2.6.18-git9

2.6.18-git9

[Andy Whitcroft]

I have a couple of power boxes which no longer seem to be able to
compile mainline as of 2.6.18-git9.  The build fails as below:

  CC      drivers/usb/core/driver.o
drivers/usb/core/driver.c: In function `usb_probe_device':
drivers/usb/core/driver.c:168: error: structure has no member named
`pm_usage_cnt'
drivers/usb/core/driver.c: In function `usb_driver_claim_interface':
drivers/usb/core/driver.c:305: error: structure has no member named
`pm_mutex'
drivers/usb/core/driver.c:309: error: structure has no member named
`pm_mutex'
drivers/usb/core/driver.c: In function `usb_driver_release_interface':
drivers/usb/core/driver.c:358: error: structure has no member named
`pm_mutex'
drivers/usb/core/driver.c:362: error: structure has no member named
`pm_mutex'
make[3]: *** [drivers/usb/core/driver.o] Error 1
make[2]: *** [drivers/usb/core] Error 2
make[1]: *** [drivers/usb] Error 2
make: *** [drivers] Error 2

This seems to be realated to the changes in this commit:

	commit 645daaab0b6adc35c1838df2a82f9d729fdb1767
    	usbcore: add autosuspend/autoresume infrastructure

The machines in question do not have CONFIG_PM set.  It seems that the
definition of pm_mutex et al in struct usb_device are contingent on that
defined, but the use of it in usb_driver_release_interface() are not.

Its not clear to me if suspending in this context is quite the same as
full suspend to disk that I associate with PM so I am unsure which bit
is in error.  Will retest with those references under CONFIG_PM.

Alan?

-apw

Re: 2.6.18-git9

[Alan Stern]

On Thu, 28 Sep 2006, Andy Whitcroft wrote:

> I have a couple of power boxes which no longer seem to be able to
> compile mainline as of 2.6.18-git9.  The build fails as below:
> 
>   CC      drivers/usb/core/driver.o
> drivers/usb/core/driver.c: In function `usb_probe_device':
> drivers/usb/core/driver.c:168: error: structure has no member named
> `pm_usage_cnt'
> drivers/usb/core/driver.c: In function `usb_driver_claim_interface':
> drivers/usb/core/driver.c:305: error: structure has no member named
> `pm_mutex'
> drivers/usb/core/driver.c:309: error: structure has no member named
> `pm_mutex'
> drivers/usb/core/driver.c: In function `usb_driver_release_interface':
> drivers/usb/core/driver.c:358: error: structure has no member named
> `pm_mutex'
> drivers/usb/core/driver.c:362: error: structure has no member named
> `pm_mutex'
> make[3]: *** [drivers/usb/core/driver.o] Error 1
> make[2]: *** [drivers/usb/core] Error 2
> make[1]: *** [drivers/usb] Error 2
> make: *** [drivers] Error 2
> 
> This seems to be realated to the changes in this commit:
> 
> 	commit 645daaab0b6adc35c1838df2a82f9d729fdb1767
>     	usbcore: add autosuspend/autoresume infrastructure
> 
> The machines in question do not have CONFIG_PM set.  It seems that the
> definition of pm_mutex et al in struct usb_device are contingent on that
> defined, but the use of it in usb_driver_release_interface() are not.
> 
> Its not clear to me if suspending in this context is quite the same as
> full suspend to disk that I associate with PM so I am unsure which bit
> is in error.  Will retest with those references under CONFIG_PM.
> 
> Alan?

This problem was noticed in -mm and the patch to fix it was sent in very 
recently; it hasn't filtered upstream yet.  The patch itself is here:

http://marc.theaimsgroup.com/?l=linux-usb-devel&m=115929663530629&w=2

Alan Stern