Re: [PATCH V4 2/2] mtd: spi-nor: Add driver for Cadence Quad SPI Flash Controller.

[Graham Moore]

On 05/20/2015 05:26 PM, Brian Norris wrote:
> On Mon, Mar 23, 2015 at 08:36:22AM -0500, Graham Moore wrote:
...
I'm fixing the sparse, smatch, unused variables, etc.
...
> > +	if (n_tx && opcode == SPINOR_OP_WD_EVCR &&
> > +	    !(txbuf[0] & EVCR_QUAD_EN_MICRON)) {
> > +		struct cqspi_flash_pdata *f_pdata;
> > +
> > +		f_pdata = &cqspi->f_pdata[cqspi->current_cs];
> > +		f_pdata->quad_mode_set = 1;
>
> No, please don't do this. The whole point of the spi-nor layer is that
> we don't want low-level drivers having to snoop every type of flash
> command. We want to have proper feedback between spi-nor.c and your
> driver, so you *know* when you're using a quad-enabled device.
>
> Food for thought: what happens for a non-Micron SPI flash that uses quad
> mode?
>
> Is it sufficient to just check if nor->flash_read == SPI_NOR_QUAD?

I *said* it was ugly :)

Here's the deal.  The Cadence controller has its own quad mode setting, 
which must match the setting of the chip (or the quad-mode command.) 
Before the addition of micron_quad_enable(), spi-nor was using a 
quad-mode read command.  That was all good, because we could use 
nor->flash_read == SPI_NOR_QUAD to set the Cadence controller properly.

However, micron_quad_enable() sets quad mode in the Micron chip.  At 
that point, all communication with the chip must be in quad mode, even 
the very next read-status command.  The micron_quad_enable() function 
reads status and EVCR register to see if the quad-mode command 'took'. 
This all happens before nor->flash_read is set to SPI_NOR_QUAD.  If I 
don't set the Cadence controller into quad mode immediately after the 
chip is set into quad mode, then status is garbage and the probe fails.

I agree that any other chip with a quad mode, separate from quad 
commands, will be jacked up too.

So...how can I know when to set the Cadence controller into quad mode?
 Perhaps the micron_quad_enable() should set nor->flash_read = 
SPI_NOR_QUAD as soon as it sends the command to the chip?

...

-Graham
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/