Dear Tim Robert, Doran, Calvin Guan,
I got your answer with some different ideas :
[1] Not continue to implement [Tim Robert & Doron]
? - Reason :
??? * Access to and operation of device memory must be thoroughly tested
??? * Exposing device memory as a disk is a bad bad idea [Doron]
[2] Continue to implement [Calvin Guan]
? # Reason :
??? * We can implement within 2 days in case of understanding (SCSI commands)
I’d like to make clear my requirement . I didn’t intend to verify my memory device
by only implementation of disk on PC. But here’s the story :
[1] For verification of my memory device, I separated into 2 parts :
??? 1.1.? Test by using multiple READ/WRITE commands ( can access configuration space, any memory space )
??? 1.2?? Implement disk on PC
So, my memory device is already tested carefully in 1.1 . But I need to implement
1.2 because it’s easy to show the DEMO to customer . So, I don’t think that I must change my idea or ask my customer to change the request .
? ? ? ? ? ?? I have some experience about working with USB and I also know SCSI commands ( INQUIRY, READ10, WRITE10 ) . However , this is a story :
? [Host PC]--------------> [USB 2.0 device]
In USB2.0 device, it implements functions and flows of receiving commands( INQUIRY, READ10, WRITE10 …) from Host and returning the data . This is controlled by ATAPI ( I remember ) . This means : When USB device receive CBW ( 13Bytes) , it must decode to know what commands it receives and the states that it’s inside .
??? With my current project about PCIe, I need that Host PC is active in controlling device and device only receive READ/WRITE commands .It doesn’t understand READ10 or WRITE10 but Host PC has driver for controlling such issues .
//------------------
I know that? the sample of WDK “RAM disk driver” also does something like that ,
but I don’t know what to modify from that sample for my requirement.
If possible, would you please instruct me some steps to do ?
Thank you very much !
Best Regards
HanNguyen
— On Sat, 6/18/11, Calvin Guan (news) wrote:
From: Calvin Guan (news)
Subject: Re: [ntdev] About import PCI Express Memory into disk on PC
To: “Windows System Software Devs Interest List”
Date: Saturday, June 18, 2011, 12:56 AM
Not to say if it’s the idea is good or bad… but a basic memory-back storport or scsi miniport is very very very easy.?One has moderate experience in windows driver and knows basic scsi commands (inquiry, read10 write10, read capacity?and such) should need no more than 2 days to get it working (able to format, read and write, suspend, resume…) by typing every line of?code by bare fingers.
?If you were just try to exercise the target access on your pci devices, you can just do it from windbg by using the bang dd and bang ed commands.
?Good luck,
Calvin?
On Fri, Jun 17, 2011 at 10:34 AM, Tim Roberts wrote:
Nhat Han wrote:
>
> I got it . I also heard about Mini Storport. However, I don’t have any
> guidance document or sample code to start my project .
> If you have experience in this type of project, please kindly show me
> some documents and guidance .
> ?I’m new to windows driver development and it’s hard to look up
> everything by myself .
>
Of course it’s hard. ?Drivers are not easy to write. ?If you are new to
Windows driver development AND inexperienced with disk systems, then you
are 8 or 10 months away from having a working StorPort driver to
accomplish this task. ?A simple memory test driver could be completed in
3 or 4 weeks, even with the KMDF learning curve. ?An experienced driver
writer could have something running in half that time.
> >>[From Tim Roberts*]* If you want to exercise the memory on your
> board, then the right way is to write a small driver that does exactly
> that. ?Just write some memory tests.
>
> I got the requirements to make such driver . So , I must investigate
> how to do it . Please, kindly help me if you experienced this .
>
Doron is exactly right. ?Your requirements are flawed, and as an
engineer it is your responsibility to point this out, rather than
blindly follow misguided instructions.
Here is a bad requirement:
1. Memory on the device should be exposed as a disk drive so memory can
be exercised
Here is a good requirement:
1. Access to and operation of device memory must be thoroughly tested.
–
Tim Roberts, xxxxx@probo.com
Providenza & Boekelheide, Inc.
—
NTDEV is sponsored by OSR
For our schedule of WDF, WDM, debugging and other seminars visit:
http://www.osr.com/seminars
To unsubscribe, visit the List Server section of OSR Online at http://www.osronline.com/page.cfm?name=ListServer
—
NTDEV is sponsored by OSR
For our schedule of WDF, WDM, debugging and other seminars visit:
http://www.osr.com/seminars
To unsubscribe, visit the List Server section of OSR Online at http://www.osronline.com/page.cfm?name=ListServer