Messages in this thread | | | Date | Fri, 6 Mar 2020 15:54:45 +0200 | From | Andy Shevchenko <> | Subject | Re: [PATCH 0/6] i2c: designeware: Add Baikal-T1 SoC DW I2C specifics support |
| |
First of all, I got only 3 out of 6 patches. Are you sure you properly prepared the series?
On Fri, Mar 06, 2020 at 04:19:49PM +0300, Sergey.Semin@baikalelectronics.ru wrote: > From: Serge Semin <fancer.lancer@gmail.com>
Same comment as per DMA series, try next time to link the cover letter to the series correctly.
> There are three DW I2C controllers embedded into the Baikal-T1 SoC. Two > of them are normal with standard DW I2C IP-core configurations and registers > accessible over normal MMIO space - so they are acceptable by the available > DW I2C driver with no modification.
> But there is a third, which is a bit > different. Its registers are indirectly accessed be means of "command/data > in/data out" registers tuple. In order to have it also supported by the DW > I2C driver, we must modify the code a bit. This is a main purpose of this > patchset. > > First of all traditionally we replaced the legacy plain text-based dt-binding > file with yaml-based one. Then we found and fixed a bug in the DW I2C FIFO size > detection algorithm which tried to do it too early before dw_readl/dw_writel > methods could be used.
So far so good (looks like, I think colleagues of mine and myself will review individual patches later on).
> Finally we introduced a platform-specific flag > ACCESS_INDIRECT, which would enable the indirect access to the DW I2C registers > implemented for one of the Baikal-T1 SoC DW I2C controllers. See the commit > message of the corresponding patch for details.
This is quite questionable. In Intel SoCs we have indirect I²C controllers to access (inside PMIC, for example). The approach used to do that is usually to have an IPC mechanism and specific bus controller driver. See i2c-cht-wc.c for instance.
I'm not sure if it makes a lot of duplication and if actually switching I²C DesignWare driver to regmap API will solve it. At least that is the second approach I would consider.
But I'll wait others to comment on this. We have to settle the solution before going further.
> This patchset is rebased and tested on the mainline Linux kernel 5.6-rc4: > commit 98d54f81e36b ("Linux 5.6-rc4").
`git format-patch --base ...` should do the job.
> Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru> > Signed-off-by: Alexey Malahov <Alexey.Malahov@baikalelectronics.ru>
Same comment as per UART patch. Who is the Alexey in relation to the work done?
-- With Best Regards, Andy Shevchenko
| |