Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > linux.kernel > #1275012 > unrolled thread
| Started by | Sinan Kaya <okaya@codeaurora.org> |
|---|---|
| First post | 2015-11-23 03:30 +0100 |
| Last post | 2015-12-11 10:40 +0100 |
| Articles | 19 — 3 participants |
Back to article view | Back to linux.kernel
[PATCH V7 0/3] dma: add Qualcomm Technologies HIDMA driver Sinan Kaya <okaya@codeaurora.org> - 2015-11-23 03:30 +0100
[PATCH V7 2/3] dma: add Qualcomm Technologies HIDMA management driver Sinan Kaya <okaya@codeaurora.org> - 2015-11-23 03:30 +0100
Re: [PATCH V7 2/3] dma: add Qualcomm Technologies HIDMA management driver Andy Shevchenko <andy.shevchenko@gmail.com> - 2015-11-23 19:10 +0100
Re: [PATCH V7 2/3] dma: add Qualcomm Technologies HIDMA management driver Sinan Kaya <okaya@codeaurora.org> - 2015-11-23 19:50 +0100
Re: [PATCH V7 2/3] dma: add Qualcomm Technologies HIDMA management driver Vinod Koul <vinod.koul@intel.com> - 2015-11-30 09:20 +0100
Re: [PATCH V7 2/3] dma: add Qualcomm Technologies HIDMA management driver Sinan Kaya <okaya@codeaurora.org> - 2015-11-30 15:50 +0100
Re: [PATCH V7 2/3] dma: add Qualcomm Technologies HIDMA management driver Vinod Koul <vinod.koul@intel.com> - 2015-12-01 04:20 +0100
Re: [PATCH V7 2/3] dma: add Qualcomm Technologies HIDMA management driver Sinan Kaya <okaya@codeaurora.org> - 2015-12-02 06:00 +0100
Re: [PATCH V7 2/3] dma: add Qualcomm Technologies HIDMA management driver Vinod Koul <vinod.koul@intel.com> - 2015-12-05 09:00 +0100
Re: [PATCH V7 2/3] dma: add Qualcomm Technologies HIDMA management driver Sinan Kaya <okaya@codeaurora.org> - 2015-12-08 15:40 +0100
Re: [PATCH V7 3/3] dma: add Qualcomm Technologies HIDMA channel driver Vinod Koul <vinod.koul@intel.com> - 2015-11-30 10:00 +0100
Re: [PATCH V7 3/3] dma: add Qualcomm Technologies HIDMA channel driver Sinan Kaya <okaya@codeaurora.org> - 2015-11-30 21:10 +0100
Re: [PATCH V7 3/3] dma: add Qualcomm Technologies HIDMA channel driver Vinod Koul <vinod.koul@intel.com> - 2015-12-01 12:40 +0100
Re: [PATCH V7 3/3] dma: add Qualcomm Technologies HIDMA channel driver Sinan Kaya <okaya@codeaurora.org> - 2015-12-01 22:20 +0100
Re: [PATCH V7 3/3] dma: add Qualcomm Technologies HIDMA channel driver Sinan Kaya <okaya@codeaurora.org> - 2015-12-02 20:10 +0100
Re: [PATCH V7 3/3] dma: add Qualcomm Technologies HIDMA channel driver Vinod Koul <vinod.koul@intel.com> - 2015-12-05 09:00 +0100
Re: [PATCH V7 3/3] dma: add Qualcomm Technologies HIDMA channel driver Sinan Kaya <okaya@codeaurora.org> - 2015-12-08 15:40 +0100
Re: [PATCH V7 3/3] dma: add Qualcomm Technologies HIDMA channel driver Sinan Kaya <okaya@codeaurora.org> - 2015-12-10 21:20 +0100
Re: [PATCH V7 3/3] dma: add Qualcomm Technologies HIDMA channel driver Vinod Koul <vinod.koul@intel.com> - 2015-12-11 10:40 +0100
| From | Sinan Kaya <okaya@codeaurora.org> |
|---|---|
| Date | 2015-11-23 03:30 +0100 |
| Subject | [PATCH V7 0/3] dma: add Qualcomm Technologies HIDMA driver |
| Message-ID | <qxP3c-3W5-3@gated-at.bofh.it> |
The Qualcomm Technologies HIDMA device has been designed
to support virtualization technology. The driver has been
divided into two to follow the hardware design.
1. HIDMA Management driver
2. HIDMA Channel driver
Each HIDMA HW consists of multiple channels. These channels
share some set of common parameters. These parameters are
initialized by the management driver during power up.
Same management driver is used for monitoring the execution
of the channels. Management driver can change the performance
behavior dynamically such as bandwidth allocation and
prioritization in the future.
The management driver is executed in hypervisor context and
is the main management entity for all channels provided by
the device.
Changes from V6: (https://lkml.org/lkml/2015/11/22/111)
* none
Changes from V6: (https://lkml.org/lkml/2015/11/22/112)
* remove extra _ in documentation
* remove extra parenthesis in assignments
* create subdirectories for channels in sysfs
Changes from V6: (https://lkml.org/lkml/2015/11/22/113)
* remove extra parenthesis
Sinan Kaya (3):
dma: qcom_bam_dma: move to qcom directory
dma: add Qualcomm Technologies HIDMA management driver
dma: add Qualcomm Technologies HIDMA channel driver
.../ABI/testing/sysfs-platform-hidma-mgmt | 97 +++
.../devicetree/bindings/dma/qcom_hidma.txt | 18 +
.../devicetree/bindings/dma/qcom_hidma_mgmt.txt | 61 ++
drivers/dma/Kconfig | 11 +-
drivers/dma/Makefile | 2 +-
drivers/dma/qcom/Kconfig | 29 +
drivers/dma/qcom/Makefile | 4 +
drivers/dma/{qcom_bam_dma.c => qcom/bam_dma.c} | 4 +-
drivers/dma/qcom/hidma.c | 706 ++++++++++++++++
drivers/dma/qcom/hidma.h | 157 ++++
drivers/dma/qcom/hidma_dbg.c | 217 +++++
drivers/dma/qcom/hidma_ll.c | 924 +++++++++++++++++++++
drivers/dma/qcom/hidma_mgmt.c | 308 +++++++
drivers/dma/qcom/hidma_mgmt.h | 39 +
drivers/dma/qcom/hidma_mgmt_sys.c | 299 +++++++
15 files changed, 2864 insertions(+), 12 deletions(-)
create mode 100644 Documentation/ABI/testing/sysfs-platform-hidma-mgmt
create mode 100644 Documentation/devicetree/bindings/dma/qcom_hidma.txt
create mode 100644 Documentation/devicetree/bindings/dma/qcom_hidma_mgmt.txt
create mode 100644 drivers/dma/qcom/Kconfig
create mode 100644 drivers/dma/qcom/Makefile
rename drivers/dma/{qcom_bam_dma.c => qcom/bam_dma.c} (99%)
create mode 100644 drivers/dma/qcom/hidma.c
create mode 100644 drivers/dma/qcom/hidma.h
create mode 100644 drivers/dma/qcom/hidma_dbg.c
create mode 100644 drivers/dma/qcom/hidma_ll.c
create mode 100644 drivers/dma/qcom/hidma_mgmt.c
create mode 100644 drivers/dma/qcom/hidma_mgmt.h
create mode 100644 drivers/dma/qcom/hidma_mgmt_sys.c
--
Qualcomm Technologies, Inc. on behalf of Qualcomm Innovation Center, Inc.
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a Linux Foundation Collaborative Project
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
[toc] | [next] | [standalone]
| From | Sinan Kaya <okaya@codeaurora.org> |
|---|---|
| Date | 2015-11-23 03:30 +0100 |
| Subject | [PATCH V7 2/3] dma: add Qualcomm Technologies HIDMA management driver |
| Message-ID | <qxP3c-3W5-9@gated-at.bofh.it> |
| In reply to | #1275012 |
The Qualcomm Technologies HIDMA device has been designed
to support virtualization technology. The driver has been
divided into two to follow the hardware design.
1. HIDMA Management driver
2. HIDMA Channel driver
Each HIDMA HW consists of multiple channels. These channels
share some set of common parameters. These parameters are
initialized by the management driver during power up.
Same management driver is used for monitoring the execution
of the channels. Management driver can change the performance
behavior dynamically such as bandwidth allocation and
prioritization.
The management driver is executed in hypervisor context and
is the main management entity for all channels provided by
the device.
Signed-off-by: Sinan Kaya <okaya@codeaurora.org>
---
.../ABI/testing/sysfs-platform-hidma-mgmt | 97 +++++++
.../devicetree/bindings/dma/qcom_hidma_mgmt.txt | 61 ++++
drivers/dma/qcom/Kconfig | 11 +
drivers/dma/qcom/Makefile | 1 +
drivers/dma/qcom/hidma_mgmt.c | 308 +++++++++++++++++++++
drivers/dma/qcom/hidma_mgmt.h | 39 +++
drivers/dma/qcom/hidma_mgmt_sys.c | 299 ++++++++++++++++++++
7 files changed, 816 insertions(+)
create mode 100644 Documentation/ABI/testing/sysfs-platform-hidma-mgmt
create mode 100644 Documentation/devicetree/bindings/dma/qcom_hidma_mgmt.txt
create mode 100644 drivers/dma/qcom/hidma_mgmt.c
create mode 100644 drivers/dma/qcom/hidma_mgmt.h
create mode 100644 drivers/dma/qcom/hidma_mgmt_sys.c
diff --git a/Documentation/ABI/testing/sysfs-platform-hidma-mgmt b/Documentation/ABI/testing/sysfs-platform-hidma-mgmt
new file mode 100644
index 0000000..4fc6876
--- /dev/null
+++ b/Documentation/ABI/testing/sysfs-platform-hidma-mgmt
@@ -0,0 +1,97 @@
+What: /sys/devices/platform/hidma-mgmt*/chan*/priority
+ /sys/devices/platform/QCOM8060:*/chan*/priority
+Date: Nov 2015
+KernelVersion: 4.4
+Contact: "Sinan Kaya <okaya@cudeaurora.org>"
+Description:
+ Contains either 0 or 1 and indicates if the DMA channel is a
+ low priority (0) or high priority (1) channel.
+
+What: /sys/devices/platform/hidma-mgmt*/chan*/weight
+ /sys/devices/platform/QCOM8060:*/chan*/weight
+Date: Nov 2015
+KernelVersion: 4.4
+Contact: "Sinan Kaya <okaya@cudeaurora.org>"
+Description:
+ Contains 0..15 and indicates the weight of the channel among
+ equal priority channels during round robin scheduling.
+
+What: /sys/devices/platform/hidma-mgmt*/chreset_timeout_cycles
+ /sys/devices/platform/QCOM8060:*/chreset_timeout_cycles
+Date: Nov 2015
+KernelVersion: 4.4
+Contact: "Sinan Kaya <okaya@cudeaurora.org>"
+Description:
+ Contains the platform specific cycle value to wait after a
+ reset command is issued. If the value is chosen too short,
+ then the HW will issue a reset failure interrupt. The value
+ is platform specific and should not be changed without
+ consultance.
+
+What: /sys/devices/platform/hidma-mgmt*/dma_channels
+ /sys/devices/platform/QCOM8060:*/dma_channels
+Date: Nov 2015
+KernelVersion: 4.4
+Contact: "Sinan Kaya <okaya@cudeaurora.org>"
+Description:
+ Contains the number of dma channels supported by one instance
+ of HIDMA hardware. The value may change from chip to chip.
+
+What: /sys/devices/platform/hidma-mgmt*/hw_version_major
+ /sys/devices/platform/QCOM8060:*/hw_version_major
+Date: Nov 2015
+KernelVersion: 4.4
+Contact: "Sinan Kaya <okaya@cudeaurora.org>"
+Description:
+ Version number major for the hardware.
+
+What: /sys/devices/platform/hidma-mgmt*/hw_version_minor
+ /sys/devices/platform/QCOM8060:*/hw_version_minor
+Date: Nov 2015
+KernelVersion: 4.4
+Contact: "Sinan Kaya <okaya@cudeaurora.org>"
+Description:
+ Version number minor for the hardware.
+
+What: /sys/devices/platform/hidma-mgmt*/max_rd_xactions
+ /sys/devices/platform/QCOM8060:*/max_rd_xactions
+Date: Nov 2015
+KernelVersion: 4.4
+Contact: "Sinan Kaya <okaya@cudeaurora.org>"
+Description:
+ Contains a value between 0 and 31. Maximum number of
+ read transactions that can be issued back to back.
+ Choosing a higher number gives better performance but
+ can also cause performance reduction to other peripherals
+ sharing the same bus.
+
+What: /sys/devices/platform/hidma-mgmt*/max_read_request
+ /sys/devices/platform/QCOM8060:*/max_read_request
+Date: Nov 2015
+KernelVersion: 4.4
+Contact: "Sinan Kaya <okaya@cudeaurora.org>"
+Description:
+ Size of each read request. The value needs to be a power
+ of two and can be between 128 and 1024.
+
+What: /sys/devices/platform/hidma-mgmt*/max_wr_xactions
+ /sys/devices/platform/QCOM8060:*/max_wr_xactions
+Date: Nov 2015
+KernelVersion: 4.4
+Contact: "Sinan Kaya <okaya@cudeaurora.org>"
+Description:
+ Contains a value between 0 and 31. Maximum number of
+ write transactions that can be issued back to back.
+ Choosing a higher number gives better performance but
+ can also cause performance reduction to other peripherals
+ sharing the same bus.
+
+
+What: /sys/devices/platform/hidma-mgmt*/max_write_request
+ /sys/devices/platform/QCOM8060:*/max_write_request
+Date: Nov 2015
+KernelVersion: 4.4
+Contact: "Sinan Kaya <okaya@cudeaurora.org>"
+Description:
+ Size of each write request. The value needs to be a power
+ of two and can be between 128 and 1024.
diff --git a/Documentation/devicetree/bindings/dma/qcom_hidma_mgmt.txt b/Documentation/devicetree/bindings/dma/qcom_hidma_mgmt.txt
new file mode 100644
index 0000000..eb053b9
--- /dev/null
+++ b/Documentation/devicetree/bindings/dma/qcom_hidma_mgmt.txt
@@ -0,0 +1,61 @@
+Qualcomm Technologies HIDMA Management interface
+
+The Qualcomm Technologies HIDMA device has been designed
+to support virtualization technology. The driver has been
+divided into two to follow the hardware design. The management
+driver is executed in hypervisor context and is the main
+management entity for all channels provided by the device.
+
+Each HIDMA HW consists of multiple channels. These channels
+share some set of common parameters. These parameters are
+initialized by the management driver during power up.
+Same management driver is used for monitoring the execution
+of the channels. Management driver can change the performance
+behavior dynamically such as bandwidth allocation and
+prioritization.
+
+All channel devices get probed in the hypervisor
+context during power up. They show up as DMA engine
+DMA channels. Then, before starting the virtualization; each
+channel device is unbound from the hypervisor by VFIO
+and assign to the guest machine for control.
+
+This management driver will be used by the system
+admin to monitor/reset the execution state of the DMA
+channels. This will be the management interface.
+
+
+Required properties:
+- compatible: "qcom,hidma-mgmt-1.0";
+- reg: Address range for DMA device
+- dma-channels: Number of channels supported by this DMA controller.
+- max-write-burst-bytes: Maximum write burst in bytes. A memcpy requested is
+ fragmented to multiples of this amount.
+- max-read-burst-bytes: Maximum read burst in bytes. A memcpy request is
+ fragmented to multiples of this amount.
+- max-write-transactions: Maximum write transactions to perform in a burst
+- max-read-transactions: Maximum read transactions to perform in a burst
+- channel-reset-timeout-cycles: Channel reset timeout in cycles for this SOC.
+- channel-priority: Priority of the channel.
+ Each dma channel share the same HW bandwidth with other dma channels.
+ If two requests reach to the HW at the same time from a low priority and
+ high priority channel, high priority channel will claim the bus.
+ 0=low priority, 1=high priority
+- channel-weight: Round robin weight of the channel
+ Since there are only two priority levels supported, scheduling among
+ the equal priority channels is done via weights.
+
+Example:
+
+ hidma-mgmt@f9984000 = {
+ compatible = "qcom,hidma-mgmt-1.0";
+ reg = <0xf9984000 0x15000>;
+ dma-channels = 6;
+ max-write-burst-bytes = 1024;
+ max-read-burst-bytes = 1024;
+ max-write-transactions = 31;
+ max-read-transactions = 31;
+ channel-reset-timeout-cycles = 0x500;
+ channel-priority = < 1 1 0 0 0 0>;
+ channel-weight = < 1 13 10 3 4 5>;
+ };
diff --git a/drivers/dma/qcom/Kconfig b/drivers/dma/qcom/Kconfig
index f17c272..ebe5b8c 100644
--- a/drivers/dma/qcom/Kconfig
+++ b/drivers/dma/qcom/Kconfig
@@ -6,3 +6,14 @@ config QCOM_BAM_DMA
---help---
Enable support for the QCOM BAM DMA controller. This controller
provides DMA capabilities for a variety of on-chip devices.
+
+config QCOM_HIDMA_MGMT
+ tristate "Qualcomm Technologies HIDMA Management support"
+ select DMA_ENGINE
+ help
+ Enable support for the Qualcomm Technologies HIDMA Management.
+ Each DMA device requires one management interface driver
+ for basic initialization before QCOM_HIDMA channel driver can
+ start managing the channels. In a virtualized environment,
+ the guest OS would run QCOM_HIDMA channel driver and the
+ hypervisor would run the QCOM_HIDMA_MGMT management driver.
diff --git a/drivers/dma/qcom/Makefile b/drivers/dma/qcom/Makefile
index f612ae3..1a5a96d 100644
--- a/drivers/dma/qcom/Makefile
+++ b/drivers/dma/qcom/Makefile
@@ -1 +1,2 @@
obj-$(CONFIG_QCOM_BAM_DMA) += bam_dma.o
+obj-$(CONFIG_QCOM_HIDMA_MGMT) += hidma_mgmt.o hidma_mgmt_sys.o
diff --git a/drivers/dma/qcom/hidma_mgmt.c b/drivers/dma/qcom/hidma_mgmt.c
new file mode 100644
index 0000000..ea78a4d
--- /dev/null
+++ b/drivers/dma/qcom/hidma_mgmt.c
@@ -0,0 +1,308 @@
+/*
+ * Qualcomm Technologies HIDMA DMA engine Management interface
+ *
+ * Copyright (c) 2015, The Linux Foundation. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ */
+
+#include <linux/dmaengine.h>
+#include <linux/acpi.h>
+#include <linux/of.h>
+#include <linux/property.h>
+#include <linux/interrupt.h>
+#include <linux/platform_device.h>
+#include <linux/module.h>
+#include <linux/uaccess.h>
+#include <linux/slab.h>
+#include <linux/pm_runtime.h>
+#include <linux/bitops.h>
+
+#include "hidma_mgmt.h"
+
+#define QOS_N_OFFSET 0x300
+#define CFG_OFFSET 0x400
+#define MAX_BUS_REQ_LEN_OFFSET 0x41C
+#define MAX_XACTIONS_OFFSET 0x420
+#define HW_VERSION_OFFSET 0x424
+#define CHRESET_TIMEOUT_OFFSET 0x418
+
+#define MAX_WR_XACTIONS_MASK GENMASK(4, 0)
+#define MAX_RD_XACTIONS_MASK GENMASK(4, 0)
+#define WEIGHT_MASK GENMASK(6, 0)
+#define MAX_BUS_REQ_LEN_MASK GENMASK(15, 0)
+#define CHRESET_TIMEOUUT_MASK GENMASK(19, 0)
+
+#define MAX_WR_XACTIONS_BIT_POS 16
+#define MAX_BUS_WR_REQ_BIT_POS 16
+#define WRR_BIT_POS 8
+#define PRIORITY_BIT_POS 15
+
+#define AUTOSUSPEND_TIMEOUT 2000
+#define MAX_CHANNEL_WEIGHT 15
+
+int hidma_mgmt_setup(struct hidma_mgmt_dev *mgmtdev)
+{
+ unsigned int i;
+ u32 val;
+
+ if (!is_power_of_2(mgmtdev->max_write_request) ||
+ (mgmtdev->max_write_request < 128) ||
+ (mgmtdev->max_write_request > 1024)) {
+ dev_err(&mgmtdev->pdev->dev, "invalid write request %d\n",
+ mgmtdev->max_write_request);
+ return -EINVAL;
+ }
+
+ if (!is_power_of_2(mgmtdev->max_read_request) ||
+ (mgmtdev->max_read_request < 128) ||
+ (mgmtdev->max_read_request > 1024)) {
+ dev_err(&mgmtdev->pdev->dev, "invalid read request %d\n",
+ mgmtdev->max_read_request);
+ return -EINVAL;
+ }
+
+ if (mgmtdev->max_wr_xactions > MAX_WR_XACTIONS_MASK) {
+ dev_err(&mgmtdev->pdev->dev,
+ "max_wr_xactions cannot be bigger than %ld\n",
+ MAX_WR_XACTIONS_MASK);
+ return -EINVAL;
+ }
+
+ if (mgmtdev->max_rd_xactions > MAX_RD_XACTIONS_MASK) {
+ dev_err(&mgmtdev->pdev->dev,
+ "max_rd_xactions cannot be bigger than %ld\n",
+ MAX_RD_XACTIONS_MASK);
+ return -EINVAL;
+ }
+
+ for (i = 0; i < mgmtdev->dma_channels; i++) {
+ if (mgmtdev->priority[i] > 1) {
+ dev_err(&mgmtdev->pdev->dev,
+ "priority can be 0 or 1\n");
+ return -EINVAL;
+ }
+
+ if (mgmtdev->weight[i] > MAX_CHANNEL_WEIGHT) {
+ dev_err(&mgmtdev->pdev->dev,
+ "max value of weight can be %d.\n",
+ MAX_CHANNEL_WEIGHT);
+ return -EINVAL;
+ }
+
+ /* weight needs to be at least one */
+ if (mgmtdev->weight[i] == 0)
+ mgmtdev->weight[i] = 1;
+ }
+
+ pm_runtime_get_sync(&mgmtdev->pdev->dev);
+ val = readl(mgmtdev->virtaddr + MAX_BUS_REQ_LEN_OFFSET);
+ val &= ~(MAX_BUS_REQ_LEN_MASK << MAX_BUS_WR_REQ_BIT_POS);
+ val |= mgmtdev->max_write_request << MAX_BUS_WR_REQ_BIT_POS;
+ val &= ~MAX_BUS_REQ_LEN_MASK;
+ val |= mgmtdev->max_read_request;
+ writel(val, mgmtdev->virtaddr + MAX_BUS_REQ_LEN_OFFSET);
+
+ val = readl(mgmtdev->virtaddr + MAX_XACTIONS_OFFSET);
+ val &= ~(MAX_WR_XACTIONS_MASK << MAX_WR_XACTIONS_BIT_POS);
+ val |= mgmtdev->max_wr_xactions << MAX_WR_XACTIONS_BIT_POS;
+ val &= ~MAX_RD_XACTIONS_MASK;
+ val |= mgmtdev->max_rd_xactions;
+ writel(val, mgmtdev->virtaddr + MAX_XACTIONS_OFFSET);
+
+ mgmtdev->hw_version = readl(mgmtdev->virtaddr + HW_VERSION_OFFSET);
+ mgmtdev->hw_version_major = (mgmtdev->hw_version >> 28) & 0xF;
+ mgmtdev->hw_version_minor = (mgmtdev->hw_version >> 16) & 0xF;
+
+ for (i = 0; i < mgmtdev->dma_channels; i++) {
+ val = readl(mgmtdev->virtaddr + QOS_N_OFFSET + (4 * i));
+ val &= ~(1 << PRIORITY_BIT_POS);
+ val |= (mgmtdev->priority[i] & 0x1) << PRIORITY_BIT_POS;
+ val &= ~(WEIGHT_MASK << WRR_BIT_POS);
+ val |= (mgmtdev->weight[i] & WEIGHT_MASK) << WRR_BIT_POS;
+ writel(val, mgmtdev->virtaddr + QOS_N_OFFSET + (4 * i));
+ }
+
+ val = readl(mgmtdev->virtaddr + CHRESET_TIMEOUT_OFFSET);
+ val &= ~CHRESET_TIMEOUUT_MASK;
+ val |= mgmtdev->chreset_timeout_cycles & CHRESET_TIMEOUUT_MASK;
+ writel(val, mgmtdev->virtaddr + CHRESET_TIMEOUT_OFFSET);
+
+ pm_runtime_mark_last_busy(&mgmtdev->pdev->dev);
+ pm_runtime_put_autosuspend(&mgmtdev->pdev->dev);
+ return 0;
+}
+EXPORT_SYMBOL_GPL(hidma_mgmt_setup);
+
+static int hidma_mgmt_probe(struct platform_device *pdev)
+{
+ struct hidma_mgmt_dev *mgmtdev;
+ struct resource *res;
+ void __iomem *virtaddr;
+ int irq;
+ int rc;
+ u32 val;
+
+ pm_runtime_set_autosuspend_delay(&pdev->dev, AUTOSUSPEND_TIMEOUT);
+ pm_runtime_use_autosuspend(&pdev->dev);
+ pm_runtime_set_active(&pdev->dev);
+ pm_runtime_enable(&pdev->dev);
+ pm_runtime_get_sync(&pdev->dev);
+
+ res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+ virtaddr = devm_ioremap_resource(&pdev->dev, res);
+ if (IS_ERR(virtaddr)) {
+ rc = -ENOMEM;
+ goto out;
+ }
+
+ irq = platform_get_irq(pdev, 0);
+ if (irq < 0) {
+ dev_err(&pdev->dev, "irq resources not found\n");
+ rc = irq;
+ goto out;
+ }
+
+ mgmtdev = devm_kzalloc(&pdev->dev, sizeof(*mgmtdev), GFP_KERNEL);
+ if (!mgmtdev) {
+ rc = -ENOMEM;
+ goto out;
+ }
+
+ mgmtdev->pdev = pdev;
+ mgmtdev->addrsize = resource_size(res);
+ mgmtdev->virtaddr = virtaddr;
+
+ rc = device_property_read_u32(&pdev->dev, "dma-channels",
+ &mgmtdev->dma_channels);
+ if (rc) {
+ dev_err(&pdev->dev, "number of channels missing\n");
+ goto out;
+ }
+
+ rc = device_property_read_u32(&pdev->dev,
+ "channel-reset-timeout-cycles",
+ &mgmtdev->chreset_timeout_cycles);
+ if (rc) {
+ dev_err(&pdev->dev, "channel reset timeout missing\n");
+ goto out;
+ }
+
+ rc = device_property_read_u32(&pdev->dev, "max-write-burst-bytes",
+ &mgmtdev->max_write_request);
+ if (rc) {
+ dev_err(&pdev->dev, "max-write-burst-bytes missing\n");
+ goto out;
+ }
+
+ rc = device_property_read_u32(&pdev->dev, "max-read-burst-bytes",
+ &mgmtdev->max_read_request);
+ if (rc) {
+ dev_err(&pdev->dev, "max-read-burst-bytes missing\n");
+ goto out;
+ }
+
+ rc = device_property_read_u32(&pdev->dev, "max-write-transactions",
+ &mgmtdev->max_wr_xactions);
+ if (rc) {
+ dev_err(&pdev->dev, "max-write-transactions missing\n");
+ goto out;
+ }
+
+ rc = device_property_read_u32(&pdev->dev, "max-read-transactions",
+ &mgmtdev->max_rd_xactions);
+ if (rc) {
+ dev_err(&pdev->dev, "max-read-transactions missing\n");
+ goto out;
+ }
+
+ mgmtdev->priority = devm_kcalloc(&pdev->dev,
+ mgmtdev->dma_channels, sizeof(*mgmtdev->priority), GFP_KERNEL);
+ if (!mgmtdev->priority) {
+ rc = -ENOMEM;
+ goto out;
+ }
+
+ mgmtdev->weight = devm_kcalloc(&pdev->dev,
+ mgmtdev->dma_channels, sizeof(*mgmtdev->weight), GFP_KERNEL);
+ if (!mgmtdev->weight) {
+ rc = -ENOMEM;
+ goto out;
+ }
+
+ rc = device_property_read_u32_array(&pdev->dev, "channel-priority",
+ mgmtdev->priority, mgmtdev->dma_channels);
+ if (rc) {
+ dev_err(&pdev->dev, "channel-priority missing\n");
+ goto out;
+ }
+
+ rc = device_property_read_u32_array(&pdev->dev, "channel-weight",
+ mgmtdev->weight, mgmtdev->dma_channels);
+ if (rc) {
+ dev_err(&pdev->dev, "channel-weight missing\n");
+ goto out;
+ }
+
+ rc = hidma_mgmt_setup(mgmtdev);
+ if (rc) {
+ dev_err(&pdev->dev, "setup failed\n");
+ goto out;
+ }
+
+ /* start the HW */
+ val = readl(mgmtdev->virtaddr + CFG_OFFSET);
+ val |= 1;
+ writel(val, mgmtdev->virtaddr + CFG_OFFSET);
+
+ rc = hidma_mgmt_init_sys(mgmtdev);
+ if (rc) {
+ dev_err(&pdev->dev, "sysfs setup failed\n");
+ goto out;
+ }
+
+ dev_info(&pdev->dev,
+ "HW rev: %d.%d @ %pa with %d physical channels\n",
+ mgmtdev->hw_version_major, mgmtdev->hw_version_minor,
+ &res->start, mgmtdev->dma_channels);
+
+ platform_set_drvdata(pdev, mgmtdev);
+ pm_runtime_mark_last_busy(&pdev->dev);
+ pm_runtime_put_autosuspend(&pdev->dev);
+ return 0;
+out:
+ pm_runtime_put_sync_suspend(&pdev->dev);
+ pm_runtime_disable(&pdev->dev);
+ return rc;
+}
+
+#if IS_ENABLED(CONFIG_ACPI)
+static const struct acpi_device_id hidma_mgmt_acpi_ids[] = {
+ {"QCOM8060"},
+ {},
+};
+#endif
+
+static const struct of_device_id hidma_mgmt_match[] = {
+ { .compatible = "qcom,hidma-mgmt-1.0", },
+ {},
+};
+MODULE_DEVICE_TABLE(of, hidma_mgmt_match);
+
+static struct platform_driver hidma_mgmt_driver = {
+ .probe = hidma_mgmt_probe,
+ .driver = {
+ .name = "hidma-mgmt",
+ .of_match_table = hidma_mgmt_match,
+ .acpi_match_table = ACPI_PTR(hidma_mgmt_acpi_ids),
+ },
+};
+module_platform_driver(hidma_mgmt_driver);
+MODULE_LICENSE("GPL v2");
diff --git a/drivers/dma/qcom/hidma_mgmt.h b/drivers/dma/qcom/hidma_mgmt.h
new file mode 100644
index 0000000..f7daf33
--- /dev/null
+++ b/drivers/dma/qcom/hidma_mgmt.h
@@ -0,0 +1,39 @@
+/*
+ * Qualcomm Technologies HIDMA Management common header
+ *
+ * Copyright (c) 2015, The Linux Foundation. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ */
+
+struct hidma_mgmt_dev {
+ u8 hw_version_major;
+ u8 hw_version_minor;
+
+ u32 max_wr_xactions;
+ u32 max_rd_xactions;
+ u32 max_write_request;
+ u32 max_read_request;
+ u32 dma_channels;
+ u32 chreset_timeout_cycles;
+ u32 hw_version;
+ u32 *priority;
+ u32 *weight;
+
+ /* Hardware device constants */
+ void __iomem *virtaddr;
+ resource_size_t addrsize;
+
+ struct kobject **chroots;
+ struct platform_device *pdev;
+};
+
+int hidma_mgmt_init_sys(struct hidma_mgmt_dev *dev);
+int hidma_mgmt_setup(struct hidma_mgmt_dev *mgmtdev);
diff --git a/drivers/dma/qcom/hidma_mgmt_sys.c b/drivers/dma/qcom/hidma_mgmt_sys.c
new file mode 100644
index 0000000..a4a6414
--- /dev/null
+++ b/drivers/dma/qcom/hidma_mgmt_sys.c
@@ -0,0 +1,299 @@
+/*
+ * Qualcomm Technologies HIDMA Management SYS interface
+ *
+ * Copyright (c) 2015, The Linux Foundation. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ */
+
+#include <linux/sysfs.h>
+#include <linux/platform_device.h>
+
+#include "hidma_mgmt.h"
+
+struct hidma_chan_attr {
+ struct hidma_mgmt_dev *mdev;
+ int index;
+ struct kobj_attribute attr;
+};
+
+struct hidma_mgmt_fileinfo {
+ char *name;
+ int mode;
+ int (*get)(struct hidma_mgmt_dev *mdev);
+ int (*set)(struct hidma_mgmt_dev *mdev, u64 val);
+};
+
+#define IMPLEMENT_GETSET(name) \
+static int get_##name(struct hidma_mgmt_dev *mdev) \
+{ \
+ return mdev->name; \
+} \
+static int set_##name(struct hidma_mgmt_dev *mdev, u64 val) \
+{ \
+ u64 tmp; \
+ int rc; \
+ \
+ tmp = mdev->name; \
+ mdev->name = val; \
+ rc = hidma_mgmt_setup(mdev); \
+ if (rc) \
+ mdev->name = tmp; \
+ return rc; \
+}
+
+#define DECLARE_ATTRIBUTE(name, mode) \
+ {#name, mode, get_##name, set_##name}
+
+IMPLEMENT_GETSET(hw_version_major)
+IMPLEMENT_GETSET(hw_version_minor)
+IMPLEMENT_GETSET(max_wr_xactions)
+IMPLEMENT_GETSET(max_rd_xactions)
+IMPLEMENT_GETSET(max_write_request)
+IMPLEMENT_GETSET(max_read_request)
+IMPLEMENT_GETSET(dma_channels)
+IMPLEMENT_GETSET(chreset_timeout_cycles)
+
+static int set_priority(struct hidma_mgmt_dev *mdev, unsigned int i, u64 val)
+{
+ u64 tmp;
+ int rc;
+
+ if (i > mdev->dma_channels)
+ return -EINVAL;
+
+ tmp = mdev->priority[i];
+ mdev->priority[i] = val;
+ rc = hidma_mgmt_setup(mdev);
+ if (rc)
+ mdev->priority[i] = tmp;
+ return rc;
+}
+
+static int set_weight(struct hidma_mgmt_dev *mdev, unsigned int i, u64 val)
+{
+ u64 tmp;
+ int rc;
+
+ if (i > mdev->dma_channels)
+ return -EINVAL;
+
+ tmp = mdev->weight[i];
+ mdev->weight[i] = val;
+ rc = hidma_mgmt_setup(mdev);
+ if (rc)
+ mdev->weight[i] = tmp;
+ return rc;
+}
+
+static struct hidma_mgmt_fileinfo hidma_mgmt_files[] = {
+ DECLARE_ATTRIBUTE(hw_version_major, S_IRUGO),
+ DECLARE_ATTRIBUTE(hw_version_minor, S_IRUGO),
+ DECLARE_ATTRIBUTE(dma_channels, S_IRUGO),
+ DECLARE_ATTRIBUTE(chreset_timeout_cycles, S_IRUGO),
+ DECLARE_ATTRIBUTE(max_wr_xactions, (S_IRUGO|S_IWUGO)),
+ DECLARE_ATTRIBUTE(max_rd_xactions, (S_IRUGO|S_IWUGO)),
+ DECLARE_ATTRIBUTE(max_write_request, (S_IRUGO|S_IWUGO)),
+ DECLARE_ATTRIBUTE(max_read_request, (S_IRUGO|S_IWUGO)),
+};
+
+static ssize_t show_values(struct device *dev, struct device_attribute *attr,
+ char *buf)
+{
+ struct platform_device *pdev = to_platform_device(dev);
+ struct hidma_mgmt_dev *mdev = platform_get_drvdata(pdev);
+ unsigned int i;
+
+ buf[0] = 0;
+
+ for (i = 0; i < ARRAY_SIZE(hidma_mgmt_files); i++) {
+ if (strcmp(attr->attr.name, hidma_mgmt_files[i].name) == 0) {
+ sprintf(buf, "%d\n", hidma_mgmt_files[i].get(mdev));
+ goto done;
+ }
+ }
+done:
+ return strlen(buf);
+}
+
+static ssize_t set_values(struct device *dev, struct device_attribute *attr,
+ const char *buf, size_t count)
+{
+ struct platform_device *pdev = to_platform_device(dev);
+ struct hidma_mgmt_dev *mdev = platform_get_drvdata(pdev);
+ unsigned long tmp;
+ unsigned int i;
+ int rc;
+
+ rc = kstrtoul(buf, 0, &tmp);
+ if (rc)
+ return rc;
+
+ for (i = 0; i < ARRAY_SIZE(hidma_mgmt_files); i++) {
+ if (strcmp(attr->attr.name, hidma_mgmt_files[i].name) == 0) {
+ rc = hidma_mgmt_files[i].set(mdev, tmp);
+ if (rc)
+ return rc;
+
+ goto done;
+ }
+ }
+done:
+ return count;
+}
+
+static ssize_t show_values_channel(struct kobject *kobj,
+ struct kobj_attribute *attr, char *buf)
+{
+ struct hidma_chan_attr *chattr;
+ struct hidma_mgmt_dev *mdev;
+
+ buf[0] = 0;
+ chattr = container_of(attr, struct hidma_chan_attr, attr);
+ mdev = chattr->mdev;
+ if (strcmp(attr->attr.name, "priority") == 0) {
+ sprintf(buf, "%d\n", mdev->priority[chattr->index]);
+ goto done;
+ }
+
+ if (strcmp(attr->attr.name, "weight") == 0) {
+ sprintf(buf, "%d\n", mdev->weight[chattr->index]);
+ goto done;
+ }
+
+done:
+ return strlen(buf);
+}
+
+static ssize_t set_values_channel(struct kobject *kobj,
+ struct kobj_attribute *attr, const char *buf, size_t count)
+{
+ struct hidma_chan_attr *chattr;
+ struct hidma_mgmt_dev *mdev;
+ unsigned long tmp;
+ int rc;
+
+ chattr = container_of(attr, struct hidma_chan_attr, attr);
+ mdev = chattr->mdev;
+
+ rc = kstrtoul(buf, 0, &tmp);
+ if (rc)
+ return rc;
+
+ if (strcmp(attr->attr.name, "priority") == 0) {
+ rc = set_priority(mdev, chattr->index, tmp);
+ if (rc)
+ return rc;
+ goto done;
+ }
+
+ if (strcmp(attr->attr.name, "weight") == 0) {
+ rc = set_weight(mdev, chattr->index, tmp);
+ if (rc)
+ return rc;
+ }
+done:
+ return count;
+}
+
+static int create_sysfs_entry(struct hidma_mgmt_dev *dev, char *name, int mode)
+{
+ struct device_attribute *attrs;
+ char *name_copy;
+
+ attrs = devm_kmalloc(&dev->pdev->dev,
+ sizeof(struct device_attribute), GFP_KERNEL);
+ if (!attrs)
+ return -ENOMEM;
+
+ name_copy = devm_kstrdup(&dev->pdev->dev, name, GFP_KERNEL);
+ if (!name_copy)
+ return -ENOMEM;
+
+ attrs->attr.name = name_copy;
+ attrs->attr.mode = mode;
+ attrs->show = show_values;
+ attrs->store = set_values;
+ sysfs_attr_init(&attrs->attr);
+
+ return device_create_file(&dev->pdev->dev, attrs);
+}
+
+static int create_sysfs_entry_channel(struct hidma_mgmt_dev *mdev, char *name,
+ int mode, int index, struct kobject *parent)
+{
+ struct hidma_chan_attr *chattr;
+ char *name_copy;
+
+ chattr = devm_kmalloc(&mdev->pdev->dev, sizeof(*chattr), GFP_KERNEL);
+ if (!chattr)
+ return -ENOMEM;
+
+ name_copy = devm_kstrdup(&mdev->pdev->dev, name, GFP_KERNEL);
+ if (!name_copy)
+ return -ENOMEM;
+
+ chattr->mdev = mdev;
+ chattr->index = index;
+ chattr->attr.attr.name = name_copy;
+ chattr->attr.attr.mode = mode;
+ chattr->attr.show = show_values_channel;
+ chattr->attr.store = set_values_channel;
+ sysfs_attr_init(&chattr->attr.attr);
+
+ return sysfs_create_file(parent, &chattr->attr.attr);
+}
+
+int hidma_mgmt_init_sys(struct hidma_mgmt_dev *mdev)
+{
+ unsigned int i;
+ int rc;
+ int required;
+
+ required = sizeof(*mdev->chroots) * mdev->dma_channels;
+ mdev->chroots = devm_kmalloc(&mdev->pdev->dev, required, GFP_KERNEL);
+ if (!mdev->chroots)
+ return -ENOMEM;
+
+ /* create each channel directory here */
+ for (i = 0; i < mdev->dma_channels; i++) {
+ char name[10];
+
+ snprintf(name, sizeof(name), "chan%d", i);
+ mdev->chroots[i] = kobject_create_and_add(name,
+ &mdev->pdev->dev.kobj);
+ if (!mdev->chroots[i])
+ return -ENOMEM;
+ }
+
+ /* populate common parameters */
+ for (i = 0; i < ARRAY_SIZE(hidma_mgmt_files); i++) {
+ rc = create_sysfs_entry(mdev, hidma_mgmt_files[i].name,
+ hidma_mgmt_files[i].mode);
+ if (rc)
+ return rc;
+ }
+
+ /* populate parameters that are per channel */
+ for (i = 0; i < mdev->dma_channels; i++) {
+ rc = create_sysfs_entry_channel(mdev, "priority",
+ (S_IRUGO|S_IWUGO), i, mdev->chroots[i]);
+ if (rc)
+ return rc;
+
+ rc = create_sysfs_entry_channel(mdev, "weight",
+ (S_IRUGO|S_IWUGO), i, mdev->chroots[i]);
+ if (rc)
+ return rc;
+ }
+
+ return 0;
+}
+EXPORT_SYMBOL_GPL(hidma_mgmt_init_sys);
--
Qualcomm Technologies, Inc. on behalf of Qualcomm Innovation Center, Inc.
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a Linux Foundation Collaborative Project
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
[toc] | [prev] | [next] | [standalone]
| From | Andy Shevchenko <andy.shevchenko@gmail.com> |
|---|---|
| Date | 2015-11-23 19:10 +0100 |
| Subject | Re: [PATCH V7 2/3] dma: add Qualcomm Technologies HIDMA management driver |
| Message-ID | <qy3IS-5o5-27@gated-at.bofh.it> |
| In reply to | #1275014 |
On Mon, Nov 23, 2015 at 4:28 AM, Sinan Kaya <okaya@codeaurora.org> wrote:
> The Qualcomm Technologies HIDMA device has been designed
> to support virtualization technology. The driver has been
> divided into two to follow the hardware design.
>
> 1. HIDMA Management driver
> 2. HIDMA Channel driver
>
> Each HIDMA HW consists of multiple channels. These channels
> share some set of common parameters. These parameters are
> initialized by the management driver during power up.
> Same management driver is used for monitoring the execution
> of the channels. Management driver can change the performance
> behavior dynamically such as bandwidth allocation and
> prioritization.
>
> The management driver is executed in hypervisor context and
> is the main management entity for all channels provided by
> the device.
>
Thanks for an update!
Looks cool! Though still few errors spelling and style nitpicks (you
may ignore them if you wish) below.
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
> Signed-off-by: Sinan Kaya <okaya@codeaurora.org>
> ---
> .../ABI/testing/sysfs-platform-hidma-mgmt | 97 +++++++
> .../devicetree/bindings/dma/qcom_hidma_mgmt.txt | 61 ++++
> drivers/dma/qcom/Kconfig | 11 +
> drivers/dma/qcom/Makefile | 1 +
> drivers/dma/qcom/hidma_mgmt.c | 308 +++++++++++++++++++++
> drivers/dma/qcom/hidma_mgmt.h | 39 +++
> drivers/dma/qcom/hidma_mgmt_sys.c | 299 ++++++++++++++++++++
> 7 files changed, 816 insertions(+)
> create mode 100644 Documentation/ABI/testing/sysfs-platform-hidma-mgmt
> create mode 100644 Documentation/devicetree/bindings/dma/qcom_hidma_mgmt.txt
> create mode 100644 drivers/dma/qcom/hidma_mgmt.c
> create mode 100644 drivers/dma/qcom/hidma_mgmt.h
> create mode 100644 drivers/dma/qcom/hidma_mgmt_sys.c
>
> diff --git a/Documentation/ABI/testing/sysfs-platform-hidma-mgmt b/Documentation/ABI/testing/sysfs-platform-hidma-mgmt
> new file mode 100644
> index 0000000..4fc6876
> --- /dev/null
> +++ b/Documentation/ABI/testing/sysfs-platform-hidma-mgmt
> @@ -0,0 +1,97 @@
> +What: /sys/devices/platform/hidma-mgmt*/chan*/priority
> + /sys/devices/platform/QCOM8060:*/chan*/priority
> +Date: Nov 2015
> +KernelVersion: 4.4
> +Contact: "Sinan Kaya <okaya@cudeaurora.org>"
> +Description:
> + Contains either 0 or 1 and indicates if the DMA channel is a
> + low priority (0) or high priority (1) channel.
> +
> +What: /sys/devices/platform/hidma-mgmt*/chan*/weight
> + /sys/devices/platform/QCOM8060:*/chan*/weight
> +Date: Nov 2015
> +KernelVersion: 4.4
> +Contact: "Sinan Kaya <okaya@cudeaurora.org>"
> +Description:
> + Contains 0..15 and indicates the weight of the channel among
> + equal priority channels during round robin scheduling.
> +
> +What: /sys/devices/platform/hidma-mgmt*/chreset_timeout_cycles
> + /sys/devices/platform/QCOM8060:*/chreset_timeout_cycles
> +Date: Nov 2015
> +KernelVersion: 4.4
> +Contact: "Sinan Kaya <okaya@cudeaurora.org>"
> +Description:
> + Contains the platform specific cycle value to wait after a
> + reset command is issued. If the value is chosen too short,
> + then the HW will issue a reset failure interrupt. The value
> + is platform specific and should not be changed without
> + consultance.
> +
> +What: /sys/devices/platform/hidma-mgmt*/dma_channels
> + /sys/devices/platform/QCOM8060:*/dma_channels
> +Date: Nov 2015
> +KernelVersion: 4.4
> +Contact: "Sinan Kaya <okaya@cudeaurora.org>"
> +Description:
> + Contains the number of dma channels supported by one instance
> + of HIDMA hardware. The value may change from chip to chip.
> +
> +What: /sys/devices/platform/hidma-mgmt*/hw_version_major
> + /sys/devices/platform/QCOM8060:*/hw_version_major
> +Date: Nov 2015
> +KernelVersion: 4.4
> +Contact: "Sinan Kaya <okaya@cudeaurora.org>"
> +Description:
> + Version number major for the hardware.
> +
> +What: /sys/devices/platform/hidma-mgmt*/hw_version_minor
> + /sys/devices/platform/QCOM8060:*/hw_version_minor
> +Date: Nov 2015
> +KernelVersion: 4.4
> +Contact: "Sinan Kaya <okaya@cudeaurora.org>"
> +Description:
> + Version number minor for the hardware.
> +
> +What: /sys/devices/platform/hidma-mgmt*/max_rd_xactions
> + /sys/devices/platform/QCOM8060:*/max_rd_xactions
> +Date: Nov 2015
> +KernelVersion: 4.4
> +Contact: "Sinan Kaya <okaya@cudeaurora.org>"
> +Description:
> + Contains a value between 0 and 31. Maximum number of
> + read transactions that can be issued back to back.
> + Choosing a higher number gives better performance but
> + can also cause performance reduction to other peripherals
> + sharing the same bus.
> +
> +What: /sys/devices/platform/hidma-mgmt*/max_read_request
> + /sys/devices/platform/QCOM8060:*/max_read_request
> +Date: Nov 2015
> +KernelVersion: 4.4
> +Contact: "Sinan Kaya <okaya@cudeaurora.org>"
> +Description:
> + Size of each read request. The value needs to be a power
> + of two and can be between 128 and 1024.
> +
> +What: /sys/devices/platform/hidma-mgmt*/max_wr_xactions
> + /sys/devices/platform/QCOM8060:*/max_wr_xactions
> +Date: Nov 2015
> +KernelVersion: 4.4
> +Contact: "Sinan Kaya <okaya@cudeaurora.org>"
> +Description:
> + Contains a value between 0 and 31. Maximum number of
> + write transactions that can be issued back to back.
> + Choosing a higher number gives better performance but
> + can also cause performance reduction to other peripherals
> + sharing the same bus.
> +
> +
> +What: /sys/devices/platform/hidma-mgmt*/max_write_request
> + /sys/devices/platform/QCOM8060:*/max_write_request
> +Date: Nov 2015
> +KernelVersion: 4.4
> +Contact: "Sinan Kaya <okaya@cudeaurora.org>"
> +Description:
> + Size of each write request. The value needs to be a power
> + of two and can be between 128 and 1024.
> diff --git a/Documentation/devicetree/bindings/dma/qcom_hidma_mgmt.txt b/Documentation/devicetree/bindings/dma/qcom_hidma_mgmt.txt
> new file mode 100644
> index 0000000..eb053b9
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/dma/qcom_hidma_mgmt.txt
> @@ -0,0 +1,61 @@
> +Qualcomm Technologies HIDMA Management interface
> +
> +The Qualcomm Technologies HIDMA device has been designed
> +to support virtualization technology. The driver has been
> +divided into two to follow the hardware design. The management
> +driver is executed in hypervisor context and is the main
> +management entity for all channels provided by the device.
> +
> +Each HIDMA HW consists of multiple channels. These channels
> +share some set of common parameters. These parameters are
> +initialized by the management driver during power up.
> +Same management driver is used for monitoring the execution
> +of the channels. Management driver can change the performance
> +behavior dynamically such as bandwidth allocation and
> +prioritization.
> +
> +All channel devices get probed in the hypervisor
> +context during power up. They show up as DMA engine
> +DMA channels. Then, before starting the virtualization; each
> +channel device is unbound from the hypervisor by VFIO
> +and assign to the guest machine for control.
> +
> +This management driver will be used by the system
> +admin to monitor/reset the execution state of the DMA
> +channels. This will be the management interface.
> +
> +
> +Required properties:
> +- compatible: "qcom,hidma-mgmt-1.0";
> +- reg: Address range for DMA device
> +- dma-channels: Number of channels supported by this DMA controller.
> +- max-write-burst-bytes: Maximum write burst in bytes. A memcpy requested is
> + fragmented to multiples of this amount.
> +- max-read-burst-bytes: Maximum read burst in bytes. A memcpy request is
> + fragmented to multiples of this amount.
> +- max-write-transactions: Maximum write transactions to perform in a burst
> +- max-read-transactions: Maximum read transactions to perform in a burst
> +- channel-reset-timeout-cycles: Channel reset timeout in cycles for this SOC.
> +- channel-priority: Priority of the channel.
> + Each dma channel share the same HW bandwidth with other dma channels.
> + If two requests reach to the HW at the same time from a low priority and
> + high priority channel, high priority channel will claim the bus.
> + 0=low priority, 1=high priority
> +- channel-weight: Round robin weight of the channel
> + Since there are only two priority levels supported, scheduling among
> + the equal priority channels is done via weights.
> +
> +Example:
> +
> + hidma-mgmt@f9984000 = {
> + compatible = "qcom,hidma-mgmt-1.0";
> + reg = <0xf9984000 0x15000>;
> + dma-channels = 6;
> + max-write-burst-bytes = 1024;
> + max-read-burst-bytes = 1024;
> + max-write-transactions = 31;
> + max-read-transactions = 31;
> + channel-reset-timeout-cycles = 0x500;
> + channel-priority = < 1 1 0 0 0 0>;
> + channel-weight = < 1 13 10 3 4 5>;
I don't know if it's okay, but I saw format of arrays like <X Y Z> (no
head space).
> + };
> diff --git a/drivers/dma/qcom/Kconfig b/drivers/dma/qcom/Kconfig
> index f17c272..ebe5b8c 100644
> --- a/drivers/dma/qcom/Kconfig
> +++ b/drivers/dma/qcom/Kconfig
> @@ -6,3 +6,14 @@ config QCOM_BAM_DMA
> ---help---
> Enable support for the QCOM BAM DMA controller. This controller
> provides DMA capabilities for a variety of on-chip devices.
> +
> +config QCOM_HIDMA_MGMT
> + tristate "Qualcomm Technologies HIDMA Management support"
> + select DMA_ENGINE
> + help
> + Enable support for the Qualcomm Technologies HIDMA Management.
> + Each DMA device requires one management interface driver
> + for basic initialization before QCOM_HIDMA channel driver can
> + start managing the channels. In a virtualized environment,
> + the guest OS would run QCOM_HIDMA channel driver and the
> + hypervisor would run the QCOM_HIDMA_MGMT management driver.
> diff --git a/drivers/dma/qcom/Makefile b/drivers/dma/qcom/Makefile
> index f612ae3..1a5a96d 100644
> --- a/drivers/dma/qcom/Makefile
> +++ b/drivers/dma/qcom/Makefile
> @@ -1 +1,2 @@
> obj-$(CONFIG_QCOM_BAM_DMA) += bam_dma.o
> +obj-$(CONFIG_QCOM_HIDMA_MGMT) += hidma_mgmt.o hidma_mgmt_sys.o
> diff --git a/drivers/dma/qcom/hidma_mgmt.c b/drivers/dma/qcom/hidma_mgmt.c
> new file mode 100644
> index 0000000..ea78a4d
> --- /dev/null
> +++ b/drivers/dma/qcom/hidma_mgmt.c
> @@ -0,0 +1,308 @@
> +/*
> + * Qualcomm Technologies HIDMA DMA engine Management interface
> + *
> + * Copyright (c) 2015, The Linux Foundation. All rights reserved.
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License version 2 and
> + * only version 2 as published by the Free Software Foundation.
> + *
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
> + * GNU General Public License for more details.
> + */
> +
> +#include <linux/dmaengine.h>
> +#include <linux/acpi.h>
> +#include <linux/of.h>
> +#include <linux/property.h>
> +#include <linux/interrupt.h>
> +#include <linux/platform_device.h>
> +#include <linux/module.h>
> +#include <linux/uaccess.h>
> +#include <linux/slab.h>
> +#include <linux/pm_runtime.h>
> +#include <linux/bitops.h>
> +
> +#include "hidma_mgmt.h"
> +
> +#define QOS_N_OFFSET 0x300
> +#define CFG_OFFSET 0x400
> +#define MAX_BUS_REQ_LEN_OFFSET 0x41C
> +#define MAX_XACTIONS_OFFSET 0x420
> +#define HW_VERSION_OFFSET 0x424
> +#define CHRESET_TIMEOUT_OFFSET 0x418
> +
> +#define MAX_WR_XACTIONS_MASK GENMASK(4, 0)
> +#define MAX_RD_XACTIONS_MASK GENMASK(4, 0)
> +#define WEIGHT_MASK GENMASK(6, 0)
> +#define MAX_BUS_REQ_LEN_MASK GENMASK(15, 0)
> +#define CHRESET_TIMEOUUT_MASK GENMASK(19, 0)
Spelling?
> +
> +#define MAX_WR_XACTIONS_BIT_POS 16
> +#define MAX_BUS_WR_REQ_BIT_POS 16
> +#define WRR_BIT_POS 8
> +#define PRIORITY_BIT_POS 15
> +
> +#define AUTOSUSPEND_TIMEOUT 2000
> +#define MAX_CHANNEL_WEIGHT 15
> +
> +int hidma_mgmt_setup(struct hidma_mgmt_dev *mgmtdev)
> +{
> + unsigned int i;
> + u32 val;
> +
> + if (!is_power_of_2(mgmtdev->max_write_request) ||
> + (mgmtdev->max_write_request < 128) ||
> + (mgmtdev->max_write_request > 1024)) {
> + dev_err(&mgmtdev->pdev->dev, "invalid write request %d\n",
> + mgmtdev->max_write_request);
> + return -EINVAL;
> + }
> +
> + if (!is_power_of_2(mgmtdev->max_read_request) ||
> + (mgmtdev->max_read_request < 128) ||
> + (mgmtdev->max_read_request > 1024)) {
> + dev_err(&mgmtdev->pdev->dev, "invalid read request %d\n",
> + mgmtdev->max_read_request);
> + return -EINVAL;
> + }
> +
> + if (mgmtdev->max_wr_xactions > MAX_WR_XACTIONS_MASK) {
> + dev_err(&mgmtdev->pdev->dev,
> + "max_wr_xactions cannot be bigger than %ld\n",
> + MAX_WR_XACTIONS_MASK);
> + return -EINVAL;
> + }
> +
> + if (mgmtdev->max_rd_xactions > MAX_RD_XACTIONS_MASK) {
> + dev_err(&mgmtdev->pdev->dev,
> + "max_rd_xactions cannot be bigger than %ld\n",
> + MAX_RD_XACTIONS_MASK);
> + return -EINVAL;
> + }
> +
> + for (i = 0; i < mgmtdev->dma_channels; i++) {
> + if (mgmtdev->priority[i] > 1) {
> + dev_err(&mgmtdev->pdev->dev,
> + "priority can be 0 or 1\n");
> + return -EINVAL;
> + }
> +
> + if (mgmtdev->weight[i] > MAX_CHANNEL_WEIGHT) {
> + dev_err(&mgmtdev->pdev->dev,
> + "max value of weight can be %d.\n",
> + MAX_CHANNEL_WEIGHT);
> + return -EINVAL;
> + }
> +
> + /* weight needs to be at least one */
> + if (mgmtdev->weight[i] == 0)
> + mgmtdev->weight[i] = 1;
> + }
> +
> + pm_runtime_get_sync(&mgmtdev->pdev->dev);
> + val = readl(mgmtdev->virtaddr + MAX_BUS_REQ_LEN_OFFSET);
> + val &= ~(MAX_BUS_REQ_LEN_MASK << MAX_BUS_WR_REQ_BIT_POS);
> + val |= mgmtdev->max_write_request << MAX_BUS_WR_REQ_BIT_POS;
> + val &= ~MAX_BUS_REQ_LEN_MASK;
> + val |= mgmtdev->max_read_request;
> + writel(val, mgmtdev->virtaddr + MAX_BUS_REQ_LEN_OFFSET);
> +
> + val = readl(mgmtdev->virtaddr + MAX_XACTIONS_OFFSET);
> + val &= ~(MAX_WR_XACTIONS_MASK << MAX_WR_XACTIONS_BIT_POS);
> + val |= mgmtdev->max_wr_xactions << MAX_WR_XACTIONS_BIT_POS;
> + val &= ~MAX_RD_XACTIONS_MASK;
> + val |= mgmtdev->max_rd_xactions;
> + writel(val, mgmtdev->virtaddr + MAX_XACTIONS_OFFSET);
> +
> + mgmtdev->hw_version = readl(mgmtdev->virtaddr + HW_VERSION_OFFSET);
> + mgmtdev->hw_version_major = (mgmtdev->hw_version >> 28) & 0xF;
> + mgmtdev->hw_version_minor = (mgmtdev->hw_version >> 16) & 0xF;
> +
> + for (i = 0; i < mgmtdev->dma_channels; i++) {
> + val = readl(mgmtdev->virtaddr + QOS_N_OFFSET + (4 * i));
> + val &= ~(1 << PRIORITY_BIT_POS);
> + val |= (mgmtdev->priority[i] & 0x1) << PRIORITY_BIT_POS;
> + val &= ~(WEIGHT_MASK << WRR_BIT_POS);
> + val |= (mgmtdev->weight[i] & WEIGHT_MASK) << WRR_BIT_POS;
> + writel(val, mgmtdev->virtaddr + QOS_N_OFFSET + (4 * i));
> + }
> +
> + val = readl(mgmtdev->virtaddr + CHRESET_TIMEOUT_OFFSET);
> + val &= ~CHRESET_TIMEOUUT_MASK;
Ditto. _TIMEOUT_?
> + val |= mgmtdev->chreset_timeout_cycles & CHRESET_TIMEOUUT_MASK;
Ditto.
> + writel(val, mgmtdev->virtaddr + CHRESET_TIMEOUT_OFFSET);
> +
> + pm_runtime_mark_last_busy(&mgmtdev->pdev->dev);
> + pm_runtime_put_autosuspend(&mgmtdev->pdev->dev);
> + return 0;
> +}
> +EXPORT_SYMBOL_GPL(hidma_mgmt_setup);
> +
> +static int hidma_mgmt_probe(struct platform_device *pdev)
> +{
> + struct hidma_mgmt_dev *mgmtdev;
> + struct resource *res;
> + void __iomem *virtaddr;
> + int irq;
> + int rc;
> + u32 val;
> +
> + pm_runtime_set_autosuspend_delay(&pdev->dev, AUTOSUSPEND_TIMEOUT);
> + pm_runtime_use_autosuspend(&pdev->dev);
> + pm_runtime_set_active(&pdev->dev);
> + pm_runtime_enable(&pdev->dev);
> + pm_runtime_get_sync(&pdev->dev);
> +
> + res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
> + virtaddr = devm_ioremap_resource(&pdev->dev, res);
> + if (IS_ERR(virtaddr)) {
> + rc = -ENOMEM;
> + goto out;
> + }
> +
> + irq = platform_get_irq(pdev, 0);
> + if (irq < 0) {
> + dev_err(&pdev->dev, "irq resources not found\n");
> + rc = irq;
> + goto out;
> + }
> +
> + mgmtdev = devm_kzalloc(&pdev->dev, sizeof(*mgmtdev), GFP_KERNEL);
> + if (!mgmtdev) {
> + rc = -ENOMEM;
> + goto out;
> + }
> +
> + mgmtdev->pdev = pdev;
> + mgmtdev->addrsize = resource_size(res);
> + mgmtdev->virtaddr = virtaddr;
> +
> + rc = device_property_read_u32(&pdev->dev, "dma-channels",
> + &mgmtdev->dma_channels);
> + if (rc) {
> + dev_err(&pdev->dev, "number of channels missing\n");
> + goto out;
> + }
> +
> + rc = device_property_read_u32(&pdev->dev,
> + "channel-reset-timeout-cycles",
> + &mgmtdev->chreset_timeout_cycles);
> + if (rc) {
> + dev_err(&pdev->dev, "channel reset timeout missing\n");
> + goto out;
> + }
> +
> + rc = device_property_read_u32(&pdev->dev, "max-write-burst-bytes",
> + &mgmtdev->max_write_request);
> + if (rc) {
> + dev_err(&pdev->dev, "max-write-burst-bytes missing\n");
> + goto out;
> + }
> +
> + rc = device_property_read_u32(&pdev->dev, "max-read-burst-bytes",
> + &mgmtdev->max_read_request);
> + if (rc) {
> + dev_err(&pdev->dev, "max-read-burst-bytes missing\n");
> + goto out;
> + }
> +
> + rc = device_property_read_u32(&pdev->dev, "max-write-transactions",
> + &mgmtdev->max_wr_xactions);
> + if (rc) {
> + dev_err(&pdev->dev, "max-write-transactions missing\n");
> + goto out;
> + }
> +
> + rc = device_property_read_u32(&pdev->dev, "max-read-transactions",
> + &mgmtdev->max_rd_xactions);
> + if (rc) {
> + dev_err(&pdev->dev, "max-read-transactions missing\n");
> + goto out;
> + }
> +
> + mgmtdev->priority = devm_kcalloc(&pdev->dev,
> + mgmtdev->dma_channels, sizeof(*mgmtdev->priority), GFP_KERNEL);
> + if (!mgmtdev->priority) {
> + rc = -ENOMEM;
> + goto out;
> + }
> +
> + mgmtdev->weight = devm_kcalloc(&pdev->dev,
> + mgmtdev->dma_channels, sizeof(*mgmtdev->weight), GFP_KERNEL);
> + if (!mgmtdev->weight) {
> + rc = -ENOMEM;
> + goto out;
> + }
> +
> + rc = device_property_read_u32_array(&pdev->dev, "channel-priority",
> + mgmtdev->priority, mgmtdev->dma_channels);
> + if (rc) {
> + dev_err(&pdev->dev, "channel-priority missing\n");
> + goto out;
> + }
> +
> + rc = device_property_read_u32_array(&pdev->dev, "channel-weight",
> + mgmtdev->weight, mgmtdev->dma_channels);
> + if (rc) {
> + dev_err(&pdev->dev, "channel-weight missing\n");
> + goto out;
> + }
> +
> + rc = hidma_mgmt_setup(mgmtdev);
> + if (rc) {
> + dev_err(&pdev->dev, "setup failed\n");
> + goto out;
> + }
> +
> + /* start the HW */
> + val = readl(mgmtdev->virtaddr + CFG_OFFSET);
> + val |= 1;
> + writel(val, mgmtdev->virtaddr + CFG_OFFSET);
> +
> + rc = hidma_mgmt_init_sys(mgmtdev);
> + if (rc) {
> + dev_err(&pdev->dev, "sysfs setup failed\n");
> + goto out;
> + }
> +
> + dev_info(&pdev->dev,
> + "HW rev: %d.%d @ %pa with %d physical channels\n",
> + mgmtdev->hw_version_major, mgmtdev->hw_version_minor,
> + &res->start, mgmtdev->dma_channels);
> +
> + platform_set_drvdata(pdev, mgmtdev);
> + pm_runtime_mark_last_busy(&pdev->dev);
> + pm_runtime_put_autosuspend(&pdev->dev);
> + return 0;
> +out:
> + pm_runtime_put_sync_suspend(&pdev->dev);
> + pm_runtime_disable(&pdev->dev);
> + return rc;
> +}
> +
> +#if IS_ENABLED(CONFIG_ACPI)
> +static const struct acpi_device_id hidma_mgmt_acpi_ids[] = {
> + {"QCOM8060"},
> + {},
> +};
> +#endif
> +
> +static const struct of_device_id hidma_mgmt_match[] = {
> + { .compatible = "qcom,hidma-mgmt-1.0", },
> + {},
> +};
> +MODULE_DEVICE_TABLE(of, hidma_mgmt_match);
> +
> +static struct platform_driver hidma_mgmt_driver = {
> + .probe = hidma_mgmt_probe,
> + .driver = {
> + .name = "hidma-mgmt",
> + .of_match_table = hidma_mgmt_match,
> + .acpi_match_table = ACPI_PTR(hidma_mgmt_acpi_ids),
> + },
> +};
> +module_platform_driver(hidma_mgmt_driver);
> +MODULE_LICENSE("GPL v2");
> diff --git a/drivers/dma/qcom/hidma_mgmt.h b/drivers/dma/qcom/hidma_mgmt.h
> new file mode 100644
> index 0000000..f7daf33
> --- /dev/null
> +++ b/drivers/dma/qcom/hidma_mgmt.h
> @@ -0,0 +1,39 @@
> +/*
> + * Qualcomm Technologies HIDMA Management common header
> + *
> + * Copyright (c) 2015, The Linux Foundation. All rights reserved.
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License version 2 and
> + * only version 2 as published by the Free Software Foundation.
> + *
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
> + * GNU General Public License for more details.
> + */
> +
> +struct hidma_mgmt_dev {
> + u8 hw_version_major;
> + u8 hw_version_minor;
> +
> + u32 max_wr_xactions;
> + u32 max_rd_xactions;
> + u32 max_write_request;
> + u32 max_read_request;
> + u32 dma_channels;
> + u32 chreset_timeout_cycles;
> + u32 hw_version;
> + u32 *priority;
> + u32 *weight;
> +
> + /* Hardware device constants */
> + void __iomem *virtaddr;
> + resource_size_t addrsize;
> +
> + struct kobject **chroots;
> + struct platform_device *pdev;
> +};
> +
> +int hidma_mgmt_init_sys(struct hidma_mgmt_dev *dev);
> +int hidma_mgmt_setup(struct hidma_mgmt_dev *mgmtdev);
> diff --git a/drivers/dma/qcom/hidma_mgmt_sys.c b/drivers/dma/qcom/hidma_mgmt_sys.c
> new file mode 100644
> index 0000000..a4a6414
> --- /dev/null
> +++ b/drivers/dma/qcom/hidma_mgmt_sys.c
> @@ -0,0 +1,299 @@
> +/*
> + * Qualcomm Technologies HIDMA Management SYS interface
> + *
> + * Copyright (c) 2015, The Linux Foundation. All rights reserved.
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License version 2 and
> + * only version 2 as published by the Free Software Foundation.
> + *
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
> + * GNU General Public License for more details.
> + */
> +
> +#include <linux/sysfs.h>
> +#include <linux/platform_device.h>
> +
> +#include "hidma_mgmt.h"
> +
> +struct hidma_chan_attr {
> + struct hidma_mgmt_dev *mdev;
> + int index;
> + struct kobj_attribute attr;
> +};
> +
> +struct hidma_mgmt_fileinfo {
> + char *name;
> + int mode;
> + int (*get)(struct hidma_mgmt_dev *mdev);
> + int (*set)(struct hidma_mgmt_dev *mdev, u64 val);
> +};
> +
> +#define IMPLEMENT_GETSET(name) \
> +static int get_##name(struct hidma_mgmt_dev *mdev) \
> +{ \
> + return mdev->name; \
> +} \
> +static int set_##name(struct hidma_mgmt_dev *mdev, u64 val) \
> +{ \
> + u64 tmp; \
> + int rc; \
> + \
> + tmp = mdev->name; \
> + mdev->name = val; \
> + rc = hidma_mgmt_setup(mdev); \
> + if (rc) \
> + mdev->name = tmp; \
> + return rc; \
> +}
> +
> +#define DECLARE_ATTRIBUTE(name, mode) \
> + {#name, mode, get_##name, set_##name}
> +
> +IMPLEMENT_GETSET(hw_version_major)
> +IMPLEMENT_GETSET(hw_version_minor)
> +IMPLEMENT_GETSET(max_wr_xactions)
> +IMPLEMENT_GETSET(max_rd_xactions)
> +IMPLEMENT_GETSET(max_write_request)
> +IMPLEMENT_GETSET(max_read_request)
> +IMPLEMENT_GETSET(dma_channels)
> +IMPLEMENT_GETSET(chreset_timeout_cycles)
> +
> +static int set_priority(struct hidma_mgmt_dev *mdev, unsigned int i, u64 val)
> +{
> + u64 tmp;
> + int rc;
> +
> + if (i > mdev->dma_channels)
> + return -EINVAL;
> +
> + tmp = mdev->priority[i];
> + mdev->priority[i] = val;
> + rc = hidma_mgmt_setup(mdev);
> + if (rc)
> + mdev->priority[i] = tmp;
> + return rc;
> +}
> +
> +static int set_weight(struct hidma_mgmt_dev *mdev, unsigned int i, u64 val)
> +{
> + u64 tmp;
> + int rc;
> +
> + if (i > mdev->dma_channels)
> + return -EINVAL;
> +
> + tmp = mdev->weight[i];
> + mdev->weight[i] = val;
> + rc = hidma_mgmt_setup(mdev);
> + if (rc)
> + mdev->weight[i] = tmp;
> + return rc;
> +}
> +
> +static struct hidma_mgmt_fileinfo hidma_mgmt_files[] = {
> + DECLARE_ATTRIBUTE(hw_version_major, S_IRUGO),
> + DECLARE_ATTRIBUTE(hw_version_minor, S_IRUGO),
> + DECLARE_ATTRIBUTE(dma_channels, S_IRUGO),
> + DECLARE_ATTRIBUTE(chreset_timeout_cycles, S_IRUGO),
> + DECLARE_ATTRIBUTE(max_wr_xactions, (S_IRUGO|S_IWUGO)),
> + DECLARE_ATTRIBUTE(max_rd_xactions, (S_IRUGO|S_IWUGO)),
> + DECLARE_ATTRIBUTE(max_write_request, (S_IRUGO|S_IWUGO)),
> + DECLARE_ATTRIBUTE(max_read_request, (S_IRUGO|S_IWUGO)),
> +};
> +
> +static ssize_t show_values(struct device *dev, struct device_attribute *attr,
> + char *buf)
> +{
> + struct platform_device *pdev = to_platform_device(dev);
> + struct hidma_mgmt_dev *mdev = platform_get_drvdata(pdev);
> + unsigned int i;
> +
> + buf[0] = 0;
> +
> + for (i = 0; i < ARRAY_SIZE(hidma_mgmt_files); i++) {
> + if (strcmp(attr->attr.name, hidma_mgmt_files[i].name) == 0) {
> + sprintf(buf, "%d\n", hidma_mgmt_files[i].get(mdev));
> + goto done;
break; ?
> + }
> + }
> +done:
> + return strlen(buf);
> +}
> +
> +static ssize_t set_values(struct device *dev, struct device_attribute *attr,
> + const char *buf, size_t count)
> +{
> + struct platform_device *pdev = to_platform_device(dev);
> + struct hidma_mgmt_dev *mdev = platform_get_drvdata(pdev);
> + unsigned long tmp;
> + unsigned int i;
> + int rc;
> +
> + rc = kstrtoul(buf, 0, &tmp);
> + if (rc)
> + return rc;
> +
> + for (i = 0; i < ARRAY_SIZE(hidma_mgmt_files); i++) {
> + if (strcmp(attr->attr.name, hidma_mgmt_files[i].name) == 0) {
> + rc = hidma_mgmt_files[i].set(mdev, tmp);
> + if (rc)
> + return rc;
> +
> + goto done;
Ditto.
> + }
> + }
> +done:
> + return count;
> +}
> +
> +static ssize_t show_values_channel(struct kobject *kobj,
> + struct kobj_attribute *attr, char *buf)
> +{
> + struct hidma_chan_attr *chattr;
> + struct hidma_mgmt_dev *mdev;
> +
> + buf[0] = 0;
> + chattr = container_of(attr, struct hidma_chan_attr, attr);
> + mdev = chattr->mdev;
> + if (strcmp(attr->attr.name, "priority") == 0) {
> + sprintf(buf, "%d\n", mdev->priority[chattr->index]);
> + goto done;
> + }
> +
> + if (strcmp(attr->attr.name, "weight") == 0) {
> + sprintf(buf, "%d\n", mdev->weight[chattr->index]);
> + goto done;
> + }
> +
> +done:
> + return strlen(buf);
> +}
> +
> +static ssize_t set_values_channel(struct kobject *kobj,
> + struct kobj_attribute *attr, const char *buf, size_t count)
> +{
> + struct hidma_chan_attr *chattr;
> + struct hidma_mgmt_dev *mdev;
> + unsigned long tmp;
> + int rc;
> +
> + chattr = container_of(attr, struct hidma_chan_attr, attr);
> + mdev = chattr->mdev;
> +
> + rc = kstrtoul(buf, 0, &tmp);
> + if (rc)
> + return rc;
> +
> + if (strcmp(attr->attr.name, "priority") == 0) {
> + rc = set_priority(mdev, chattr->index, tmp);
> + if (rc)
> + return rc;
> + goto done;
> + }
> +
> + if (strcmp(attr->attr.name, "weight") == 0) {
> + rc = set_weight(mdev, chattr->index, tmp);
> + if (rc)
> + return rc;
> + }
> +done:
> + return count;
> +}
> +
> +static int create_sysfs_entry(struct hidma_mgmt_dev *dev, char *name, int mode)
> +{
> + struct device_attribute *attrs;
> + char *name_copy;
> +
> + attrs = devm_kmalloc(&dev->pdev->dev,
> + sizeof(struct device_attribute), GFP_KERNEL);
> + if (!attrs)
> + return -ENOMEM;
> +
> + name_copy = devm_kstrdup(&dev->pdev->dev, name, GFP_KERNEL);
> + if (!name_copy)
> + return -ENOMEM;
> +
> + attrs->attr.name = name_copy;
> + attrs->attr.mode = mode;
> + attrs->show = show_values;
> + attrs->store = set_values;
> + sysfs_attr_init(&attrs->attr);
> +
> + return device_create_file(&dev->pdev->dev, attrs);
> +}
> +
> +static int create_sysfs_entry_channel(struct hidma_mgmt_dev *mdev, char *name,
> + int mode, int index, struct kobject *parent)
> +{
> + struct hidma_chan_attr *chattr;
> + char *name_copy;
> +
> + chattr = devm_kmalloc(&mdev->pdev->dev, sizeof(*chattr), GFP_KERNEL);
> + if (!chattr)
> + return -ENOMEM;
> +
> + name_copy = devm_kstrdup(&mdev->pdev->dev, name, GFP_KERNEL);
> + if (!name_copy)
> + return -ENOMEM;
> +
> + chattr->mdev = mdev;
> + chattr->index = index;
> + chattr->attr.attr.name = name_copy;
> + chattr->attr.attr.mode = mode;
> + chattr->attr.show = show_values_channel;
> + chattr->attr.store = set_values_channel;
> + sysfs_attr_init(&chattr->attr.attr);
> +
> + return sysfs_create_file(parent, &chattr->attr.attr);
> +}
> +
> +int hidma_mgmt_init_sys(struct hidma_mgmt_dev *mdev)
> +{
> + unsigned int i;
> + int rc;
> + int required;
> +
> + required = sizeof(*mdev->chroots) * mdev->dma_channels;
> + mdev->chroots = devm_kmalloc(&mdev->pdev->dev, required, GFP_KERNEL);
> + if (!mdev->chroots)
> + return -ENOMEM;
> +
> + /* create each channel directory here */
> + for (i = 0; i < mdev->dma_channels; i++) {
> + char name[10];
> +
> + snprintf(name, sizeof(name), "chan%d", i);
> + mdev->chroots[i] = kobject_create_and_add(name,
> + &mdev->pdev->dev.kobj);
> + if (!mdev->chroots[i])
> + return -ENOMEM;
> + }
> +
> + /* populate common parameters */
> + for (i = 0; i < ARRAY_SIZE(hidma_mgmt_files); i++) {
> + rc = create_sysfs_entry(mdev, hidma_mgmt_files[i].name,
> + hidma_mgmt_files[i].mode);
> + if (rc)
> + return rc;
> + }
> +
> + /* populate parameters that are per channel */
> + for (i = 0; i < mdev->dma_channels; i++) {
> + rc = create_sysfs_entry_channel(mdev, "priority",
> + (S_IRUGO|S_IWUGO), i, mdev->chroots[i]);
> + if (rc)
> + return rc;
> +
> + rc = create_sysfs_entry_channel(mdev, "weight",
> + (S_IRUGO|S_IWUGO), i, mdev->chroots[i]);
> + if (rc)
> + return rc;
> + }
> +
> + return 0;
> +}
> +EXPORT_SYMBOL_GPL(hidma_mgmt_init_sys);
> --
> Qualcomm Technologies, Inc. on behalf of Qualcomm Innovation Center, Inc.
> Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a Linux Foundation Collaborative Project
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
--
With Best Regards,
Andy Shevchenko
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
[toc] | [prev] | [next] | [standalone]
| From | Sinan Kaya <okaya@codeaurora.org> |
|---|---|
| Date | 2015-11-23 19:50 +0100 |
| Subject | Re: [PATCH V7 2/3] dma: add Qualcomm Technologies HIDMA management driver |
| Message-ID | <qy4lA-5FS-21@gated-at.bofh.it> |
| In reply to | #1275700 |
On 11/23/2015 1:06 PM, Andy Shevchenko wrote: > Thanks for an update! > > Looks cool! Though still few errors spelling and style nitpicks (you > may ignore them if you wish) below. > > Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com> Thanks, I'll post an updated version next week. -- Sinan Kaya Qualcomm Technologies, Inc. on behalf of Qualcomm Innovation Center, Inc. Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a Linux Foundation Collaborative Project -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
[toc] | [prev] | [next] | [standalone]
| From | Vinod Koul <vinod.koul@intel.com> |
|---|---|
| Date | 2015-11-30 09:20 +0100 |
| Subject | Re: [PATCH V7 2/3] dma: add Qualcomm Technologies HIDMA management driver |
| Message-ID | <qArQJ-14d-5@gated-at.bofh.it> |
| In reply to | #1275014 |
On Sun, Nov 22, 2015 at 09:28:24PM -0500, Sinan Kaya wrote:
> +++ b/Documentation/ABI/testing/sysfs-platform-hidma-mgmt
> @@ -0,0 +1,97 @@
> +What: /sys/devices/platform/hidma-mgmt*/chan*/priority
> + /sys/devices/platform/QCOM8060:*/chan*/priority
> +Date: Nov 2015
> +KernelVersion: 4.4
> +Contact: "Sinan Kaya <okaya@cudeaurora.org>"
> +Description:
> + Contains either 0 or 1 and indicates if the DMA channel is a
> + low priority (0) or high priority (1) channel.
What is purpose of adding sysfs entries here ?
> +
> +#define QOS_N_OFFSET 0x300
> +#define CFG_OFFSET 0x400
> +#define MAX_BUS_REQ_LEN_OFFSET 0x41C
> +#define MAX_XACTIONS_OFFSET 0x420
> +#define HW_VERSION_OFFSET 0x424
> +#define CHRESET_TIMEOUT_OFFSET 0x418
> +
> +#define MAX_WR_XACTIONS_MASK GENMASK(4, 0)
> +#define MAX_RD_XACTIONS_MASK GENMASK(4, 0)
> +#define WEIGHT_MASK GENMASK(6, 0)
> +#define MAX_BUS_REQ_LEN_MASK GENMASK(15, 0)
> +#define CHRESET_TIMEOUUT_MASK GENMASK(19, 0)
> +
> +#define MAX_WR_XACTIONS_BIT_POS 16
> +#define MAX_BUS_WR_REQ_BIT_POS 16
> +#define WRR_BIT_POS 8
> +#define PRIORITY_BIT_POS 15
> +
> +#define AUTOSUSPEND_TIMEOUT 2000
> +#define MAX_CHANNEL_WEIGHT 15
These names are quite generic and prone to collide with generic names,
please prefix them with your driver name
> +}
> +EXPORT_SYMBOL_GPL(hidma_mgmt_setup);
Why is this exported or rather who would be users of this?
> +static int hidma_mgmt_probe(struct platform_device *pdev)
> +{
> + struct hidma_mgmt_dev *mgmtdev;
> + struct resource *res;
> + void __iomem *virtaddr;
> + int irq;
> + int rc;
> + u32 val;
> +
> + pm_runtime_set_autosuspend_delay(&pdev->dev, AUTOSUSPEND_TIMEOUT);
> + pm_runtime_use_autosuspend(&pdev->dev);
> + pm_runtime_set_active(&pdev->dev);
> + pm_runtime_enable(&pdev->dev);
at this time pm core will treat device as fully enabled and pm methods can
be invoked, but you are not ready yet right. Typically these are done at the
end of the probe unless you have a reason...
> +static ssize_t show_values(struct device *dev, struct device_attribute *attr,
> + char *buf)
Please fix the coding style here and other places as well. Specifically
please read Chapter 2
--
~Vinod
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
[toc] | [prev] | [next] | [standalone]
| From | Sinan Kaya <okaya@codeaurora.org> |
|---|---|
| Date | 2015-11-30 15:50 +0100 |
| Subject | Re: [PATCH V7 2/3] dma: add Qualcomm Technologies HIDMA management driver |
| Message-ID | <qAxWa-4RR-11@gated-at.bofh.it> |
| In reply to | #1279662 |
On 11/30/2015 3:21 AM, Vinod Koul wrote:
> On Sun, Nov 22, 2015 at 09:28:24PM -0500, Sinan Kaya wrote:
>
>> +++ b/Documentation/ABI/testing/sysfs-platform-hidma-mgmt
>> @@ -0,0 +1,97 @@
>> +What: /sys/devices/platform/hidma-mgmt*/chan*/priority
>> + /sys/devices/platform/QCOM8060:*/chan*/priority
>> +Date: Nov 2015
>> +KernelVersion: 4.4
>> +Contact: "Sinan Kaya <okaya@cudeaurora.org>"
>> +Description:
>> + Contains either 0 or 1 and indicates if the DMA channel is a
>> + low priority (0) or high priority (1) channel.
>
> What is purpose of adding sysfs entries here ?
>
The goal is to run hidma channels in the guest machines and have
management driver adjust the runtime characteristics of the channels
such as round robing weight, priority, max read request etc. All of
these are runtime configurable.
There will be a userspace application that will configure this on behalf
of the system administrator via sysfs. System administrator will decide
how to allocate hardware resources to the guest machines.
>> +
>> +#define QOS_N_OFFSET 0x300
>> +#define CFG_OFFSET 0x400
>> +#define MAX_BUS_REQ_LEN_OFFSET 0x41C
>> +#define MAX_XACTIONS_OFFSET 0x420
>> +#define HW_VERSION_OFFSET 0x424
>> +#define CHRESET_TIMEOUT_OFFSET 0x418
>> +
>> +#define MAX_WR_XACTIONS_MASK GENMASK(4, 0)
>> +#define MAX_RD_XACTIONS_MASK GENMASK(4, 0)
>> +#define WEIGHT_MASK GENMASK(6, 0)
>> +#define MAX_BUS_REQ_LEN_MASK GENMASK(15, 0)
>> +#define CHRESET_TIMEOUUT_MASK GENMASK(19, 0)
>> +
>> +#define MAX_WR_XACTIONS_BIT_POS 16
>> +#define MAX_BUS_WR_REQ_BIT_POS 16
>> +#define WRR_BIT_POS 8
>> +#define PRIORITY_BIT_POS 15
>> +
>> +#define AUTOSUSPEND_TIMEOUT 2000
>> +#define MAX_CHANNEL_WEIGHT 15
>
> These names are quite generic and prone to collide with generic names,
> please prefix them with your driver name
>
OK
>> +}
>> +EXPORT_SYMBOL_GPL(hidma_mgmt_setup);
>
> Why is this exported or rather who would be users of this?
This driver consists of two files as hidma_mgmt.c and hidma_mgmt_sys.c.
I'm calling hidma_mgmt_setup from the hidma_mgmt_sys.c file when a
parameter such as priority and weight is changed to reconfigure the
hardware. I got linker errors without this export as this function is in
hidma_mgmt.c file.
>
>> +static int hidma_mgmt_probe(struct platform_device *pdev)
>> +{
>> + struct hidma_mgmt_dev *mgmtdev;
>> + struct resource *res;
>> + void __iomem *virtaddr;
>> + int irq;
>> + int rc;
>> + u32 val;
>> +
>> + pm_runtime_set_autosuspend_delay(&pdev->dev, AUTOSUSPEND_TIMEOUT);
>> + pm_runtime_use_autosuspend(&pdev->dev);
>> + pm_runtime_set_active(&pdev->dev);
>> + pm_runtime_enable(&pdev->dev);
>
> at this time pm core will treat device as fully enabled and pm methods can
> be invoked, but you are not ready yet right. Typically these are done at the
> end of the probe unless you have a reason...
I need it here because the clocks are declared as ACPI power resources.
The kernel is turning off all power resources during initialization. In
order for this code to touch the hardware, I need to call enable so that
clocks are enabled once again.
>
>> +static ssize_t show_values(struct device *dev, struct device_attribute *attr,
>> + char *buf)
>
> Please fix the coding style here and other places as well.
what's the problem here?
> Specifically
> please read Chapter 2
>
Why is checkpatch not complaining about any of the coding style issues?
I'm checking my code with checkpatch before submitting. Is there any
other tool that would catch this?
--
Sinan Kaya
Qualcomm Technologies, Inc. on behalf of Qualcomm Innovation Center, Inc.
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a
Linux Foundation Collaborative Project
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
[toc] | [prev] | [next] | [standalone]
| From | Vinod Koul <vinod.koul@intel.com> |
|---|---|
| Date | 2015-12-01 04:20 +0100 |
| Subject | Re: [PATCH V7 2/3] dma: add Qualcomm Technologies HIDMA management driver |
| Message-ID | <qAJDX-44H-3@gated-at.bofh.it> |
| In reply to | #1279946 |
On Mon, Nov 30, 2015 at 09:42:01AM -0500, Sinan Kaya wrote:
> >> +static int hidma_mgmt_probe(struct platform_device *pdev)
> >> +{
> >> + struct hidma_mgmt_dev *mgmtdev;
> >> + struct resource *res;
> >> + void __iomem *virtaddr;
> >> + int irq;
> >> + int rc;
> >> + u32 val;
> >> +
> >> + pm_runtime_set_autosuspend_delay(&pdev->dev, AUTOSUSPEND_TIMEOUT);
> >> + pm_runtime_use_autosuspend(&pdev->dev);
> >> + pm_runtime_set_active(&pdev->dev);
> >> + pm_runtime_enable(&pdev->dev);
> >
> > at this time pm core will treat device as fully enabled and pm methods can
> > be invoked, but you are not ready yet right. Typically these are done at the
> > end of the probe unless you have a reason...
>
> I need it here because the clocks are declared as ACPI power resources.
> The kernel is turning off all power resources during initialization. In
> order for this code to touch the hardware, I need to call enable so that
> clocks are enabled once again.
The question is are you ready in your driver routines to be invoked by pm
core?
>
> >
> >> +static ssize_t show_values(struct device *dev, struct device_attribute *attr,
> >> + char *buf)
> >
> > Please fix the coding style here and other places as well.
>
> what's the problem here?
>
> > Specifically
> > please read Chapter 2
> >
>
> Why is checkpatch not complaining about any of the coding style issues?
> I'm checking my code with checkpatch before submitting. Is there any
> other tool that would catch this?
So did you read the Chapter 2.. Quoting here
"Statements longer than 80 columns will be broken into sensible chunks,
unless
exceeding 80 columns significantly increases readability and does not hide
information. Descendants are always substantially shorter than the parent
and
are placed substantially to the right. The same applies to function headers
with a long argument list. However, never break user-visible strings such as
printk messages, because that breaks the ability to grep for them."
Your breaking lines is not placed substantially to the right..
I do not think checkpatch is checking this..
--
~Vinod
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
[toc] | [prev] | [next] | [standalone]
| From | Sinan Kaya <okaya@codeaurora.org> |
|---|---|
| Date | 2015-12-02 06:00 +0100 |
| Subject | Re: [PATCH V7 2/3] dma: add Qualcomm Technologies HIDMA management driver |
| Message-ID | <qB7Gh-2rO-1@gated-at.bofh.it> |
| In reply to | #1280535 |
On 11/30/2015 10:17 PM, Vinod Koul wrote:
> On Mon, Nov 30, 2015 at 09:42:01AM -0500, Sinan Kaya wrote:
>
>>>> +static int hidma_mgmt_probe(struct platform_device *pdev)
>>>> +{
>>>> + struct hidma_mgmt_dev *mgmtdev;
>>>> + struct resource *res;
>>>> + void __iomem *virtaddr;
>>>> + int irq;
>>>> + int rc;
>>>> + u32 val;
>>>> +
>>>> + pm_runtime_set_autosuspend_delay(&pdev->dev, AUTOSUSPEND_TIMEOUT);
>>>> + pm_runtime_use_autosuspend(&pdev->dev);
>>>> + pm_runtime_set_active(&pdev->dev);
>>>> + pm_runtime_enable(&pdev->dev);
>>>
>>> at this time pm core will treat device as fully enabled and pm methods can
>>> be invoked, but you are not ready yet right. Typically these are done at the
>>> end of the probe unless you have a reason...
>>
>> I need it here because the clocks are declared as ACPI power resources.
>> The kernel is turning off all power resources during initialization. In
>> order for this code to touch the hardware, I need to call enable so that
>> clocks are enabled once again.
>
> The question is are you ready in your driver routines to be invoked by pm
> core?
>
I don't have any support for suspend and resume PM APIs. The only PM
interface I support is PM runtime. PM can turn on/off the clocks based
on the reference counts it maintains after get/set APIs. Since PM is
turning off the clocks during power up before my driver load, I do need
to grab this lock to re-enable it during HW initialization. Then, let PM
turn off the clocks again after the AUTOSUSPEND_TIMEOUT when I'm done.
Is there any other interaction with the PM that I'm not aware of?
>>
>>>
>>>> +static ssize_t show_values(struct device *dev, struct device_attribute *attr,
>>>> + char *buf)
>>>
>>> Please fix the coding style here and other places as well.
>>
>> what's the problem here?
>>
>>> Specifically
>>> please read Chapter 2
>>>
>>
>> Why is checkpatch not complaining about any of the coding style issues?
>> I'm checking my code with checkpatch before submitting. Is there any
>> other tool that would catch this?
>
> So did you read the Chapter 2.. Quoting here
>
I did read the chapter 2. Maybe, my lack of native english speaking but
I don't get from this sentence that function parameters need to be
aligned to the opening paranthesis.
> Descendants are always substantially shorter than the parent
> and are placed substantially to the right. The same applies to function headers
> with a long argument list.
>
I ran Lindent and manually cleaned up the junk it introduced. The result
is this
static ssize_t show_values(struct device *dev, struct device_attribute
*attr,
char *buf)
> Your breaking lines is not placed substantially to the right..
> I do not think checkpatch is checking this..
>
--
Sinan Kaya
Qualcomm Technologies, Inc. on behalf of Qualcomm Innovation Center, Inc.
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a
Linux Foundation Collaborative Project
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
[toc] | [prev] | [next] | [standalone]
| From | Vinod Koul <vinod.koul@intel.com> |
|---|---|
| Date | 2015-12-05 09:00 +0100 |
| Subject | Re: [PATCH V7 2/3] dma: add Qualcomm Technologies HIDMA management driver |
| Message-ID | <qCfV8-6f3-13@gated-at.bofh.it> |
| In reply to | #1281445 |
On Tue, Dec 01, 2015 at 11:57:34PM -0500, Sinan Kaya wrote:
> On 11/30/2015 10:17 PM, Vinod Koul wrote:
> > On Mon, Nov 30, 2015 at 09:42:01AM -0500, Sinan Kaya wrote:
> >
> >>>> +static int hidma_mgmt_probe(struct platform_device *pdev)
> >>>> +{
> >>>> + struct hidma_mgmt_dev *mgmtdev;
> >>>> + struct resource *res;
> >>>> + void __iomem *virtaddr;
> >>>> + int irq;
> >>>> + int rc;
> >>>> + u32 val;
> >>>> +
> >>>> + pm_runtime_set_autosuspend_delay(&pdev->dev, AUTOSUSPEND_TIMEOUT);
> >>>> + pm_runtime_use_autosuspend(&pdev->dev);
> >>>> + pm_runtime_set_active(&pdev->dev);
> >>>> + pm_runtime_enable(&pdev->dev);
> >>>
> >>> at this time pm core will treat device as fully enabled and pm methods can
> >>> be invoked, but you are not ready yet right. Typically these are done at the
> >>> end of the probe unless you have a reason...
> >>
> >> I need it here because the clocks are declared as ACPI power resources.
> >> The kernel is turning off all power resources during initialization. In
> >> order for this code to touch the hardware, I need to call enable so that
> >> clocks are enabled once again.
> >
> > The question is are you ready in your driver routines to be invoked by pm
> > core?
> >
>
> I don't have any support for suspend and resume PM APIs. The only PM
> interface I support is PM runtime. PM can turn on/off the clocks based
> on the reference counts it maintains after get/set APIs. Since PM is
> turning off the clocks during power up before my driver load, I do need
> to grab this lock to re-enable it during HW initialization. Then, let PM
> turn off the clocks again after the AUTOSUSPEND_TIMEOUT when I'm done.
>
> Is there any other interaction with the PM that I'm not aware of?
No this is fine. The the runtime_resume will be onvoked and it will request
resources are those set before you enable the device?
--
~Vinod
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
[toc] | [prev] | [next] | [standalone]
| From | Sinan Kaya <okaya@codeaurora.org> |
|---|---|
| Date | 2015-12-08 15:40 +0100 |
| Subject | Re: [PATCH V7 2/3] dma: add Qualcomm Technologies HIDMA management driver |
| Message-ID | <qDrAS-3yq-31@gated-at.bofh.it> |
| In reply to | #1284489 |
On 12/5/2015 2:58 AM, Vinod Koul wrote:
> On Tue, Dec 01, 2015 at 11:57:34PM -0500, Sinan Kaya wrote:
>> On 11/30/2015 10:17 PM, Vinod Koul wrote:
>>> On Mon, Nov 30, 2015 at 09:42:01AM -0500, Sinan Kaya wrote:
>>>
>>>>>> +static int hidma_mgmt_probe(struct platform_device *pdev)
>>>>>> +{
>>>>>> + struct hidma_mgmt_dev *mgmtdev;
>>>>>> + struct resource *res;
>>>>>> + void __iomem *virtaddr;
>>>>>> + int irq;
>>>>>> + int rc;
>>>>>> + u32 val;
>>>>>> +
>>>>>> + pm_runtime_set_autosuspend_delay(&pdev->dev, AUTOSUSPEND_TIMEOUT);
>>>>>> + pm_runtime_use_autosuspend(&pdev->dev);
>>>>>> + pm_runtime_set_active(&pdev->dev);
>>>>>> + pm_runtime_enable(&pdev->dev);
>>>>>
>>>>> at this time pm core will treat device as fully enabled and pm methods can
>>>>> be invoked, but you are not ready yet right. Typically these are done at the
>>>>> end of the probe unless you have a reason...
>>>>
>>>> I need it here because the clocks are declared as ACPI power resources.
>>>> The kernel is turning off all power resources during initialization. In
>>>> order for this code to touch the hardware, I need to call enable so that
>>>> clocks are enabled once again.
>>>
>>> The question is are you ready in your driver routines to be invoked by pm
>>> core?
>>>
>>
>> I don't have any support for suspend and resume PM APIs. The only PM
>> interface I support is PM runtime. PM can turn on/off the clocks based
>> on the reference counts it maintains after get/set APIs. Since PM is
>> turning off the clocks during power up before my driver load, I do need
>> to grab this lock to re-enable it during HW initialization. Then, let PM
>> turn off the clocks again after the AUTOSUSPEND_TIMEOUT when I'm done.
>>
>> Is there any other interaction with the PM that I'm not aware of?
>
> No this is fine. The the runtime_resume will be onvoked and it will request
> resources are those set before you enable the device?
>
Yes, the only resource that this device needs for power management is
the ACPI power resources. The device does not support suspend/resume via
traditional _PS0 and _PS3 calls. ACPI power resources are initialized
during power up while ACPI is being enumerated.
The probing of the HIDMA driver happens much afterwards.
--
Sinan Kaya
Qualcomm Technologies, Inc. on behalf of Qualcomm Innovation Center, Inc.
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a
Linux Foundation Collaborative Project
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
[toc] | [prev] | [next] | [standalone]
| From | Vinod Koul <vinod.koul@intel.com> |
|---|---|
| Date | 2015-11-30 10:00 +0100 |
| Subject | Re: [PATCH V7 3/3] dma: add Qualcomm Technologies HIDMA channel driver |
| Message-ID | <qAsts-1kv-1@gated-at.bofh.it> |
| In reply to | #1275012 |
On Sun, Nov 22, 2015 at 09:28:25PM -0500, Sinan Kaya wrote:
> This patch adds support for hidma engine. The driver consists
> of two logical blocks. The DMA engine interface and the
> low-level interface. The hardware only supports memcpy/memset
> and this driver only support memcpy interface. HW and driver
> doesn't support slave interface.
>
> Signed-off-by: Sinan Kaya <okaya@codeaurora.org>
> ---
> .../devicetree/bindings/dma/qcom_hidma.txt | 18 +
> drivers/dma/qcom/Kconfig | 10 +
> drivers/dma/qcom/Makefile | 2 +
> drivers/dma/qcom/hidma.c | 706 ++++++++++++++++
> drivers/dma/qcom/hidma.h | 157 ++++
> drivers/dma/qcom/hidma_dbg.c | 217 +++++
> drivers/dma/qcom/hidma_ll.c | 924 +++++++++++++++++++++
> 7 files changed, 2034 insertions(+)
This was one of the reasons for slow review of this. I started few times but
large patches made this getting pushed back
Pls help reviewers by splitting your code which looking at this could have
been done
> +#include <linux/dmaengine.h>
> +#include <linux/dma-mapping.h>
> +#include <linux/err.h>
> +#include <linux/init.h>
> +#include <linux/interrupt.h>
> +#include <linux/list.h>
> +#include <linux/module.h>
> +#include <linux/platform_device.h>
> +#include <linux/slab.h>
> +#include <linux/spinlock.h>
> +#include <linux/of_dma.h>
> +#include <linux/property.h>
> +#include <linux/delay.h>
> +#include <linux/highmem.h>
> +#include <linux/io.h>
> +#include <linux/sched.h>
> +#include <linux/wait.h>
> +#include <linux/acpi.h>
> +#include <linux/irq.h>
> +#include <linux/atomic.h>
> +#include <linux/pm_runtime.h>
Do you need so many headers...?
> +MODULE_PARM_DESC(event_channel_idx,
> + "event channel index array for the notifications");
Can you explain this parameter in detail pls
> +static void hidma_callback(void *data)
> +{
> + struct hidma_desc *mdesc = data;
> + struct hidma_chan *mchan = to_hidma_chan(mdesc->desc.chan);
> + struct dma_device *ddev = mchan->chan.device;
> + struct hidma_dev *dmadev = to_hidma_dev(ddev);
> + unsigned long irqflags;
> + bool queued = false;
> +
> + spin_lock_irqsave(&mchan->lock, irqflags);
> + if (mdesc->node.next) {
> + /* Delete from the active list, add to completed list */
> + list_move_tail(&mdesc->node, &mchan->completed);
> + queued = true;
> + }
> + spin_unlock_irqrestore(&mchan->lock, irqflags);
> +
> + hidma_process_completed(dmadev);
> +
> + if (queued) {
> + pm_runtime_mark_last_busy(dmadev->ddev.dev);
> + pm_runtime_put_autosuspend(dmadev->ddev.dev);
> + }
> +}
Callback is typically referred to client callback upon dma completion, can
you explain what you are trying to do here
> +static int hidma_chan_init(struct hidma_dev *dmadev, u32 dma_sig)
> +{
> + struct hidma_chan *mchan;
> + struct dma_device *ddev;
> +
> + mchan = devm_kzalloc(dmadev->ddev.dev, sizeof(*mchan), GFP_KERNEL);
> + if (!mchan)
> + return -ENOMEM;
> +
> + ddev = &dmadev->ddev;
> + mchan->dma_sig = dma_sig;
> + mchan->dmadev = dmadev;
> + mchan->chan.device = ddev;
> + dma_cookie_init(&mchan->chan);
> +
> + INIT_LIST_HEAD(&mchan->free);
> + INIT_LIST_HEAD(&mchan->prepared);
> + INIT_LIST_HEAD(&mchan->active);
> + INIT_LIST_HEAD(&mchan->completed);
> +
> + spin_lock_init(&mchan->lock);
> + list_add_tail(&mchan->chan.device_node, &ddev->channels);
> + dmadev->ddev.chancnt++;
Have you looked at vchan implementation and moving list handling to this
layer ?
> + return 0;
> +}
> +
> +static void hidma_issue_pending(struct dma_chan *dmach)
> +{
> + struct hidma_chan *mchan = to_hidma_chan(dmach);
> + struct hidma_dev *dmadev = mchan->dmadev;
> +
> + /* PM will be released in hidma_callback function. */
> + pm_runtime_get_sync(dmadev->ddev.dev);
Oh no, this is not allowed. pm_runtime_get_sync() can sleep and
issue_pending can be invoked from atomic context.
> +static enum dma_status hidma_tx_status(struct dma_chan *dmach,
> + dma_cookie_t cookie, struct dma_tx_state *txstate)
> +{
> + struct hidma_chan *mchan = to_hidma_chan(dmach);
> + enum dma_status ret;
> +
> + if (mchan->paused)
> + ret = DMA_PAUSED;
This is not quite right. The status query is for a descriptor and NOT for
channel. Here if the descriptor queried was running then it would be paused
for the rest pending txn, it would be DMA_IN_PROGRESS.
> +static dma_cookie_t hidma_tx_submit(struct dma_async_tx_descriptor *txd)
> +{
> + struct hidma_chan *mchan = to_hidma_chan(txd->chan);
> + struct hidma_dev *dmadev = mchan->dmadev;
> + struct hidma_desc *mdesc;
> + unsigned long irqflags;
> + dma_cookie_t cookie;
> +
> + if (!hidma_ll_isenabled(dmadev->lldev))
> + return -ENODEV;
is this not too late for this check, you should fail this on prepare
method
> +static int hidma_alloc_chan_resources(struct dma_chan *dmach)
> +{
> + struct hidma_chan *mchan = to_hidma_chan(dmach);
> + struct hidma_dev *dmadev = mchan->dmadev;
> + struct hidma_desc *mdesc, *tmp;
> + unsigned long irqflags;
> + LIST_HEAD(descs);
> + unsigned int i;
> + int rc = 0;
> +
> + if (mchan->allocated)
> + return 0;
> +
> + /* Alloc descriptors for this channel */
> + for (i = 0; i < dmadev->nr_descriptors; i++) {
> + mdesc = kzalloc(sizeof(struct hidma_desc), GFP_KERNEL);
GFP_NOWAIT pls, this can be called from atomic context
> + if (!mdesc) {
> + rc = -ENOMEM;
> + break;
> + }
> + dma_async_tx_descriptor_init(&mdesc->desc, dmach);
> + mdesc->desc.flags = DMA_CTRL_ACK;
what is the purpose of setting this flag here
> +static void hidma_free_chan_resources(struct dma_chan *dmach)
> +{
> + struct hidma_chan *mchan = to_hidma_chan(dmach);
> + struct hidma_dev *mdma = mchan->dmadev;
> + struct hidma_desc *mdesc, *tmp;
> + unsigned long irqflags;
> + LIST_HEAD(descs);
> +
> + if (!list_empty(&mchan->prepared) || !list_empty(&mchan->active) ||
> + !list_empty(&mchan->completed)) {
> + /*
> + * We have unfinished requests waiting.
> + * Terminate the request from the hardware.
that sounds as bug.. free should be called when txn are completed/aborted
> + */
> + hidma_cleanup_pending_tre(mdma->lldev, ERR_INFO_SW,
> + ERR_CODE_UNEXPECTED_TERMINATE);
> +
> + /* Give enough time for completions to be called. */
> + msleep(100);
should we not poll/read after delay, also we are still in atomic context
> + }
> +
> + spin_lock_irqsave(&mchan->lock, irqflags);
> + /* Channel must be idle */
sorry I do not like that assumption
> + /* return all user requests */
> + list_for_each_entry_safe(mdesc, tmp, &list, node) {
> + struct dma_async_tx_descriptor *txd = &mdesc->desc;
> + dma_async_tx_callback callback = mdesc->desc.callback;
> + void *param = mdesc->desc.callback_param;
> + enum dma_status status;
> +
> + dma_descriptor_unmap(txd);
> +
> + status = hidma_ll_status(dmadev->lldev, mdesc->tre_ch);
> + /*
> + * The API requires that no submissions are done from a
> + * callback, so we don't need to drop the lock here
> + */
That is correct assumption for async_tx and which is deprecated now.
In fact the converse is true for rest!
> +static int hidma_pause(struct dma_chan *chan)
> +{
> + struct hidma_chan *mchan;
> + struct hidma_dev *dmadev;
> +
> + mchan = to_hidma_chan(chan);
> + dmadev = to_hidma_dev(mchan->chan.device);
> + if (!mchan->paused) {
> + pm_runtime_get_sync(dmadev->ddev.dev);
> + if (hidma_ll_pause(dmadev->lldev))
> + dev_warn(dmadev->ddev.dev, "channel did not stop\n");
> + mchan->paused = true;
> + pm_runtime_mark_last_busy(dmadev->ddev.dev);
> + pm_runtime_put_autosuspend(dmadev->ddev.dev);
> + }
> + return 0;
> +}
This is interesting, we associate pause/resume for slave dma ops and not for
memcpy ops, what is the reason for adding pause/resume here?
> +static int hidma_remove(struct platform_device *pdev)
> +{
> + struct hidma_dev *dmadev = platform_get_drvdata(pdev);
> +
> + pm_runtime_get_sync(dmadev->ddev.dev);
> + dma_async_device_unregister(&dmadev->ddev);
> + hidma_debug_uninit(dmadev);
> + hidma_ll_uninit(dmadev->lldev);
> + hidma_free(dmadev);
> +
> + dev_info(&pdev->dev, "HI-DMA engine removed\n");
> + pm_runtime_put_sync_suspend(&pdev->dev);
> + pm_runtime_disable(&pdev->dev);
and your irq is still active and can invoke your irq handler!
> +
> + return 0;
> +}
> +
> +#if IS_ENABLED(CONFIG_ACPI)
> +static const struct acpi_device_id hidma_acpi_ids[] = {
> + {"QCOM8061"},
Qcom and ACPI, that is v interesting combination !
--
~Vinod
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
[toc] | [prev] | [next] | [standalone]
| From | Sinan Kaya <okaya@codeaurora.org> |
|---|---|
| Date | 2015-11-30 21:10 +0100 |
| Subject | Re: [PATCH V7 3/3] dma: add Qualcomm Technologies HIDMA channel driver |
| Message-ID | <qACVQ-8dD-15@gated-at.bofh.it> |
| In reply to | #1279698 |
On 11/30/2015 3:59 AM, Vinod Koul wrote:
> On Sun, Nov 22, 2015 at 09:28:25PM -0500, Sinan Kaya wrote:
>> This patch adds support for hidma engine. The driver consists
>> of two logical blocks. The DMA engine interface and the
>> low-level interface. The hardware only supports memcpy/memset
>> and this driver only support memcpy interface. HW and driver
>> doesn't support slave interface.
>>
>> Signed-off-by: Sinan Kaya <okaya@codeaurora.org>
>> ---
>> .../devicetree/bindings/dma/qcom_hidma.txt | 18 +
>> drivers/dma/qcom/Kconfig | 10 +
>> drivers/dma/qcom/Makefile | 2 +
>> drivers/dma/qcom/hidma.c | 706 ++++++++++++++++
>> drivers/dma/qcom/hidma.h | 157 ++++
>> drivers/dma/qcom/hidma_dbg.c | 217 +++++
>> drivers/dma/qcom/hidma_ll.c | 924 +++++++++++++++++++++
>> 7 files changed, 2034 insertions(+)
>
> This was one of the reasons for slow review of this. I started few times but
> large patches made this getting pushed back
>
> Pls help reviewers by splitting your code which looking at this could have
> been done
I have split the debugfs support from this patch to its own patch. Any
other idea on how else you'd break the code? I can take one more step
and separate the lower layer from the OS layer by using stub functions
on the initial commit.
>
>> +#include <linux/dmaengine.h>
>> +#include <linux/dma-mapping.h>
>> +#include <linux/err.h>
>> +#include <linux/init.h>
>> +#include <linux/interrupt.h>
>> +#include <linux/list.h>
>> +#include <linux/module.h>
>> +#include <linux/platform_device.h>
>> +#include <linux/slab.h>
>> +#include <linux/spinlock.h>
>> +#include <linux/of_dma.h>
>> +#include <linux/property.h>
>> +#include <linux/delay.h>
>> +#include <linux/highmem.h>
>> +#include <linux/io.h>
>> +#include <linux/sched.h>
>> +#include <linux/wait.h>
>> +#include <linux/acpi.h>
>> +#include <linux/irq.h>
>> +#include <linux/atomic.h>
>> +#include <linux/pm_runtime.h>
>
> Do you need so many headers...?
probably not, let me see what I can do about them.
>
>> +MODULE_PARM_DESC(event_channel_idx,
>> + "event channel index array for the notifications");
>
> Can you explain this parameter in detail pls
OK. I added the following comment to the code.
/*
* Each DMA channel is associated with an event channel for interrupt
* delivery. The event channel index usually comes from the firmware through
* ACPI/DT. When a HIDMA channel is executed in the guest machine
context (QEMU)
* the device tree gets auto-generated based on the memory and IRQ resources
* this driver uses on the host machine. Any device specific parameter
such as
* event-channel gets ignored by the QEMU.
* We are using this command line parameter to pass the event channel
index to
* the guest machine.
*/
s
>
>> +static void hidma_callback(void *data)
>> +{
>> + struct hidma_desc *mdesc = data;
>> + struct hidma_chan *mchan = to_hidma_chan(mdesc->desc.chan);
>> + struct dma_device *ddev = mchan->chan.device;
>> + struct hidma_dev *dmadev = to_hidma_dev(ddev);
>> + unsigned long irqflags;
>> + bool queued = false;
>> +
>> + spin_lock_irqsave(&mchan->lock, irqflags);
>> + if (mdesc->node.next) {
>> + /* Delete from the active list, add to completed list */
>> + list_move_tail(&mdesc->node, &mchan->completed);
>> + queued = true;
>> + }
>> + spin_unlock_irqrestore(&mchan->lock, irqflags);
>> +
>> + hidma_process_completed(dmadev);
>> +
>> + if (queued) {
>> + pm_runtime_mark_last_busy(dmadev->ddev.dev);
>> + pm_runtime_put_autosuspend(dmadev->ddev.dev);
>> + }
>> +}
>
> Callback is typically referred to client callback upon dma completion, can
> you explain what you are trying to do here
When a DMA transfer completes, the hidma_callback function in the
hidma.c file gets called from the lower layer (hidma_ll.c) in order to
move this request from active queue into the completed queue.
hidma_process_completed eventually calls the "client callback" in it.
The PM stuff is for guaranteeing that clocks are not turned off while HW
is working on it.
I grab the PM lock in issue pending and release it in the hidma_callback.
>
>> +static int hidma_chan_init(struct hidma_dev *dmadev, u32 dma_sig)
>> +{
>> + struct hidma_chan *mchan;
>> + struct dma_device *ddev;
>> +
>> + mchan = devm_kzalloc(dmadev->ddev.dev, sizeof(*mchan), GFP_KERNEL);
>> + if (!mchan)
>> + return -ENOMEM;
>> +
>> + ddev = &dmadev->ddev;
>> + mchan->dma_sig = dma_sig;
>> + mchan->dmadev = dmadev;
>> + mchan->chan.device = ddev;
>> + dma_cookie_init(&mchan->chan);
>> +
>> + INIT_LIST_HEAD(&mchan->free);
>> + INIT_LIST_HEAD(&mchan->prepared);
>> + INIT_LIST_HEAD(&mchan->active);
>> + INIT_LIST_HEAD(&mchan->completed);
>> +
>> + spin_lock_init(&mchan->lock);
>> + list_add_tail(&mchan->chan.device_node, &ddev->channels);
>> + dmadev->ddev.chancnt++;
>
> Have you looked at vchan implementation and moving list handling to this
> layer ?
I'll add vchan support later on another revision. The original code had
vchan support in it. It became a terminology mess. So, I removed it for
the moment.
Using vchan, I can take multiple clients and serve them in a single
channel later.
>
>> + return 0;
>> +}
>> +
>> +static void hidma_issue_pending(struct dma_chan *dmach)
>> +{
>> + struct hidma_chan *mchan = to_hidma_chan(dmach);
>> + struct hidma_dev *dmadev = mchan->dmadev;
>> +
>> + /* PM will be released in hidma_callback function. */
>> + pm_runtime_get_sync(dmadev->ddev.dev);
>
> Oh no, this is not allowed. pm_runtime_get_sync() can sleep and
> issue_pending can be invoked from atomic context.
OK. I didn't know that. I'll try pm_runtime_get first. If it fails, I'll
post a task to do the real job with pm_runtime_get_sync.
The clocks may be off by the time the request is made. I need to grab
the PM lock before accessing the HW registers.
>
>> +static enum dma_status hidma_tx_status(struct dma_chan *dmach,
>> + dma_cookie_t cookie, struct dma_tx_state *txstate)
>> +{
>> + struct hidma_chan *mchan = to_hidma_chan(dmach);
>> + enum dma_status ret;
>> +
>> + if (mchan->paused)
>> + ret = DMA_PAUSED;
>
> This is not quite right. The status query is for a descriptor and NOT for
> channel. Here if the descriptor queried was running then it would be paused
> for the rest pending txn, it would be DMA_IN_PROGRESS.
The channel can be paused by the hypervisor. If it is paused, then no
other transaction will go through including the pending requests. That's
why, I'm checking the state first.
>
>> +static dma_cookie_t hidma_tx_submit(struct dma_async_tx_descriptor *txd)
>> +{
>> + struct hidma_chan *mchan = to_hidma_chan(txd->chan);
>> + struct hidma_dev *dmadev = mchan->dmadev;
>> + struct hidma_desc *mdesc;
>> + unsigned long irqflags;
>> + dma_cookie_t cookie;
>> +
>> + if (!hidma_ll_isenabled(dmadev->lldev))
>> + return -ENODEV;
>
> is this not too late for this check, you should fail this on prepare
> method
The channels can be paused by the hypervisor at runtime after the guest
OS boot. The client might have allocated the channels during guest
machine boot. If a channel is paused and client makes a request, client
will never get the callback. By placing this check, I'm looking at the
runtime status of the channel and rejecting the transmit request right
ahead.
>
>
>> +static int hidma_alloc_chan_resources(struct dma_chan *dmach)
>> +{
>> + struct hidma_chan *mchan = to_hidma_chan(dmach);
>> + struct hidma_dev *dmadev = mchan->dmadev;
>> + struct hidma_desc *mdesc, *tmp;
>> + unsigned long irqflags;
>> + LIST_HEAD(descs);
>> + unsigned int i;
>> + int rc = 0;
>> +
>> + if (mchan->allocated)
>> + return 0;
>> +
>> + /* Alloc descriptors for this channel */
>> + for (i = 0; i < dmadev->nr_descriptors; i++) {
>> + mdesc = kzalloc(sizeof(struct hidma_desc), GFP_KERNEL);
>
> GFP_NOWAIT pls, this can be called from atomic context
I'll change it, but why would anyone allocate a channel in an interrupt
handler?
>
>> + if (!mdesc) {
>> + rc = -ENOMEM;
>> + break;
>> + }
>> + dma_async_tx_descriptor_init(&mdesc->desc, dmach);
>> + mdesc->desc.flags = DMA_CTRL_ACK;
>
> what is the purpose of setting this flag here
It means that the code only supports interrupt. Maybe, you can correct
me here. I couldn't see a very useful info about the DMA engine flags.
My understanding is that DMA_CTRL_ACK means user wants an interrupt
based callback.
If DMA_CTRL_ACK is not specified, then user wants to poll for the
results. The current code only supports the interrupt based delivery for
callbacks. Polling support is another to-do in my list for small sized
transfers.
>
>> +static void hidma_free_chan_resources(struct dma_chan *dmach)
>> +{
>> + struct hidma_chan *mchan = to_hidma_chan(dmach);
>> + struct hidma_dev *mdma = mchan->dmadev;
>> + struct hidma_desc *mdesc, *tmp;
>> + unsigned long irqflags;
>> + LIST_HEAD(descs);
>> +
>> + if (!list_empty(&mchan->prepared) || !list_empty(&mchan->active) ||
>> + !list_empty(&mchan->completed)) {
>> + /*
>> + * We have unfinished requests waiting.
>> + * Terminate the request from the hardware.
>
> that sounds as bug.. free should be called when txn are completed/aborted
Let me see what other drivers are doing...
>
>> + */
>> + hidma_cleanup_pending_tre(mdma->lldev, ERR_INFO_SW,
>> + ERR_CODE_UNEXPECTED_TERMINATE);
>> +
>> + /* Give enough time for completions to be called. */
>> + msleep(100);
>
> should we not poll/read after delay, also we are still in atomic context
>
>> + }
>> +
>> + spin_lock_irqsave(&mchan->lock, irqflags);
>> + /* Channel must be idle */
>
> sorry I do not like that assumption
>
>> + /* return all user requests */
>> + list_for_each_entry_safe(mdesc, tmp, &list, node) {
>> + struct dma_async_tx_descriptor *txd = &mdesc->desc;
>> + dma_async_tx_callback callback = mdesc->desc.callback;
>> + void *param = mdesc->desc.callback_param;
>> + enum dma_status status;
>> +
>> + dma_descriptor_unmap(txd);
>> +
>> + status = hidma_ll_status(dmadev->lldev, mdesc->tre_ch);
>> + /*
>> + * The API requires that no submissions are done from a
>> + * callback, so we don't need to drop the lock here
>> + */
>
> That is correct assumption for async_tx and which is deprecated now.
> In fact the converse is true for rest!
I copied this function from another driver. The cookie/unmap business
seemed scary to me. Can you point me to a good/recent implementation if
this is deprecated?
>
>> +static int hidma_pause(struct dma_chan *chan)
>> +{
>> + struct hidma_chan *mchan;
>> + struct hidma_dev *dmadev;
>> +
>> + mchan = to_hidma_chan(chan);
>> + dmadev = to_hidma_dev(mchan->chan.device);
>> + if (!mchan->paused) {
>> + pm_runtime_get_sync(dmadev->ddev.dev);
>> + if (hidma_ll_pause(dmadev->lldev))
>> + dev_warn(dmadev->ddev.dev, "channel did not stop\n");
>> + mchan->paused = true;
>> + pm_runtime_mark_last_busy(dmadev->ddev.dev);
>> + pm_runtime_put_autosuspend(dmadev->ddev.dev);
>> + }
>> + return 0;
>> +}
>
> This is interesting, we associate pause/resume for slave dma ops and not for
> memcpy ops, what is the reason for adding pause/resume here?
Why is it limited to the slave device only? I can queue a bunch of
requests. I can pause and resume their execution with the current
implementation.
>
>> +static int hidma_remove(struct platform_device *pdev)
>> +{
>> + struct hidma_dev *dmadev = platform_get_drvdata(pdev);
>> +
>> + pm_runtime_get_sync(dmadev->ddev.dev);
>> + dma_async_device_unregister(&dmadev->ddev);
>> + hidma_debug_uninit(dmadev);
>> + hidma_ll_uninit(dmadev->lldev);
>> + hidma_free(dmadev);
>> +
>> + dev_info(&pdev->dev, "HI-DMA engine removed\n");
>> + pm_runtime_put_sync_suspend(&pdev->dev);
>> + pm_runtime_disable(&pdev->dev);
>
> and your irq is still active and can invoke your irq handler!
>
why? I shutdown the IRQ in hidma_ll_uninit and request the interrupt
with devm_request_irq. As soon as this driver is removed, the interrupt
should be released by the managed code.
>> +
>> + return 0;
>> +}
>> +
>> +#if IS_ENABLED(CONFIG_ACPI)
>> +static const struct acpi_device_id hidma_acpi_ids[] = {
>> + {"QCOM8061"},
> Qcom and ACPI, that is v interesting combination !
>
>
Thanks for the detailed review.
--
Sinan Kaya
Qualcomm Technologies, Inc. on behalf of Qualcomm Innovation Center, Inc.
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a
Linux Foundation Collaborative Project
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
[toc] | [prev] | [next] | [standalone]
| From | Vinod Koul <vinod.koul@intel.com> |
|---|---|
| Date | 2015-12-01 12:40 +0100 |
| Subject | Re: [PATCH V7 3/3] dma: add Qualcomm Technologies HIDMA channel driver |
| Message-ID | <qARrQ-Ak-13@gated-at.bofh.it> |
| In reply to | #1280244 |
On Mon, Nov 30, 2015 at 03:06:36PM -0500, Sinan Kaya wrote:
> I have split the debugfs support from this patch to its own patch. Any
> other idea on how else you'd break the code? I can take one more step
> and separate the lower layer from the OS layer by using stub functions
> on the initial commit.
Yes the ll.c can be a separate patch and no need to add placeholders as
makefile can be last
> >
> >> +static enum dma_status hidma_tx_status(struct dma_chan *dmach,
> >> + dma_cookie_t cookie, struct dma_tx_state *txstate)
> >> +{
> >> + struct hidma_chan *mchan = to_hidma_chan(dmach);
> >> + enum dma_status ret;
> >> +
> >> + if (mchan->paused)
> >> + ret = DMA_PAUSED;
> >
> > This is not quite right. The status query is for a descriptor and NOT for
> > channel. Here if the descriptor queried was running then it would be paused
> > for the rest pending txn, it would be DMA_IN_PROGRESS.
>
> The channel can be paused by the hypervisor. If it is paused, then no
> other transaction will go through including the pending requests. That's
> why, I'm checking the state first.
You are missing the point. Channel can be paused, yes but the descriptor
is in queue and is not paused. The descriptor running is paused, yes.
There is subtle difference between these
> >> +static dma_cookie_t hidma_tx_submit(struct dma_async_tx_descriptor *txd)
> >> +{
> >> + struct hidma_chan *mchan = to_hidma_chan(txd->chan);
> >> + struct hidma_dev *dmadev = mchan->dmadev;
> >> + struct hidma_desc *mdesc;
> >> + unsigned long irqflags;
> >> + dma_cookie_t cookie;
> >> +
> >> + if (!hidma_ll_isenabled(dmadev->lldev))
> >> + return -ENODEV;
> >
> > is this not too late for this check, you should fail this on prepare
> > method
>
> The channels can be paused by the hypervisor at runtime after the guest
> OS boot. The client might have allocated the channels during guest
> machine boot. If a channel is paused and client makes a request, client
> will never get the callback. By placing this check, I'm looking at the
> runtime status of the channel and rejecting the transmit request right
> ahead.
In this case you have request already given to you, here the request is
submitted to the pending queue, if hypervisor has paused you, so the entire
txn queue is paused. But from API semantics I would argue that this should be
accepted and suffer the same fate as already submitted txns
>
> >
> >
> >> +static int hidma_alloc_chan_resources(struct dma_chan *dmach)
> >> +{
> >> + struct hidma_chan *mchan = to_hidma_chan(dmach);
> >> + struct hidma_dev *dmadev = mchan->dmadev;
> >> + struct hidma_desc *mdesc, *tmp;
> >> + unsigned long irqflags;
> >> + LIST_HEAD(descs);
> >> + unsigned int i;
> >> + int rc = 0;
> >> +
> >> + if (mchan->allocated)
> >> + return 0;
> >> +
> >> + /* Alloc descriptors for this channel */
> >> + for (i = 0; i < dmadev->nr_descriptors; i++) {
> >> + mdesc = kzalloc(sizeof(struct hidma_desc), GFP_KERNEL);
> >
> > GFP_NOWAIT pls, this can be called from atomic context
>
> I'll change it, but why would anyone allocate a channel in an interrupt
> handler?
remember this is dmaengine, where people traditionally wanted to offload
from cpu and were not allowed to sleep.
I think am okay with this one..
>
> >
> >> + if (!mdesc) {
> >> + rc = -ENOMEM;
> >> + break;
> >> + }
> >> + dma_async_tx_descriptor_init(&mdesc->desc, dmach);
> >> + mdesc->desc.flags = DMA_CTRL_ACK;
> >
> > what is the purpose of setting this flag here
>
> It means that the code only supports interrupt. Maybe, you can correct
> me here. I couldn't see a very useful info about the DMA engine flags.
> My understanding is that DMA_CTRL_ACK means user wants an interrupt
> based callback.
Right user wants, you are not user though
>
> If DMA_CTRL_ACK is not specified, then user wants to poll for the
> results. The current code only supports the interrupt based delivery for
> callbacks. Polling support is another to-do in my list for small sized
> transfers.
See the documentation for flags and usage. The point here is that user
should set this. I know some old driver do this but that is not right
>
> >
> >> +static void hidma_free_chan_resources(struct dma_chan *dmach)
> >> +{
> >> + struct hidma_chan *mchan = to_hidma_chan(dmach);
> >> + struct hidma_dev *mdma = mchan->dmadev;
> >> + struct hidma_desc *mdesc, *tmp;
> >> + unsigned long irqflags;
> >> + LIST_HEAD(descs);
> >> +
> >> + if (!list_empty(&mchan->prepared) || !list_empty(&mchan->active) ||
> >> + !list_empty(&mchan->completed)) {
> >> + /*
> >> + * We have unfinished requests waiting.
> >> + * Terminate the request from the hardware.
> >
> > that sounds as bug.. free should be called when txn are completed/aborted
>
> Let me see what other drivers are doing...
>
> >
> >> + */
> >> + hidma_cleanup_pending_tre(mdma->lldev, ERR_INFO_SW,
> >> + ERR_CODE_UNEXPECTED_TERMINATE);
> >> +
> >> + /* Give enough time for completions to be called. */
> >> + msleep(100);
> >
> > should we not poll/read after delay, also we are still in atomic context
> >
> >> + }
> >> +
> >> + spin_lock_irqsave(&mchan->lock, irqflags);
> >> + /* Channel must be idle */
> >
> > sorry I do not like that assumption
> >
> >> + /* return all user requests */
> >> + list_for_each_entry_safe(mdesc, tmp, &list, node) {
> >> + struct dma_async_tx_descriptor *txd = &mdesc->desc;
> >> + dma_async_tx_callback callback = mdesc->desc.callback;
> >> + void *param = mdesc->desc.callback_param;
> >> + enum dma_status status;
> >> +
> >> + dma_descriptor_unmap(txd);
> >> +
> >> + status = hidma_ll_status(dmadev->lldev, mdesc->tre_ch);
> >> + /*
> >> + * The API requires that no submissions are done from a
> >> + * callback, so we don't need to drop the lock here
> >> + */
> >
> > That is correct assumption for async_tx and which is deprecated now.
> > In fact the converse is true for rest!
>
> I copied this function from another driver. The cookie/unmap business
> seemed scary to me. Can you point me to a good/recent implementation if
> this is deprecated?
>
> >
> >> +static int hidma_pause(struct dma_chan *chan)
> >> +{
> >> + struct hidma_chan *mchan;
> >> + struct hidma_dev *dmadev;
> >> +
> >> + mchan = to_hidma_chan(chan);
> >> + dmadev = to_hidma_dev(mchan->chan.device);
> >> + if (!mchan->paused) {
> >> + pm_runtime_get_sync(dmadev->ddev.dev);
> >> + if (hidma_ll_pause(dmadev->lldev))
> >> + dev_warn(dmadev->ddev.dev, "channel did not stop\n");
> >> + mchan->paused = true;
> >> + pm_runtime_mark_last_busy(dmadev->ddev.dev);
> >> + pm_runtime_put_autosuspend(dmadev->ddev.dev);
> >> + }
> >> + return 0;
> >> +}
> >
> > This is interesting, we associate pause/resume for slave dma ops and not for
> > memcpy ops, what is the reason for adding pause/resume here?
>
> Why is it limited to the slave device only? I can queue a bunch of
> requests. I can pause and resume their execution with the current
> implementation.
Traditionally no one paused memcpy... People want the copy to be done with
as fast as possible :)
>
> >
> >> +static int hidma_remove(struct platform_device *pdev)
> >> +{
> >> + struct hidma_dev *dmadev = platform_get_drvdata(pdev);
> >> +
> >> + pm_runtime_get_sync(dmadev->ddev.dev);
> >> + dma_async_device_unregister(&dmadev->ddev);
> >> + hidma_debug_uninit(dmadev);
> >> + hidma_ll_uninit(dmadev->lldev);
> >> + hidma_free(dmadev);
> >> +
> >> + dev_info(&pdev->dev, "HI-DMA engine removed\n");
> >> + pm_runtime_put_sync_suspend(&pdev->dev);
> >> + pm_runtime_disable(&pdev->dev);
> >
> > and your irq is still active and can invoke your irq handler!
> >
>
> why? I shutdown the IRQ in hidma_ll_uninit and request the interrupt
> with devm_request_irq. As soon as this driver is removed, the interrupt
> should be released by the managed code.
but there is a delay between that and free getting called. Since you don't
free up explicitly you can still have irq and tasklets scheduled.
Recommendation is that free up irq or ensure none are scheduled and
disabled
--
~Vinod
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
[toc] | [prev] | [next] | [standalone]
| From | Sinan Kaya <okaya@codeaurora.org> |
|---|---|
| Date | 2015-12-01 22:20 +0100 |
| Subject | Re: [PATCH V7 3/3] dma: add Qualcomm Technologies HIDMA channel driver |
| Message-ID | <qB0v7-6w7-5@gated-at.bofh.it> |
| In reply to | #1280800 |
On 12/1/2015 6:34 AM, Vinod Koul wrote:
> On Mon, Nov 30, 2015 at 03:06:36PM -0500, Sinan Kaya wrote:
>> I have split the debugfs support from this patch to its own patch. Any
>> other idea on how else you'd break the code? I can take one more step
>> and separate the lower layer from the OS layer by using stub functions
>> on the initial commit.
>
> Yes the ll.c can be a separate patch and no need to add placeholders as
> makefile can be last
>
ok. I'll create a new patch file with the Makefile and hidma_ll.c only.
>>>
>>>> +static enum dma_status hidma_tx_status(struct dma_chan *dmach,
>>>> + dma_cookie_t cookie, struct dma_tx_state *txstate)
>>>> +{
>>>> + struct hidma_chan *mchan = to_hidma_chan(dmach);
>>>> + enum dma_status ret;
>>>> +
>>>> + if (mchan->paused)
>>>> + ret = DMA_PAUSED;
>>>
>>> This is not quite right. The status query is for a descriptor and NOT for
>>> channel. Here if the descriptor queried was running then it would be paused
>>> for the rest pending txn, it would be DMA_IN_PROGRESS.
>>
>> The channel can be paused by the hypervisor. If it is paused, then no
>> other transaction will go through including the pending requests. That's
>> why, I'm checking the state first.
>
> You are missing the point. Channel can be paused, yes but the descriptor
> is in queue and is not paused. The descriptor running is paused, yes.
> There is subtle difference between these
I'll follow your recommendation. PAUSE for the currently active
descriptor and DMA_IN_PROGRESS for the rest.
>
>>>> +static dma_cookie_t hidma_tx_submit(struct dma_async_tx_descriptor *txd)
>>>> +{
>>>> + struct hidma_chan *mchan = to_hidma_chan(txd->chan);
>>>> + struct hidma_dev *dmadev = mchan->dmadev;
>>>> + struct hidma_desc *mdesc;
>>>> + unsigned long irqflags;
>>>> + dma_cookie_t cookie;
>>>> +
>>>> + if (!hidma_ll_isenabled(dmadev->lldev))
>>>> + return -ENODEV;
>>>
>>> is this not too late for this check, you should fail this on prepare
>>> method
>>
>> The channels can be paused by the hypervisor at runtime after the guest
>> OS boot. The client might have allocated the channels during guest
>> machine boot. If a channel is paused and client makes a request, client
>> will never get the callback. By placing this check, I'm looking at the
>> runtime status of the channel and rejecting the transmit request right
>> ahead.
>
> In this case you have request already given to you, here the request is
> submitted to the pending queue,
No, it is not. Putting into the pending queue is done after this check
not before.
I'm rejecting the queue request directly here as HIDMA channel is no
longer functional. I'd rather do active error handling rather than
passive error handling. Debugging of passive faults are much more difficult.
> if hypervisor has paused you, so the entire
> txn queue is paused. But from API semantics I would argue that this should be
> accepted and suffer the same fate as already submitted txns
>
>>
>>>
>>>
>>>> +static int hidma_alloc_chan_resources(struct dma_chan *dmach)
>>>> +{
>>>> + struct hidma_chan *mchan = to_hidma_chan(dmach);
>>>> + struct hidma_dev *dmadev = mchan->dmadev;
>>>> + struct hidma_desc *mdesc, *tmp;
>>>> + unsigned long irqflags;
>>>> + LIST_HEAD(descs);
>>>> + unsigned int i;
>>>> + int rc = 0;
>>>> +
>>>> + if (mchan->allocated)
>>>> + return 0;
>>>> +
>>>> + /* Alloc descriptors for this channel */
>>>> + for (i = 0; i < dmadev->nr_descriptors; i++) {
>>>> + mdesc = kzalloc(sizeof(struct hidma_desc), GFP_KERNEL);
>>>
>>> GFP_NOWAIT pls, this can be called from atomic context
>>
>> I'll change it, but why would anyone allocate a channel in an interrupt
>> handler?
>
> remember this is dmaengine, where people traditionally wanted to offload
> from cpu and were not allowed to sleep.
>
> I think am okay with this one..
>
>>
>>>
>>>> + if (!mdesc) {
>>>> + rc = -ENOMEM;
>>>> + break;
>>>> + }
>>>> + dma_async_tx_descriptor_init(&mdesc->desc, dmach);
>>>> + mdesc->desc.flags = DMA_CTRL_ACK;
>>>
>>> what is the purpose of setting this flag here
>>
>> It means that the code only supports interrupt. Maybe, you can correct
>> me here. I couldn't see a very useful info about the DMA engine flags.
>> My understanding is that DMA_CTRL_ACK means user wants an interrupt
>> based callback.
>
> Right user wants, you are not user though
>
Fair enough.
>>
>> If DMA_CTRL_ACK is not specified, then user wants to poll for the
>> results. The current code only supports the interrupt based delivery for
>> callbacks. Polling support is another to-do in my list for small sized
>> transfers.
>
> See the documentation for flags and usage. The point here is that user
> should set this. I know some old driver do this but that is not right
>
I'll remove it.
>>
>>>
>>>
>>>> +static int hidma_remove(struct platform_device *pdev)
>>>> +{
>>>> + struct hidma_dev *dmadev = platform_get_drvdata(pdev);
>>>> +
>>>> + pm_runtime_get_sync(dmadev->ddev.dev);
>>>> + dma_async_device_unregister(&dmadev->ddev);
>>>> + hidma_debug_uninit(dmadev);
>>>> + hidma_ll_uninit(dmadev->lldev);
>>>> + hidma_free(dmadev);
>>>> +
>>>> + dev_info(&pdev->dev, "HI-DMA engine removed\n");
>>>> + pm_runtime_put_sync_suspend(&pdev->dev);
>>>> + pm_runtime_disable(&pdev->dev);
>>>
>>> and your irq is still active and can invoke your irq handler!
>>>
>>
>> why? I shutdown the IRQ in hidma_ll_uninit and request the interrupt
>> with devm_request_irq. As soon as this driver is removed, the interrupt
>> should be released by the managed code.
>
> but there is a delay between that and free getting called. Since you don't
> free up explicitly you can still have irq and tasklets scheduled.
>
> Recommendation is that free up irq or ensure none are scheduled and
> disabled
>
I looked at other drivers. They seem to be disabling the IRQ right after
unregistering the DMA device from DMA engine. I'll follow the same by
adding an explicit dma_free_irq call into hidma_remove function.
--
Sinan Kaya
Qualcomm Technologies, Inc. on behalf of Qualcomm Innovation Center, Inc.
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a
Linux Foundation Collaborative Project
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
[toc] | [prev] | [next] | [standalone]
| From | Sinan Kaya <okaya@codeaurora.org> |
|---|---|
| Date | 2015-12-02 20:10 +0100 |
| Subject | Re: [PATCH V7 3/3] dma: add Qualcomm Technologies HIDMA channel driver |
| Message-ID | <qBkWS-2Tg-7@gated-at.bofh.it> |
| In reply to | #1281221 |
On 12/1/2015 4:16 PM, Sinan Kaya wrote:
>>>> >>>
>>>>> >>>> +static enum dma_status hidma_tx_status(struct dma_chan *dmach,
>>>>> >>>> + dma_cookie_t cookie, struct dma_tx_state *txstate)
>>>>> >>>> +{
>>>>> >>>> + struct hidma_chan *mchan = to_hidma_chan(dmach);
>>>>> >>>> + enum dma_status ret;
>>>>> >>>> +
>>>>> >>>> + if (mchan->paused)
>>>>> >>>> + ret = DMA_PAUSED;
>>>> >>>
>>>> >>> This is not quite right. The status query is for a descriptor and NOT for
>>>> >>> channel. Here if the descriptor queried was running then it would be paused
>>>> >>> for the rest pending txn, it would be DMA_IN_PROGRESS.
>>> >>
>>> >> The channel can be paused by the hypervisor. If it is paused, then no
>>> >> other transaction will go through including the pending requests. That's
>>> >> why, I'm checking the state first.
>> >
>> > You are missing the point. Channel can be paused, yes but the descriptor
>> > is in queue and is not paused. The descriptor running is paused, yes.
>> > There is subtle difference between these
> I'll follow your recommendation. PAUSE for the currently active
> descriptor and DMA_IN_PROGRESS for the rest.
>
I'm now confused.
I looked at several DMA driver implementations.
1. They call dma_cookie_status function to see if the job is done.
2. If done, they return right ahead.
3. Otherwise, dma_cookie_status returns DMA_IN_PROGRESS.
4. Next the code checks if the channel is paused and return value is
DMA_IN_PROGRESS. The code changes return code to DMA_PAUSED.
Whereas, I was returning paused first before even checking if the
descriptor is done. Are you OK with the sequence 1..4 above?
--
Sinan Kaya
Qualcomm Technologies, Inc. on behalf of Qualcomm Innovation Center, Inc.
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a
Linux Foundation Collaborative Project
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
[toc] | [prev] | [next] | [standalone]
| From | Vinod Koul <vinod.koul@intel.com> |
|---|---|
| Date | 2015-12-05 09:00 +0100 |
| Subject | Re: [PATCH V7 3/3] dma: add Qualcomm Technologies HIDMA channel driver |
| Message-ID | <qCfV8-6f3-9@gated-at.bofh.it> |
| In reply to | #1282272 |
On Wed, Dec 02, 2015 at 02:04:05PM -0500, Sinan Kaya wrote: > >> > You are missing the point. Channel can be paused, yes but the descriptor > >> > is in queue and is not paused. The descriptor running is paused, yes. > >> > There is subtle difference between these > > I'll follow your recommendation. PAUSE for the currently active > > descriptor and DMA_IN_PROGRESS for the rest. > > > > I'm now confused. > > I looked at several DMA driver implementations. > > 1. They call dma_cookie_status function to see if the job is done. > 2. If done, they return right ahead. > 3. Otherwise, dma_cookie_status returns DMA_IN_PROGRESS. > 4. Next the code checks if the channel is paused and return value is > DMA_IN_PROGRESS. The code changes return code to DMA_PAUSED. > > Whereas, I was returning paused first before even checking if the > descriptor is done. Are you OK with the sequence 1..4 above? Yes am okay with this with slight change in 4. You should set to PAUSED only for current descriptor and not for the ones in queue -- ~Vinod -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
[toc] | [prev] | [next] | [standalone]
| From | Sinan Kaya <okaya@codeaurora.org> |
|---|---|
| Date | 2015-12-08 15:40 +0100 |
| Subject | Re: [PATCH V7 3/3] dma: add Qualcomm Technologies HIDMA channel driver |
| Message-ID | <qDrAR-3yq-11@gated-at.bofh.it> |
| In reply to | #1284487 |
On 12/5/2015 3:00 AM, Vinod Koul wrote: > On Wed, Dec 02, 2015 at 02:04:05PM -0500, Sinan Kaya wrote: >>>>> You are missing the point. Channel can be paused, yes but the descriptor >>>>> is in queue and is not paused. The descriptor running is paused, yes. >>>>> There is subtle difference between these >>> I'll follow your recommendation. PAUSE for the currently active >>> descriptor and DMA_IN_PROGRESS for the rest. >>> >> >> I'm now confused. >> >> I looked at several DMA driver implementations. >> >> 1. They call dma_cookie_status function to see if the job is done. >> 2. If done, they return right ahead. >> 3. Otherwise, dma_cookie_status returns DMA_IN_PROGRESS. >> 4. Next the code checks if the channel is paused and return value is >> DMA_IN_PROGRESS. The code changes return code to DMA_PAUSED. >> >> Whereas, I was returning paused first before even checking if the >> descriptor is done. Are you OK with the sequence 1..4 above? > > Yes am okay with this with slight change in 4. > > You should set to PAUSED only for current descriptor and not for the ones in > queue > OK, I'll work on it. -- Sinan Kaya Qualcomm Technologies, Inc. on behalf of Qualcomm Innovation Center, Inc. Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a Linux Foundation Collaborative Project -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
[toc] | [prev] | [next] | [standalone]
| From | Sinan Kaya <okaya@codeaurora.org> |
|---|---|
| Date | 2015-12-10 21:20 +0100 |
| Subject | Re: [PATCH V7 3/3] dma: add Qualcomm Technologies HIDMA channel driver |
| Message-ID | <qEfR0-2uZ-7@gated-at.bofh.it> |
| In reply to | #1284487 |
On 12/5/2015 3:00 AM, Vinod Koul wrote: > On Wed, Dec 02, 2015 at 02:04:05PM -0500, Sinan Kaya wrote: >>>>> You are missing the point. Channel can be paused, yes but the descriptor >>>>> is in queue and is not paused. The descriptor running is paused, yes. >>>>> There is subtle difference between these >>> I'll follow your recommendation. PAUSE for the currently active >>> descriptor and DMA_IN_PROGRESS for the rest. >>> >> >> I'm now confused. >> >> I looked at several DMA driver implementations. >> >> 1. They call dma_cookie_status function to see if the job is done. >> 2. If done, they return right ahead. >> 3. Otherwise, dma_cookie_status returns DMA_IN_PROGRESS. >> 4. Next the code checks if the channel is paused and return value is >> DMA_IN_PROGRESS. The code changes return code to DMA_PAUSED. >> >> Whereas, I was returning paused first before even checking if the >> descriptor is done. Are you OK with the sequence 1..4 above? > > Yes am okay with this with slight change in 4. > > You should set to PAUSED only for current descriptor and not for the ones in > queue > OK. I'll post a new version with this. Is there any other comment that needed to be addressed? -- Sinan Kaya Qualcomm Technologies, Inc. on behalf of Qualcomm Innovation Center, Inc. Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a Linux Foundation Collaborative Project -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
[toc] | [prev] | [next] | [standalone]
| From | Vinod Koul <vinod.koul@intel.com> |
|---|---|
| Date | 2015-12-11 10:40 +0100 |
| Subject | Re: [PATCH V7 3/3] dma: add Qualcomm Technologies HIDMA channel driver |
| Message-ID | <qEslc-2l8-19@gated-at.bofh.it> |
| In reply to | #1288895 |
On Thu, Dec 10, 2015 at 03:10:48PM -0500, Sinan Kaya wrote: > On 12/5/2015 3:00 AM, Vinod Koul wrote: > > On Wed, Dec 02, 2015 at 02:04:05PM -0500, Sinan Kaya wrote: > >>>>> You are missing the point. Channel can be paused, yes but the descriptor > >>>>> is in queue and is not paused. The descriptor running is paused, yes. > >>>>> There is subtle difference between these > >>> I'll follow your recommendation. PAUSE for the currently active > >>> descriptor and DMA_IN_PROGRESS for the rest. > >>> > >> > >> I'm now confused. > >> > >> I looked at several DMA driver implementations. > >> > >> 1. They call dma_cookie_status function to see if the job is done. > >> 2. If done, they return right ahead. > >> 3. Otherwise, dma_cookie_status returns DMA_IN_PROGRESS. > >> 4. Next the code checks if the channel is paused and return value is > >> DMA_IN_PROGRESS. The code changes return code to DMA_PAUSED. > >> > >> Whereas, I was returning paused first before even checking if the > >> descriptor is done. Are you OK with the sequence 1..4 above? > > > > Yes am okay with this with slight change in 4. > > > > You should set to PAUSED only for current descriptor and not for the ones in > > queue > > > > OK. I'll post a new version with this. Is there any other comment that > needed to be addressed? Looks okay to me -- ~Vinod -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
[toc] | [prev] | [standalone]
Back to top | Article view | linux.kernel
csiph-web