RE: [PATCH v1] regmap: add generic indirect regmap support

[Zhang, Tianfei]

> -----Original Message-----
> From: Mark Brown <broonie@xxxxxxxxxx>
> Sent: Wednesday, June 8, 2022 6:43 PM
> To: matthew.gerlach@xxxxxxxxxxxxxxx
> Cc: Zhang, Tianfei <tianfei.zhang@xxxxxxxxx>; gregkh@xxxxxxxxxxxxxxxxxxx;
> rafael@xxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx; Wu, Hao
> <hao.wu@xxxxxxxxx>; trix@xxxxxxxxxx; Xu, Yilun <yilun.xu@xxxxxxxxx>;
> Weight, Russell H <russell.h.weight@xxxxxxxxx>
> Subject: Re: [PATCH v1] regmap: add generic indirect regmap support
> 
> On Tue, Jun 07, 2022 at 05:27:38PM -0700, matthew.gerlach@xxxxxxxxxxxxxxx
> wrote:
> > On Tue, 7 Jun 2022, Mark Brown wrote:
> > > On Mon, Jun 06, 2022 at 09:37:55PM -0400, Tianfei Zhang wrote:
> > > > From: Matthew Gerlach <matthew.gerlach@xxxxxxxxxxxxxxx>
> 
> > > > This patch adds support for regmap APIs that are intended to be
> > > > used by the drivers of some devices which support generic indirect
> > > > register access, for example PMCI (Platform Management Control
> > > > Interface) device, HSSI (High Speed Serial Interface) device in FPGA.
> 
> > > What is "generic indirect register access"?  I'm not clear what this
> > > is intended to support...
> 
> > "indirect register access" is a RTL design pattern we use in FPGAs
> > frequently.  The design pattern involves a small number of registers
> > plus a little handshake code to access various register spaces inside
> > the FPGA fabric.  The design pattern is "generic" in the sense that
> > the same small number of registers and handshake can be used with many
> > different IP components in the FPGA.  Historically, the bit
> > definitions and handshaking was slightly different for each IP
> > component.  This is an attempt at a consistent usage across IP components.
> 
> > Would a different name help?
> 
> This wouldn't address the major problem which is...
> 
> > > > +	writel(0, ctx->base + INDIRECT_CMD_OFF);
> > > > +	ret = readl_poll_timeout((ctx->base + INDIRECT_CMD_OFF), cmd,
> > > > +				 (!cmd), INDIRECT_INT_US,
> INDIRECT_TIMEOUT_US);
> > > > +	if (ret)
> > > > +		dev_err(ctx->dev, "%s timed out on clearing cmd 0x%xn",
> > > > +__func__, cmd);
> 
> > > ...and this doesn't look particularly generic, it looks like it's
> > > for some particular controller/bridge?
> 
> ...that this appears to be entirely specific to some particular device, it's got
> things like hard coded register addresses and timeouts which mean it can't be
> reused.

Yet, this is a register access hardware controller/bridge widely used in FPGA IP blocks, like PMCI, HSSI.
How about we change the patch title like this:

regmap: add indirect register controller support