On Tue, 29 Jan 2002, Ferry Sutanto wrote:

> Hi Bill,
>
> Thank you very much for the information. Can u
> elaborate more on "invalidating the file system" ? The
> kernel gets the "device gone" interrupt, the first
> thing it does is unmounting the FFS and call
> config_detach. Is this what u mean by "invalidating
> FFS" ?

Right.

> I notice that although config_detach has finished, the
> kernel still tries to write by calling either vn_write
> or bwrite. I put little hack (a global flag == true if
> device has gone) which will bail out if the device has
> gone and I called brelse to clean up the buffer. FTP
> exit with EIO error, but the csh (shell) hangs. Does
> anybody have this problem before ?

That sounds like the right kind of premise - if the device is gone, in the
i/o routine, fail.

Not sure why the csh fails. Where was its current directory?

Take care,

Bill