From: Bjorn Andersson <bjorn.andersson@linaro.org>
To: Loic Poulain <loic.poulain@linaro.org>,
Mike Tipton <quic_mdtipton@quicinc.com>
Cc: agross@kernel.org, sboyd@kernel.org, shawn.guo@linaro.org,
linux-arm-msm@vger.kernel.org, linux-clk@vger.kernel.org
Subject: Re: [PATCH] clk: qcom: gcc-qcm2290: CLK_OPS_PARENT_ENABLE flag for rcg2 clocks
Date: Mon, 20 Dec 2021 15:41:34 -0800 [thread overview]
Message-ID: <YcEUrtQuJ5CGF2RW@ripper> (raw)
In-Reply-To: <1639994084-25447-1-git-send-email-loic.poulain@linaro.org>
On Mon 20 Dec 01:54 PST 2021, Loic Poulain wrote:
> When a rcg2 clock migrates to a new parent, both the old and new
> parent clocks must be enabled to complete the transition. This can
> be automatically performed by the clock core when a clock is flagged
> with CLK_OPS_PARENT_ENABLE.
>
> Without this, we may hit rate update failures:
> gcc_sdcc2_apps_clk_src: rcg didn't update its configuration.
> WARNING: CPU: 1 PID: 82 at drivers/clk/qcom/clk-rcg2.c:122 update_config+0xe0/0xf0
>
> Fixes: 496d1a13d405 ("clk: qcom: Add Global Clock Controller driver for QCM2290")
> Signed-off-by: Loic Poulain <loic.poulain@linaro.org>
> ---
> drivers/clk/qcom/gcc-qcm2290.c | 20 ++++++++++++++++++++
> 1 file changed, 20 insertions(+)
>
> diff --git a/drivers/clk/qcom/gcc-qcm2290.c b/drivers/clk/qcom/gcc-qcm2290.c
> index b6fa7b8..9e1d88e 100644
> --- a/drivers/clk/qcom/gcc-qcm2290.c
> +++ b/drivers/clk/qcom/gcc-qcm2290.c
> @@ -674,6 +674,7 @@ static struct clk_rcg2 gcc_usb30_prim_mock_utmi_clk_src = {
> .name = "gcc_usb30_prim_mock_utmi_clk_src",
> .parent_data = gcc_parents_0,
> .num_parents = ARRAY_SIZE(gcc_parents_0),
> + .flags = CLK_OPS_PARENT_ENABLE,
This seems like a correct fix for the obvious problem that we might end
up invoking clk_set_rate() and clk_set_parent() for these clocks while
their (and thereby themselves - in a software sense) are disabled.
However, clocks that downstream are marked "enable_safe_config", may in
addition be enabled by the hardware, behind out back. As such we must
ensure that they always have a valid configuration, we do this by
"parking" them on CXO whenever we're going to disable their parent
clocks.
Upstream we handle this by using the clk_rcg2_shared_ops, which will do
exactly this.
PS. Unfortunately these RCGs doesn't allow us to reliably implement
is_enabled() and as such clk_disable_unused() will skip parking the
unused clocks and continue to disable the PLLs feeding them (if they are
unused). I don't think this is a problem for the clocks you list here,
but we certainly have a problem with this in the dispcc. So this is
currently being discussed. For now it's expected that booting without
"clk_ignore_unused" is "broken".
Regards,
Bjorn
> .ops = &clk_rcg2_ops,
> },
> };
> @@ -710,6 +711,7 @@ static struct clk_rcg2 gcc_camss_axi_clk_src = {
> .name = "gcc_camss_axi_clk_src",
> .parent_data = gcc_parents_4,
> .num_parents = ARRAY_SIZE(gcc_parents_4),
> + .flags = CLK_OPS_PARENT_ENABLE,
> .ops = &clk_rcg2_ops,
> },
> };
> @@ -730,6 +732,7 @@ static struct clk_rcg2 gcc_camss_cci_clk_src = {
> .name = "gcc_camss_cci_clk_src",
> .parent_data = gcc_parents_9,
> .num_parents = ARRAY_SIZE(gcc_parents_9),
> + .flags = CLK_OPS_PARENT_ENABLE,
> .ops = &clk_rcg2_ops,
> },
> };
> @@ -752,6 +755,7 @@ static struct clk_rcg2 gcc_camss_csi0phytimer_clk_src = {
> .name = "gcc_camss_csi0phytimer_clk_src",
> .parent_data = gcc_parents_5,
> .num_parents = ARRAY_SIZE(gcc_parents_5),
> + .flags = CLK_OPS_PARENT_ENABLE,
> .ops = &clk_rcg2_ops,
> },
> };
> @@ -766,6 +770,7 @@ static struct clk_rcg2 gcc_camss_csi1phytimer_clk_src = {
> .name = "gcc_camss_csi1phytimer_clk_src",
> .parent_data = gcc_parents_5,
> .num_parents = ARRAY_SIZE(gcc_parents_5),
> + .flags = CLK_OPS_PARENT_ENABLE,
> .ops = &clk_rcg2_ops,
> },
> };
> @@ -854,6 +859,7 @@ static struct clk_rcg2 gcc_camss_ope_ahb_clk_src = {
> .name = "gcc_camss_ope_ahb_clk_src",
> .parent_data = gcc_parents_6,
> .num_parents = ARRAY_SIZE(gcc_parents_6),
> + .flags = CLK_OPS_PARENT_ENABLE,
> .ops = &clk_rcg2_ops,
> },
> };
> @@ -912,6 +918,7 @@ static struct clk_rcg2 gcc_camss_tfe_0_clk_src = {
> .name = "gcc_camss_tfe_0_clk_src",
> .parent_data = gcc_parents_7,
> .num_parents = ARRAY_SIZE(gcc_parents_7),
> + .flags = CLK_OPS_PARENT_ENABLE,
> .ops = &clk_rcg2_ops,
> },
> };
> @@ -936,6 +943,7 @@ static struct clk_rcg2 gcc_camss_tfe_0_csid_clk_src = {
> .name = "gcc_camss_tfe_0_csid_clk_src",
> .parent_data = gcc_parents_8,
> .num_parents = ARRAY_SIZE(gcc_parents_8),
> + .flags = CLK_OPS_PARENT_ENABLE,
> .ops = &clk_rcg2_ops,
> },
> };
> @@ -950,6 +958,7 @@ static struct clk_rcg2 gcc_camss_tfe_1_clk_src = {
> .name = "gcc_camss_tfe_1_clk_src",
> .parent_data = gcc_parents_7,
> .num_parents = ARRAY_SIZE(gcc_parents_7),
> + .flags = CLK_OPS_PARENT_ENABLE,
> .ops = &clk_rcg2_ops,
> },
> };
> @@ -964,6 +973,7 @@ static struct clk_rcg2 gcc_camss_tfe_1_csid_clk_src = {
> .name = "gcc_camss_tfe_1_csid_clk_src",
> .parent_data = gcc_parents_8,
> .num_parents = ARRAY_SIZE(gcc_parents_8),
> + .flags = CLK_OPS_PARENT_ENABLE,
> .ops = &clk_rcg2_ops,
> },
> };
> @@ -1008,6 +1018,7 @@ static struct clk_rcg2 gcc_camss_top_ahb_clk_src = {
> .name = "gcc_camss_top_ahb_clk_src",
> .parent_data = gcc_parents_4,
> .num_parents = ARRAY_SIZE(gcc_parents_4),
> + .flags = CLK_OPS_PARENT_ENABLE,
> .ops = &clk_rcg2_ops,
> },
> };
> @@ -1030,6 +1041,7 @@ static struct clk_rcg2 gcc_gp1_clk_src = {
> .name = "gcc_gp1_clk_src",
> .parent_data = gcc_parents_2,
> .num_parents = ARRAY_SIZE(gcc_parents_2),
> + .flags = CLK_OPS_PARENT_ENABLE,
> .ops = &clk_rcg2_ops,
> },
> };
> @@ -1044,6 +1056,7 @@ static struct clk_rcg2 gcc_gp2_clk_src = {
> .name = "gcc_gp2_clk_src",
> .parent_data = gcc_parents_2,
> .num_parents = ARRAY_SIZE(gcc_parents_2),
> + .flags = CLK_OPS_PARENT_ENABLE,
> .ops = &clk_rcg2_ops,
> },
> };
> @@ -1058,6 +1071,7 @@ static struct clk_rcg2 gcc_gp3_clk_src = {
> .name = "gcc_gp3_clk_src",
> .parent_data = gcc_parents_2,
> .num_parents = ARRAY_SIZE(gcc_parents_2),
> + .flags = CLK_OPS_PARENT_ENABLE,
> .ops = &clk_rcg2_ops,
> },
> };
> @@ -1078,6 +1092,7 @@ static struct clk_rcg2 gcc_pdm2_clk_src = {
> .name = "gcc_pdm2_clk_src",
> .parent_data = gcc_parents_0,
> .num_parents = ARRAY_SIZE(gcc_parents_0),
> + .flags = CLK_OPS_PARENT_ENABLE,
> .ops = &clk_rcg2_ops,
> },
> };
> @@ -1220,6 +1235,7 @@ static struct clk_rcg2 gcc_sdcc1_apps_clk_src = {
> .name = "gcc_sdcc1_apps_clk_src",
> .parent_data = gcc_parents_1,
> .num_parents = ARRAY_SIZE(gcc_parents_1),
> + .flags = CLK_OPS_PARENT_ENABLE,
> .ops = &clk_rcg2_floor_ops,
> },
> };
> @@ -1243,6 +1259,7 @@ static struct clk_rcg2 gcc_sdcc1_ice_core_clk_src = {
> .name = "gcc_sdcc1_ice_core_clk_src",
> .parent_data = gcc_parents_0,
> .num_parents = ARRAY_SIZE(gcc_parents_0),
> + .flags = CLK_OPS_PARENT_ENABLE,
> .ops = &clk_rcg2_ops,
> },
> };
> @@ -1267,6 +1284,7 @@ static struct clk_rcg2 gcc_sdcc2_apps_clk_src = {
> .name = "gcc_sdcc2_apps_clk_src",
> .parent_data = gcc_parents_12,
> .num_parents = ARRAY_SIZE(gcc_parents_12),
> + .flags = CLK_OPS_PARENT_ENABLE,
> .ops = &clk_rcg2_ops,
> },
> };
> @@ -1289,6 +1307,7 @@ static struct clk_rcg2 gcc_usb30_prim_master_clk_src = {
> .name = "gcc_usb30_prim_master_clk_src",
> .parent_data = gcc_parents_0,
> .num_parents = ARRAY_SIZE(gcc_parents_0),
> + .flags = CLK_OPS_PARENT_ENABLE,
> .ops = &clk_rcg2_ops,
> },
> };
> @@ -1303,6 +1322,7 @@ static struct clk_rcg2 gcc_usb3_prim_phy_aux_clk_src = {
> .name = "gcc_usb3_prim_phy_aux_clk_src",
> .parent_data = gcc_parents_13,
> .num_parents = ARRAY_SIZE(gcc_parents_13),
> + .flags = CLK_OPS_PARENT_ENABLE,
> .ops = &clk_rcg2_ops,
> },
> };
> --
> 2.7.4
>
next prev parent reply other threads:[~2021-12-20 23:40 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-12-20 9:54 [PATCH] clk: qcom: gcc-qcm2290: CLK_OPS_PARENT_ENABLE flag for rcg2 clocks Loic Poulain
2021-12-20 23:41 ` Bjorn Andersson [this message]
2021-12-21 10:50 ` Loic Poulain
2021-12-22 0:22 ` Bjorn Andersson
2021-12-22 10:48 ` Loic Poulain
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=YcEUrtQuJ5CGF2RW@ripper \
--to=bjorn.andersson@linaro.org \
--cc=agross@kernel.org \
--cc=linux-arm-msm@vger.kernel.org \
--cc=linux-clk@vger.kernel.org \
--cc=loic.poulain@linaro.org \
--cc=quic_mdtipton@quicinc.com \
--cc=sboyd@kernel.org \
--cc=shawn.guo@linaro.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