Re: [PATCH v2 1/8] tty: add a poll() callback in struct tty_operations

[Tal Shorer]

On Wed, Jun 14, 2017 at 11:20 AM, Tal Shorer <tal.shorer@xxxxxxxxx> wrote:
> On Wed, Jun 14, 2017 at 4:15 AM, Alan Cox <gnomes@xxxxxxxxxxxxxxxxxxx> wrote:
>> On Tue, 13 Jun 2017 09:52:07 +0300
>> Tal Shorer <tal.shorer@xxxxxxxxx> wrote:
>>
>>> If a tty driver wants to notify the user of some exceptional event,
>>> such as a usb cdc acm device set_line_coding event, it needs a way to
>>> modify the mask returned by poll() and possible also add wait queues.
>>> In order to do that, we allow the driver to supply a poll() callback
>>> of its own, which will be called in n_tty_poll().
>>>
>>> Signed-off-by: Tal Shorer <tal.shorer@xxxxxxxxx>
>>
>> You might be in any line discipline so you need to support that for each
>> ldisc that supports poll(). Also what do I do when I get an exception
>> event - what does it mean, how do I understand that, are you proposing a
>> standardized meaning ? Have you checked whether that conflicts with SuS
>> v3 or POSIX ? What will it do with existing code.
>>
>> At the very least it probably has to be something you turn on, and you
>> might then want to model it on the pty/tty interface logic and extend
>> TIOCPKT a shade.
> That would cut it, but TIOCPKT is too coupled with having a linked tty.
> I could make acm behave like a pty (accept TIOCPKT and issue the
> ctrl_status bits), but for that I need n_tty to know that packet does
> not always mean a linked tty is present, and that in case it isn't we
> take our own ctrl_status bits instead of the link's. I could write a
> small (inline?) function to fetch the correct ctrl_status bits and put
> that in n_tty. Does that make sense?
Or (maybe) better yet, I can do a hack and have the acm tty point to
itself as the link, which means n_tty doesn't have to change at all.
Any thoughts?