On Fri, Jun 15, 2018 at 12:26 AM Sascha Hauer <s.hauer@xxxxxxxxxxxxxx> wrote:
>
> On Fri, Jun 15, 2018 at 06:58:05AM +0200, Sam Ravnborg wrote:
> > Hi Andrey.
> >
> > > +
> > > +/**
> > > + * __console_get_default - Return default output console
> > > + *
> > > + * Internal function used to determine which console to use for early
> > > + * output. It has the following two use-cases:
> > > + *
> > > + * 1. PBL, where it falls back onto console_ll and whatever it is
> > > + * set up to (either putc_ll or custome callback set with
> > > + * pbl_set_putc())
> > > + *
> > > + * 2. CONSOLE_SIMPLE, where it falls back onto console_ll (which in
> > > + * this case always boils down to a putc_ll() call) until first
> > > + * (and only) console is registered via console_register().
> > > + */
> > > +struct console_device *__console_get_default(void)
> >
> > If it is used as internal only then it should be static.
> > (Have not read the the remaining patches)
> >
> > > +{
> > > + /*
> > > + * Doing on-demand initialization here as opposed to in the
> > > + * definition of console_ll above has that advantage that on
> > > + * some architecutres (e.g. AArch64) it allows us to avoid
> > > + * additional relocation and makes it possible to use console
> > > + * infrastructure before any relocation happens
> > > + */
> > > + if (unlikely(!console_ll.putc))
> > > + console_ll.putc = __console_ll_putc;
> > > +
> > > + if (IS_ENABLED(__PBL__) || !console)
> > > + return &console_ll;
> > > +
> > > + return console;
> > > +}
> > > +
> > > +/**
> > > + * __console_set_putc - Early console initalization helper
> > > + *
> > > + * @cdev: Console device to initialize
> > > + * @putcf: putc() implementation to be used for this console
> > > + * @ctx: Context to pass to putc()
> > > + *
> > > + * Internal function used to initialize early/trivial console devices
> > > + * capable of only outputting a single character
> > > + */
> > > +void __console_set_putc(struct console_device *cdev,
> > > + putc_func_t putcf, void *ctx)
> > Likewise, and so on for the reimaining __xxx functions.
> >
> >
> > > +{
> > > + cdev->putc = (void *)putcf;
> > > + cdev->ctx = ctx;
> > > +}
> > > +
> > > +/**
> > > + * __cdev_putc - Internal .putc() callback dispatcher
> > > + *
> > > + * @cdev: Console device to use
> > > + * @c: Character to print
> > > + *
> > > + * Internal .putc() callback dispatcher needed to correctly select
> > > + * which context to pass.
> > > + *
> > > + * This is needed becuase when being used in PBL in conjunction with
> > > + * pbl_set_putc(), .putc() callback is expecting to receive a void *
> > > + * context that was registered earlier.
> > > + *
> > > + * In the "normal" world, however all of the .putc() callback are
> > > + * written with expectation of receiving struct console_device * as a
> > > + * first argument.
> > > + *
> > > + * Accomodation of both of those use-cases is the purpoese of this
> > > + * function
> > > + */
> > > +void __cdev_putc(struct console_device *cdev, char c)
> > > +{
> > > + void *ctx = cdev->ctx ? : cdev;
> > Looks strange to me.
> > What is ctx assigned in the "true" case?
> > Maybe there is some subtle C rule that I have missed/forgot.
> > Point is, if this is valid code it may also make other puzzeled.
>
> It's a shortcut for a ? a : b and I like kind of like it. I'm also still
> not there that I can read fluently over it, but I think it's worth
> learning it.
>
> What puzzles me here is that the putc callback is either passed a
> context pointer or cdev itself if there is no context. I haven't read
> the full patch yet, but that is one thing that I don't like.
>
This is needed to be able to use the same generic console code with
functions like pbl_set_putc() and psci_set_putc() which, unlike the
rest of console world, take a void * context that is expected to be
passed to putc() callback.
Thanks,
Andrey Smirnov
_______________________________________________
barebox mailing list
barebox@xxxxxxxxxxxxxxxxxxx
http://lists.infradead.org/mailman/listinfo/barebox
