DEVICE_CONTROL: ; switch (irpStack->Parameters.DeviceIoControl.IoControlCode) { //Here comes our predefined custom code case IOCTL_MY_CUSTOM: // Create the URB urb = (PURB)ExAllocatePool(NonPagedPool, sizeof(struct _URB_SELECT_CONFIGURATION)); UsbBuildSelectConfigurationRequest(urb, sizeof(struct _URB_SELECT_CONFIGURATION), NULL); status = SendAwaitUrb(DeviceObject, urb); if (!NT_SUCCESS(status)) KdPrint((DRIVERNAME " - sending urb failed with status %X\n", status)); break; default: status = STATUS_INVALID_PARAMETER; break; } default: break; } return CompleteRequest(Irp, status, 0); } NTSTATUS SendAwaitUrb(PDEVICE_OBJECT fdo, PURB urb) { // SendAwaitUrb PIO_STACK_LOCATION stack; IO_STATUS_BLOCK iostatus; KEVENT event; NTSTATUS status; PIRP Irp; PDEVICE_EXTENSION pdx; PAGED_CODE(); ASSERT(KeGetCurrentIrql() == PASSIVE_LEVEL); pdx = (PDEVICE_EXTENSION) fdo->DeviceExtension; if(pdx->NextLowerDriver == NULL) return STATUS_UNSUCCESSFUL; KeInitializeEvent(&event, NotificationEvent, FALSE); Irp = IoBuildDeviceIoControlRequest(IOCTL_INTERNAL_USB_SUBMIT_URB, pdx->NextLowerDriver, NULL, 0, NULL, 0, TRUE, &event, &iostatus); if (!Irp) { KdPrint((DRIVERNAME " - Unable to allocate IRP for sending URB\n")); return STATUS_INSUFFICIENT_RESOURCES; } stack = IoGetNextIrpStackLocation(Irp); stack->Parameters.Others.Argument1 = (PVOID) urb; status = IoCallDriver(pdx->NextLowerDriver, Irp); if (status == STATUS_PENDING) { KeWaitForSingleObject(&event, Executive, KernelMode, FALSE, NULL); status = iostatus.Status; } return status; }// SendAwaitUrb Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75515
  - 8
    - IoCallDriver() returns STATUS_INVALID_DEVICE_REQUEST 0xc0000010 Hi All, I have a problem formatting and sending an Irp with Urb. I have an usb upper function filter driver and inside my custom IOCTL handler I call UsbBuildSelectConfigurationRequest to format an Urb and then I call SendAwaitUrb to sumbit the Urb. However, IoCallDriver inside SendAwaitUrb is returning STATUS_INVALID_DEVICE_REQUEST 0xc0000010. I must be wrong somewere in the code below, because as we all know the SendWaitUrb and the other code from W.Oney's book is a great source to learn and practicise. I have checked the chain Fido->Fdo->Pdo, it's ok, I believe UsbBuildSelectConfigurationRequest with NULL formats a proper setup packet... but still I am stuck and could not go ahead. Please take a look at the code below and let me know what do you think. Best regards, Dim switch (irpStack->MajorFunction) { case IRP_MJ_CREATE: DebugPrint(("Create \n")); break; case IRP_MJ_CLOSE: DebugPrint(("Close \n")); break; case IRP_MJ_CLEANUP: DebugPrint(("Cleanup \n")); break; case IRP_MJ_DEVICE_CONTROL: ; switch (irpStack->Parameters.DeviceIoControl.IoControlCode) { //Here comes our predefined custom code case IOCTL_MY_CUSTOM: // Create the URB urb = (PURB)ExAllocatePool(NonPagedPool, sizeof(struct _URB_CONTROL_VENDOR_OR_CLASS_REQUEST)); UsbBuildSelectConfigurationRequest(urb, sizeof(struct _URB_SELECT_CONFIGURATION), NULL); status = SendAwaitUrb(DeviceObject, urb); if (!NT_SUCCESS(status)) KdPrint((DRIVERNAME " - sending urb failed with status %X\n", status)); break; default: status = STATUS_INVALID_PARAMETER; break; } default: break; } return CompleteRequest(Irp, status, 0); } NTSTATUS SendAwaitUrb(PDEVICE_OBJECT fdo, PURB urb) { // SendAwaitUrb PIO_STACK_LOCATION stack; IO_STATUS_BLOCK iostatus; KEVENT event; NTSTATUS status; PIRP Irp; PDEVICE_EXTENSION pdx; PAGED_CODE(); ASSERT(KeGetCurrentIrql() == PASSIVE_LEVEL); pdx = (PDEVICE_EXTENSION) fdo->DeviceExtension; if(pdx->NextLowerDriver == NULL) return STATUS_UNSUCCESSFUL; KeInitializeEvent(&event, NotificationEvent, FALSE); Irp = IoBuildDeviceIoControlRequest(IOCTL_INTERNAL_USB_SUBMIT_URB, pdx->NextLowerDriver, NULL, 0, NULL, 0, TRUE, &event, &iostatus); if (!Irp) { KdPrint((DRIVERNAME " - Unable to allocate IRP for sending URB\n")); return STATUS_INSUFFICIENT_RESOURCES; } stack = IoGetNextIrpStackLocation(Irp); stack->Parameters.Others.Argument1 = (PVOID) urb; status = IoCallDriver(pdx->NextLowerDriver, Irp); if (status == STATUS_PENDING) { KeWaitForSingleObject(&event, Executive, KernelMode, FALSE, NULL); status = iostatus.Status; } return status; }// SendAwaitUrb Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75514
  - 9
    - Windows running without a pagefile Hi, What are the limitations of a system running windows without a pagefile? -- Thanks, Daniel Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75501
  - 10
    - Sample Mirror Driver Excel Problem. I believe this problem has been around for some years now as I did find one other posting about it from November, 2002. When the sample mirror display driver that is included with the DDK is loaded (i.e. attached to the desktop), Excel displays the following error when it is started: "Not enough system resources to display completely." I strongly suspect this error is a catch-all for several issues and that the problem here is not necessarily due to the driver's taking resource but with something the driver is, or is not, doing that Excel expects/needs. Note that Excel runs after acknowledging this error message. However, it pops up often, for example when simply clicking on a different worksheet tab. So I am hoping this problem has been solved/identified by now and I would greatly appreciate any information anyone might have about it. Cheers, Bill Wyatt. bwyatt@sonexis.com Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75499
  - 11
    - Compiling Device Drivers in C++ Hi, I would like to know how to compile a device driver written in C++. I have seen a product called DriverWorks but i want to figure out if there is another way to do so. regards, CompuMe Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75498
  - 12
    - Mounting on a directory Greetings, I'm looking for some help with mounting a file system on a directory via SetVolumeMountPoint. I have a file system that I'm porting to Windows. In general it mounts OK, but I have one strange error condition that manifests itself as follows: 1) format a partition with my file system 2) use SetVolumeMountPoint to mount the file system on an NTFS directory << at this point the file system is NOT accessible via the NTFS path >> 3) reboot the system << the file system IS accessible via the NTFS path >> As a variation, if instead of rebooting the system in step 3, I use the Disk Management tool to assign a drive letter to the partition with my file system, it somehow makes the file system accessible via the NTFS path... It feels like my file system is not handling some event properly, but I can't figure out what it is... Anybody have an idea on what's needed? Thanks in advance, -bob Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75497
  - 13
    - emachine 220 Hi All I need the "special" drivers (video etc.) for the above - tried most places - any clues/help? Thanks David Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75496
  - 14
    - emf what is this nt emf 1.003,1.006,1.007,1.008. what difference will it make if i choose 1.003 instead of 1.008. are these formats standard ones or are they undocumented. can i develop application that use these formats. why microsoft says that it has its own spool emf and another emf for for the outsiders?????? Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75490
  - 15
    - SetupDiCallClassInstaller Intiating driver install then failing. Hi, this question has to with the installation and loading of a driver. The installation code I've written works great for Windows XP but not for Windows 2K. I used w2k for the building with the latest DDK. There is no hardware involved so my application calls the Setup API functions to communicate with the PnP Manager. So the adding of the devices is done when a user adds another port, not when the user installs the software. These devices are enumerated in ENUM\ROOT\PORTS. I have a virtual serial port (Com Port) driver I've written that can have up to 255 virtual devices. I have the .inf file and the .sys file in a subdirectory of my application. When I add the first port from my application it gets to the SetupDiCallClassInstaller function and it is called. The PnP Manager then loads my driver and calls DriverEntry. DriverEntry returns STATUS_SUCCESS but the PnP Manager calls my unload routine right after this. SetupDiCallClassInstaller returns stating a reboot is required. On Windows XP the AddDevice routine is called after the successfull call to DriverEntry with no request for a reboot. If I do reboot then everything works great. What is different from wxp and w2k in this scenario? Is there something in the deviceobject that needs to be set. I have Debug Prints in the PnP dispatch routines and none are being called. setupapi.log looks like this: @ 13:54:31.914 : Install Device: Begin. @ 13:54:31.914 : Doing full install of ROOT\PORTS\0000. @ 13:54:31.914 : Install Device: Writing BASIC Logical Configs @ 13:54:31.914 : Install Device: Changing registry settings specified by INF(s). @ 13:54:31.924 : Install Device: Writing driver specific registry settings. @ 13:54:31.924 : Install Device: Installing required Windows services. @ 13:54:32.785 : Install Device: Writing drive descriptive registry settings. @ 13:54:32.795 : Install Device: Calling RunOnce/GrpConv items. @ 13:54:32.795 : Executing RunOnce to process 5 RunOnce entries. @ 13:54:32.815 : Device install finished successfully (ROOT\PORTS\0000). @ 13:54:32.815 : Install Device: End. @ 13:55:02.017 : Device install function: DIF_PROPERTYCHANGE. @ 13:55:02.027 : Executing class-installer. @ 13:55:02.027 : Completed class-installer. @ 13:55:02.027 : Executing default installer. @ 13:55:15.166 : Device required reboot: Device has problem: 31. @ 13:55:15.176 : Completed default installer. Why does w2k require a reboot and not wxp? Problem 31 states that my driver depends on another driver that is not loaded? My driver is not dependent upon another driver? Anyone have any ideas as to what is going on? -David G Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75469
  - 16
    - Help - creating thread in kernel mode Anyone can help me how to create a thread in kernel-mode that can write data to physical user memory? Thanks Arcady Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75465
  - 17
    - Question about completion routine for InternalDeviceControl. Hi, does anyone know how to have the completion routine for InternalDeviceControl set up properly so that the routine will be called when the kmdf request object is completed by lower drivers? I set it up according to kmdf help file and the driver will only execute the routine twice before device manager will stop the driver completely. And I used the following two lines WdfRequestSetCompletionRoutine (Request, completionRoutine, devExt); ret = WdfRequestSend (Request, Target, WDF_NO_SEND_OPTIONS); to register a completion routine that just print out the text message and called WdfRequestComplete(Request, Params->IoStatus.Status) to complete the request. Thanks. AL. Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75464
  - 18
    - Memory manager writing pages back to executables... I was reading an article posted in code project website. The author has noticed that his XP explorer was taking about 5 seconds to display an error message when he tried deleted a locked file or directory. He found that shell32.dll was looping around sleep for 5 times with 1 second expiration. So he decided to patch the shell32.dll using a driver so that it will loop for 0 times thus avoiding the delay. He writes "Now that we know which addresses to patch, we need to map those addresses into the kernel-mode area of the memory. This will allow us to lock the pages in memory, thus preventing them from being swapped out to disk. In this case the memory we're modifying is "backed" by shell32.dll. If it were to be swapped out, the memory manager would attempt to write the changes back to shell32.dll. ...." He also writes (in the Some Final Thoughts section) "I'm not sure why the memory manager doesn't try to write the modified pages back to shell32.dll. I suspect that it's because the entries in the Page Frame Number Database representing the physical pages we've modified always have a "reference count" of at least 1 (because they're locked). If my understanding is correct, this means they will not be moved to the "Modified List" and thus will not be written to disk by the Modified Page Writer. If anyone can expand on this, please leave a comment. " My question is: 1. Is it true that the memory manager will write pages back to the executable upon modification? Under what circumstance does the memory manager does this? 2. Where can I find more information about this topic? The original article may be found at http://www.codeproject.com/system/NoDeleteDelay.asp Tom Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75450
  - 19
    - Error compiling my code Hi all. I'm a newbie in programming device drivers and now I want to handle control-requests to my driver and I get compiler errors. I can't find the failure, can you help me please? compiler error message: "error c4028: formal parameter 1 different from declaration" here my code: // header NTSTATUS MyControl( IN PDRIVER_OBJECT Driver, IN PIRP Irp ); NTSTATUS MyDeviceControl( IN ULONG ioControlCode, OUT PIO_STATUS_BLOCK IoStatus, IN PDRIVER_OBJECT Driver ); // code NTSTATUS DriverEntry( IN PDRIVER_OBJECT DriverObject, IN PUNICODE_STRING RegistryPath ) { DriverObject->MajorFunction [IRP_MJ_DEVICE_CONTROL] = MyControl; return STATUS_SUCCESS; } NTSTATUS MyControl( IN PDRIVER_OBJECT Driver, IN PIRP Irp ) { PIO_STACK_LOCATION irpStack; ULONG ioControlCode; NTSTATUS ntstatus; ntstatus = Irp->IoStatus.Status = STATUS_SUCCESS; Irp->IoStatus.Information = 0; irpStack = IoGetCurrentIrpStackLocation(Irp); ioControlCode = irpStack->Parameters.DeviceIoControl.IoControlCode; ntstatus = MyDeviceControl( ioControlCode, &Irp->IoStatus, Driver ); IoCompleteRequest( Irp, IO_NO_INCREMENT ); return ntstatus; } NTSTATUS MyDeviceControl( IN ULONG IoControlCode, OUT PIO_STATUS_BLOCK IoStatus, IN PDRIVER_OBJECT DriverObject ) { NTSTATUS ntStatus; switch ( IoControlCode ) { case IOCTL_START_BLOCK_KBD: DriverObject->MajorFunction[IRP_MJ_READ] = MyDispatchRead; IoStatus->Status = STATUS_SUCCESS; break; case IOCTL_END_BLOCK_KBD: DriverObject->MajorFunction[IRP_MJ_READ] = MyDispatchGeneral; IoStatus->Status = STATUS_SUCCESS; break; default: IoStatus->Status = STATUS_INVALID_DEVICE_REQUEST; break; } return IoStatus->Status; } thanks macman Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75447
  - 20
    - Problems with PREFast DDK 3790.1830 (Win2003SP1) Hi all. Yesterday I migrated from DDK build 3790 (I think it's called "Windows Server 2003 DDK") to the newest build 3790.1830 ("Windows Server 2003 with SP1 DDK"). Prefast stopped working properly: I have this compiler warning if I run "prefast build -cefw" Compiling - driver.rc for i386 cl : warning D4002 : ignoring unknown option '/hotpatch' which doesn't worry me too much. The real problem is that Prefast issues some defects, but when I try to run "prefast view", it spits out a script error: Line: 422 Char: 2 Error: 'documentElement' is null or not an object Code: 0 URL: file://C:\WINDDK\3790~1.183\bin\x86\drvfast\scripts\DefectUI\DefectUI.hta If I try "prefast list", it fails with this message ----------------------------------------------------------------------------- Microsoft (R) PREfast Version 1.5.2402 for 80x86. Copyright (C) Microsoft Corporation. All rights reserved. ----------------------------------------------------------------------------- Contents of defect log: C:\WINDDK\3790~1.183\bin\x86\drvfast\defects.xml ----------------------------------------------------------------------------- while creating PREfast.DefectDefs while creating PREfast.DefectSet ----------------------------------------------------------------------------- Error 0x800a138f occurred while creating PREfast.DefectSet component: 'documentElement' is null or not an object ----------------------------------------------------------------------------- I tried the same on another machine with the same DDK, and it works, indeed. So I think there is something screwed up on my machine. Any idea where to look at? Have a nice day GV Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75446
  - 21
    - What do I have to use to compile DDK samples? Hi all. Please help to understand what are the differences between VC++ Standard and Pro/Enterprice versions? Can I compile drivers by SE version? I'm asking this question 'cause I'm trying to compile DDK driver sample and getting the errors. If I do this from command line using DDK's build.exe I get the next: BUILD: Compile and Link for i386 BUILD: Compiling and linking d:\work\msddk\ntddk\src\video\miniport\mirror BUILD: NMAKE.EXE failed - rc = 2 BUILD: Done Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75443
  - 22
    - Difference between Windows NT 5.0 and NT 5.1 booting process Hi all, I'm working on a full disk encryption driver for which I have written a lower filter to disk driver and some 16 bit assemby program (INT13 hook) which will initially decrypt all the system files. All my modules are working fine in Windows 2000 but in Windows XP it crashes with the following bug check code. PROCESS1_INITIALIZATION_FAILED (6b) Arguments: Arg1: c000007a, Indicates the NT status code that caused the failure. Arg2: 00000006, (reserved) Arg3: 00000000 Arg4: 00000000 Is there any differnence in the booting process of Windows 2000 and Windows XP in terms of reading the data from the disk at boot level. Developer Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75438
  - 23
    - Understanding NdisMAllocateMapRegisters Hi Is it required to call NdisMAllocateMapRegisters before the ndis miniport calls NdisMAllocateSharedMemory(). (as documented in the ddk)? Can't the physical address returned by calling NdisMAllocateSharedMemory() passed to the dma master on the device? My understanding is - if an Ndis Buffer passed by protocol drivers has to be mapped directly and passed to the dma master, then a NdisMStartBufferPhysicalmapping() should be called. To do this, a map register is required, hence NdisMAllocateMapregisters() should be called prior to calling NdisMstartBufferPhysixcalmapping(). But if the buffer passed by the protocol driver is copied to a buffer alloacted using NdisMAllocateSharedMemory(), then NdisMAllocateMapRegister () is not required. Please correct me if Im wrong. basically, What does NdisMAlloacteMapRegisters() do and how and when should an ndis miniport use it.? Thanks in Advance, suu. Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75430
  - 24
    - serial COM port driver Hi All, I am working on a project regarding the serial COM port. I have to design my own protocol for the serial communication. In the protocol there are some bits for Read/Write, address(bank, offset) and data to be transferred. Would anybody please tell me where I can find the Serial COM port driver, especially the driver that provides the funcation calls/APIs that allow me to call them by using my own protocol? Thanks a lot for your help. Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75425
  - 25
    - Help on NDISUIO hi everyone, i am doing a school project to obtain network adaptor information, and i am TOTALLY new with ndisuio. after reading so many articles here, i managed to use the CreatFile and NDISUIO_BINDING_DEVICES model to get a list network adaptors in my laptop... but the returned list is something like /device/xxxxxxxxxxxxxxxxxxxxxx, a serial strange number so called adaptor name, i dont understand what is that? and i am looking for something more specific, like wether it is a wireless or ethernet card...really dont have a clue on how to do that..T_T i am seeking you guys help, seems you all quite pro on this... Millions of thanks!!!! Michael Singapore... Tag: enabling tcp segmentation offload leads to poorer perfomance Tag: 75412

I am adding TSO support to a NDIS driver for win 2003. After TSO is
enabled, the driver works fine, only that the performance becomes
worse. Netperf TCP thruput drops by +30%. Logging shows that NDIS does
pass big packets to the driver and packet trace shows that the card
does split the big packets into small pieces. Nothing abnormal (no
excessive retransmission, csum error) is observed in the packet trace.
But why enabling TSO can degrade performance so much?

I am pretty new in driver development and I know it's hard to tell
without actually looking at the source code. but I really want some
suggestions, directions from those gurus out there. Are there anything
I should pay attention to? what tools, perfomance counters I can
use/watch to help trace down the root cause? Thanks a lot

Top