[9fans] How to kill a process

[9fans] How to kill a process

[Adriano Verardo]

Hi, all.

Five years ago I wrote the manager for a USB barcode reader.
Basically a derivative of the keyboard mngr, but not embedded in usbd.
Recently has been noted that after a unplug-plug of the reader, the
kernel recognizes
the reader (apparently) as usual but the mounted root of the manager is
empty.

The reason seems to be a spuriuos instance of the manager, impossible to
kill
with kill and slay too. In fact, after a reboot all works again.
Both manager and application have been never changed while the
kernel stuff is updated weekly (replica/pull).

IMHO something changed in the interface between usbd and external managers.

  Is it possible ?

In what condition a process started by usbd becomes not killable ?

Thanks in advance.

adriano

Re: [9fans] How to kill a process

[erik quanstrom]

On Thu Mar  6 12:34:27 EST 2014, adriano.verardo@mail.com wrote:
> Hi, all.
>
> Five years ago I wrote the manager for a USB barcode reader.
> Basically a derivative of the keyboard mngr, but not embedded in usbd.
> Recently has been noted that after a unplug-plug of the reader, the
> kernel recognizes
> the reader (apparently) as usual but the mounted root of the manager is
> empty.
>
> The reason seems to be a spuriuos instance of the manager, impossible to
> kill
> with kill and slay too. In fact, after a reboot all works again.
> Both manager and application have been never changed while the
> kernel stuff is updated weekly (replica/pull).
>
> IMHO something changed in the interface between usbd and external managers.
>
>   Is it possible ?
>
> In what condition a process started by usbd becomes not killable ?

recommend a little debugging.  first
	eve; acid $pid
	acid; lstk()

then
	eve; acid -l kernel -k $pid /path/to/kernel
	acid; lstk()

the idea here is to figure out what this rogue process is doing.
and why it's not killable.

- erik