From mboxrd@z Thu Jan  1 00:00:00 1970
From: Marek Vasut <marex@denx.de>
Date: Mon, 3 Aug 2015 02:16:24 +0200
Subject: [U-Boot] [PATCH 1/2] gpio: Add DW APB GPIO driver
In-Reply-To: <CAPnjgZ03Sc_WZH-Se879h3-MaE5B_5HPh9ve6_G=xz=JXGV+7Q@mail.gmail.com>
References: <1438029890-9901-1-git-send-email-marex@denx.de>
 <201508030019.50295.marex@denx.de>
 <CAPnjgZ03Sc_WZH-Se879h3-MaE5B_5HPh9ve6_G=xz=JXGV+7Q@mail.gmail.com>
Message-ID: <201508030216.24431.marex@denx.de>
List-Id: <u-boot.lists.denx.de>
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
To: u-boot@lists.denx.de

On Monday, August 03, 2015 at 01:38:28 AM, Simon Glass wrote:
> Hi Marek,

Hi Simon,

[...]

> >> > +               if (!fdtdec_get_bool(blob, node, "gpio-controller"))
> >> > +                       continue;
> >> > +
> >> > +               plat = NULL;
> >> > +               plat = calloc(1, sizeof(*plat));
> >> 
> >> I suppose this should use devm_alloc() now.
> > 
> > Is that even in u-boot/master ? Also, I'm not sure it's a good idea to
> > put this in if I use this driver in SPL.
> 
> Yes it's very new. Only in dm/master.
> 
> It's only compiled in when enabled, so you can still use it in SPL.
> 
> Up to you. At some point I think we should convert all driver allocs
> to use devm so that we know what allocation is going on.

When is this landing in master ? I don't mind either way, but I can do
a linting pass on the socfpga when things settle down too.

> >> > +               if (!plat)
> >> > +                       return -ENOMEM;
> >> > +
> >> > +               plat->base = base;
> >> > +               plat->bank = bank;
> >> > +               plat->pins = fdtdec_get_int(blob, node,
> >> > "snps,nr-gpios", 0); +               snprintf(plat->name,
> >> > sizeof(plat->name) - 1, "%s-bank%i-", +                        name,
> >> > bank);
> >> 
> >> Why such a long name? That's going to be a pain to type in the 'gpio'
> >> command.
> > 
> > Do you have a suggestion please ?
> 
> A, B, C is good if you have <=26 banks. Tegra does that.
> 
> Exynos does PA0, PA1, PB0, PC0, PC1, etc.

How do I identify which one is PA0/PA1/PA2 , shall I perform some transformation
on the register address of the GPIO block or example ? But how can I assure that 
if the next SoCFPGA has these addresses completely different, these GPIO numbers
will be stable ? Isn't it better to be explicit about the GPIO block ID then ?

Also, remember that I have an FPGA in the same package, which has a lot of I/O.

> > Also, I can as well use "gpio <operation> N" , where N is a number.

What about this ? How does indexing the GPIOs with plain number fit into the 
picture please ?

[...]