I haven't (yet) stared at the XFree86 server code in this area.
In the past, the X server was happy so long as it got proper data.
When I was last hacking in this area (Digital UNIX, early '90's), I had
a problem very similar to yours, but it turned out that the keyboard and
mouse I was using required initialization to get put into the proper state
to use with X: this occurred at system bootup, but didn't happen if you
unplugged the keyboard or mouse and plugged it back in again while the
operating system was running.
I fixed the DU device driver (which I was completely restructuring/rewriting
at the time) to detect the connection of the devices and always send the
initialization strings needed whenever a device was plugged in. Arguably
this is a device driver function (at least for the keyboard, if not the
mouse), because the keyboard is essential to the operation of the computer,
even if X is not running. Only a device driver can make such guarantees.
So you may be seeing a very similar problem (I'm not familiar with
PC keyboards and mice to be sure).
So it could be the X server: but it could equally well be that the
driver should be responsible for getting the hardware into a known
sane state. More investigation is in order before pointing fingers
in any direction.
- Jim
> From: Lincoln Stein <[EMAIL PROTECTED]>
> Date: Mon, 8 May 2000 11:27:00 -0400 (EDT)
> To: "Dunlap, Randy" <[EMAIL PROTECTED]>
> Cc: "'[EMAIL PROTECTED]'" <[EMAIL PROTECTED]>, [EMAIL PROTECTED],
> [EMAIL PROTECTED]
> Subject: RE: [linux-usb] mouse on 2.3.99pre6 not stable across suspend/res
> ume
> -----
> It's actually not just suspend/resume that's the problem. Just
> disconnecting and reconnecting the mouse causes X to ignore the device
> for the duration. My guess is that X gives up on the device once it
> returns a read error.
>
> I'm back to using the boot protocol driver in 2.2.13. Even though it
> doesn't support the scroll wheel, at least I can disconnect and
> reinsert the mouse without fear.
>
> Lincoln
>
>
> Dunlap, Randy writes:
> > Hi,
> >
> > > From: Lincoln Stein [mailto:[EMAIL PROTECTED]]
> > >
> > > The subject line says it all. When I suspend/resume with a USB
> > > mouse, the mouse device gets detached from /dev/input/mouse0 and
> > > reattached to /dev/input/mouse1. Needless to say, this confuses X
> > > tremendously.
> >
> > This seems to say to me that the input module isn't told about
> > the disconnect (from the suspend), so the resume (=> connect)
> > allocates another input/mouseN.
> >
> > [snip log]
> >
> > > I'm using standard APM. Oddly enough, the old development USB driver
> > > in the 2.2.14 kernel worked quite well across suspend/resume. I got
> > > greedy and wanted to use the scroll wheel in the intellimouse!
> >
> > Is use of the scroll wheel documented somewhere?
> >
> > ~Randy
>
> --
> ========================================================================
> Lincoln D. Stein Cold Spring Harbor Laboratory
> [EMAIL PROTECTED] Cold Spring Harbor, NY
> ========================================================================
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
--
Jim Gettys
Technology and Corporate Development
Compaq Computer Corporation
[EMAIL PROTECTED]
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
