Skip to contentSkip to conversation

vid&pid problems in usb_probe()

DISCUSSION [email protected] 2 participants 2 messages May 9, 2004

lore
Share & tools
lore
#0 lore raw

Sorry, I forgot to tell you, I have copied all the header files in "linux2.4.18/include/linux" to "/usr/include/linux". I am sure that there is no difference between "linux2.4.18/include/linux" and "/usr/include/linux"

I build this driver like this:

#gcc -c usb.c
#insmod -f usb.o //due to version dismatch, i should use -f

then plug a usb device.

Could you help me test this program? Do you have an usb device? You can write a program like this, and test if it can report the right vid & pid.
I downloaded a new kernel 2.6.5, after i started with the new kernel, lots of modules can not be loaded, so i changed back to 2.4.18, then I found my X-windows can not start. Did this error have something to do with this unstablity?

----- Original Message -----
From: Randy.Dunlap
To: [email protected]
Cc: [email protected]
Subject: Re: vid&pid problems in usb_probe()
Sent: Sun May 09 01:25:34 CST 2004

Quoted 100 lines · On Sat, 8 May 2004 17:05:24 +0800 (CST) <[email protected]> wrote:
| Thank you for replying,but I don't know what you mean,
| USB descriptor data structure is defined in 
| "/usr/include/linux/usb.h"
| 
| my kernel version is 2.4.18.

Are you saying that your driver uses /usr/include/linux/usb.h ? If so, that's wrong. It should use linux-2.4.18/include/linux/usb.h instead, and it will if you build/make the driver correctly.

So, how do you build this driver? Show the Makefile or command line that you use.

Regarding the data structure, I don't know what /usr/include/linux/usb.h on your system contains since that is not a kernel header file (and drivers shouldn't use those). I do recall, however, that many moons ago, there were some usb structs in usb.h that were not defined with __attribute__((packed)), and that could cause field offsets to be off a bit, er, byte (or more).

And what processor arch. are you doing this on?

Driver source code?

~Randy

| -----  Original Message  -----
| From: Randy.Dunlap 
| To: [email protected] 
| Cc: [email protected] 
| Subject: Re: vid&pid problems in usb_probe()
| Sent: Sat May 08 12:45:54 CST 2004
| 
| > On Sat, 8 May 2004 11:48:44 +0800 (CST) <[email protected]> wrote:
| > 
| > | 
| > | 
| > | I am writting an usb driver.You know function usb_probe(...) is used to determine whether the usbdevices just pluged in is what the driver is for.
| > 
| > a.  Please learn to use the Enter/Return key around character position
| > 70 (or before) on each line.
| > 
| > b.  what kernel version?   (*always*)
| > 
| > c.  You should ask this on the linux-usb-development mailing list:
| > [email protected]
| > 
| > 
| > | The Vid and Pid of my usb device are 0x1111 and 0x0000 respectively.
| > | 
| > | the program is :
| > | 
| > | static void* usb_probe(struct usb_device *udev, unsigned int ifnum, const struct usb_device_id *id)
| > | {
| > |     ..............
| > |     ..............
| > |     
| > |     printk("<1>Vid:%x
| Pid:%x
| ",udev->descriptor.idVendor,udev->descriptor.idProduct);
| > | 
| > |     if ((udev->descriptor.idVendor!=0x1111)
| > |          ||(udev->descriptor.idProduct!=0x0000)) return NULL;
| > | 
| > |     ..............
| > | }
| > | 
| > | when I plug the device whose vid & pid is 0x1111 & 0x0000 respectively.
| > | this Module displayed
| > | 
| > | 
| > | Vid:0
| > | Pid:201
| > | 
| > | usb.c ........ no active driver for this device;
| > | 
| > | and when I plug another device , I also got wrong vid & pid.
| > | 
| > | But when I wrote program as below:
| > | 
| > | __u16 tmp=0x1111;
| > | printk("<1>%x",tmp);
| > | 
| > | it can print "1111" on the screen. That means my syntax is correct.
| > | I mean, the problem may be at the data transfered into function usb_probe()
| > | Maybe data transfered into function usb_probe() is wrong.
| > | 
| > | I wonder where is the problem, how can i solve.
| > 
| > Seeing more (or all) of your source code could help.
| > I'm especially curious (suspicious) about your USB descriptor data
| > structures.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

#1 lore raw
Quoted 10 lines · On Sun, May 09, 2004 at 10:03:54AM +0800, [email protected] wrote:

Sorry, I forgot to tell you, I have copied all the header files in "linux2.4.18/include/linux" to "/usr/include/linux". I am sure that there is no difference between "linux2.4.18/include/linux" and "/usr/include/linux"

I build this driver like this:

#gcc -c usb.c
#insmod -f usb.o //due to version dismatch, i should use -f

then plug a usb device.

That is _not_ the proper way to build a kernel module. Please read the Linux Device Drivers for examples of how to do this the correct way (it's availble online for free.)

Quoted 5 lines

Could you help me test this program? Do you have an usb device? You can write a program like this, and test if it can report the right vid & pid.
I downloaded a new kernel 2.6.5, after i started with the new kernel, lots of modules can not be loaded, so i changed back to 2.4.18, then I found my X-windows can not start. Did this error have something to do with this unstablity?

Did you read the release notes about how to move to the 2.6 kernel? I recommend using that one for development, as the USB interface is nicer, and faster, and the build process is _vastly_ simpler.

Also, this belongs on the linux-usb-devel mailing list, not here.

greg k-h