BUS infrastructure problem

[Roman Kurakin]

Hi,

    I've found the reason. It was there where I do not expect it.
The only function's return value I didn't check was ...
<drum-roll>
... it was device_set_driver. It returns ENOMEM, all other stuff
goes well. More over it seems that at device_probe_child ()
device_set_driver do not have memory problems, but driver
methods table are broken.

Something tells me that the call of device_set_driver from
device_probe_child is intended for recovering from such case
but it doesn't work. Or this is wrong guess?

Do you think the case is closed or this is some misbehavior
that could be fixed?

rik

M. Warner Losh wrote:

>In message: <42CAAFCB.2080207 at cronyx.ru>
>            Roman Kurakin <rik at cronyx.ru> writes:
>: Hi,
>: 
>: M. Warner Losh wrote:
>: 
>: >In message: <42C57523.4050302 at cronyx.ru>
>: >            Roman Kurakin <rik at cronyx.ru> writes:
>: >:     I observe the followin strange behaviour with current: with some very
>: >: high probability after indentify callback I didn't get a probe callback. I
>: >: didn't find yet anything that could tell me why I see this. All function 
>: >: return me that all operations was successful.
>: >: 
>: >: PS. This driver is cx(4) and I am currently try to debug it in async mode
>: >: (I get strange panics while its work if I didn't get into situation with 
>: >: probe()).
>: >: 
>: >: Any ideas?
>: >
>: >I'll be happy to help you with this.
>: >  
>: >
>: The last place I get to is the call of DEVICE_PROBE macro.
>: But I do not see the call of my function.
>: There is other thing, it seems that probability highly increases
>: if the system reboots after panic and needs filesystem check.
>: 
>: This is all information I have now. I use printf as a primary
>: debug technic so I need to think how to move farther.
>: 
>: Ideas?
>
>That sounds really weird.  Add a Debugger() call and see if you wind
>up in the debugger.
>
>
>Warner
>_______________________________________________
>freebsd-current at freebsd.org mailing list
>http://lists.freebsd.org/mailman/listinfo/freebsd-current
>To unsubscribe, send any mail to "freebsd-current-unsubscribe at freebsd.org"
>  
>