Re: [RFT PATCH 2/9] drivers: qcom: rpmh-rsc: Document the register layout better

From: Evan Green <evgreen@chromium.org>
To: Doug Anderson <dianders@chromium.org>
Cc: Maulik Shah <mkshah@codeaurora.org>,
	Andy Gross <agross@kernel.org>,
	Bjorn Andersson <bjorn.andersson@linaro.org>,
	Rajendra Nayak <rnayak@codeaurora.org>,
	Matthias Kaehlcke <mka@chromium.org>,
	Stephen Boyd <swboyd@chromium.org>,
	Lina Iyer <ilina@codeaurora.org>,
	linux-arm-msm <linux-arm-msm@vger.kernel.org>,
	LKML <linux-kernel@vger.kernel.org>
Subject: Re: [RFT PATCH 2/9] drivers: qcom: rpmh-rsc: Document the register layout better
Date: Wed, 11 Mar 2020 11:49:44 -0700	[thread overview]
Message-ID: <CAE=gft7Z0aq5p5sPhBH7Q86T-pHmOL1i-E5sPBV2PQH3tFLMAQ@mail.gmail.com> (raw)
In-Reply-To: <CAD=FV=W5RGwu=ywtM7aCv3H-EpJ1i23S0awgVc8QtOsXtige4w@mail.gmail.com>

On Wed, Mar 11, 2020 at 8:28 AM Doug Anderson <dianders@chromium.org> wrote:
>
> Hi,
>
> On Wed, Mar 11, 2020 at 2:35 AM Maulik Shah <mkshah@codeaurora.org> wrote:
> >
> > Hi Doug,
> >
> > On 3/7/2020 5:29 AM, Douglas Anderson wrote:
> > > Perhaps it's just me, it took a really long time to understand what
> > > the register layout of rpmh-rsc was just from the #defines.
> > i don't understand why register layout is so important for you to understand?
> >
> > besides, i think all required registers are properly named with #define
> >
> > for e.g.
> > /* Register offsets */
> > #define RSC_DRV_IRQ_ENABLE              0x00
> > #define RSC_DRV_IRQ_STATUS              0x04
> > #define RSC_DRV_IRQ_CLEAR               0x08
> >
> > now when you want to enable/disable irq in driver code, its pretty simple to figure out
> > that we need to read/write at RSC_DRV_IRQ_ENABLE offset.
>
> It's not the specific layout that's important to me.  It's the
> relationships between everything.  In other words:
>
> a) one rpmh-rsc contains some registers and then a whole bunch of TCS blocks
>
> b) one TCS block contains some registers and space for up to 16 commands.
>
> c) each command has a handful of registers
>
> Nothing describes this in the existing code--you've gotta read through
> all the code and figure out how it's reading/writing registers to
> figure it out.
>
>
> > this seems unnecessary change to me, can you please drop this when you spin v2?
>
> In v2 I'll replace this with the prose below.  Personally I find this
> inferior to the struct definitions which are easier to read
> at-a-glance, but it seems like it would be less controversial...
>
> /*
>  * Here's a high level overview of how all the registers in RPMH work
>  * together:
>  *
>  * - The main rpmh-rsc address is the base of a register space that can
>  *   be used to find overall configuration of the hardware
>  *   (DRV_PRNT_CHLD_CONFIG).  Also found within the rpmh-rsc register
>  *   space are all the TCS blocks.  The offset of the TCS blocks is
>  *   specified in the device tree by "qcom,tcs-offset" and used to
>  *   compute tcs_base.
>  * - TCS blocks come one after another.  Type, count, and order are
>  *   specified by the device tree as "qcom,tcs-config".
>  * - Each TCS block has some registers, then space for up to 16 commands.
>  *   Note that though address space is reserved for 16 commands, fewer
>  *   might be present.  See ncpt (num cmds per TCS).
>  * - The first TCS block is special in that it has registers to control
>  *   interrupts (RSC_DRV_IRQ_XXX).  Space for these registers is reserved
>  *   in all TCS blocks to make the math easier, but only the first one
>  *   matters.
>  */
>
> I'll also move the documentation of "irq_clear", "cmd_wait_for_cmpl",
> "status", and "cmd_enable" next to the respective #defines.

Perhaps, I'm biased in the same way Doug is, but reviewing RPMh
patches is especially difficult because exactly this kind of
documentation is missing. The relationship and inheritance between
TCS, DRV, RSC, and how many of each are inside of what is never really
spelled out. This makes it difficult to reason about what new patches
are doing and whether or not it works. Having the register layout and
documentation helps convey these relationships, and helps others to
ensure that the code is correctly doing what it's intended to do.

-Evan