Thanks Neil for the review ! On 9/10/2024 3:22 PM, neil.armstrong@xxxxxxxxxx wrote:
Yes, Neil, you are right. Its actually reverse condition from HW configuration.On 10/09/2024 11:15, Mukesh Kumar Savaliya wrote:Hi Neil, On 9/9/2024 6:34 PM, neil.armstrong@xxxxxxxxxx wrote:Sure, i will add in cover letter and code check combining with fifo_disable check.On 09/09/2024 11:18, Mukesh Kumar Savaliya wrote:Hi Neil, On 9/9/2024 2:24 PM, neil.armstrong@xxxxxxxxxx wrote:Shared usecase is not supported for non GSI mode (FIFO and DMA), it should be static usecase. Dynamic sharing from two clients of two subsystems is only for GSI mode. Hope this helps ?Hi, On 06/09/2024 21:14, Mukesh Kumar Savaliya wrote:Add support to share I2C SE by two Subsystems in a mutually exclusive way.Use "qcom,shared-se" flag in a particular i2c instance node ifthe usecase requires i2c controller to be shared.I2C driver just need to mark first_msg and last_msg flag to help indicate GPI driver to take lock and unlock TRE there by protecting from concurrentaccess from other EE or Subsystem.gpi_create_i2c_tre() function at gpi.c will take care of adding Lock andUnlock TRE for the respective transfer operations.Since the GPIOs are also shared for the i2c bus between two SS, do nottouch GPIO configuration during runtime suspend and only turn off the clocks. This will allow other SS to continue to transfer the data without any disturbance over the IO lines.This doesn't answer my question about what would be the behavior if one use uses, for example, GPI DMA, and the Linux kernel FIFO mode or SE DMA ?Sure, this is why I proposed on v1 cover letter reply to add:==============><=====================================================================diff --git a/drivers/i2c/busses/i2c-qcom-geni.c b/drivers/i2c/busses/i2c-qcom-geni.cindex ee2e431601a6..a15825ea56de 100644 --- a/drivers/i2c/busses/i2c-qcom-geni.c +++ b/drivers/i2c/busses/i2c-qcom-geni.c@@ -885,7 +885,7 @@ static int geni_i2c_probe(struct platform_device *pdev)elsefifo_disable = readl_relaxed(gi2c->se.base + GENI_IF_DISABLE_RO) & FIFO_IF_DISABLE;- if (fifo_disable) { + if (gi2c->is_shared || fifo_disable) {Should be ANDING logically, as we need to combine both check. Shared usecase possible only for fifo_disable.Could you elaborate on that ? GPI DMA is totally usable even if FIFO is enabled,it's a decision took in the driver to _not_ use GPI when FIFO is enabled.
if fifo_disable = true, then FIFO registers will not be accessible, meaning its GPI mode only. And SW should decide use GPI DMA mode only.
if fifo_disable = false, it can still use GPI DMA/CPU_DMA. we need to restrict from SW side.
Provided above, i suggest to keep conditional check with ANDING.We want only GSI mode to be supported with shared SE. Because GSI mode only has GPII channel allocated to each EE. if not, then it will be misused between EEs and no way to prevent concurrency at HW level.(E.g. we use lock/unlock in GSI mode)
If so, hope you agree with the conditional check of ANDing both flags?
Neilif(gi2c->is_shared && fifo_disable) {/* FIFO is disabled, so we can only use GPI DMA */ gi2c->gpi_mode = true; ret = setup_gpi_dma(gi2c); ==============><===================================================================== Thanks, NeilBecause it seems to "fix" only the GPI DMA shared case. NeilSigned-off-by: Mukesh Kumar Savaliya <quic_msavaliy@xxxxxxxxxxx> ---drivers/i2c/busses/i2c-qcom-geni.c | 29 ++++++++++++++++++++++-------1 file changed, 22 insertions(+), 7 deletions(-)diff --git a/drivers/i2c/busses/i2c-qcom-geni.c b/drivers/i2c/busses/i2c-qcom-geni.cindex eebb0cbb6ca4..ee2e431601a6 100644 --- a/drivers/i2c/busses/i2c-qcom-geni.c +++ b/drivers/i2c/busses/i2c-qcom-geni.c @@ -1,5 +1,6 @@ // SPDX-License-Identifier: GPL-2.0// Copyright (c) 2017-2018, The Linux Foundation. All rights reserved. +// Copyright (c) 2024 Qualcomm Innovation Center, Inc. All rights reserved.#include <linux/acpi.h> #include <linux/clk.h> @@ -99,6 +100,7 @@ struct geni_i2c_dev { struct dma_chan *rx_c; bool gpi_mode; bool abort_done; + bool is_shared; }; struct geni_i2c_desc {@@ -602,6 +604,7 @@ static int geni_i2c_gpi_xfer(struct geni_i2c_dev *gi2c, struct i2c_msg msgs[], iperipheral.clk_div = itr->clk_div; peripheral.set_config = 1; peripheral.multi_msg = false; + peripheral.shared_se = gi2c->is_shared; for (i = 0; i < num; i++) { gi2c->cur =&msgs[i];@@ -612,6 +615,8 @@ static int geni_i2c_gpi_xfer(struct geni_i2c_dev *gi2c, struct i2c_msg msgs[], iif (i < num -1) peripheral.stretch = 1; + peripheral.first_msg =(i == 0); + peripheral.last_msg = (i == num - 1); peripheral.addr = msgs[i].addr; ret = geni_i2c_gpi(gi2c, &msgs[i], &config,@@ -631,8 +636,11 @@ static int geni_i2c_gpi_xfer(struct geni_i2c_dev *gi2c, struct i2c_msg msgs[], idma_async_issue_pending(gi2c->tx_c);time_left =wait_for_completion_timeout(&gi2c->done, XFER_TIMEOUT);- if (!time_left) + if (!time_left) {+ dev_err(gi2c->se.dev, "I2C timeout gpi flags:%d addr:0x%x\n",+ gi2c->cur->flags, gi2c->cur->addr); gi2c->err = -ETIMEDOUT; + } if (gi2c->err) { ret = gi2c->err;@@ -800,6 +808,11 @@ static int geni_i2c_probe(struct platform_device *pdev)gi2c->clk_freq_out = KHZ(100); }+ if (of_property_read_bool(pdev->dev.of_node, "qcom,shared-se")) {+ gi2c->is_shared = true; + dev_dbg(&pdev->dev, "Shared SE Usecase\n"); + } + if (has_acpi_companion(dev)) ACPI_COMPANION_SET(&gi2c->adap.dev, ACPI_COMPANION(dev));@@ -962,14 +975,16 @@ static int __maybe_unused geni_i2c_runtime_suspend(struct device *dev)struct geni_i2c_dev *gi2c = dev_get_drvdata(dev); disable_irq(gi2c->irq); - ret = geni_se_resources_off(&gi2c->se); - if (ret) { - enable_irq(gi2c->irq); - return ret; - + if (gi2c->is_shared) { + geni_se_clks_off(&gi2c->se); } else { - gi2c->suspended = 1; + ret = geni_se_resources_off(&gi2c->se); + if (ret) { + enable_irq(gi2c->irq); + return ret; + } } + gi2c->suspended = 1; clk_disable_unprepare(gi2c->core_clk);
