Re: Staging status of speakup

[Okash Khawaja]

On Tue, 19 Mar 2019 16:31:21 +0000
Alan Cox <gnomes@xxxxxxxxxxxxxxxxxxx> wrote:

> On Sat, 16 Mar 2019 10:35:43 +0100
> Samuel Thibault <samuel.thibault@xxxxxxxxxxxx> wrote:
> 
> > Chris Brannon, le ven. 15 mars 2019 18:19:39 -0700, a ecrit:  
> > > Okash Khawaja <okash.khawaja@xxxxxxxxx> writes:    
> > > > Finally there is an issue where text in output buffer sometimes
> > > > gets garbled on SMP systems, but we can continue working on it
> > > > after the driver is moved out of staging, if that's okay.
> > > > Basically we need a reproducer of this issue.    
> > > 
> > > What kind of reproducer do you need here?  It's straightforward to
> > > reproduce in casual use, at least with a software synthesizer.    
> > 
> > The problem is that neither Okash nor I are even casual users of
> > speakup, so we need a walk-through of the kind of operation that
> > produces the issue. It does not have to be reproducible each time
> > it is done. Perhaps (I really don't know what that bug is about
> > actually) it is a matter of putting text in the selection buffer,
> > and try to paste it 100 times, and once every 10 times it will be
> > garbled, for instance.  
> 
> paste_selection still says
> 
> /* Insert the contents of the selection buffer into the
>  * queue of the tty associated with the current console.
>  * Invoked by ioctl().
>  *
>  * Locking: called without locks. Calls the ldisc wrongly with
>  * unsafe methods,
>  */
> 
> from which I deduce that with everyone using X nobody ever bothered to
> fix it. So before you look too hard at the speakup code you might
> want to review the interaction with selection.c too.

Hi,

This is a good point. At the moment speakup uses its own set of
selection and paste functions but I am in process of changing speakup
to use these functions from drivers/tty/vt/selection.c instead. This
lack of locking will be worth watching out for.

Thanks!
Okash