From: Ziyang Huang <hzyitc@outlook.com> To: mcoquelin.stm32@gmail.com Cc: alexandre.torgue@foss.st.com, richardcochran@gmail.com, p.zabel@pengutronix.de, matthias.bgg@gmail.com, angelogioacchino.delregno@collabora.com, linux-kernel@vger.kernel.org, linux-stm32@st-md-mailman.stormreply.com, linux-arm-kernel@lists.infradead.org, netdev@vger.kernel.org, linux-mediatek@lists.infradead.org, Praveenkumar I <ipkumar@codeaurora.org>, Abhishek Sahu <absahu@codeaurora.org>, Ziyang Huang <hzyitc@outlook.com> Subject: [PATCH 5/8] clk: qcom: support for duplicate freq in RCG2 freq table Date: Sun, 21 Jan 2024 20:42:34 +0800 [thread overview] Message-ID: <TYZPR01MB5556DEA3D4740441EC561414C9762@TYZPR01MB5556.apcprd01.prod.exchangelabs.com> (raw) In-Reply-To: <TYZPR01MB55563BD6A2B78402E4BB44D4C9762@TYZPR01MB5556.apcprd01.prod.exchangelabs.com> From: Praveenkumar I <ipkumar@codeaurora.org> Currently RCG code looks up the frequency table during set rate and return the first available frequency greater than requested rate. If CLK_SET_RATE_PARENT flag is set then the set_rate request will go to its parent otherwise the clock framework will configure pre-div, m and n according to the returned frequency table entry. In this case, it is assuming that parent clock will run in the same frequency with which pre-div, m and n has been derived. But it may be possible that the parent clock supports multiple frequency and the same frequency can be derived with different pre-div, m and n values depending upon current frequency. Also, the same frequency can be derived from different parent sources and currently there is no option for having duplicate frequencies in frequency table and choosing the best one according to current rate. Now this patch adds the support for having duplicate frequencies in frequency table. During set rate, it will compare the actual rate for each entry with requested rate and will select the best entry in which the difference will be less. The existing functionality won’t be affected with this code change since this code change will hit only if frequency table has duplicate values. Change-Id: I97d9e1b55d8f3ee095f6f01729af527ba90e50e5 Signed-off-by: Abhishek Sahu <absahu@codeaurora.org> (cherry picked from commit 775e7d3b69ffc97afb5bd5a6c9c423f2f4d8a0b2) Signed-off-by: Praveenkumar I <ipkumar@codeaurora.org> Change-Id: If10193fc79a3c1375ab73597813745ff1f4df0ad Pick from https://git.codelinaro.org/clo/qsdk/oss/kernel/linux-ipq-5.4/-/commit/6dfb368bae130bee58e00ddf8330b55066e1c8c5 Signed-off-by: Ziyang Huang <hzyitc@outlook.com> --- drivers/clk/qcom/clk-rcg2.c | 86 ++++++++++++++++++++++++++++--------- 1 file changed, 66 insertions(+), 20 deletions(-) diff --git a/drivers/clk/qcom/clk-rcg2.c b/drivers/clk/qcom/clk-rcg2.c index e22baf3a7112..6141e4991fbc 100644 --- a/drivers/clk/qcom/clk-rcg2.c +++ b/drivers/clk/qcom/clk-rcg2.c @@ -209,26 +209,82 @@ clk_rcg2_recalc_rate(struct clk_hw *hw, unsigned long parent_rate) return __clk_rcg2_recalc_rate(hw, parent_rate, cfg); } -static int _freq_tbl_determine_rate(struct clk_hw *hw, const struct freq_tbl *f, - struct clk_rate_request *req, - enum freq_policy policy) +static const struct freq_tbl * +clk_rcg2_find_best_freq(struct clk_hw *hw, const struct freq_tbl *f, + unsigned long rate, enum freq_policy policy) { - unsigned long clk_flags, rate = req->rate; - struct clk_hw *p; + unsigned long req_rate = rate, best = 0, freq; struct clk_rcg2 *rcg = to_clk_rcg2(hw); int index; + u64 tmp; + const struct freq_tbl *best_ftable = NULL; switch (policy) { case FLOOR: - f = qcom_find_freq_floor(f, rate); + f = qcom_find_freq_floor(rcg->freq_tbl, rate); break; case CEIL: - f = qcom_find_freq(f, rate); + f = qcom_find_freq(rcg->freq_tbl, rate); break; default: - return -EINVAL; + return best_ftable; } + /* + * Check for duplicate frequencies in frequency table if + * CLK_SET_RATE_PARENT flag is not set + */ + if (!f || (clk_hw_get_flags(hw) & CLK_SET_RATE_PARENT) || + ((f->freq && (f + 1)->freq != f->freq))) + return f; + + /* + * Check for all the duplicate entries in frequency table and + * calculate the actual rate from current parent rate with each + * entries pre_div, m and n values. The entry, which gives the + * minimum difference in requested rate and actual rate, will be + * selected as the best one. + */ + for (freq = f->freq; freq == f->freq; f++) { + index = qcom_find_src_index(hw, rcg->parent_map, f->src); + if (index < 0) + continue; + + rate = clk_hw_get_rate(clk_hw_get_parent_by_index(hw, index)); + if (rcg->hid_width && f->pre_div) { + rate *= 2; + rate /= f->pre_div + 1; + } + + if (rcg->mnd_width && f->n) { + tmp = rate; + tmp = tmp * f->n; + do_div(tmp, f->m); + rate = tmp; + } + + if (abs(req_rate - rate) < abs(best - rate)) { + best_ftable = f; + best = rate; + + if (req_rate == rate) + break; + } + } + + return best_ftable; +} + +static int _freq_tbl_determine_rate(struct clk_hw *hw, const struct freq_tbl *f, + struct clk_rate_request *req, + enum freq_policy policy) +{ + unsigned long clk_flags, rate = req->rate; + struct clk_hw *p; + struct clk_rcg2 *rcg = to_clk_rcg2(hw); + int index; + + f = clk_rcg2_find_best_freq(hw, f, rate, policy); if (!f) return -EINVAL; @@ -360,17 +416,7 @@ static int __clk_rcg2_set_rate(struct clk_hw *hw, unsigned long rate, struct clk_rcg2 *rcg = to_clk_rcg2(hw); const struct freq_tbl *f; - switch (policy) { - case FLOOR: - f = qcom_find_freq_floor(rcg->freq_tbl, rate); - break; - case CEIL: - f = qcom_find_freq(rcg->freq_tbl, rate); - break; - default: - return -EINVAL; - } - + f = clk_rcg2_find_best_freq(hw, rcg->freq_tbl, rate, policy); if (!f) return -EINVAL; @@ -1032,7 +1078,7 @@ static int clk_rcg2_shared_set_rate(struct clk_hw *hw, unsigned long rate, struct clk_rcg2 *rcg = to_clk_rcg2(hw); const struct freq_tbl *f; - f = qcom_find_freq(rcg->freq_tbl, rate); + f = clk_rcg2_find_best_freq(hw, rcg->freq_tbl, rate, CEIL); if (!f) return -EINVAL; -- 2.40.1
WARNING: multiple messages have this Message-ID (diff)
From: Ziyang Huang <hzyitc@outlook.com> To: mcoquelin.stm32@gmail.com Cc: alexandre.torgue@foss.st.com, richardcochran@gmail.com, p.zabel@pengutronix.de, matthias.bgg@gmail.com, angelogioacchino.delregno@collabora.com, linux-kernel@vger.kernel.org, linux-stm32@st-md-mailman.stormreply.com, linux-arm-kernel@lists.infradead.org, netdev@vger.kernel.org, linux-mediatek@lists.infradead.org, Praveenkumar I <ipkumar@codeaurora.org>, Abhishek Sahu <absahu@codeaurora.org>, Ziyang Huang <hzyitc@outlook.com> Subject: [PATCH 5/8] clk: qcom: support for duplicate freq in RCG2 freq table Date: Sun, 21 Jan 2024 20:42:34 +0800 [thread overview] Message-ID: <TYZPR01MB5556DEA3D4740441EC561414C9762@TYZPR01MB5556.apcprd01.prod.exchangelabs.com> (raw) In-Reply-To: <TYZPR01MB55563BD6A2B78402E4BB44D4C9762@TYZPR01MB5556.apcprd01.prod.exchangelabs.com> From: Praveenkumar I <ipkumar@codeaurora.org> Currently RCG code looks up the frequency table during set rate and return the first available frequency greater than requested rate. If CLK_SET_RATE_PARENT flag is set then the set_rate request will go to its parent otherwise the clock framework will configure pre-div, m and n according to the returned frequency table entry. In this case, it is assuming that parent clock will run in the same frequency with which pre-div, m and n has been derived. But it may be possible that the parent clock supports multiple frequency and the same frequency can be derived with different pre-div, m and n values depending upon current frequency. Also, the same frequency can be derived from different parent sources and currently there is no option for having duplicate frequencies in frequency table and choosing the best one according to current rate. Now this patch adds the support for having duplicate frequencies in frequency table. During set rate, it will compare the actual rate for each entry with requested rate and will select the best entry in which the difference will be less. The existing functionality won’t be affected with this code change since this code change will hit only if frequency table has duplicate values. Change-Id: I97d9e1b55d8f3ee095f6f01729af527ba90e50e5 Signed-off-by: Abhishek Sahu <absahu@codeaurora.org> (cherry picked from commit 775e7d3b69ffc97afb5bd5a6c9c423f2f4d8a0b2) Signed-off-by: Praveenkumar I <ipkumar@codeaurora.org> Change-Id: If10193fc79a3c1375ab73597813745ff1f4df0ad Pick from https://git.codelinaro.org/clo/qsdk/oss/kernel/linux-ipq-5.4/-/commit/6dfb368bae130bee58e00ddf8330b55066e1c8c5 Signed-off-by: Ziyang Huang <hzyitc@outlook.com> --- drivers/clk/qcom/clk-rcg2.c | 86 ++++++++++++++++++++++++++++--------- 1 file changed, 66 insertions(+), 20 deletions(-) diff --git a/drivers/clk/qcom/clk-rcg2.c b/drivers/clk/qcom/clk-rcg2.c index e22baf3a7112..6141e4991fbc 100644 --- a/drivers/clk/qcom/clk-rcg2.c +++ b/drivers/clk/qcom/clk-rcg2.c @@ -209,26 +209,82 @@ clk_rcg2_recalc_rate(struct clk_hw *hw, unsigned long parent_rate) return __clk_rcg2_recalc_rate(hw, parent_rate, cfg); } -static int _freq_tbl_determine_rate(struct clk_hw *hw, const struct freq_tbl *f, - struct clk_rate_request *req, - enum freq_policy policy) +static const struct freq_tbl * +clk_rcg2_find_best_freq(struct clk_hw *hw, const struct freq_tbl *f, + unsigned long rate, enum freq_policy policy) { - unsigned long clk_flags, rate = req->rate; - struct clk_hw *p; + unsigned long req_rate = rate, best = 0, freq; struct clk_rcg2 *rcg = to_clk_rcg2(hw); int index; + u64 tmp; + const struct freq_tbl *best_ftable = NULL; switch (policy) { case FLOOR: - f = qcom_find_freq_floor(f, rate); + f = qcom_find_freq_floor(rcg->freq_tbl, rate); break; case CEIL: - f = qcom_find_freq(f, rate); + f = qcom_find_freq(rcg->freq_tbl, rate); break; default: - return -EINVAL; + return best_ftable; } + /* + * Check for duplicate frequencies in frequency table if + * CLK_SET_RATE_PARENT flag is not set + */ + if (!f || (clk_hw_get_flags(hw) & CLK_SET_RATE_PARENT) || + ((f->freq && (f + 1)->freq != f->freq))) + return f; + + /* + * Check for all the duplicate entries in frequency table and + * calculate the actual rate from current parent rate with each + * entries pre_div, m and n values. The entry, which gives the + * minimum difference in requested rate and actual rate, will be + * selected as the best one. + */ + for (freq = f->freq; freq == f->freq; f++) { + index = qcom_find_src_index(hw, rcg->parent_map, f->src); + if (index < 0) + continue; + + rate = clk_hw_get_rate(clk_hw_get_parent_by_index(hw, index)); + if (rcg->hid_width && f->pre_div) { + rate *= 2; + rate /= f->pre_div + 1; + } + + if (rcg->mnd_width && f->n) { + tmp = rate; + tmp = tmp * f->n; + do_div(tmp, f->m); + rate = tmp; + } + + if (abs(req_rate - rate) < abs(best - rate)) { + best_ftable = f; + best = rate; + + if (req_rate == rate) + break; + } + } + + return best_ftable; +} + +static int _freq_tbl_determine_rate(struct clk_hw *hw, const struct freq_tbl *f, + struct clk_rate_request *req, + enum freq_policy policy) +{ + unsigned long clk_flags, rate = req->rate; + struct clk_hw *p; + struct clk_rcg2 *rcg = to_clk_rcg2(hw); + int index; + + f = clk_rcg2_find_best_freq(hw, f, rate, policy); if (!f) return -EINVAL; @@ -360,17 +416,7 @@ static int __clk_rcg2_set_rate(struct clk_hw *hw, unsigned long rate, struct clk_rcg2 *rcg = to_clk_rcg2(hw); const struct freq_tbl *f; - switch (policy) { - case FLOOR: - f = qcom_find_freq_floor(rcg->freq_tbl, rate); - break; - case CEIL: - f = qcom_find_freq(rcg->freq_tbl, rate); - break; - default: - return -EINVAL; - } - + f = clk_rcg2_find_best_freq(hw, rcg->freq_tbl, rate, policy); if (!f) return -EINVAL; @@ -1032,7 +1078,7 @@ static int clk_rcg2_shared_set_rate(struct clk_hw *hw, unsigned long rate, struct clk_rcg2 *rcg = to_clk_rcg2(hw); const struct freq_tbl *f; - f = qcom_find_freq(rcg->freq_tbl, rate); + f = clk_rcg2_find_best_freq(hw, rcg->freq_tbl, rate, CEIL); if (!f) return -EINVAL; -- 2.40.1 _______________________________________________ linux-arm-kernel mailing list linux-arm-kernel@lists.infradead.org http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
next prev parent reply other threads:[~2024-01-21 12:43 UTC|newest] Thread overview: 70+ messages / expand[flat|nested] mbox.gz Atom feed top 2024-01-21 12:40 [PATCH 0/8] ipq5018: enable ethernet support Ziyang Huang 2024-01-21 12:40 ` Ziyang Huang 2024-01-21 12:42 ` [PATCH 1/8] net: phy: Introduce Qualcomm IPQ5018 internal PHY driver Ziyang Huang 2024-01-21 12:42 ` Ziyang Huang 2024-01-21 16:19 ` Andrew Lunn 2024-01-21 16:19 ` Andrew Lunn 2024-01-22 15:37 ` Ziyang Huang 2024-01-22 15:37 ` Ziyang Huang 2024-01-22 17:18 ` Andrew Lunn 2024-01-22 17:18 ` Andrew Lunn 2024-01-23 15:38 ` Ziyang Huang 2024-01-23 15:38 ` Ziyang Huang 2024-01-23 23:15 ` Andrew Lunn 2024-01-23 23:15 ` Andrew Lunn 2024-01-21 12:42 ` [PATCH 2/8] phy: Introduce Qualcomm ethernet uniphy driver Ziyang Huang 2024-01-21 12:42 ` Ziyang Huang 2024-01-23 15:58 ` Ziyang Huang 2024-01-23 15:58 ` Ziyang Huang 2024-01-23 23:25 ` Andrew Lunn 2024-01-23 23:25 ` Andrew Lunn 2024-01-21 12:42 ` [PATCH 3/8] net: stmmac: Introduce Qualcomm IPQ50xx DWMAC driver Ziyang Huang 2024-01-21 12:42 ` Ziyang Huang 2024-01-24 5:54 ` kernel test robot 2024-01-24 5:54 ` kernel test robot 2024-01-24 9:40 ` kernel test robot 2024-01-24 9:40 ` kernel test robot 2024-01-21 12:42 ` [PATCH 4/8] clk: qcom: gcc-ipq5018: correct gcc_gmac0_sys_clk reg Ziyang Huang 2024-01-21 12:42 ` Ziyang Huang 2024-01-21 16:28 ` Andrew Lunn 2024-01-21 16:28 ` Andrew Lunn 2024-01-22 15:39 ` Ziyang Huang 2024-01-22 15:39 ` Ziyang Huang 2024-01-21 12:42 ` Ziyang Huang [this message] 2024-01-21 12:42 ` [PATCH 5/8] clk: qcom: support for duplicate freq in RCG2 freq table Ziyang Huang 2024-01-21 16:57 ` Andrew Lunn 2024-01-21 16:57 ` Andrew Lunn 2024-01-22 16:35 ` Ziyang Huang 2024-01-22 16:35 ` Ziyang Huang 2024-01-22 17:34 ` Andrew Lunn 2024-01-22 17:34 ` Andrew Lunn 2024-01-23 15:43 ` Ziyang Huang 2024-01-23 15:43 ` Ziyang Huang 2024-01-22 7:55 ` Krzysztof Kozlowski 2024-01-22 7:55 ` Krzysztof Kozlowski 2024-01-22 14:48 ` Ziyang Huang 2024-01-22 14:48 ` Ziyang Huang 2024-01-21 12:42 ` [PATCH 6/8] net: mdio: ipq4019: support reset control Ziyang Huang 2024-01-21 12:42 ` Ziyang Huang 2024-01-21 16:35 ` Andrew Lunn 2024-01-21 16:35 ` Andrew Lunn 2024-01-22 15:52 ` Ziyang Huang 2024-01-22 15:52 ` Ziyang Huang 2024-01-21 12:42 ` [PATCH 7/8] arm64: dts: qcom: ipq5018: enable ethernet support Ziyang Huang 2024-01-21 12:42 ` Ziyang Huang 2024-01-21 16:45 ` Andrew Lunn 2024-01-21 16:45 ` Andrew Lunn 2024-01-22 15:52 ` Ziyang Huang 2024-01-22 15:52 ` Ziyang Huang 2024-01-22 17:27 ` Andrew Lunn 2024-01-22 17:27 ` Andrew Lunn 2024-01-21 12:42 ` [PATCH 8/8] arm64: dts: qcom: ipq5018-rdp432-c2: " Ziyang Huang 2024-01-21 12:42 ` Ziyang Huang 2024-01-22 7:54 ` Krzysztof Kozlowski 2024-01-22 7:54 ` Krzysztof Kozlowski 2024-01-24 0:53 ` kernel test robot 2024-01-24 0:53 ` kernel test robot 2024-01-21 15:51 ` [PATCH 0/8] ipq5018: " Andrew Lunn 2024-01-21 15:51 ` Andrew Lunn 2024-01-22 14:45 ` Ziyang Huang 2024-01-22 14:45 ` Ziyang Huang
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=TYZPR01MB5556DEA3D4740441EC561414C9762@TYZPR01MB5556.apcprd01.prod.exchangelabs.com \ --to=hzyitc@outlook.com \ --cc=absahu@codeaurora.org \ --cc=alexandre.torgue@foss.st.com \ --cc=angelogioacchino.delregno@collabora.com \ --cc=ipkumar@codeaurora.org \ --cc=linux-arm-kernel@lists.infradead.org \ --cc=linux-kernel@vger.kernel.org \ --cc=linux-mediatek@lists.infradead.org \ --cc=linux-stm32@st-md-mailman.stormreply.com \ --cc=matthias.bgg@gmail.com \ --cc=mcoquelin.stm32@gmail.com \ --cc=netdev@vger.kernel.org \ --cc=p.zabel@pengutronix.de \ --cc=richardcochran@gmail.com \ /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: linkBe sure your reply has a Subject: header at the top and a blank line before the message body.
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.