[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <de9f281c-6bbb-2686-a809-f53302d1a3e0@ti.com>
Date: Mon, 21 Jan 2019 11:30:57 +0530
From: Vignesh R <vigneshr@...com>
To: "Tudor.Ambarus@...rochip.com" <Tudor.Ambarus@...rochip.com>,
"bbrezillon@...nel.org" <bbrezillon@...nel.org>
CC: "marek.vasut@...il.com" <marek.vasut@...il.com>,
"robh+dt@...nel.org" <robh+dt@...nel.org>,
"linux-mtd@...ts.infradead.org" <linux-mtd@...ts.infradead.org>,
"devicetree@...r.kernel.org" <devicetree@...r.kernel.org>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v3 2/2] mtd: spi-nor: cadence-quadspi: Add support for
Octal SPI controller
On 20/01/19 8:58 PM, Tudor.Ambarus@...rochip.com wrote:
> Vignesh,
>
> On 01/20/2019 04:56 PM, Tudor.Ambarus@...rochip.com wrote:
>> Looks good.
>
> one more thing: in cqspi_read_setup() I see that the dummy cycles are adjusted
> if (f_pdata->inst_width != CQSPI_INST_TYPE_QUAD).
>
> Should you adjust the dummy cycles for octo too?
Driver does not claim support for SNOR_HWCAPS_PP_4_4_4 mode or
SNOR_HWCAPS_PP_8_8_8 mode. Therefore inst_width is always be
CQSPI_INST_TYPE_SINGLE (cqspi_set_protocol() always sets inst_width and
addr_width to CQSPI_INST_TYPE_SINGLE).
Code to support 4_4_4 or 8_8_8 mode is incomplete in the driver. Also,
using mode bits to satisfy dummy clk cycles does not seem right. My
guess is driver is broken, if inst_width is indeed CQSPI_INST_TYPE_QUAD
as we end up sending additional 8 dummy clock cycles over what is asked
in nor->read_dummy.
I will plan to fix these as part of moving driver over to spi-mem. But
above code should not impact 1-1-8 mode support.
> I tried to find a datasheet for the controller, without success though ...
>
You can find Octal SPI controller datasheet in AM65x SoC TRM here:
http://www.ti.com/lit/ug/spruid7b/spruid7b.pdf
Page 9449: 12.3.2 Octal Serial Peripheral Interface (OSPI)
--
Regards
Vignesh
Powered by blists - more mailing lists