Re: [patch 05/14] mfd: PCAP2 driver

From: Daniel Ribeiro <drwyrm-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
To: David Brownell <david-b-yBeKhBN/0LDR7s880joybQ@public.gmane.org>
Cc: Stefan Schmidt <stefan-4Bgg8jF3iZdg9hUCZPvPmw@public.gmane.org>,
	eric.y.miao-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org,
	linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
	spi-devel-general-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org,
	sameo-RWuK6r/cQWRpLGFMi4vTTA@public.gmane.org,
	linux-arm-kernel-xIg/pKzrS19vn6HldHNs0ANdhmdF6hFW@public.gmane.org,
	Stefan Schmidt
	<stefan-OrPQZGeq07wqhVmZOOOmNx2eb7JE58TQ@public.gmane.org>
Subject: Re: [patch 05/14] mfd: PCAP2 driver
Date: Sun, 23 Nov 2008 01:38:57 -0200	[thread overview]
Message-ID: <1227411537.3148.22.camel@brutus> (raw)
In-Reply-To: <200811221819.53186.david-b-yBeKhBN/0LDR7s880joybQ@public.gmane.org>

Em Sáb, 2008-11-22 às 18:19 -0800, David Brownell escreveu:
> SPI-specific bits:
> 
>  - I'd have to see the patch.  The last one I saw still didn't
>    list a SPI_MASTER dependency for the core MFD module.

The next one will.

>  - You should make ezx_pcap_write() and ezx_pcap_read() switch
>    over to spi_write_then_read(), to ensure it's never doing
>    DMA to/from the stack.  I see a byte-order dependency too...

I tried spi_write_then_read before, but it didn't work. I supposed it
was because it was doing 2 transfers as the second transfer rx_buf
always came zeroed. I see that commit
f9b90e39cbc5c4d6ef60022fd1f25d541df0aad1 changed it to do a single
transfer, so i will try it again.

>  - For general paranoia, the probe() should abort if pcap.spi is
>    already set ... and its cleanup path, plus remove(), should
>    null that pointer.

Ok.

>  - If you're going to mark the probe() as __devinit, then mark
>    the remove() as __devexit and use __devexit_p() in the driver
>    struct.

Ok. Shouldn't i use __init instead?

> Other comments about the pcap2 core:

> - Those show_regs/store_regs calls would IMO make more sense
>    in debugfs than in sysfs.

I will just remove those for now.

>  - The set_vreg() stuff would seem to make more sense in a
>    regulator subdevice and driver ... but that's more of a
>    general driver structure thing, and might be fixed later.
> 
>  - Andrew seems to always want a comment explaining why the
>    IRQ handler for I2C and SPI devices needs to queue_work().
>    Maybe the threaded IRQ stuff will help there...
> 
>  - The mask_event()/unmask_event() stuff looks like you're
>    more or less reinventing a baby "struct irq_chip", with
>    register_event() instead of request_irq().

> Re the IRQ stuff, this looks more like what i2c/chips/menelaus.c
> did than mfd/twl4030-irq.c ... in general I think it's better to
> pursue the latter approach, making genirq handle such stuff.
> (Even though it's kind of awkward to use it for I2C or SPI based
> interrupt controllers just now.)

I didn't know i could use genirq with spi. In fact i was using genirq
before, when the driver was using ssp.c. Now, looking at twl4030-irq.c
its clear how i should have done this. Thanks!!

>  - And the ADC sysfs support isn't supporting hwmon models.
>    (Neither does the twl4030 ADC support, but that's not been
>    submitted for mainline yet either...)

I will look into this after i finish the irq stuff.

This is a lot of stuff, thanks for the review. :)

-- 
Daniel Ribeiro

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
spi-devel-general mailing list
spi-devel-general@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/spi-devel-general