cvs commit: src/sys/i386/conf GENERIC src/sys/amd64/conf GENERIC

Ian Dowse iedowse at maths.tcd.ie
Wed Mar 16 02:17:06 PST 2005


In message <4237F33D.6040100 at portaone.com>, Maxim Sobolev writes:
>Doug Barton wrote:
>> On Wed, 16 Mar 2005, Ian Dowse wrote:
>>>  Enable ehci by default on i386 and amd64. It had got to the stage
>>>  where having this disabled was actually hurting us, since so many
>>>  BIOSes include legacy USB emulation that takes control of all usb
>>>  ports and only the ehci driver knows how to disable it.
>> 
>> Given all the bad things that happen to my system when I compile ehci 
>> into the kernel, such as lost usb signals from my docking station, and 
>> instant panics when I plug in certain usb 2 devices, I have to say I'm 
>> not sure this is the best course of action. Has this been widely 
>> discussed somewhere that I've not kept up on?
>
>I have the same worry. The problem with ehci that I am personally aware 
>of is that it causes all USB devices that have been attached during the 
>boot process in the presence of efci are forced to detach and attach 
>again just near the point when root fs is to be mounted. This may cause 
>troubles with booting off USB mass storage or cd-rom devices.

Sorry - I'll send a message to -current to let people know. If it
causes too many problems then maybe it is not worth enabling yet.

I agree that the re-attach behaviour is a problem - ideally we would
need to somehow defer probing the usb busses until all PCI functions
in the current PCI device have been attached.

Doug, can you point me at some details of the problems you mention?

Ian


More information about the cvs-src mailing list