All of lore.kernel.org
 help / color / mirror / Atom feed
From: Zong Li <zong.li@sifive.com>
To: Sean Anderson <seanga2@gmail.com>
Cc: rick@andestech.com, Leo Liang <ycliang@andestech.com>,
	 Bin Meng <bmeng.cn@gmail.com>,
	sjg@chromium.org, Green Wan <green.wan@sifive.com>,
	 Paul Walmsley <paul.walmsley@sifive.com>,
	u-boot@lists.denx.de
Subject: Re: [PATCH 2/2] board: sifive: use ccache driver instead of helper function
Date: Thu, 29 Jul 2021 20:26:55 +0800	[thread overview]
Message-ID: <CANXhq0qFu-MfMgLOsO_he9N=xXoKcqAtCsR5P8S_r09_t4epDA@mail.gmail.com> (raw)
In-Reply-To: <d2431440-668d-7beb-9e56-eafb730f986d@gmail.com>

On Wed, Jul 28, 2021 at 11:18 PM Sean Anderson <seanga2@gmail.com> wrote:
>
> On 7/28/21 3:25 AM, Zong Li wrote:
> > On Wed, Jul 28, 2021 at 12:29 PM Sean Anderson <seanga2@gmail.com> wrote:
> >>
> >> On 7/27/21 4:54 AM, Zong Li wrote:
> >>> Invokes the generic cache_enable interface to execute the relative
> >>> implementation in SiFive ccache driver.
> >>>
> >>> Signed-off-by: Zong Li <zong.li@sifive.com>
> >>> ---
> >>>    arch/riscv/cpu/fu540/Kconfig              |  1 +
> >>>    arch/riscv/cpu/fu540/cache.c              | 62 ++++++++---------------
> >>>    arch/riscv/cpu/fu740/Kconfig              |  1 +
> >>>    arch/riscv/cpu/fu740/cache.c              | 60 +++++++---------------
> >>>    arch/riscv/include/asm/arch-fu540/cache.h |  2 +-
> >>>    arch/riscv/include/asm/arch-fu740/cache.h |  2 +-
> >>>    board/sifive/unleashed/unleashed.c        | 10 +---
> >>>    board/sifive/unmatched/unmatched.c        |  9 +---
> >>>    8 files changed, 45 insertions(+), 102 deletions(-)
> >>>
> >>> diff --git a/arch/riscv/cpu/fu540/Kconfig b/arch/riscv/cpu/fu540/Kconfig
> >>> index 05463b2625..1f50b823ed 100644
> >>> --- a/arch/riscv/cpu/fu540/Kconfig
> >>> +++ b/arch/riscv/cpu/fu540/Kconfig
> >>> @@ -19,6 +19,7 @@ config SIFIVE_FU540
> >>>        imply SMP
> >>>        imply CLK_SIFIVE
> >>>        imply CLK_SIFIVE_PRCI
> >>> +     imply SIFIVE_CACHE_CCACHE
> >>>        imply SIFIVE_SERIAL
> >>>        imply MACB
> >>>        imply MII
> >>> diff --git a/arch/riscv/cpu/fu540/cache.c b/arch/riscv/cpu/fu540/cache.c
> >>> index 0fc4ef6c00..3c754c4614 100644
> >>> --- a/arch/riscv/cpu/fu540/cache.c
> >>> +++ b/arch/riscv/cpu/fu540/cache.c
> >>> @@ -1,55 +1,33 @@
> >>>    // SPDX-License-Identifier: GPL-2.0+
> >>>    /*
> >>> - * Copyright (C) 2020 SiFive, Inc
> >>> + * Copyright (C) 2020 - 2021 SiFive, Inc
> >>>     *
> >>>     * Authors:
> >>>     *   Pragnesh Patel <pragnesh.patel@sifive.com>
> >>>     */
> >>>
> >>>    #include <common.h>
> >>> -#include <asm/global_data.h>
> >>> -#include <asm/io.h>
> >>> -#include <linux/bitops.h>
> >>> +#include <cache.h>
> >>> +#include <dm.h>
> >>>
> >>> -/* Register offsets */
> >>> -#define L2_CACHE_CONFIG      0x000
> >>> -#define L2_CACHE_ENABLE      0x008
> >>> -
> >>> -#define MASK_NUM_WAYS        GENMASK(15, 8)
> >>> -#define NUM_WAYS_SHIFT       8
> >>> -
> >>> -DECLARE_GLOBAL_DATA_PTR;
> >>> -
> >>> -int cache_enable_ways(void)
> >>> +int sifive_ccache_enable_ways(void)
> >>>    {
> >>
> >> Is there any reason this function is duplicated? See below for further comments.
> >
> > Sorry, I don't completely understand about duplication here. Do you
> > mean why we need this function? or why is it present in both unleashed
> > and unmatched?
>
> Why it is present in both. Shouldn't it be present in some shared file?
>

I considered that before, the places to put this ccache function might
be either 'arch/riscv/lib/' or 'arch/riscv/cpu/generic', these two
folders put some common stuff for all riscv platforms, but this ccache
function is actually used in sifive platform only, I'm not sure if it
is suitable to put the function into them, so I just followed the
original tree. OTOH, the compiler will drop the unused code, so it
might be OK to put it into the folders above, and not influence other
platforms. Do you have any preference or any recommendation?

> >
> >>
> >>> -     const void *blob = gd->fdt_blob;
> >>> -     int node;
> >>> -     fdt_addr_t base;
> >>> -     u32 config;
> >>> -     u32 ways;
> >>> -
> >>> -     volatile u32 *enable;
> >>> -
> >>> -     node = fdt_node_offset_by_compatible(blob, -1,
> >>> -                                          "sifive,fu540-c000-ccache");
> >>> -
> >>> -     if (node < 0)
> >>> -             return node;
> >>> -
> >>> -     base = fdtdec_get_addr_size_auto_parent(blob, 0, node, "reg", 0,
> >>> -                                             NULL, false);
> >>> -     if (base == FDT_ADDR_T_NONE)
> >>> -             return FDT_ADDR_T_NONE;
> >>> -
> >>> -     config = readl((volatile u32 *)base + L2_CACHE_CONFIG);
> >>> -     ways = (config & MASK_NUM_WAYS) >> NUM_WAYS_SHIFT;
> >>> -
> >>> -     enable = (volatile u32 *)(base + L2_CACHE_ENABLE);
> >>> +     struct udevice *dev;
> >>> +     int ret;
> >>> +
> >>> +     ret = uclass_get_device_by_driver(UCLASS_CACHE,
> >>> +                                       DM_DRIVER_GET(sifive_ccache),
> >>> +                                       &dev);
> >>> +     if (ret) {
> >>> +             pr_debug("%s: could not enable cache ways\n", __func__);
> >>> +             return ret;
> >>> +     }
> >>> +
> >>> +     ret = cache_enable(dev);
> >>> +     if (ret) {
> >>> +             pr_debug("%s: ccache enable filed\n", __func__);
> >>> +             return ret;
> >>> +     }
> >>>
> >>> -     /* memory barrier */
> >>> -     mb();
> >>> -     (*enable) = ways - 1;
> >>> -     /* memory barrier */
> >>> -     mb();
> >>>        return 0;
> >>>    }
> >>> diff --git a/arch/riscv/cpu/fu740/Kconfig b/arch/riscv/cpu/fu740/Kconfig
> >>> index 8e54310b9c..87d8016c88 100644
> >>> --- a/arch/riscv/cpu/fu740/Kconfig
> >>> +++ b/arch/riscv/cpu/fu740/Kconfig
> >>> @@ -19,6 +19,7 @@ config SIFIVE_FU740
> >>>        imply SMP
> >>>        imply CLK_SIFIVE
> >>>        imply CLK_SIFIVE_PRCI
> >>> +     imply SIFIVE_CACHE_CCACHE
> >>>        imply SIFIVE_SERIAL
> >>>        imply MACB
> >>>        imply MII
> >>> diff --git a/arch/riscv/cpu/fu740/cache.c b/arch/riscv/cpu/fu740/cache.c
> >>> index 680955c9e3..1d6c295d98 100644
> >>> --- a/arch/riscv/cpu/fu740/cache.c
> >>> +++ b/arch/riscv/cpu/fu740/cache.c
> >>> @@ -7,49 +7,27 @@
> >>>     */
> >>>
> >>>    #include <common.h>
> >>> -#include <asm/io.h>
> >>> -#include <linux/bitops.h>
> >>> -#include <asm/global_data.h>
> >>> +#include <cache.h>
> >>> +#include <dm.h>
> >>>
> >>> -/* Register offsets */
> >>> -#define L2_CACHE_CONFIG      0x000
> >>> -#define L2_CACHE_ENABLE      0x008
> >>> -
> >>> -#define MASK_NUM_WAYS        GENMASK(15, 8)
> >>> -#define NUM_WAYS_SHIFT       8
> >>> -
> >>> -DECLARE_GLOBAL_DATA_PTR;
> >>> -
> >>> -int cache_enable_ways(void)
> >>> +int sifive_ccache_enable_ways(void)
> >>>    {
> >>> -     const void *blob = gd->fdt_blob;
> >>> -     int node;
> >>> -     fdt_addr_t base;
> >>> -     u32 config;
> >>> -     u32 ways;
> >>> -
> >>> -     volatile u32 *enable;
> >>> -
> >>> -     node = fdt_node_offset_by_compatible(blob, -1,
> >>> -                                          "sifive,fu740-c000-ccache");
> >>> -
> >>> -     if (node < 0)
> >>> -             return node;
> >>> -
> >>> -     base = fdtdec_get_addr_size_auto_parent(blob, 0, node, "reg", 0,
> >>> -                                             NULL, false);
> >>> -     if (base == FDT_ADDR_T_NONE)
> >>> -             return FDT_ADDR_T_NONE;
> >>> -
> >>> -     config = readl((volatile u32 *)base + L2_CACHE_CONFIG);
> >>> -     ways = (config & MASK_NUM_WAYS) >> NUM_WAYS_SHIFT;
> >>> -
> >>> -     enable = (volatile u32 *)(base + L2_CACHE_ENABLE);
> >>> +     struct udevice *dev;
> >>> +     int ret;
> >>> +
> >>> +     ret = uclass_get_device_by_driver(UCLASS_CACHE,
> >>> +                                       DM_DRIVER_GET(sifive_ccache),
> >>> +                                       &dev);
> >>> +     if (ret) {
> >>> +             pr_debug("%s: could not enable cache ways\n", __func__);
> >>> +             return ret;
> >>
> >> return log_msg_ret(...);
> >>
> >> (and you don't have to add a __func__ parameter if you use that function)
> >
> > Okay, let me change the function. Thanks.
> >
> >>
> >>> +     }
> >>> +
> >>> +     ret = cache_enable(dev);
> >>> +     if (ret) {
> >>> +             pr_debug("%s: ccache enable filed\n", __func__);
> >>
> >> failed; ditto
> >
> > okay.
> >
> >>
> >>> +             return ret;
> >>> +     }
> >>>
> >>> -     /* memory barrier */
> >>> -     mb();
> >>> -     (*enable) = ways - 1;
> >>> -     /* memory barrier */
> >>> -     mb();
> >>
> >> Is this handled by the __iowmb/__iormb in writel/readl in the driver?
> >>
> >
> > Yes, the ccache driver uses writel to write the register, and writel
> > will invoke __iowmb.
> >
> >>>        return 0;
> >>>    }
> >>> diff --git a/arch/riscv/include/asm/arch-fu540/cache.h b/arch/riscv/include/asm/arch-fu540/cache.h
> >>> index 135a17c679..c252eb64d1 100644
> >>> --- a/arch/riscv/include/asm/arch-fu540/cache.h
> >>> +++ b/arch/riscv/include/asm/arch-fu540/cache.h
> >>> @@ -9,6 +9,6 @@
> >>>    #ifndef _CACHE_SIFIVE_H
> >>>    #define _CACHE_SIFIVE_H
> >>>
> >>> -int cache_enable_ways(void);
> >>> +int sifive_ccache_enable_ways(void);
> >>>
> >>>    #endif /* _CACHE_SIFIVE_H */
> >>> diff --git a/arch/riscv/include/asm/arch-fu740/cache.h b/arch/riscv/include/asm/arch-fu740/cache.h
> >>> index 7d4fe9942b..8c456e3658 100644
> >>> --- a/arch/riscv/include/asm/arch-fu740/cache.h
> >>> +++ b/arch/riscv/include/asm/arch-fu740/cache.h
> >>> @@ -9,6 +9,6 @@
> >>>    #ifndef _CACHE_SIFIVE_H
> >>>    #define _CACHE_SIFIVE_H
> >>>
> >>> -int cache_enable_ways(void);
> >>> +int sifive_ccache_enable_ways(void);
> >>>
> >>>    #endif /* _CACHE_SIFIVE_H */
> >>> diff --git a/board/sifive/unleashed/unleashed.c b/board/sifive/unleashed/unleashed.c
> >>> index 43027f0b54..12e61ec85f 100644
> >>> --- a/board/sifive/unleashed/unleashed.c
> >>> +++ b/board/sifive/unleashed/unleashed.c
> >>> @@ -126,14 +126,6 @@ void *board_fdt_blob_setup(void)
> >>>
> >>>    int board_init(void)
> >>>    {
> >>> -     int ret;
> >>> -
> >>>        /* enable all cache ways */
> >>> -     ret = cache_enable_ways();
> >>> -     if (ret) {
> >>> -             debug("%s: could not enable cache ways\n", __func__);
> >>> -             return ret;
> >>
> >> ditto, though here I would just skip the error message since you reported it above already
> >>
> >
> > It seems to me that this part was dropped in this patch.
>
> Ah, whoops. Ignore this comment, then.
>
> >
> >
> >>> -     }
> >>> -
> >>> -     return 0;
> >>> +     return sifive_ccache_enable_ways();
> >>>    }
> >>> diff --git a/board/sifive/unmatched/unmatched.c b/board/sifive/unmatched/unmatched.c
> >>> index 2f5629b578..d27c4d3e88 100644
> >>> --- a/board/sifive/unmatched/unmatched.c
> >>> +++ b/board/sifive/unmatched/unmatched.c
> >>> @@ -23,13 +23,6 @@ void *board_fdt_blob_setup(void)
> >>>
> >>>    int board_init(void)
> >>>    {
> >>> -     int ret;
> >>> -
> >>>        /* enable all cache ways */
> >>> -     ret = cache_enable_ways();
> >>> -     if (ret) {
> >>> -             debug("%s: could not enable cache ways\n", __func__);
> >>
> >> ditto
> >
> > Okay.
> >
> >>
> >>> -             return ret;
> >>> -     }
> >>> -     return 0;
> >>> +     return sifive_ccache_enable_ways();
> >>>    }
> >>>
> >>
> >> --Sean
>

  reply	other threads:[~2021-07-29 12:27 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-07-27  8:54 [PATCH 1/2] cache: add sifive composable cache driver Zong Li
2021-07-27  8:54 ` [PATCH 2/2] board: sifive: use ccache driver instead of helper function Zong Li
2021-07-28  4:29   ` Sean Anderson
2021-07-28  7:25     ` Zong Li
2021-07-28 15:18       ` Sean Anderson
2021-07-29 12:26         ` Zong Li [this message]
2021-07-30  9:16           ` Zong Li
2021-07-28  4:23 ` [PATCH 1/2] cache: add sifive composable cache driver Sean Anderson
2021-07-28 10:40   ` Zong Li

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to='CANXhq0qFu-MfMgLOsO_he9N=xXoKcqAtCsR5P8S_r09_t4epDA@mail.gmail.com' \
    --to=zong.li@sifive.com \
    --cc=bmeng.cn@gmail.com \
    --cc=green.wan@sifive.com \
    --cc=paul.walmsley@sifive.com \
    --cc=rick@andestech.com \
    --cc=seanga2@gmail.com \
    --cc=sjg@chromium.org \
    --cc=u-boot@lists.denx.de \
    --cc=ycliang@andestech.com \
    --subject='Re: [PATCH 2/2] board: sifive: use ccache driver instead of helper function' \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.