From: "Brandt, Oliver - Lenze" <Oliver.Brandt@lenze.com>
To: "linux-mmc@vger.kernel.org" <linux-mmc@vger.kernel.org>
Cc: "ulfh@kernel.org" <ulfh@kernel.org>,
"jh80.chung@samsung.com" <jh80.chung@samsung.com>,
"shawn.lin@rock-chips.com" <shawn.lin@rock-chips.com>,
"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH v2] mmc: dw_mmc-pltfm: Fix usage of phase_map for SoCFPGA
Date: Thu, 12 Nov 2026 16:40:00 +0000 [thread overview]
Message-ID: <541c0004866a37c4ae94114b53932e087ca2889a.camel@lenze.com> (raw)
In-Reply-To: <140317dff0feb9d62b580b1e3170ccb10b8af181.camel@lenze.com>
On Thu, 2026-11-12 at 09:45 +0000, Brandt, Oliver - Lenze wrote:
> With a kernel 7.3.3 we saw the following on a SoCFPGA (Cyclone 5) based
> board during initialization when using a high-speed SD-card:
>
> dw_mmc ff704000.mmc: DW MMC controller at irq 53,32 bit host data width,1024 deep fifo
> dw_mmc ff704000.mmc: Got CD GPIO
> dw_mmc ff704000.mmc: Got WP GPIO
> mmc_host mmc0: Bus speed = 50000000Hz (req 400000Hz, actual 396825HZ div = 63)
> mmc_host mmc0: Bus speed = 50000000Hz (req 50000000Hz, actual 50000000HZ div = 0)
> mmc0: error -110 whilst initialising SD card
> mmc_host mmc0: Bus speed = 50000000Hz (req 300000Hz, actual 297619HZ div = 84)
> mmc_host mmc0: Bus speed = 50000000Hz (req 50000000Hz, actual 50000000HZ div = 0)
> mmc0: error -110 whilst initialising SD card
> mmc_host mmc0: Bus speed = 50000000Hz (req 200000Hz, actual 200000HZ div = 125)
> mmc_host mmc0: Bus speed = 50000000Hz (req 50000000Hz, actual 50000000HZ div = 0)
> mmc0: error -110 whilst initialising SD card
> mmc_host mmc0: Bus speed = 50000000Hz (req 100000Hz, actual 100000HZ div = 250)
> mmc_host mmc0: Bus speed = 50000000Hz (req 50000000Hz, actual 50000000HZ div = 0)
> mmc0: error -110 whilst initialising SD card
>
> The SD-card was not usable in the following.
>
> Reason: Since commit 3f1628baa51e ("mmc: dw_mmc-pltfm: Use phase_map for
> SoCFPGA clock phase configuration") host->phase_map is used in order to
> get the phase settings for MMC_TIMING_SD_HS. This is done in drv_data->init
> (dw_mci_socfpga_priv_init()) which is called before dw_mci_init_host(). As
> host->phase_map will be initialized only during dw_mci_init_host(), we'll
> call dw_mci_socfpga_priv_init() without the needed data.
>
> Do it like other users of dw_mci_platfom and use drv_data->set_ios.
>
> Fixes: 3f1628baa51e ("mmc: dw_mmc-pltfm: Use phase_map for SoCFPGA clock phase configuration")
> Cc: stable@vger.kernel.org
> Signed-off-by: Oliver Brandt <oliver.brandt@lenze.com>
> ---
> Changes in v2:
> - Add 'return' in case of missing regmap (Sashiko AI).
> - Avoid repeated execution (Sashiko AI).
> - Write register 'hs_timing' even if MMC_TIMING_SD_HS is not requested
> (Sashiko AI).
> - Link to v1: https://lore.kernel.org/linux-mmc/2a03a222c4a9d330ebc7157de09ec33f72875f28.camel@lenze.com/T/#u
>
> drivers/mmc/host/dw_mmc-pltfm.c | 14 +++++++-------
> 1 file changed, 7 insertions(+), 7 deletions(-)
>
> diff --git a/drivers/mmc/host/dw_mmc-pltfm.c b/drivers/mmc/host/dw_mmc-pltfm.c
> index cf38bb118dc2c..b51574a0a1fe1 100644
> --- a/drivers/mmc/host/dw_mmc-pltfm.c
> +++ b/drivers/mmc/host/dw_mmc-pltfm.c
> @@ -56,22 +56,23 @@ int dw_mci_pltfm_register(struct platform_device *pdev,
> }
> EXPORT_SYMBOL_GPL(dw_mci_pltfm_register);
>
> -static int dw_mci_socfpga_priv_init(struct dw_mci *host)
> +static void dw_mci_socfpga_priv_set_ios(struct dw_mci *host, struct mmc_ios *ios)
> {
> struct device_node *np = host->dev->of_node;
> struct mmc_clk_phase phase;
> struct regmap *sys_mgr_base_addr;
> u32 reg_offset, reg_shift;
> int hs_timing;
> + static int phase_already_set;
>
> phase = host->phase_map.phase[MMC_TIMING_SD_HS];
> - if (!phase.valid)
> - return 0;
> + if (!phase.valid || phase_already_set)
> + return;
>
> sys_mgr_base_addr = altr_sysmgr_regmap_lookup_by_phandle(np, "altr,sysmgr-syscon");
> if (IS_ERR(sys_mgr_base_addr)) {
> dev_warn(host->dev, "clk-phase-sd-hs was specified, but failed to find altr,sys-mgr regmap!\n");
> - return 0;
> + return;
> }
>
> of_property_read_u32_index(np, "altr,sysmgr-syscon", 1, ®_offset);
> @@ -82,12 +83,11 @@ static int dw_mci_socfpga_priv_init(struct dw_mci *host)
>
> hs_timing = SYSMGR_SDMMC_CTRL_SET(phase.in_deg, phase.out_deg, reg_shift);
> regmap_write(sys_mgr_base_addr, reg_offset, hs_timing);
> -
> - return 0;
> + phase_already_set = 1;
> }
>
> static const struct dw_mci_drv_data socfpga_drv_data = {
> - .init = dw_mci_socfpga_priv_init,
> + .set_ios = dw_mci_socfpga_priv_set_ios,
> };
>
> static const struct of_device_id dw_mci_pltfm_match[] = {
Sashiko pointed out that using a static variable is "not a good
idea". Even if, at the moment, we don't have any in-tree DTS with
more than one instance of the controller, a cleaner solution
seems to be needed.
I've already prepared a new (hopefully better) patch. This one
should be ignored. Sorry for the noise.
Regards,
Oliver
prev parent reply other threads:[~2026-11-12 16:40 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-11-12 9:45 Brandt, Oliver - Lenze
2026-11-12 16:40 ` Brandt, Oliver - Lenze [this message]
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=541c0004866a37c4ae94114b53932e087ca2889a.camel@lenze.com \
--to=oliver.brandt@lenze.com \
--cc=jh80.chung@samsung.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mmc@vger.kernel.org \
--cc=shawn.lin@rock-chips.com \
--cc=ulfh@kernel.org \
/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
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®