Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > linux.kernel > #1330460 > unrolled thread
| Started by | Mathieu Poirier <mathieu.poirier@linaro.org> |
|---|---|
| First post | 2016-02-09 17:50 +0100 |
| Last post | 2016-02-19 21:50 +0100 |
| Articles | 6 — 2 participants |
Back to article view | Back to linux.kernel
This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by
below is the oldest one visible, not the original post.
Re: [PATCH v5 1/3] mailbox: Add support for APM X-Gene platform mailbox driver Mathieu Poirier <mathieu.poirier@linaro.org> - 2016-02-09 17:50 +0100
Re: [PATCH v5 1/3] mailbox: Add support for APM X-Gene platform mailbox driver Duc Dang <dhdang@apm.com> - 2016-02-10 04:50 +0100
Re: [PATCH v5 1/3] mailbox: Add support for APM X-Gene platform mailbox driver Mathieu Poirier <mathieu.poirier@linaro.org> - 2016-02-10 15:50 +0100
Re: [PATCH v5 1/3] mailbox: Add support for APM X-Gene platform mailbox driver Duc Dang <dhdang@apm.com> - 2016-02-13 04:30 +0100
[PATCH v6 0/3] mailbox: Add APM X-Gene platform mailbox driver Duc Dang <dhdang@apm.com> - 2016-02-13 04:50 +0100
Re: [PATCH v6 0/3] mailbox: Add APM X-Gene platform mailbox driver Duc Dang <dhdang@apm.com> - 2016-02-19 21:50 +0100
| From | Mathieu Poirier <mathieu.poirier@linaro.org> |
|---|---|
| Date | 2016-02-09 17:50 +0100 |
| Subject | Re: [PATCH v5 1/3] mailbox: Add support for APM X-Gene platform mailbox driver |
| Message-ID | <r0jEf-6IG-33@gated-at.bofh.it> |
On 8 February 2016 at 15:04, Duc Dang <dhdang@apm.com> wrote:
> X-Gene mailbox controller provides 8 mailbox channels, with
> each channel has a dedicated interrupt line.
>
> Signed-off-by: Feng Kan <fkan@apm.com>
> Signed-off-by: Duc Dang <dhdang@apm.com>
> ---
> Changes since v4:
> - Rebase over v4.5-rc1
> - Fix section mistmatch warning by removing
> __init in slimpro_mbox_probe declaration
> - Correctly print channel number when
> there is no IRQ for that channel
>
> Changes since v3:
> - Rebase over v4.4
> - Remove 'id' in slimpro_mbox_chan structure
> - Remove small functions that are only called once
> and fold them into the other callers
> - Remove void* pointer type cast
> - Relax the number of mailbox IRQs condition
> - Use subsys_initcall to guarantee mailbox driver
> will be registered before any other dependent driver
> is loaded.
>
> drivers/mailbox/Kconfig | 9 ++
> drivers/mailbox/Makefile | 2 +
> drivers/mailbox/mailbox-xgene-slimpro.c | 264 ++++++++++++++++++++++++++++++++
> 3 files changed, 275 insertions(+)
> create mode 100644 drivers/mailbox/mailbox-xgene-slimpro.c
>
> diff --git a/drivers/mailbox/Kconfig b/drivers/mailbox/Kconfig
> index 546d05f..678e434 100644
> --- a/drivers/mailbox/Kconfig
> +++ b/drivers/mailbox/Kconfig
> @@ -85,4 +85,13 @@ config MAILBOX_TEST
> Test client to help with testing new Controller driver
> implementations.
>
> +config XGENE_SLIMPRO_MBOX
> + tristate "APM SoC X-Gene SLIMpro Mailbox Controller"
> + depends on ARCH_XGENE
> + help
> + An implementation of the APM X-Gene Interprocessor Communication
> + Mailbox (IPCM) between the ARM 64-bit cores and SLIMpro controller.
> + It is used to send short messages between ARM64-bit cores and
> + the SLIMpro Management Engine, primarily for PM. Say Y here if you
> + want to use the APM X-Gene SLIMpro IPCM support.
> endif
> diff --git a/drivers/mailbox/Makefile b/drivers/mailbox/Makefile
> index 92435ef..b602ef8 100644
> --- a/drivers/mailbox/Makefile
> +++ b/drivers/mailbox/Makefile
> @@ -17,3 +17,5 @@ obj-$(CONFIG_ALTERA_MBOX) += mailbox-altera.o
> obj-$(CONFIG_BCM2835_MBOX) += bcm2835-mailbox.o
>
> obj-$(CONFIG_STI_MBOX) += mailbox-sti.o
> +
> +obj-$(CONFIG_XGENE_SLIMPRO_MBOX) += mailbox-xgene-slimpro.o
> diff --git a/drivers/mailbox/mailbox-xgene-slimpro.c b/drivers/mailbox/mailbox-xgene-slimpro.c
> new file mode 100644
> index 0000000..0ea1eb8
> --- /dev/null
> +++ b/drivers/mailbox/mailbox-xgene-slimpro.c
> @@ -0,0 +1,264 @@
> +/*
> + * APM X-Gene SLIMpro MailBox Driver
> + *
> + * Copyright (c) 2015, Applied Micro Circuits Corporation
> + * Author: Feng Kan fkan@apm.com
> + *
> + * This program is free software; you can redistribute it and/or
> + * modify it under the terms of the GNU General Public License as
> + * published by the Free Software Foundation; either version 2 of
> + * the License, or (at your option) any later version.
> + *
> + * 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.
> + *
> + * You should have received a copy of the GNU General Public License
> + * along with this program; if not, see <http://www.gnu.org/licenses/>.
> + *
> + */
> +#include <linux/acpi.h>
> +#include <linux/delay.h>
> +#include <linux/interrupt.h>
> +#include <linux/io.h>
> +#include <linux/mailbox_controller.h>
> +#include <linux/module.h>
> +#include <linux/of.h>
> +#include <linux/platform_device.h>
> +#include <linux/spinlock.h>
> +
> +#define MBOX_CON_NAME "slimpro-mbox"
> +#define MBOX_REG_SET_OFFSET 0x1000
> +#define MBOX_CNT 8
> +#define MBOX_STATUS_AVAIL_MASK BIT(16)
> +#define MBOX_STATUS_ACK_MASK BIT(0)
> +
> +/* Configuration and Status Registers */
> +#define REG_DB_IN 0x00
> +#define REG_DB_DIN0 0x04
> +#define REG_DB_DIN1 0x08
> +#define REG_DB_OUT 0x10
> +#define REG_DB_DOUT0 0x14
> +#define REG_DB_DOUT1 0x18
> +#define REG_DB_STAT 0x20
> +#define REG_DB_STATMASK 0x24
> +
> +struct slimpro_mbox_chan {
> + struct device *dev;
> + struct mbox_chan *chan;
> + void __iomem *reg;
> + int irq;
> + u32 rx_msg[3];
> +};
> +
> +struct slimpro_mbox {
> + struct mbox_controller mb_ctrl;
> + struct slimpro_mbox_chan mc[MBOX_CNT];
> + struct mbox_chan chans[MBOX_CNT];
> +};
The above two structure declaration could use kernel style documentation.
> +
> +static void mb_chan_send_msg(struct slimpro_mbox_chan *mb_chan, u32 *msg)
> +{
> + writel(msg[1], mb_chan->reg + REG_DB_DOUT0);
> + writel(msg[2], mb_chan->reg + REG_DB_DOUT1);
> + writel(msg[0], mb_chan->reg + REG_DB_OUT);
> +}
> +
> +static void mb_chan_recv_msg(struct slimpro_mbox_chan *mb_chan)
> +{
> + mb_chan->rx_msg[1] = readl(mb_chan->reg + REG_DB_DIN0);
> + mb_chan->rx_msg[2] = readl(mb_chan->reg + REG_DB_DIN1);
> + mb_chan->rx_msg[0] = readl(mb_chan->reg + REG_DB_IN);
> +}
Is there a requirement that operations in the above two functions be
ordered? If not significant gains could be achieved by using
writel/readl_relaxed(). Same comment for the rest of this driver.
Thanks,
Mathieu
> +
> +static int mb_chan_status_ack(struct slimpro_mbox_chan *mb_chan)
> +{
> + u32 val = readl(mb_chan->reg + REG_DB_STAT);
> +
> + if (val & MBOX_STATUS_ACK_MASK) {
> + writel(MBOX_STATUS_ACK_MASK, mb_chan->reg + REG_DB_STAT);
> + return 1;
> + }
> + return 0;
> +}
> +
> +static int mb_chan_status_avail(struct slimpro_mbox_chan *mb_chan)
> +{
> + u32 val = readl(mb_chan->reg + REG_DB_STAT);
> +
> + if (val & MBOX_STATUS_AVAIL_MASK) {
> + mb_chan_recv_msg(mb_chan);
> + writel(MBOX_STATUS_AVAIL_MASK, mb_chan->reg + REG_DB_STAT);
> + return 1;
> + }
> + return 0;
> +}
> +
> +static irqreturn_t slimpro_mbox_irq(int irq, void *id)
> +{
> + struct slimpro_mbox_chan *mb_chan = id;
> +
> + if (mb_chan_status_ack(mb_chan))
> + mbox_chan_txdone(mb_chan->chan, 0);
> +
> + if (mb_chan_status_avail(mb_chan))
> + mbox_chan_received_data(mb_chan->chan, mb_chan->rx_msg);
> +
> + return IRQ_HANDLED;
> +}
> +
> +static int slimpro_mbox_send_data(struct mbox_chan *chan, void *msg)
> +{
> + struct slimpro_mbox_chan *mb_chan = chan->con_priv;
> +
> + mb_chan_send_msg(mb_chan, msg);
> + return 0;
> +}
> +
> +static int slimpro_mbox_startup(struct mbox_chan *chan)
> +{
> + struct slimpro_mbox_chan *mb_chan = chan->con_priv;
> + int rc;
> + u32 val;
> +
> + rc = devm_request_irq(mb_chan->dev, mb_chan->irq, slimpro_mbox_irq, 0,
> + MBOX_CON_NAME, mb_chan);
> + if (unlikely(rc)) {
> + dev_err(mb_chan->dev, "failed to register mailbox interrupt %d\n",
> + mb_chan->irq);
> + return rc;
> + }
> +
> + /* Enable HW interrupt */
> + writel(MBOX_STATUS_ACK_MASK | MBOX_STATUS_AVAIL_MASK,
> + mb_chan->reg + REG_DB_STAT);
> + /* Unmask doorbell status interrupt */
> + val = readl(mb_chan->reg + REG_DB_STATMASK);
> + val &= ~(MBOX_STATUS_ACK_MASK | MBOX_STATUS_AVAIL_MASK);
> + writel(val, mb_chan->reg + REG_DB_STATMASK);
> +
> + return 0;
> +}
> +
> +static void slimpro_mbox_shutdown(struct mbox_chan *chan)
> +{
> + struct slimpro_mbox_chan *mb_chan = chan->con_priv;
> + u32 val;
> +
> + /* Mask doorbell status interrupt */
> + val = readl(mb_chan->reg + REG_DB_STATMASK);
> + val |= (MBOX_STATUS_ACK_MASK | MBOX_STATUS_AVAIL_MASK);
> + writel(val, mb_chan->reg + REG_DB_STATMASK);
> +
> + devm_free_irq(mb_chan->dev, mb_chan->irq, mb_chan);
> +}
> +
> +static struct mbox_chan_ops slimpro_mbox_ops = {
> + .send_data = slimpro_mbox_send_data,
> + .startup = slimpro_mbox_startup,
> + .shutdown = slimpro_mbox_shutdown,
> +};
> +
> +static int slimpro_mbox_probe(struct platform_device *pdev)
> +{
> + struct slimpro_mbox *ctx;
> + struct resource *regs;
> + void __iomem *mb_base;
> + int rc;
> + int i;
> +
> + ctx = devm_kzalloc(&pdev->dev, sizeof(struct slimpro_mbox), GFP_KERNEL);
> + if (IS_ERR(ctx))
> + return PTR_ERR(ctx);
> +
> + platform_set_drvdata(pdev, ctx);
> +
> + regs = platform_get_resource(pdev, IORESOURCE_MEM, 0);
> + mb_base = devm_ioremap(&pdev->dev, regs->start, resource_size(regs));
> + if (IS_ERR(mb_base))
> + return PTR_ERR(mb_base);
> +
> + /* Setup mailbox links */
> + for (i = 0; i < MBOX_CNT; i++) {
> + ctx->mc[i].irq = platform_get_irq(pdev, i);
> + if (ctx->mc[i].irq < 0) {
> + if (i == 0) {
> + dev_err(&pdev->dev, "no available IRQ\n");
> + return -EINVAL;
> + }
> + dev_info(&pdev->dev, "no IRQ for channel %d\n", i);
> + break;
> + }
> +
> + ctx->mc[i].dev = &pdev->dev;
> + ctx->mc[i].reg = mb_base + i * MBOX_REG_SET_OFFSET;
> + ctx->mc[i].chan = &ctx->chans[i];
> + ctx->chans[i].con_priv = &ctx->mc[i];
> + }
> +
> + /* Setup mailbox controller */
> + ctx->mb_ctrl.dev = &pdev->dev;
> + ctx->mb_ctrl.chans = ctx->chans;
> + ctx->mb_ctrl.txdone_irq = true;
> + ctx->mb_ctrl.ops = &slimpro_mbox_ops;
> + ctx->mb_ctrl.num_chans = i;
> +
> + rc = mbox_controller_register(&ctx->mb_ctrl);
> + if (rc) {
> + dev_err(&pdev->dev,
> + "APM X-Gene SLIMpro MailBox register failed:%d\n", rc);
> + return rc;
> + }
> +
> + dev_info(&pdev->dev, "APM X-Gene SLIMpro MailBox registered\n");
> + return 0;
> +}
> +
> +static int slimpro_mbox_remove(struct platform_device *pdev)
> +{
> + struct slimpro_mbox *smb = platform_get_drvdata(pdev);
> +
> + mbox_controller_unregister(&smb->mb_ctrl);
> + return 0;
> +}
> +
> +static const struct of_device_id slimpro_of_match[] = {
> + {.compatible = "apm,xgene-slimpro-mbox" },
> + { },
> +};
> +MODULE_DEVICE_TABLE(of, slimpro_of_match);
> +
> +#ifdef CONFIG_ACPI
> +static const struct acpi_device_id slimpro_acpi_ids[] = {
> + {"APMC0D01", 0},
> + {}
> +};
> +MODULE_DEVICE_TABLE(acpi, slimpro_acpi_ids);
> +#endif
> +
> +static struct platform_driver slimpro_mbox_driver = {
> + .probe = slimpro_mbox_probe,
> + .remove = slimpro_mbox_remove,
> + .driver = {
> + .name = "xgene-slimpro-mbox",
> + .of_match_table = of_match_ptr(slimpro_of_match),
> + .acpi_match_table = ACPI_PTR(slimpro_acpi_ids)
> + },
> +};
> +
> +static int __init slimpro_mbox_init(void)
> +{
> + return platform_driver_register(&slimpro_mbox_driver);
> +}
> +
> +static void __exit slimpro_mbox_exit(void)
> +{
> + platform_driver_unregister(&slimpro_mbox_driver);
> +}
> +
> +subsys_initcall(slimpro_mbox_init);
> +module_exit(slimpro_mbox_exit);
> +
> +MODULE_DESCRIPTION("APM X-Gene SLIMpro Mailbox Driver");
> +MODULE_LICENSE("GPL");
> --
> 1.9.1
>
[toc] | [next] | [standalone]
| From | Duc Dang <dhdang@apm.com> |
|---|---|
| Date | 2016-02-10 04:50 +0100 |
| Message-ID | <r0tWW-5d2-3@gated-at.bofh.it> |
| In reply to | #1330460 |
On Tue, Feb 9, 2016 at 8:40 AM, Mathieu Poirier
<mathieu.poirier@linaro.org> wrote:
> On 8 February 2016 at 15:04, Duc Dang <dhdang@apm.com> wrote:
>> X-Gene mailbox controller provides 8 mailbox channels, with
>> each channel has a dedicated interrupt line.
>>
>> Signed-off-by: Feng Kan <fkan@apm.com>
>> Signed-off-by: Duc Dang <dhdang@apm.com>
>> ---
>> Changes since v4:
>> - Rebase over v4.5-rc1
>> - Fix section mistmatch warning by removing
>> __init in slimpro_mbox_probe declaration
>> - Correctly print channel number when
>> there is no IRQ for that channel
>>
>> Changes since v3:
>> - Rebase over v4.4
>> - Remove 'id' in slimpro_mbox_chan structure
>> - Remove small functions that are only called once
>> and fold them into the other callers
>> - Remove void* pointer type cast
>> - Relax the number of mailbox IRQs condition
>> - Use subsys_initcall to guarantee mailbox driver
>> will be registered before any other dependent driver
>> is loaded.
>>
>> drivers/mailbox/Kconfig | 9 ++
>> drivers/mailbox/Makefile | 2 +
>> drivers/mailbox/mailbox-xgene-slimpro.c | 264 ++++++++++++++++++++++++++++++++
>> 3 files changed, 275 insertions(+)
>> create mode 100644 drivers/mailbox/mailbox-xgene-slimpro.c
>>
>> diff --git a/drivers/mailbox/Kconfig b/drivers/mailbox/Kconfig
>> index 546d05f..678e434 100644
>> --- a/drivers/mailbox/Kconfig
>> +++ b/drivers/mailbox/Kconfig
>> @@ -85,4 +85,13 @@ config MAILBOX_TEST
>> Test client to help with testing new Controller driver
>> implementations.
>>
>> +config XGENE_SLIMPRO_MBOX
>> + tristate "APM SoC X-Gene SLIMpro Mailbox Controller"
>> + depends on ARCH_XGENE
>> + help
>> + An implementation of the APM X-Gene Interprocessor Communication
>> + Mailbox (IPCM) between the ARM 64-bit cores and SLIMpro controller.
>> + It is used to send short messages between ARM64-bit cores and
>> + the SLIMpro Management Engine, primarily for PM. Say Y here if you
>> + want to use the APM X-Gene SLIMpro IPCM support.
>> endif
>> diff --git a/drivers/mailbox/Makefile b/drivers/mailbox/Makefile
>> index 92435ef..b602ef8 100644
>> --- a/drivers/mailbox/Makefile
>> +++ b/drivers/mailbox/Makefile
>> @@ -17,3 +17,5 @@ obj-$(CONFIG_ALTERA_MBOX) += mailbox-altera.o
>> obj-$(CONFIG_BCM2835_MBOX) += bcm2835-mailbox.o
>>
>> obj-$(CONFIG_STI_MBOX) += mailbox-sti.o
>> +
>> +obj-$(CONFIG_XGENE_SLIMPRO_MBOX) += mailbox-xgene-slimpro.o
>> diff --git a/drivers/mailbox/mailbox-xgene-slimpro.c b/drivers/mailbox/mailbox-xgene-slimpro.c
>> new file mode 100644
>> index 0000000..0ea1eb8
>> --- /dev/null
>> +++ b/drivers/mailbox/mailbox-xgene-slimpro.c
>> @@ -0,0 +1,264 @@
>> +/*
>> + * APM X-Gene SLIMpro MailBox Driver
>> + *
>> + * Copyright (c) 2015, Applied Micro Circuits Corporation
>> + * Author: Feng Kan fkan@apm.com
>> + *
>> + * This program is free software; you can redistribute it and/or
>> + * modify it under the terms of the GNU General Public License as
>> + * published by the Free Software Foundation; either version 2 of
>> + * the License, or (at your option) any later version.
>> + *
>> + * 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.
>> + *
>> + * You should have received a copy of the GNU General Public License
>> + * along with this program; if not, see <http://www.gnu.org/licenses/>.
>> + *
>> + */
>> +#include <linux/acpi.h>
>> +#include <linux/delay.h>
>> +#include <linux/interrupt.h>
>> +#include <linux/io.h>
>> +#include <linux/mailbox_controller.h>
>> +#include <linux/module.h>
>> +#include <linux/of.h>
>> +#include <linux/platform_device.h>
>> +#include <linux/spinlock.h>
>> +
>> +#define MBOX_CON_NAME "slimpro-mbox"
>> +#define MBOX_REG_SET_OFFSET 0x1000
>> +#define MBOX_CNT 8
>> +#define MBOX_STATUS_AVAIL_MASK BIT(16)
>> +#define MBOX_STATUS_ACK_MASK BIT(0)
>> +
>> +/* Configuration and Status Registers */
>> +#define REG_DB_IN 0x00
>> +#define REG_DB_DIN0 0x04
>> +#define REG_DB_DIN1 0x08
>> +#define REG_DB_OUT 0x10
>> +#define REG_DB_DOUT0 0x14
>> +#define REG_DB_DOUT1 0x18
>> +#define REG_DB_STAT 0x20
>> +#define REG_DB_STATMASK 0x24
>> +
>> +struct slimpro_mbox_chan {
>> + struct device *dev;
>> + struct mbox_chan *chan;
>> + void __iomem *reg;
>> + int irq;
>> + u32 rx_msg[3];
>> +};
>> +
>> +struct slimpro_mbox {
>> + struct mbox_controller mb_ctrl;
>> + struct slimpro_mbox_chan mc[MBOX_CNT];
>> + struct mbox_chan chans[MBOX_CNT];
>> +};
>
> The above two structure declaration could use kernel style documentation.
Can you please clarify or give an example? I looked into several files
in drivers/mailbox/ and drivers/irqchip but could not find what I need
to tweak.
>
>> +
>> +static void mb_chan_send_msg(struct slimpro_mbox_chan *mb_chan, u32 *msg)
>> +{
>> + writel(msg[1], mb_chan->reg + REG_DB_DOUT0);
>> + writel(msg[2], mb_chan->reg + REG_DB_DOUT1);
>> + writel(msg[0], mb_chan->reg + REG_DB_OUT);
>> +}
>> +
>> +static void mb_chan_recv_msg(struct slimpro_mbox_chan *mb_chan)
>> +{
>> + mb_chan->rx_msg[1] = readl(mb_chan->reg + REG_DB_DIN0);
>> + mb_chan->rx_msg[2] = readl(mb_chan->reg + REG_DB_DIN1);
>> + mb_chan->rx_msg[0] = readl(mb_chan->reg + REG_DB_IN);
>> +}
>
> Is there a requirement that operations in the above two functions be
> ordered? If not significant gains could be achieved by using
> writel/readl_relaxed(). Same comment for the rest of this driver.
The 'writes' need to be in order as writing to REG_DB_OUT will trigger
interrupt, so I want the other 2 'writes' complete before it.
After the receiver get the message, it will clear door bell available
interrupt request, this will also trigger a door bell acknowledge
interrupt. So the 'read' to get the message and the 'write' to clear
the door bell available interrupt must be in order as well.
Regards,
Duc Dang.
>
> Thanks,
> Mathieu
>
>> +
>> +static int mb_chan_status_ack(struct slimpro_mbox_chan *mb_chan)
>> +{
>> + u32 val = readl(mb_chan->reg + REG_DB_STAT);
>> +
>> + if (val & MBOX_STATUS_ACK_MASK) {
>> + writel(MBOX_STATUS_ACK_MASK, mb_chan->reg + REG_DB_STAT);
>> + return 1;
>> + }
>> + return 0;
>> +}
>> +
>> +static int mb_chan_status_avail(struct slimpro_mbox_chan *mb_chan)
>> +{
>> + u32 val = readl(mb_chan->reg + REG_DB_STAT);
>> +
>> + if (val & MBOX_STATUS_AVAIL_MASK) {
>> + mb_chan_recv_msg(mb_chan);
>> + writel(MBOX_STATUS_AVAIL_MASK, mb_chan->reg + REG_DB_STAT);
>> + return 1;
>> + }
>> + return 0;
>> +}
>> +
>> +static irqreturn_t slimpro_mbox_irq(int irq, void *id)
>> +{
>> + struct slimpro_mbox_chan *mb_chan = id;
>> +
>> + if (mb_chan_status_ack(mb_chan))
>> + mbox_chan_txdone(mb_chan->chan, 0);
>> +
>> + if (mb_chan_status_avail(mb_chan))
>> + mbox_chan_received_data(mb_chan->chan, mb_chan->rx_msg);
>> +
>> + return IRQ_HANDLED;
>> +}
>> +
>> +static int slimpro_mbox_send_data(struct mbox_chan *chan, void *msg)
>> +{
>> + struct slimpro_mbox_chan *mb_chan = chan->con_priv;
>> +
>> + mb_chan_send_msg(mb_chan, msg);
>> + return 0;
>> +}
>> +
>> +static int slimpro_mbox_startup(struct mbox_chan *chan)
>> +{
>> + struct slimpro_mbox_chan *mb_chan = chan->con_priv;
>> + int rc;
>> + u32 val;
>> +
>> + rc = devm_request_irq(mb_chan->dev, mb_chan->irq, slimpro_mbox_irq, 0,
>> + MBOX_CON_NAME, mb_chan);
>> + if (unlikely(rc)) {
>> + dev_err(mb_chan->dev, "failed to register mailbox interrupt %d\n",
>> + mb_chan->irq);
>> + return rc;
>> + }
>> +
>> + /* Enable HW interrupt */
>> + writel(MBOX_STATUS_ACK_MASK | MBOX_STATUS_AVAIL_MASK,
>> + mb_chan->reg + REG_DB_STAT);
>> + /* Unmask doorbell status interrupt */
>> + val = readl(mb_chan->reg + REG_DB_STATMASK);
>> + val &= ~(MBOX_STATUS_ACK_MASK | MBOX_STATUS_AVAIL_MASK);
>> + writel(val, mb_chan->reg + REG_DB_STATMASK);
>> +
>> + return 0;
>> +}
>> +
>> +static void slimpro_mbox_shutdown(struct mbox_chan *chan)
>> +{
>> + struct slimpro_mbox_chan *mb_chan = chan->con_priv;
>> + u32 val;
>> +
>> + /* Mask doorbell status interrupt */
>> + val = readl(mb_chan->reg + REG_DB_STATMASK);
>> + val |= (MBOX_STATUS_ACK_MASK | MBOX_STATUS_AVAIL_MASK);
>> + writel(val, mb_chan->reg + REG_DB_STATMASK);
>> +
>> + devm_free_irq(mb_chan->dev, mb_chan->irq, mb_chan);
>> +}
>> +
>> +static struct mbox_chan_ops slimpro_mbox_ops = {
>> + .send_data = slimpro_mbox_send_data,
>> + .startup = slimpro_mbox_startup,
>> + .shutdown = slimpro_mbox_shutdown,
>> +};
>> +
>> +static int slimpro_mbox_probe(struct platform_device *pdev)
>> +{
>> + struct slimpro_mbox *ctx;
>> + struct resource *regs;
>> + void __iomem *mb_base;
>> + int rc;
>> + int i;
>> +
>> + ctx = devm_kzalloc(&pdev->dev, sizeof(struct slimpro_mbox), GFP_KERNEL);
>> + if (IS_ERR(ctx))
>> + return PTR_ERR(ctx);
>> +
>> + platform_set_drvdata(pdev, ctx);
>> +
>> + regs = platform_get_resource(pdev, IORESOURCE_MEM, 0);
>> + mb_base = devm_ioremap(&pdev->dev, regs->start, resource_size(regs));
>> + if (IS_ERR(mb_base))
>> + return PTR_ERR(mb_base);
>> +
>> + /* Setup mailbox links */
>> + for (i = 0; i < MBOX_CNT; i++) {
>> + ctx->mc[i].irq = platform_get_irq(pdev, i);
>> + if (ctx->mc[i].irq < 0) {
>> + if (i == 0) {
>> + dev_err(&pdev->dev, "no available IRQ\n");
>> + return -EINVAL;
>> + }
>> + dev_info(&pdev->dev, "no IRQ for channel %d\n", i);
>> + break;
>> + }
>> +
>> + ctx->mc[i].dev = &pdev->dev;
>> + ctx->mc[i].reg = mb_base + i * MBOX_REG_SET_OFFSET;
>> + ctx->mc[i].chan = &ctx->chans[i];
>> + ctx->chans[i].con_priv = &ctx->mc[i];
>> + }
>> +
>> + /* Setup mailbox controller */
>> + ctx->mb_ctrl.dev = &pdev->dev;
>> + ctx->mb_ctrl.chans = ctx->chans;
>> + ctx->mb_ctrl.txdone_irq = true;
>> + ctx->mb_ctrl.ops = &slimpro_mbox_ops;
>> + ctx->mb_ctrl.num_chans = i;
>> +
>> + rc = mbox_controller_register(&ctx->mb_ctrl);
>> + if (rc) {
>> + dev_err(&pdev->dev,
>> + "APM X-Gene SLIMpro MailBox register failed:%d\n", rc);
>> + return rc;
>> + }
>> +
>> + dev_info(&pdev->dev, "APM X-Gene SLIMpro MailBox registered\n");
>> + return 0;
>> +}
>> +
>> +static int slimpro_mbox_remove(struct platform_device *pdev)
>> +{
>> + struct slimpro_mbox *smb = platform_get_drvdata(pdev);
>> +
>> + mbox_controller_unregister(&smb->mb_ctrl);
>> + return 0;
>> +}
>> +
>> +static const struct of_device_id slimpro_of_match[] = {
>> + {.compatible = "apm,xgene-slimpro-mbox" },
>> + { },
>> +};
>> +MODULE_DEVICE_TABLE(of, slimpro_of_match);
>> +
>> +#ifdef CONFIG_ACPI
>> +static const struct acpi_device_id slimpro_acpi_ids[] = {
>> + {"APMC0D01", 0},
>> + {}
>> +};
>> +MODULE_DEVICE_TABLE(acpi, slimpro_acpi_ids);
>> +#endif
>> +
>> +static struct platform_driver slimpro_mbox_driver = {
>> + .probe = slimpro_mbox_probe,
>> + .remove = slimpro_mbox_remove,
>> + .driver = {
>> + .name = "xgene-slimpro-mbox",
>> + .of_match_table = of_match_ptr(slimpro_of_match),
>> + .acpi_match_table = ACPI_PTR(slimpro_acpi_ids)
>> + },
>> +};
>> +
>> +static int __init slimpro_mbox_init(void)
>> +{
>> + return platform_driver_register(&slimpro_mbox_driver);
>> +}
>> +
>> +static void __exit slimpro_mbox_exit(void)
>> +{
>> + platform_driver_unregister(&slimpro_mbox_driver);
>> +}
>> +
>> +subsys_initcall(slimpro_mbox_init);
>> +module_exit(slimpro_mbox_exit);
>> +
>> +MODULE_DESCRIPTION("APM X-Gene SLIMpro Mailbox Driver");
>> +MODULE_LICENSE("GPL");
>> --
>> 1.9.1
>>
[toc] | [prev] | [next] | [standalone]
| From | Mathieu Poirier <mathieu.poirier@linaro.org> |
|---|---|
| Date | 2016-02-10 15:50 +0100 |
| Message-ID | <r0EfE-3qm-23@gated-at.bofh.it> |
| In reply to | #1330880 |
On 9 February 2016 at 20:46, Duc Dang <dhdang@apm.com> wrote:
> On Tue, Feb 9, 2016 at 8:40 AM, Mathieu Poirier
> <mathieu.poirier@linaro.org> wrote:
>> On 8 February 2016 at 15:04, Duc Dang <dhdang@apm.com> wrote:
>>> X-Gene mailbox controller provides 8 mailbox channels, with
>>> each channel has a dedicated interrupt line.
>>>
>>> Signed-off-by: Feng Kan <fkan@apm.com>
>>> Signed-off-by: Duc Dang <dhdang@apm.com>
>>> ---
>>> Changes since v4:
>>> - Rebase over v4.5-rc1
>>> - Fix section mistmatch warning by removing
>>> __init in slimpro_mbox_probe declaration
>>> - Correctly print channel number when
>>> there is no IRQ for that channel
>>>
>>> Changes since v3:
>>> - Rebase over v4.4
>>> - Remove 'id' in slimpro_mbox_chan structure
>>> - Remove small functions that are only called once
>>> and fold them into the other callers
>>> - Remove void* pointer type cast
>>> - Relax the number of mailbox IRQs condition
>>> - Use subsys_initcall to guarantee mailbox driver
>>> will be registered before any other dependent driver
>>> is loaded.
>>>
>>> drivers/mailbox/Kconfig | 9 ++
>>> drivers/mailbox/Makefile | 2 +
>>> drivers/mailbox/mailbox-xgene-slimpro.c | 264 ++++++++++++++++++++++++++++++++
>>> 3 files changed, 275 insertions(+)
>>> create mode 100644 drivers/mailbox/mailbox-xgene-slimpro.c
>>>
>>> diff --git a/drivers/mailbox/Kconfig b/drivers/mailbox/Kconfig
>>> index 546d05f..678e434 100644
>>> --- a/drivers/mailbox/Kconfig
>>> +++ b/drivers/mailbox/Kconfig
>>> @@ -85,4 +85,13 @@ config MAILBOX_TEST
>>> Test client to help with testing new Controller driver
>>> implementations.
>>>
>>> +config XGENE_SLIMPRO_MBOX
>>> + tristate "APM SoC X-Gene SLIMpro Mailbox Controller"
>>> + depends on ARCH_XGENE
>>> + help
>>> + An implementation of the APM X-Gene Interprocessor Communication
>>> + Mailbox (IPCM) between the ARM 64-bit cores and SLIMpro controller.
>>> + It is used to send short messages between ARM64-bit cores and
>>> + the SLIMpro Management Engine, primarily for PM. Say Y here if you
>>> + want to use the APM X-Gene SLIMpro IPCM support.
>>> endif
>>> diff --git a/drivers/mailbox/Makefile b/drivers/mailbox/Makefile
>>> index 92435ef..b602ef8 100644
>>> --- a/drivers/mailbox/Makefile
>>> +++ b/drivers/mailbox/Makefile
>>> @@ -17,3 +17,5 @@ obj-$(CONFIG_ALTERA_MBOX) += mailbox-altera.o
>>> obj-$(CONFIG_BCM2835_MBOX) += bcm2835-mailbox.o
>>>
>>> obj-$(CONFIG_STI_MBOX) += mailbox-sti.o
>>> +
>>> +obj-$(CONFIG_XGENE_SLIMPRO_MBOX) += mailbox-xgene-slimpro.o
>>> diff --git a/drivers/mailbox/mailbox-xgene-slimpro.c b/drivers/mailbox/mailbox-xgene-slimpro.c
>>> new file mode 100644
>>> index 0000000..0ea1eb8
>>> --- /dev/null
>>> +++ b/drivers/mailbox/mailbox-xgene-slimpro.c
>>> @@ -0,0 +1,264 @@
>>> +/*
>>> + * APM X-Gene SLIMpro MailBox Driver
>>> + *
>>> + * Copyright (c) 2015, Applied Micro Circuits Corporation
>>> + * Author: Feng Kan fkan@apm.com
>>> + *
>>> + * This program is free software; you can redistribute it and/or
>>> + * modify it under the terms of the GNU General Public License as
>>> + * published by the Free Software Foundation; either version 2 of
>>> + * the License, or (at your option) any later version.
>>> + *
>>> + * 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.
>>> + *
>>> + * You should have received a copy of the GNU General Public License
>>> + * along with this program; if not, see <http://www.gnu.org/licenses/>.
>>> + *
>>> + */
>>> +#include <linux/acpi.h>
>>> +#include <linux/delay.h>
>>> +#include <linux/interrupt.h>
>>> +#include <linux/io.h>
>>> +#include <linux/mailbox_controller.h>
>>> +#include <linux/module.h>
>>> +#include <linux/of.h>
>>> +#include <linux/platform_device.h>
>>> +#include <linux/spinlock.h>
>>> +
>>> +#define MBOX_CON_NAME "slimpro-mbox"
>>> +#define MBOX_REG_SET_OFFSET 0x1000
>>> +#define MBOX_CNT 8
>>> +#define MBOX_STATUS_AVAIL_MASK BIT(16)
>>> +#define MBOX_STATUS_ACK_MASK BIT(0)
>>> +
>>> +/* Configuration and Status Registers */
>>> +#define REG_DB_IN 0x00
>>> +#define REG_DB_DIN0 0x04
>>> +#define REG_DB_DIN1 0x08
>>> +#define REG_DB_OUT 0x10
>>> +#define REG_DB_DOUT0 0x14
>>> +#define REG_DB_DOUT1 0x18
>>> +#define REG_DB_STAT 0x20
>>> +#define REG_DB_STATMASK 0x24
>>> +
>>> +struct slimpro_mbox_chan {
>>> + struct device *dev;
>>> + struct mbox_chan *chan;
>>> + void __iomem *reg;
>>> + int irq;
>>> + u32 rx_msg[3];
>>> +};
>>> +
>>> +struct slimpro_mbox {
>>> + struct mbox_controller mb_ctrl;
>>> + struct slimpro_mbox_chan mc[MBOX_CNT];
>>> + struct mbox_chan chans[MBOX_CNT];
>>> +};
>>
>> The above two structure declaration could use kernel style documentation.
>
> Can you please clarify or give an example? I looked into several files
> in drivers/mailbox/ and drivers/irqchip but could not find what I need
> to tweak.
http://lxr.free-electrons.com/source/drivers/mailbox/mailbox-sti.c#L42
>
>>
>>> +
>>> +static void mb_chan_send_msg(struct slimpro_mbox_chan *mb_chan, u32 *msg)
>>> +{
>>> + writel(msg[1], mb_chan->reg + REG_DB_DOUT0);
>>> + writel(msg[2], mb_chan->reg + REG_DB_DOUT1);
>>> + writel(msg[0], mb_chan->reg + REG_DB_OUT);
>>> +}
>>> +
>>> +static void mb_chan_recv_msg(struct slimpro_mbox_chan *mb_chan)
>>> +{
>>> + mb_chan->rx_msg[1] = readl(mb_chan->reg + REG_DB_DIN0);
>>> + mb_chan->rx_msg[2] = readl(mb_chan->reg + REG_DB_DIN1);
>>> + mb_chan->rx_msg[0] = readl(mb_chan->reg + REG_DB_IN);
>>> +}
>>
>> Is there a requirement that operations in the above two functions be
>> ordered? If not significant gains could be achieved by using
>> writel/readl_relaxed(). Same comment for the rest of this driver.
>
> The 'writes' need to be in order as writing to REG_DB_OUT will trigger
> interrupt, so I want the other 2 'writes' complete before it.
>
> After the receiver get the message, it will clear door bell available
> interrupt request, this will also trigger a door bell acknowledge
> interrupt. So the 'read' to get the message and the 'write' to clear
> the door bell available interrupt must be in order as well.
>
> Regards,
> Duc Dang.
>>
>> Thanks,
>> Mathieu
>>
>>> +
>>> +static int mb_chan_status_ack(struct slimpro_mbox_chan *mb_chan)
>>> +{
>>> + u32 val = readl(mb_chan->reg + REG_DB_STAT);
>>> +
>>> + if (val & MBOX_STATUS_ACK_MASK) {
>>> + writel(MBOX_STATUS_ACK_MASK, mb_chan->reg + REG_DB_STAT);
>>> + return 1;
>>> + }
>>> + return 0;
>>> +}
>>> +
>>> +static int mb_chan_status_avail(struct slimpro_mbox_chan *mb_chan)
>>> +{
>>> + u32 val = readl(mb_chan->reg + REG_DB_STAT);
>>> +
>>> + if (val & MBOX_STATUS_AVAIL_MASK) {
>>> + mb_chan_recv_msg(mb_chan);
>>> + writel(MBOX_STATUS_AVAIL_MASK, mb_chan->reg + REG_DB_STAT);
>>> + return 1;
>>> + }
>>> + return 0;
>>> +}
>>> +
>>> +static irqreturn_t slimpro_mbox_irq(int irq, void *id)
>>> +{
>>> + struct slimpro_mbox_chan *mb_chan = id;
>>> +
>>> + if (mb_chan_status_ack(mb_chan))
>>> + mbox_chan_txdone(mb_chan->chan, 0);
>>> +
>>> + if (mb_chan_status_avail(mb_chan))
>>> + mbox_chan_received_data(mb_chan->chan, mb_chan->rx_msg);
>>> +
>>> + return IRQ_HANDLED;
>>> +}
>>> +
>>> +static int slimpro_mbox_send_data(struct mbox_chan *chan, void *msg)
>>> +{
>>> + struct slimpro_mbox_chan *mb_chan = chan->con_priv;
>>> +
>>> + mb_chan_send_msg(mb_chan, msg);
>>> + return 0;
>>> +}
>>> +
>>> +static int slimpro_mbox_startup(struct mbox_chan *chan)
>>> +{
>>> + struct slimpro_mbox_chan *mb_chan = chan->con_priv;
>>> + int rc;
>>> + u32 val;
>>> +
>>> + rc = devm_request_irq(mb_chan->dev, mb_chan->irq, slimpro_mbox_irq, 0,
>>> + MBOX_CON_NAME, mb_chan);
>>> + if (unlikely(rc)) {
>>> + dev_err(mb_chan->dev, "failed to register mailbox interrupt %d\n",
>>> + mb_chan->irq);
>>> + return rc;
>>> + }
>>> +
>>> + /* Enable HW interrupt */
>>> + writel(MBOX_STATUS_ACK_MASK | MBOX_STATUS_AVAIL_MASK,
>>> + mb_chan->reg + REG_DB_STAT);
>>> + /* Unmask doorbell status interrupt */
>>> + val = readl(mb_chan->reg + REG_DB_STATMASK);
>>> + val &= ~(MBOX_STATUS_ACK_MASK | MBOX_STATUS_AVAIL_MASK);
>>> + writel(val, mb_chan->reg + REG_DB_STATMASK);
>>> +
>>> + return 0;
>>> +}
>>> +
>>> +static void slimpro_mbox_shutdown(struct mbox_chan *chan)
>>> +{
>>> + struct slimpro_mbox_chan *mb_chan = chan->con_priv;
>>> + u32 val;
>>> +
>>> + /* Mask doorbell status interrupt */
>>> + val = readl(mb_chan->reg + REG_DB_STATMASK);
>>> + val |= (MBOX_STATUS_ACK_MASK | MBOX_STATUS_AVAIL_MASK);
>>> + writel(val, mb_chan->reg + REG_DB_STATMASK);
>>> +
>>> + devm_free_irq(mb_chan->dev, mb_chan->irq, mb_chan);
>>> +}
>>> +
>>> +static struct mbox_chan_ops slimpro_mbox_ops = {
>>> + .send_data = slimpro_mbox_send_data,
>>> + .startup = slimpro_mbox_startup,
>>> + .shutdown = slimpro_mbox_shutdown,
>>> +};
>>> +
>>> +static int slimpro_mbox_probe(struct platform_device *pdev)
>>> +{
>>> + struct slimpro_mbox *ctx;
>>> + struct resource *regs;
>>> + void __iomem *mb_base;
>>> + int rc;
>>> + int i;
>>> +
>>> + ctx = devm_kzalloc(&pdev->dev, sizeof(struct slimpro_mbox), GFP_KERNEL);
>>> + if (IS_ERR(ctx))
>>> + return PTR_ERR(ctx);
>>> +
>>> + platform_set_drvdata(pdev, ctx);
>>> +
>>> + regs = platform_get_resource(pdev, IORESOURCE_MEM, 0);
>>> + mb_base = devm_ioremap(&pdev->dev, regs->start, resource_size(regs));
>>> + if (IS_ERR(mb_base))
>>> + return PTR_ERR(mb_base);
>>> +
>>> + /* Setup mailbox links */
>>> + for (i = 0; i < MBOX_CNT; i++) {
>>> + ctx->mc[i].irq = platform_get_irq(pdev, i);
>>> + if (ctx->mc[i].irq < 0) {
>>> + if (i == 0) {
>>> + dev_err(&pdev->dev, "no available IRQ\n");
>>> + return -EINVAL;
>>> + }
>>> + dev_info(&pdev->dev, "no IRQ for channel %d\n", i);
>>> + break;
>>> + }
>>> +
>>> + ctx->mc[i].dev = &pdev->dev;
>>> + ctx->mc[i].reg = mb_base + i * MBOX_REG_SET_OFFSET;
>>> + ctx->mc[i].chan = &ctx->chans[i];
>>> + ctx->chans[i].con_priv = &ctx->mc[i];
>>> + }
>>> +
>>> + /* Setup mailbox controller */
>>> + ctx->mb_ctrl.dev = &pdev->dev;
>>> + ctx->mb_ctrl.chans = ctx->chans;
>>> + ctx->mb_ctrl.txdone_irq = true;
>>> + ctx->mb_ctrl.ops = &slimpro_mbox_ops;
>>> + ctx->mb_ctrl.num_chans = i;
>>> +
>>> + rc = mbox_controller_register(&ctx->mb_ctrl);
>>> + if (rc) {
>>> + dev_err(&pdev->dev,
>>> + "APM X-Gene SLIMpro MailBox register failed:%d\n", rc);
>>> + return rc;
>>> + }
>>> +
>>> + dev_info(&pdev->dev, "APM X-Gene SLIMpro MailBox registered\n");
>>> + return 0;
>>> +}
>>> +
>>> +static int slimpro_mbox_remove(struct platform_device *pdev)
>>> +{
>>> + struct slimpro_mbox *smb = platform_get_drvdata(pdev);
>>> +
>>> + mbox_controller_unregister(&smb->mb_ctrl);
>>> + return 0;
>>> +}
>>> +
>>> +static const struct of_device_id slimpro_of_match[] = {
>>> + {.compatible = "apm,xgene-slimpro-mbox" },
>>> + { },
>>> +};
>>> +MODULE_DEVICE_TABLE(of, slimpro_of_match);
>>> +
>>> +#ifdef CONFIG_ACPI
>>> +static const struct acpi_device_id slimpro_acpi_ids[] = {
>>> + {"APMC0D01", 0},
>>> + {}
>>> +};
>>> +MODULE_DEVICE_TABLE(acpi, slimpro_acpi_ids);
>>> +#endif
>>> +
>>> +static struct platform_driver slimpro_mbox_driver = {
>>> + .probe = slimpro_mbox_probe,
>>> + .remove = slimpro_mbox_remove,
>>> + .driver = {
>>> + .name = "xgene-slimpro-mbox",
>>> + .of_match_table = of_match_ptr(slimpro_of_match),
>>> + .acpi_match_table = ACPI_PTR(slimpro_acpi_ids)
>>> + },
>>> +};
>>> +
>>> +static int __init slimpro_mbox_init(void)
>>> +{
>>> + return platform_driver_register(&slimpro_mbox_driver);
>>> +}
>>> +
>>> +static void __exit slimpro_mbox_exit(void)
>>> +{
>>> + platform_driver_unregister(&slimpro_mbox_driver);
>>> +}
>>> +
>>> +subsys_initcall(slimpro_mbox_init);
>>> +module_exit(slimpro_mbox_exit);
>>> +
>>> +MODULE_DESCRIPTION("APM X-Gene SLIMpro Mailbox Driver");
>>> +MODULE_LICENSE("GPL");
>>> --
>>> 1.9.1
>>>
[toc] | [prev] | [next] | [standalone]
| From | Duc Dang <dhdang@apm.com> |
|---|---|
| Date | 2016-02-13 04:30 +0100 |
| Message-ID | <r1z4d-7Jv-1@gated-at.bofh.it> |
| In reply to | #1331232 |
On Wed, Feb 10, 2016 at 6:41 AM, Mathieu Poirier
<mathieu.poirier@linaro.org> wrote:
> On 9 February 2016 at 20:46, Duc Dang <dhdang@apm.com> wrote:
>> On Tue, Feb 9, 2016 at 8:40 AM, Mathieu Poirier
>> <mathieu.poirier@linaro.org> wrote:
>>> On 8 February 2016 at 15:04, Duc Dang <dhdang@apm.com> wrote:
>>>> X-Gene mailbox controller provides 8 mailbox channels, with
>>>> each channel has a dedicated interrupt line.
>>>>
>>>> Signed-off-by: Feng Kan <fkan@apm.com>
>>>> Signed-off-by: Duc Dang <dhdang@apm.com>
>>>> ---
>>>> Changes since v4:
>>>> - Rebase over v4.5-rc1
>>>> - Fix section mistmatch warning by removing
>>>> __init in slimpro_mbox_probe declaration
>>>> - Correctly print channel number when
>>>> there is no IRQ for that channel
>>>>
>>>> Changes since v3:
>>>> - Rebase over v4.4
>>>> - Remove 'id' in slimpro_mbox_chan structure
>>>> - Remove small functions that are only called once
>>>> and fold them into the other callers
>>>> - Remove void* pointer type cast
>>>> - Relax the number of mailbox IRQs condition
>>>> - Use subsys_initcall to guarantee mailbox driver
>>>> will be registered before any other dependent driver
>>>> is loaded.
>>>>
>>>> drivers/mailbox/Kconfig | 9 ++
>>>> drivers/mailbox/Makefile | 2 +
>>>> drivers/mailbox/mailbox-xgene-slimpro.c | 264 ++++++++++++++++++++++++++++++++
>>>> 3 files changed, 275 insertions(+)
>>>> create mode 100644 drivers/mailbox/mailbox-xgene-slimpro.c
>>>>
>>>> diff --git a/drivers/mailbox/Kconfig b/drivers/mailbox/Kconfig
>>>> index 546d05f..678e434 100644
>>>> --- a/drivers/mailbox/Kconfig
>>>> +++ b/drivers/mailbox/Kconfig
>>>> @@ -85,4 +85,13 @@ config MAILBOX_TEST
>>>> Test client to help with testing new Controller driver
>>>> implementations.
>>>>
>>>> +config XGENE_SLIMPRO_MBOX
>>>> + tristate "APM SoC X-Gene SLIMpro Mailbox Controller"
>>>> + depends on ARCH_XGENE
>>>> + help
>>>> + An implementation of the APM X-Gene Interprocessor Communication
>>>> + Mailbox (IPCM) between the ARM 64-bit cores and SLIMpro controller.
>>>> + It is used to send short messages between ARM64-bit cores and
>>>> + the SLIMpro Management Engine, primarily for PM. Say Y here if you
>>>> + want to use the APM X-Gene SLIMpro IPCM support.
>>>> endif
>>>> diff --git a/drivers/mailbox/Makefile b/drivers/mailbox/Makefile
>>>> index 92435ef..b602ef8 100644
>>>> --- a/drivers/mailbox/Makefile
>>>> +++ b/drivers/mailbox/Makefile
>>>> @@ -17,3 +17,5 @@ obj-$(CONFIG_ALTERA_MBOX) += mailbox-altera.o
>>>> obj-$(CONFIG_BCM2835_MBOX) += bcm2835-mailbox.o
>>>>
>>>> obj-$(CONFIG_STI_MBOX) += mailbox-sti.o
>>>> +
>>>> +obj-$(CONFIG_XGENE_SLIMPRO_MBOX) += mailbox-xgene-slimpro.o
>>>> diff --git a/drivers/mailbox/mailbox-xgene-slimpro.c b/drivers/mailbox/mailbox-xgene-slimpro.c
>>>> new file mode 100644
>>>> index 0000000..0ea1eb8
>>>> --- /dev/null
>>>> +++ b/drivers/mailbox/mailbox-xgene-slimpro.c
>>>> @@ -0,0 +1,264 @@
>>>> +/*
>>>> + * APM X-Gene SLIMpro MailBox Driver
>>>> + *
>>>> + * Copyright (c) 2015, Applied Micro Circuits Corporation
>>>> + * Author: Feng Kan fkan@apm.com
>>>> + *
>>>> + * This program is free software; you can redistribute it and/or
>>>> + * modify it under the terms of the GNU General Public License as
>>>> + * published by the Free Software Foundation; either version 2 of
>>>> + * the License, or (at your option) any later version.
>>>> + *
>>>> + * 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.
>>>> + *
>>>> + * You should have received a copy of the GNU General Public License
>>>> + * along with this program; if not, see <http://www.gnu.org/licenses/>.
>>>> + *
>>>> + */
>>>> +#include <linux/acpi.h>
>>>> +#include <linux/delay.h>
>>>> +#include <linux/interrupt.h>
>>>> +#include <linux/io.h>
>>>> +#include <linux/mailbox_controller.h>
>>>> +#include <linux/module.h>
>>>> +#include <linux/of.h>
>>>> +#include <linux/platform_device.h>
>>>> +#include <linux/spinlock.h>
>>>> +
>>>> +#define MBOX_CON_NAME "slimpro-mbox"
>>>> +#define MBOX_REG_SET_OFFSET 0x1000
>>>> +#define MBOX_CNT 8
>>>> +#define MBOX_STATUS_AVAIL_MASK BIT(16)
>>>> +#define MBOX_STATUS_ACK_MASK BIT(0)
>>>> +
>>>> +/* Configuration and Status Registers */
>>>> +#define REG_DB_IN 0x00
>>>> +#define REG_DB_DIN0 0x04
>>>> +#define REG_DB_DIN1 0x08
>>>> +#define REG_DB_OUT 0x10
>>>> +#define REG_DB_DOUT0 0x14
>>>> +#define REG_DB_DOUT1 0x18
>>>> +#define REG_DB_STAT 0x20
>>>> +#define REG_DB_STATMASK 0x24
>>>> +
>>>> +struct slimpro_mbox_chan {
>>>> + struct device *dev;
>>>> + struct mbox_chan *chan;
>>>> + void __iomem *reg;
>>>> + int irq;
>>>> + u32 rx_msg[3];
>>>> +};
>>>> +
>>>> +struct slimpro_mbox {
>>>> + struct mbox_controller mb_ctrl;
>>>> + struct slimpro_mbox_chan mc[MBOX_CNT];
>>>> + struct mbox_chan chans[MBOX_CNT];
>>>> +};
>>>
>>> The above two structure declaration could use kernel style documentation.
>>
>> Can you please clarify or give an example? I looked into several files
>> in drivers/mailbox/ and drivers/irqchip but could not find what I need
>> to tweak.
>
> http://lxr.free-electrons.com/source/drivers/mailbox/mailbox-sti.c#L42
I am modifying the structure defintion to add description and will
post a new version shortly.
Regards,
Duc Dang.
>
>>
>>>
>>>> +
>>>> +static void mb_chan_send_msg(struct slimpro_mbox_chan *mb_chan, u32 *msg)
>>>> +{
>>>> + writel(msg[1], mb_chan->reg + REG_DB_DOUT0);
>>>> + writel(msg[2], mb_chan->reg + REG_DB_DOUT1);
>>>> + writel(msg[0], mb_chan->reg + REG_DB_OUT);
>>>> +}
>>>> +
>>>> +static void mb_chan_recv_msg(struct slimpro_mbox_chan *mb_chan)
>>>> +{
>>>> + mb_chan->rx_msg[1] = readl(mb_chan->reg + REG_DB_DIN0);
>>>> + mb_chan->rx_msg[2] = readl(mb_chan->reg + REG_DB_DIN1);
>>>> + mb_chan->rx_msg[0] = readl(mb_chan->reg + REG_DB_IN);
>>>> +}
>>>
>>> Is there a requirement that operations in the above two functions be
>>> ordered? If not significant gains could be achieved by using
>>> writel/readl_relaxed(). Same comment for the rest of this driver.
>>
>> The 'writes' need to be in order as writing to REG_DB_OUT will trigger
>> interrupt, so I want the other 2 'writes' complete before it.
>>
>> After the receiver get the message, it will clear door bell available
>> interrupt request, this will also trigger a door bell acknowledge
>> interrupt. So the 'read' to get the message and the 'write' to clear
>> the door bell available interrupt must be in order as well.
>>
>> Regards,
>> Duc Dang.
>>>
>>> Thanks,
>>> Mathieu
>>>
>>>> +
>>>> +static int mb_chan_status_ack(struct slimpro_mbox_chan *mb_chan)
>>>> +{
>>>> + u32 val = readl(mb_chan->reg + REG_DB_STAT);
>>>> +
>>>> + if (val & MBOX_STATUS_ACK_MASK) {
>>>> + writel(MBOX_STATUS_ACK_MASK, mb_chan->reg + REG_DB_STAT);
>>>> + return 1;
>>>> + }
>>>> + return 0;
>>>> +}
>>>> +
>>>> +static int mb_chan_status_avail(struct slimpro_mbox_chan *mb_chan)
>>>> +{
>>>> + u32 val = readl(mb_chan->reg + REG_DB_STAT);
>>>> +
>>>> + if (val & MBOX_STATUS_AVAIL_MASK) {
>>>> + mb_chan_recv_msg(mb_chan);
>>>> + writel(MBOX_STATUS_AVAIL_MASK, mb_chan->reg + REG_DB_STAT);
>>>> + return 1;
>>>> + }
>>>> + return 0;
>>>> +}
>>>> +
>>>> +static irqreturn_t slimpro_mbox_irq(int irq, void *id)
>>>> +{
>>>> + struct slimpro_mbox_chan *mb_chan = id;
>>>> +
>>>> + if (mb_chan_status_ack(mb_chan))
>>>> + mbox_chan_txdone(mb_chan->chan, 0);
>>>> +
>>>> + if (mb_chan_status_avail(mb_chan))
>>>> + mbox_chan_received_data(mb_chan->chan, mb_chan->rx_msg);
>>>> +
>>>> + return IRQ_HANDLED;
>>>> +}
>>>> +
>>>> +static int slimpro_mbox_send_data(struct mbox_chan *chan, void *msg)
>>>> +{
>>>> + struct slimpro_mbox_chan *mb_chan = chan->con_priv;
>>>> +
>>>> + mb_chan_send_msg(mb_chan, msg);
>>>> + return 0;
>>>> +}
>>>> +
>>>> +static int slimpro_mbox_startup(struct mbox_chan *chan)
>>>> +{
>>>> + struct slimpro_mbox_chan *mb_chan = chan->con_priv;
>>>> + int rc;
>>>> + u32 val;
>>>> +
>>>> + rc = devm_request_irq(mb_chan->dev, mb_chan->irq, slimpro_mbox_irq, 0,
>>>> + MBOX_CON_NAME, mb_chan);
>>>> + if (unlikely(rc)) {
>>>> + dev_err(mb_chan->dev, "failed to register mailbox interrupt %d\n",
>>>> + mb_chan->irq);
>>>> + return rc;
>>>> + }
>>>> +
>>>> + /* Enable HW interrupt */
>>>> + writel(MBOX_STATUS_ACK_MASK | MBOX_STATUS_AVAIL_MASK,
>>>> + mb_chan->reg + REG_DB_STAT);
>>>> + /* Unmask doorbell status interrupt */
>>>> + val = readl(mb_chan->reg + REG_DB_STATMASK);
>>>> + val &= ~(MBOX_STATUS_ACK_MASK | MBOX_STATUS_AVAIL_MASK);
>>>> + writel(val, mb_chan->reg + REG_DB_STATMASK);
>>>> +
>>>> + return 0;
>>>> +}
>>>> +
>>>> +static void slimpro_mbox_shutdown(struct mbox_chan *chan)
>>>> +{
>>>> + struct slimpro_mbox_chan *mb_chan = chan->con_priv;
>>>> + u32 val;
>>>> +
>>>> + /* Mask doorbell status interrupt */
>>>> + val = readl(mb_chan->reg + REG_DB_STATMASK);
>>>> + val |= (MBOX_STATUS_ACK_MASK | MBOX_STATUS_AVAIL_MASK);
>>>> + writel(val, mb_chan->reg + REG_DB_STATMASK);
>>>> +
>>>> + devm_free_irq(mb_chan->dev, mb_chan->irq, mb_chan);
>>>> +}
>>>> +
>>>> +static struct mbox_chan_ops slimpro_mbox_ops = {
>>>> + .send_data = slimpro_mbox_send_data,
>>>> + .startup = slimpro_mbox_startup,
>>>> + .shutdown = slimpro_mbox_shutdown,
>>>> +};
>>>> +
>>>> +static int slimpro_mbox_probe(struct platform_device *pdev)
>>>> +{
>>>> + struct slimpro_mbox *ctx;
>>>> + struct resource *regs;
>>>> + void __iomem *mb_base;
>>>> + int rc;
>>>> + int i;
>>>> +
>>>> + ctx = devm_kzalloc(&pdev->dev, sizeof(struct slimpro_mbox), GFP_KERNEL);
>>>> + if (IS_ERR(ctx))
>>>> + return PTR_ERR(ctx);
>>>> +
>>>> + platform_set_drvdata(pdev, ctx);
>>>> +
>>>> + regs = platform_get_resource(pdev, IORESOURCE_MEM, 0);
>>>> + mb_base = devm_ioremap(&pdev->dev, regs->start, resource_size(regs));
>>>> + if (IS_ERR(mb_base))
>>>> + return PTR_ERR(mb_base);
>>>> +
>>>> + /* Setup mailbox links */
>>>> + for (i = 0; i < MBOX_CNT; i++) {
>>>> + ctx->mc[i].irq = platform_get_irq(pdev, i);
>>>> + if (ctx->mc[i].irq < 0) {
>>>> + if (i == 0) {
>>>> + dev_err(&pdev->dev, "no available IRQ\n");
>>>> + return -EINVAL;
>>>> + }
>>>> + dev_info(&pdev->dev, "no IRQ for channel %d\n", i);
>>>> + break;
>>>> + }
>>>> +
>>>> + ctx->mc[i].dev = &pdev->dev;
>>>> + ctx->mc[i].reg = mb_base + i * MBOX_REG_SET_OFFSET;
>>>> + ctx->mc[i].chan = &ctx->chans[i];
>>>> + ctx->chans[i].con_priv = &ctx->mc[i];
>>>> + }
>>>> +
>>>> + /* Setup mailbox controller */
>>>> + ctx->mb_ctrl.dev = &pdev->dev;
>>>> + ctx->mb_ctrl.chans = ctx->chans;
>>>> + ctx->mb_ctrl.txdone_irq = true;
>>>> + ctx->mb_ctrl.ops = &slimpro_mbox_ops;
>>>> + ctx->mb_ctrl.num_chans = i;
>>>> +
>>>> + rc = mbox_controller_register(&ctx->mb_ctrl);
>>>> + if (rc) {
>>>> + dev_err(&pdev->dev,
>>>> + "APM X-Gene SLIMpro MailBox register failed:%d\n", rc);
>>>> + return rc;
>>>> + }
>>>> +
>>>> + dev_info(&pdev->dev, "APM X-Gene SLIMpro MailBox registered\n");
>>>> + return 0;
>>>> +}
>>>> +
>>>> +static int slimpro_mbox_remove(struct platform_device *pdev)
>>>> +{
>>>> + struct slimpro_mbox *smb = platform_get_drvdata(pdev);
>>>> +
>>>> + mbox_controller_unregister(&smb->mb_ctrl);
>>>> + return 0;
>>>> +}
>>>> +
>>>> +static const struct of_device_id slimpro_of_match[] = {
>>>> + {.compatible = "apm,xgene-slimpro-mbox" },
>>>> + { },
>>>> +};
>>>> +MODULE_DEVICE_TABLE(of, slimpro_of_match);
>>>> +
>>>> +#ifdef CONFIG_ACPI
>>>> +static const struct acpi_device_id slimpro_acpi_ids[] = {
>>>> + {"APMC0D01", 0},
>>>> + {}
>>>> +};
>>>> +MODULE_DEVICE_TABLE(acpi, slimpro_acpi_ids);
>>>> +#endif
>>>> +
>>>> +static struct platform_driver slimpro_mbox_driver = {
>>>> + .probe = slimpro_mbox_probe,
>>>> + .remove = slimpro_mbox_remove,
>>>> + .driver = {
>>>> + .name = "xgene-slimpro-mbox",
>>>> + .of_match_table = of_match_ptr(slimpro_of_match),
>>>> + .acpi_match_table = ACPI_PTR(slimpro_acpi_ids)
>>>> + },
>>>> +};
>>>> +
>>>> +static int __init slimpro_mbox_init(void)
>>>> +{
>>>> + return platform_driver_register(&slimpro_mbox_driver);
>>>> +}
>>>> +
>>>> +static void __exit slimpro_mbox_exit(void)
>>>> +{
>>>> + platform_driver_unregister(&slimpro_mbox_driver);
>>>> +}
>>>> +
>>>> +subsys_initcall(slimpro_mbox_init);
>>>> +module_exit(slimpro_mbox_exit);
>>>> +
>>>> +MODULE_DESCRIPTION("APM X-Gene SLIMpro Mailbox Driver");
>>>> +MODULE_LICENSE("GPL");
>>>> --
>>>> 1.9.1
>>>>
[toc] | [prev] | [next] | [standalone]
| From | Duc Dang <dhdang@apm.com> |
|---|---|
| Date | 2016-02-13 04:50 +0100 |
| Subject | [PATCH v6 0/3] mailbox: Add APM X-Gene platform mailbox driver |
| Message-ID | <r1znz-7PE-3@gated-at.bofh.it> |
| In reply to | #1331232 |
APM X-Gene SoC has a mailbox controller that provides
communication mechanism for X-Gene Arm64 cores to communicate
with X-Gene SoC's Cortex M3 (SLIMpro) processor.
X-Gene mailbox controller provides 8 mailbox channels, with
each channel has a dedicated interrupt line.
Changes since v5:
- Add more description into SlimPRO
mailbox data structure
Changes since v4:
- Rebase over v4.5-rc1
- Fix section mismatch warning during compiling
- Correctly print channel number when there is
no IRQ for that channel
- Change node name to mailbox@10540000
- Correct the number of IRQs in documentation
Changes since v3:
- Rebase over v4.4
- Remove 'id' in slimpro_mbox_chan structure
- Remove functions that are only called once
and fold them into the other callers
- Remove void* pointer type cast
- Relax the number of mailbox IRQs condition
- Fix error and address comment in documentation
(xgene-slimpro-mailbox.txt)
Changes since v2:
- Rebase Feng's patch set over v4.3-rc5
- Remove uneccessary 'inline' in function definition
- Use module_platform_driver instead of subsys_initcall
- Minor coding stype clean up
Changes since v1:
- Add ACPI support
- Use defines for reg offset
Duc Dang (3):
mailbox: Add support for APM X-Gene platform mailbox driver
Documentation: mailbox: Add APM X-Gene SLIMpro mailbox dts
documentation
arm64: dts: mailbox device tree node for APM X-Gene platform.
.../bindings/mailbox/xgene-slimpro-mailbox.txt | 35 +++
arch/arm64/boot/dts/apm/apm-storm.dtsi | 14 +
drivers/mailbox/Kconfig | 9 +
drivers/mailbox/Makefile | 2 +
drivers/mailbox/mailbox-xgene-slimpro.c | 284 +++++++++++++++++++++
5 files changed, 344 insertions(+)
create mode 100644 Documentation/devicetree/bindings/mailbox/xgene-slimpro-mailbox.txt
create mode 100644 drivers/mailbox/mailbox-xgene-slimpro.c
--
1.9.1
[toc] | [prev] | [next] | [standalone]
| From | Duc Dang <dhdang@apm.com> |
|---|---|
| Date | 2016-02-19 21:50 +0100 |
| Subject | Re: [PATCH v6 0/3] mailbox: Add APM X-Gene platform mailbox driver |
| Message-ID | <r409X-2uF-3@gated-at.bofh.it> |
| In reply to | #1333273 |
On Fri, Feb 12, 2016 at 7:39 PM, Duc Dang <dhdang@apm.com> wrote: > APM X-Gene SoC has a mailbox controller that provides > communication mechanism for X-Gene Arm64 cores to communicate > with X-Gene SoC's Cortex M3 (SLIMpro) processor. > > X-Gene mailbox controller provides 8 mailbox channels, with > each channel has a dedicated interrupt line. Hi Jassi, Are you fine with this version of the patch? Regards, Duc Dang. > > Changes since v5: > - Add more description into SlimPRO > mailbox data structure > > Changes since v4: > - Rebase over v4.5-rc1 > - Fix section mismatch warning during compiling > - Correctly print channel number when there is > no IRQ for that channel > - Change node name to mailbox@10540000 > - Correct the number of IRQs in documentation > > Changes since v3: > - Rebase over v4.4 > - Remove 'id' in slimpro_mbox_chan structure > - Remove functions that are only called once > and fold them into the other callers > - Remove void* pointer type cast > - Relax the number of mailbox IRQs condition > - Fix error and address comment in documentation > (xgene-slimpro-mailbox.txt) > > Changes since v2: > - Rebase Feng's patch set over v4.3-rc5 > - Remove uneccessary 'inline' in function definition > - Use module_platform_driver instead of subsys_initcall > - Minor coding stype clean up > > Changes since v1: > - Add ACPI support > - Use defines for reg offset > > Duc Dang (3): > mailbox: Add support for APM X-Gene platform mailbox driver > Documentation: mailbox: Add APM X-Gene SLIMpro mailbox dts > documentation > arm64: dts: mailbox device tree node for APM X-Gene platform. > > .../bindings/mailbox/xgene-slimpro-mailbox.txt | 35 +++ > arch/arm64/boot/dts/apm/apm-storm.dtsi | 14 + > drivers/mailbox/Kconfig | 9 + > drivers/mailbox/Makefile | 2 + > drivers/mailbox/mailbox-xgene-slimpro.c | 284 +++++++++++++++++++++ > 5 files changed, 344 insertions(+) > create mode 100644 Documentation/devicetree/bindings/mailbox/xgene-slimpro-mailbox.txt > create mode 100644 drivers/mailbox/mailbox-xgene-slimpro.c > > -- > 1.9.1 >
[toc] | [prev] | [standalone]
Back to top | Article view | linux.kernel
csiph-web