Skip to contentSkip to conversation

Add Aspeed SSIF BMC driver

PATCH Superseded Quan Nguyen 8 participants 35 messages May 19, 2021 – Jun 25, 2021

Subsystems (6)

v3 of 10 Superseded Latest is v10 →

lore
Share & tools
lore
#0 lore raw

v3 cover Add Aspeed SSIF BMC driver

This series add support for the Aspeed specific SSIF BMC driver which is to perform in-band IPMI communication with the host in management (BMC) side.

v3:
+ Switched binding doc to use DT schema format [Rob]
+ Splited into generic ssif_bmc and aspeed-specific [Corey, Joel]
+ Removed redundant license info [Joel]
+ Switched to use traditional if-else [Joel]
+ Removed unused ssif_bmc_ioctl() [Joel]
+ Made handle_request()/complete_response() to return void [Joel]
+ Refactored send_ssif_bmc_response()/receive_ssif_bmc_request() [Corey]
+ Remove mutex [Corey]
+ Use spin_lock/unlock_irqsave/restore in callback [Corey]
+ Removed the unnecessary memset [Corey]
+ Switch to use dev_err() [Corey]
+ Combine mask/unmask two interrupts together [Corey]
+ Fixed unhandled Tx done with NAK [Quan]
+ Late ack'ed Tx done w/wo Ack irq [Quan]
+ Use aspeed-specific exported aspeed_set_slave_busy() when slave busy
to fix the deadlock [Graeme, Philipp, Quan]
+ Clean buffer for last multipart read [Quan]
+ Handle unknown incoming command [Quan]

v2:
+ Fixed compiling error with COMPILE_TEST for arc

Quan Nguyen (7):
i2c: i2c-core-smbus: Expose PEC calculate function for generic use
ipmi: ssif_bmc: Add SSIF BMC driver
i2c: aspeed: Fix unhandled Tx done with NAK
i2c: aspeed: Acknowledge Tx done w/wo ACK irq late
i2c: aspeed: Add aspeed_set_slave_busy()
ipmi: ssif_bmc: Add Aspeed SSIF BMC driver
bindings: ipmi: Add binding for Aspeed SSIF BMC driver

 .../bindings/ipmi/aspeed-ssif-bmc.yaml        |  33 +
 drivers/char/ipmi/Kconfig                     |  22 +
 drivers/char/ipmi/Makefile                    |   2 +
 drivers/char/ipmi/ssif_bmc.c                  | 605 ++++++++++++++++++
 drivers/char/ipmi/ssif_bmc.h                  |  93 +++
 drivers/char/ipmi/ssif_bmc_aspeed.c           |  75 +++
 drivers/i2c/busses/i2c-aspeed.c               |  51 +-
 drivers/i2c/i2c-core-smbus.c                  |  12 +-
 include/linux/i2c.h                           |   1 +
 9 files changed, 884 insertions(+), 10 deletions(-)
 create mode 100644 Documentation/devicetree/bindings/ipmi/aspeed-ssif-bmc.yaml
 create mode 100644 drivers/char/ipmi/ssif_bmc.c
 create mode 100644 drivers/char/ipmi/ssif_bmc.h
 create mode 100644 drivers/char/ipmi/ssif_bmc_aspeed.c
Signature · 1 line
2.28.0
#1 lore raw

PATCH v3 1/7 i2c: i2c-core-smbus: Expose PEC calculate function for generic use

Expose the PEC calculation i2c_smbus_pec() for generic use.

---
 drivers/i2c/i2c-core-smbus.c | 12 ++++++++++--
 include/linux/i2c.h          |  1 +
 2 files changed, 11 insertions(+), 2 deletions(-)
drivers/i2c/i2c-core-smbus.c +10 -2
index d2d32c0fd8c3..e5b2d1465e7e 100644
@@ -37,8 +37,15 @@ static u8 crc8(u16 data)
 	return (u8)(data >> 8);
 }
 
-/* Incremental CRC8 over count bytes in the array pointed to by p */
-static u8 i2c_smbus_pec(u8 crc, u8 *p, size_t count)
+/**
+ * i2c_smbus_pec - Incremental CRC8 over the given input data array
+ * @crc: previous return crc8 value
+ * @p: pointer to data buffer.
+ * @count: number of bytes in data buffer.
+ *
+ * Incremental CRC8 over count bytes in the array pointed to by p
+ */
+u8 i2c_smbus_pec(u8 crc, u8 *p, size_t count)
 {
 	int i;
 
@@ -46,6 +53,7 @@ static u8 i2c_smbus_pec(u8 crc, u8 *p, size_t count)
 		crc = crc8((crc ^ p[i]) << 8);
 	return crc;
 }
+EXPORT_SYMBOL(i2c_smbus_pec);
 
 /* Assume a 7-bit address, which is reasonable for SMBus */
 static u8 i2c_smbus_msg_pec(u8 pec, struct i2c_msg *msg)
include/linux/i2c.h +1 -0
index e8f2ac8c9c3d..2494a007dfda 100644
@@ -147,6 +147,7 @@ s32 __i2c_smbus_xfer(struct i2c_adapter *adapter, u16 addr,
 /* Now follow the 'nice' access routines. These also document the calling
    conventions of i2c_smbus_xfer. */
 
+u8 i2c_smbus_pec(u8 crc, u8 *p, size_t count);
 s32 i2c_smbus_read_byte(const struct i2c_client *client);
 s32 i2c_smbus_write_byte(const struct i2c_client *client, u8 value);
 s32 i2c_smbus_read_byte_data(const struct i2c_client *client, u8 command);
Signature · 1 line
2.28.0
#2 lore raw

Re: [PATCH v3 1/7] i2c: i2c-core-smbus: Expose PEC calculate function for generic use

Quoted 3 lines · Quan Nguyen

Expose the PEC calculation i2c_smbus_pec() for generic use.

You are the second one who wanted this exported. I agree it makes sense for slave drivers. So, I'll skip the required user because two are in flight. Applied to for-next, thanks!

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: not available
URL: <http://lists.ozlabs.org/pipermail/linux-aspeed/attachments/20210625/99eb8b53/attachment.sig>

#3 lore raw

PATCH v3 2/7 ipmi: ssif_bmc: Add SSIF BMC driver

In reply to Quan Nguyen

The SMBus system interface (SSIF) IPMI BMC driver can be used to perform in-band IPMI communication with their host in management (BMC) side.

---
v3:
+ Removed redundant license info [Joel]
+ Switched to use traditional if-else [Joel]
+ Removed unused ssif_bmc_ioctl() [Joel]
+ Made handle_request()/complete_response() to return void [Joel]
+ Refactored send_ssif_bmc_response()/receive_ssif_bmc_request() [Corey]
+ Removed mutex [Corey]
+ Use spin_lock/unlock_irqsave/restore in callback [Corey]
+ Removed the unnecessary memset [Corey]
+ Switch to use dev_err() [Corey]

v2:
+ Fixed compiling error with COMPILE_TEST for arc

 drivers/char/ipmi/Kconfig    |  11 +
 drivers/char/ipmi/Makefile   |   1 +
 drivers/char/ipmi/ssif_bmc.c | 605 +++++++++++++++++++++++++++++++++++
 drivers/char/ipmi/ssif_bmc.h |  93 ++++++
 4 files changed, 710 insertions(+)
 create mode 100644 drivers/char/ipmi/ssif_bmc.c
 create mode 100644 drivers/char/ipmi/ssif_bmc.h
drivers/char/ipmi/Kconfig +11 -0
index 07847d9a459a..ad5c5161bcd6 100644
@@ -133,6 +133,17 @@ config ASPEED_BT_IPMI_BMC
 	  found on Aspeed SOCs (AST2400 and AST2500). The driver
 	  implements the BMC side of the BT interface.
 
+config SSIF_IPMI_BMC
+	tristate "SSIF IPMI BMC driver"
+	select I2C
+	select I2C_SLAVE
+	help
+	  This enables the IPMI SMBus system interface (SSIF) at the
+	  management (BMC) side.
+
+	  The driver implements the BMC side of the SMBus system
+	  interface (SSIF).
+
 config IPMB_DEVICE_INTERFACE
 	tristate 'IPMB Interface handler'
 	depends on I2C
drivers/char/ipmi/Makefile +1 -0
index 0822adc2ec41..d04a214d74c4 100644
@@ -27,3 +27,4 @@ obj-$(CONFIG_ASPEED_BT_IPMI_BMC) += bt-bmc.o
 obj-$(CONFIG_ASPEED_KCS_IPMI_BMC) += kcs_bmc_aspeed.o
 obj-$(CONFIG_NPCM7XX_KCS_IPMI_BMC) += kcs_bmc_npcm7xx.o
 obj-$(CONFIG_IPMB_DEVICE_INTERFACE) += ipmb_dev_int.o
+obj-$(CONFIG_SSIF_IPMI_BMC) += ssif_bmc.o
drivers/char/ipmi/ssif_bmc.c new file +605 -0

605 lines not on this page · raw message

drivers/char/ipmi/ssif_bmc.h new file +93 -0

93 lines not on this page · raw message

Signature · 1 line
2.28.0
#4 lore raw

Re: [PATCH v3 2/7] ipmi: ssif_bmc: Add SSIF BMC driver

Quoted 85 lines · Quan Nguyen

The SMBus system interface (SSIF) IPMI BMC driver can be used to perform in-band IPMI communication with their host in management (BMC) side.

---
v3:
+ Removed redundant license info [Joel]
+ Switched to use traditional if-else [Joel]
+ Removed unused ssif_bmc_ioctl() [Joel]
+ Made handle_request()/complete_response() to return void [Joel]
+ Refactored send_ssif_bmc_response()/receive_ssif_bmc_request() [Corey]
+ Removed mutex [Corey]
+ Use spin_lock/unlock_irqsave/restore in callback [Corey]
+ Removed the unnecessary memset [Corey]
+ Switch to use dev_err() [Corey]

v2:
+ Fixed compiling error with COMPILE_TEST for arc

 drivers/char/ipmi/Kconfig    |  11 +
 drivers/char/ipmi/Makefile   |   1 +
 drivers/char/ipmi/ssif_bmc.c | 605 +++++++++++++++++++++++++++++++++++
 drivers/char/ipmi/ssif_bmc.h |  93 ++++++
 4 files changed, 710 insertions(+)
 create mode 100644 drivers/char/ipmi/ssif_bmc.c
 create mode 100644 drivers/char/ipmi/ssif_bmc.h
drivers/char/ipmi/Kconfig +11 -0
index 07847d9a459a..ad5c5161bcd6 100644
@@ -133,6 +133,17 @@ config ASPEED_BT_IPMI_BMC
 	  found on Aspeed SOCs (AST2400 and AST2500). The driver
 	  implements the BMC side of the BT interface.
 
+config SSIF_IPMI_BMC
+	tristate "SSIF IPMI BMC driver"
+	select I2C
+	select I2C_SLAVE
+	help
+	  This enables the IPMI SMBus system interface (SSIF) at the
+	  management (BMC) side.
+
+	  The driver implements the BMC side of the SMBus system
+	  interface (SSIF).
+
 config IPMB_DEVICE_INTERFACE
 	tristate 'IPMB Interface handler'
 	depends on I2C
drivers/char/ipmi/Makefile +1 -0
index 0822adc2ec41..d04a214d74c4 100644
@@ -27,3 +27,4 @@ obj-$(CONFIG_ASPEED_BT_IPMI_BMC) += bt-bmc.o
 obj-$(CONFIG_ASPEED_KCS_IPMI_BMC) += kcs_bmc_aspeed.o
 obj-$(CONFIG_NPCM7XX_KCS_IPMI_BMC) += kcs_bmc_npcm7xx.o
 obj-$(CONFIG_IPMB_DEVICE_INTERFACE) += ipmb_dev_int.o
+obj-$(CONFIG_SSIF_IPMI_BMC) += ssif_bmc.o
drivers/char/ipmi/ssif_bmc.c new file +21 -0
new file mode 100644
index 000000000000..d0ea7059b1db
@@ -0,0 +1,605 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * The driver for BMC side of SSIF interface
+ *
+ * Copyright (c) 2021, Ampere Computing LLC
+ *
+ */
+
+#include <linux/i2c.h>
+#include <linux/miscdevice.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/platform_device.h>
+#include <linux/poll.h>
+#include <linux/sched.h>
+#include <linux/mutex.h>
+#include <linux/spinlock.h>
+
+#include "ssif_bmc.h"
+
+/* Handle SSIF message that will be sent to user */

drivers/char/ipmi/ssif_bmc.c line 31

Quoted 10 lines
drivers/char/ipmi/ssif_bmc.c new file +10 -0
+static ssize_t ssif_bmc_read(struct file *file, char __user *buf, size_t count, loff_t *ppos)
+{
+	struct ssif_bmc_ctx *ssif_bmc = to_ssif_bmc(file);
+	struct ssif_msg msg;
+	unsigned long flags;
+	ssize_t ret;
+
+	spin_lock_irqsave(&ssif_bmc->lock, flags);
+	while (!ssif_bmc->request_available) {
+		if (file->f_flags & O_NONBLOCK) {

There is no real need to check this under the lock. It would simplify the end of this function if you checked this outside the lock.

drivers/char/ipmi/ssif_bmc.c line 44

Quoted 13 lines
drivers/char/ipmi/ssif_bmc.c new file +13 -0
+			ret = -EAGAIN;
+			goto out_unlock;
+		}
+		spin_unlock_irqrestore(&ssif_bmc->lock, flags);
+		ret = wait_event_interruptible(ssif_bmc->wait_queue,
+					       ssif_bmc->request_available);
+		if (ret)
+			return ret;
+		spin_lock_irqsave(&ssif_bmc->lock, flags);
+	}
+
+	count = min_t(ssize_t, ssif_msg_len(&ssif_bmc->request),
+		      min_t(ssize_t, sizeof(struct ssif_msg), count));

Just curious, what happens if userland supplies a buffer that is too small? Is data lost? In that case, I would think returning an error here would be the better thing, as opposed to silently truncating the message.

Quoted 25 lines
drivers/char/ipmi/ssif_bmc.c new file +25 -0
+	memcpy(&msg, &ssif_bmc->request, count);
+	ssif_bmc->request_available = false;
+	spin_unlock_irqrestore(&ssif_bmc->lock, flags);
+
+	ret = copy_to_user(buf, &msg, count);
+	goto out;
+
+out_unlock:
+	spin_unlock_irqrestore(&ssif_bmc->lock, flags);
+out:
+	return (ret < 0) ? ret : count;
+}
+
+/* Handle SSIF message that is written by user */
+static ssize_t ssif_bmc_write(struct file *file, const char __user *buf, size_t count,
+			      loff_t *ppos)
+{
+	struct ssif_bmc_ctx *ssif_bmc = to_ssif_bmc(file);
+	struct ssif_msg msg;
+	unsigned long flags;
+	ssize_t ret;
+
+	if (count > sizeof(struct ssif_msg))
+		return -EINVAL;
+

drivers/char/ipmi/ssif_bmc.c line 79

Quoted 10 lines
drivers/char/ipmi/ssif_bmc.c new file +10 -0
+	ret = copy_from_user(&msg, buf, count);
+	if (ret)
+		return ret;
+
+	if (!msg.len || count < ssif_msg_len(&msg))
+		return -EINVAL;
+
+	spin_lock_irqsave(&ssif_bmc->lock, flags);
+	while (ssif_bmc->response_in_progress) {
+		if (file->f_flags & O_NONBLOCK) {

Same comment as O_NONBLOCK before.

Quoted 245 lines
drivers/char/ipmi/ssif_bmc.c new file +245 -0
+			ret = -EAGAIN;
+			goto out_unlock;
+		}
+		spin_unlock_irqrestore(&ssif_bmc->lock, flags);
+		ret = wait_event_interruptible(ssif_bmc->wait_queue,
+					       !ssif_bmc->response_in_progress);
+		if (ret)
+			goto out;
+		spin_lock_irqsave(&ssif_bmc->lock, flags);
+	}
+
+	memcpy(&ssif_bmc->response, &msg, count);
+	ssif_bmc->is_singlepart_read = (ssif_msg_len(&msg) <= MAX_PAYLOAD_PER_TRANSACTION + 1);
+	ssif_bmc->response_in_progress = true;
+	if (ssif_bmc->set_ssif_bmc_status)
+		ssif_bmc->set_ssif_bmc_status(ssif_bmc, SSIF_BMC_READY);
+
+out_unlock:
+	spin_unlock_irqrestore(&ssif_bmc->lock, flags);
+
+out:
+	return (ret < 0) ? ret : count;
+}
+
+static int ssif_bmc_open(struct inode *inode, struct file *file)
+{
+	struct ssif_bmc_ctx *ssif_bmc = to_ssif_bmc(file);
+	int ret = 0;
+
+	spin_lock_irq(&ssif_bmc->lock);
+	if (!ssif_bmc->running)
+		ssif_bmc->running = 1;
+	else
+		ret = -EBUSY;
+	spin_unlock_irq(&ssif_bmc->lock);
+
+	return ret;
+}
+
+static unsigned int ssif_bmc_poll(struct file *file, poll_table *wait)
+{
+	struct ssif_bmc_ctx *ssif_bmc = to_ssif_bmc(file);
+	unsigned int mask = 0;
+
+	poll_wait(file, &ssif_bmc->wait_queue, wait);
+
+	spin_lock_irq(&ssif_bmc->lock);
+	/*
+	 * The request message is now available so userspace application can
+	 * get the request
+	 */
+	if (ssif_bmc->request_available)
+		mask |= POLLIN;
+
+	spin_unlock_irq(&ssif_bmc->lock);
+
+	return mask;
+}
+
+static int ssif_bmc_release(struct inode *inode, struct file *file)
+{
+	struct ssif_bmc_ctx *ssif_bmc = to_ssif_bmc(file);
+
+	spin_lock_irq(&ssif_bmc->lock);
+	ssif_bmc->running = 0;
+	spin_unlock_irq(&ssif_bmc->lock);
+
+	return 0;
+}
+
+/*
+ * System calls to device interface for user apps
+ */
+static const struct file_operations ssif_bmc_fops = {
+	.owner		= THIS_MODULE,
+	.open		= ssif_bmc_open,
+	.read		= ssif_bmc_read,
+	.write		= ssif_bmc_write,
+	.release	= ssif_bmc_release,
+	.poll		= ssif_bmc_poll,
+};
+
+/* Called with ssif_bmc->lock held. */
+static void handle_request(struct ssif_bmc_ctx *ssif_bmc)
+{
+	if (ssif_bmc->set_ssif_bmc_status)
+		ssif_bmc->set_ssif_bmc_status(ssif_bmc, SSIF_BMC_BUSY);
+
+	/* Request message is available to process */
+	ssif_bmc->request_available = true;
+	/*
+	 * This is the new READ request.
+	 */
+	wake_up_all(&ssif_bmc->wait_queue);
+}
+
+/* Called with ssif_bmc->lock held. */
+static void complete_response(struct ssif_bmc_ctx *ssif_bmc)
+{
+	/* Invalidate response in buffer to denote it having been sent. */
+	ssif_bmc->response.len = 0;
+	ssif_bmc->response_in_progress = false;
+	ssif_bmc->nbytes_processed = 0;
+	ssif_bmc->remain_len = 0;
+	wake_up_all(&ssif_bmc->wait_queue);
+}
+
+static void set_multipart_response_buffer(struct ssif_bmc_ctx *ssif_bmc, u8 *val)
+{
+	u8 response_len = 0;
+	int idx = 0;
+	u8 data_len;
+
+	data_len = ssif_bmc->response.len;
+	switch (ssif_bmc->smbus_cmd) {
+	case SSIF_IPMI_MULTIPART_READ_START:
+		/*
+		 * Read Start length is 32 bytes.
+		 * Read Start transfer first 30 bytes of IPMI response
+		 * and 2 special code 0x00, 0x01.
+		 */
+		*val = MAX_PAYLOAD_PER_TRANSACTION;
+		ssif_bmc->remain_len = data_len - MAX_IPMI_DATA_PER_START_TRANSACTION;
+		ssif_bmc->block_num = 0;
+
+		ssif_bmc->response_buf[idx++] = 0x00; /* Start Flag */
+		ssif_bmc->response_buf[idx++] = 0x01; /* Start Flag */
+		ssif_bmc->response_buf[idx++] = ssif_bmc->response.netfn_lun;
+		ssif_bmc->response_buf[idx++] = ssif_bmc->response.cmd;
+		ssif_bmc->response_buf[idx++] = ssif_bmc->response.payload[0];
+
+		response_len = MAX_PAYLOAD_PER_TRANSACTION - idx;
+
+		memcpy(&ssif_bmc->response_buf[idx], &ssif_bmc->response.payload[1],
+		       response_len);
+		break;
+
+	case SSIF_IPMI_MULTIPART_READ_MIDDLE:
+		/*
+		 * IPMI READ Middle or READ End messages can carry up to 31 bytes
+		 * IPMI data plus block number byte.
+		 */
+		if (ssif_bmc->remain_len < MAX_IPMI_DATA_PER_MIDDLE_TRANSACTION) {
+			/*
+			 * This is READ End message
+			 *  Return length is the remaining response data length
+			 *  plus block number
+			 *  Block number 0xFF is to indicate this is last message
+			 *
+			 */
+			*val = ssif_bmc->remain_len + 1;
+			ssif_bmc->block_num = 0xFF;
+			ssif_bmc->response_buf[idx++] = ssif_bmc->block_num;
+			response_len = ssif_bmc->remain_len;
+			/* Clean the buffer */
+			memset(&ssif_bmc->response_buf[idx], 0, MAX_PAYLOAD_PER_TRANSACTION - idx);
+		} else {
+			/*
+			 * This is READ Middle message
+			 *  Response length is the maximum SMBUS transfer length
+			 *  Block number byte is incremented
+			 * Return length is maximum SMBUS transfer length
+			 */
+			*val = MAX_PAYLOAD_PER_TRANSACTION;
+			ssif_bmc->remain_len -= MAX_IPMI_DATA_PER_MIDDLE_TRANSACTION;
+			response_len = MAX_IPMI_DATA_PER_MIDDLE_TRANSACTION;
+			ssif_bmc->response_buf[idx++] = ssif_bmc->block_num;
+			ssif_bmc->block_num++;
+		}
+
+		memcpy(&ssif_bmc->response_buf[idx],
+		       ssif_bmc->response.payload + 1 + ssif_bmc->nbytes_processed,
+		       response_len);
+		break;
+
+	default:
+		/* Do not expect to go to this case */
+		dev_err(&ssif_bmc->client->dev,
+			"Error: Unexpected SMBus command received 0x%x\n", ssif_bmc->smbus_cmd);
+		break;
+	}
+
+	ssif_bmc->nbytes_processed += response_len;
+}
+
+/* Process the IPMI response that will be read by master */
+static void handle_read_processed(struct ssif_bmc_ctx *ssif_bmc, u8 *val)
+{
+	u8 *buf;
+	u8 pec_len, addr, len;
+	u8 pec = 0;
+
+	pec_len = ssif_bmc->pec_support ? 1 : 0;
+	/* PEC - Start Read Address */
+	addr = GET_8BIT_ADDR(ssif_bmc->client->addr);
+	pec = i2c_smbus_pec(pec, &addr, 1);
+	/* PEC - SSIF Command */
+	pec = i2c_smbus_pec(pec, &ssif_bmc->smbus_cmd, 1);
+	/* PEC - Restart Write Address */
+	addr = addr | 0x01;
+	pec = i2c_smbus_pec(pec, &addr, 1);
+
+	if (ssif_bmc->is_singlepart_read) {
+		/* Single-part Read processing */
+		buf = (u8 *)&ssif_bmc->response;
+
+		if (ssif_bmc->response.len && ssif_bmc->msg_idx < ssif_bmc->response.len) {
+			ssif_bmc->msg_idx++;
+			*val = buf[ssif_bmc->msg_idx];
+		} else if (ssif_bmc->response.len && ssif_bmc->msg_idx == ssif_bmc->response.len) {
+			ssif_bmc->msg_idx++;
+			*val = i2c_smbus_pec(pec, buf, ssif_msg_len(&ssif_bmc->response));
+		} else {
+			*val = 0;
+		}
+		/* Invalidate response buffer to denote it is sent */
+		if (ssif_bmc->msg_idx + 1 >= (ssif_msg_len(&ssif_bmc->response) + pec_len))
+			complete_response(ssif_bmc);
+	} else {
+		/* Multi-part Read processing */
+		switch (ssif_bmc->smbus_cmd) {
+		case SSIF_IPMI_MULTIPART_READ_START:
+		case SSIF_IPMI_MULTIPART_READ_MIDDLE:
+			buf = (u8 *)&ssif_bmc->response_buf;
+			*val = buf[ssif_bmc->msg_idx];
+			ssif_bmc->msg_idx++;
+			break;
+		default:
+			/* Do not expect to go to this case */
+			dev_err(&ssif_bmc->client->dev,
+				"Error: Unexpected SMBus command received 0x%x\n",
+				ssif_bmc->smbus_cmd);
+			break;
+		}
+		len = (ssif_bmc->block_num == 0xFF) ?
+		       ssif_bmc->remain_len + 1 : MAX_PAYLOAD_PER_TRANSACTION;
+		if (ssif_bmc->msg_idx == (len + 1)) {
+			pec = i2c_smbus_pec(pec, &len, 1);
+			*val = i2c_smbus_pec(pec, ssif_bmc->response_buf, len);
+		}
+		/* Invalidate response buffer to denote last response is sent */
+		if (ssif_bmc->block_num == 0xFF &&
+		    ssif_bmc->msg_idx > (ssif_bmc->remain_len + pec_len)) {
+			complete_response(ssif_bmc);
+		}

drivers/char/ipmi/ssif_bmc.c line 334

Quoted 10 lines
drivers/char/ipmi/ssif_bmc.c new file +10 -0
+	}
+}
+
+static void handle_write_received(struct ssif_bmc_ctx *ssif_bmc, u8 *val)
+{
+	u8 *buf;
+	u8 smbus_cmd;
+
+	buf = (u8 *)&ssif_bmc->request;
+	if (ssif_bmc->msg_idx >= sizeof(struct ssif_msg))

Shouldn't you abort the message if you get too much data? Otherwise you get silent truncation.

drivers/char/ipmi/ssif_bmc.c line 344

Quoted 10 lines
drivers/char/ipmi/ssif_bmc.c new file +10 -0
+		return;
+
+	smbus_cmd = ssif_bmc->smbus_cmd;
+	switch (smbus_cmd) {
+	case SSIF_IPMI_SINGLEPART_WRITE:
+		/* Single-part write */
+		buf[ssif_bmc->msg_idx - 1] = *val;
+		ssif_bmc->msg_idx++;
+
+		break;

Generally you put the space after the break, not before. You did this right before, but didn't from here down in the file.

drivers/char/ipmi/ssif_bmc.c line 348

Quoted 4 lines
drivers/char/ipmi/ssif_bmc.c new file +4 -0
+	case SSIF_IPMI_MULTIPART_WRITE_START:
+		/* Reset length to zero */
+		if (ssif_bmc->msg_idx == 1)
+			ssif_bmc->request.len = 0;

What happens if you get this and you are in the middle of a message?

drivers/char/ipmi/ssif_bmc.c line 352

Quoted 4 lines
drivers/char/ipmi/ssif_bmc.c new file +4 -0
+
+		fallthrough;
+	case SSIF_IPMI_MULTIPART_WRITE_MIDDLE:
+	case SSIF_IPMI_MULTIPART_WRITE_END:

What happens if you get these and you are not in a message?

Quoted 355 lines
drivers/char/ipmi/ssif_bmc.c new file +253 -0
+		/* Multi-part write, 2nd byte received is length */
+		if (ssif_bmc->msg_idx == 1) {
+			ssif_bmc->request.len += *val;
+			ssif_bmc->recv_len = *val;
+		} else {
+			buf[ssif_bmc->msg_idx - 1 +
+			    ssif_bmc->request.len - ssif_bmc->recv_len]	= *val;
+		}
+
+		ssif_bmc->msg_idx++;
+
+		break;
+	default:
+		/* Do not expect to go to this case */
+		dev_err(&ssif_bmc->client->dev,
+			"Error: Unexpected SMBus command received 0x%x\n", ssif_bmc->smbus_cmd);
+		break;
+	}
+}
+
+static bool validate_pec(struct ssif_bmc_ctx *ssif_bmc)
+{
+	u8 rpec = 0, cpec = 0;
+	bool ret = true;
+	u8 addr, index;
+	u8 *buf;
+
+	buf = (u8 *)&ssif_bmc->request;
+	switch (ssif_bmc->smbus_cmd) {
+	case SSIF_IPMI_SINGLEPART_WRITE:
+		if ((ssif_bmc->msg_idx - 1) == ssif_msg_len(&ssif_bmc->request)) {
+			/* PEC is not included */
+			ssif_bmc->pec_support = false;
+			return true;
+		}
+
+		if ((ssif_bmc->msg_idx - 1) != (ssif_msg_len(&ssif_bmc->request) + 1))
+			goto error;
+
+		/* PEC is included */
+		ssif_bmc->pec_support = true;
+		rpec = buf[ssif_bmc->msg_idx - 2];
+		addr = GET_8BIT_ADDR(ssif_bmc->client->addr);
+		cpec = i2c_smbus_pec(cpec, &addr, 1);
+		cpec = i2c_smbus_pec(cpec, &ssif_bmc->smbus_cmd, 1);
+		cpec = i2c_smbus_pec(cpec, buf, ssif_msg_len(&ssif_bmc->request));
+		if (rpec != cpec) {
+			dev_err(&ssif_bmc->client->dev, "Bad PEC 0x%02x vs. 0x%02x\n", rpec, cpec);
+			ret = false;
+		}
+
+		break;
+	case SSIF_IPMI_MULTIPART_WRITE_START:
+	case SSIF_IPMI_MULTIPART_WRITE_MIDDLE:
+	case SSIF_IPMI_MULTIPART_WRITE_END:
+		index = ssif_bmc->request.len - ssif_bmc->recv_len;
+		if ((ssif_bmc->msg_idx - 1 + index) == ssif_msg_len(&ssif_bmc->request)) {
+			/* PEC is not included */
+			ssif_bmc->pec_support = false;
+			return true;
+		}
+
+		if ((ssif_bmc->msg_idx - 1 + index) != (ssif_msg_len(&ssif_bmc->request) + 1))
+			goto error;
+
+		/* PEC is included */
+		ssif_bmc->pec_support = true;
+		rpec = buf[ssif_bmc->msg_idx - 2 + index];
+		addr = GET_8BIT_ADDR(ssif_bmc->client->addr);
+		cpec = i2c_smbus_pec(cpec, &addr, 1);
+		cpec = i2c_smbus_pec(cpec, &ssif_bmc->smbus_cmd, 1);
+		cpec = i2c_smbus_pec(cpec, &ssif_bmc->recv_len, 1);
+		/* As SMBus specification does not allow the length
+		 * (byte count) in the Write-Block protocol to be zero.
+		 * Therefore, it is illegal to have the last Middle
+		 * transaction in the sequence carry 32-bytes and have
+		 * a length of ?0? in the End transaction.
+		 * But some users may try to use this way and we should
+		 * prevent ssif_bmc driver broken in this case.
+		 */
+		if (ssif_bmc->recv_len != 0)
+			cpec = i2c_smbus_pec(cpec, buf + 1 + index, ssif_bmc->recv_len);
+
+		if (rpec != cpec) {
+			dev_err(&ssif_bmc->client->dev, "Bad PEC 0x%02x vs. 0x%02x\n", rpec, cpec);
+			ret = false;
+		}
+
+		break;
+	default:
+		break;
+	}
+
+	return ret;
+error:
+	/* Do not expect to go to this case */
+	dev_err(&ssif_bmc->client->dev,
+		"Error: Unexpected length received %d\n", ssif_msg_len(&ssif_bmc->request));
+
+	return false;
+}
+
+static void complete_write_received(struct ssif_bmc_ctx *ssif_bmc)
+{
+	u8 cmd = ssif_bmc->smbus_cmd;
+
+	/* A BMC that receives an invalid PEC shall drop the data for the write
+	 * transaction and any further transactions (read or write) until
+	 * the next valid read or write Start transaction is received
+	 */
+	if (!validate_pec(ssif_bmc)) {
+		dev_err(&ssif_bmc->client->dev, "Received invalid PEC\n");
+		return;
+	}
+
+	if (cmd == SSIF_IPMI_SINGLEPART_WRITE || cmd == SSIF_IPMI_MULTIPART_WRITE_END)
+		handle_request(ssif_bmc);
+}
+
+static void initialize_transfer(struct ssif_bmc_ctx *ssif_bmc, u8 *val)
+{
+	/* SMBUS command can vary (single or multi-part) */
+	ssif_bmc->smbus_cmd = *val;
+	ssif_bmc->msg_idx++;
+
+	if (ssif_bmc->smbus_cmd == SSIF_IPMI_SINGLEPART_WRITE ||
+	    ssif_bmc->smbus_cmd == SSIF_IPMI_MULTIPART_WRITE_START) {
+		/*
+		 * The response can be delayed in BMC causing host SSIF driver
+		 * to timeout and send a new request once BMC slave is ready.
+		 * In that case check for pending response and clear it
+		 */
+		if (ssif_bmc->response_in_progress) {
+			dev_err(&ssif_bmc->client->dev,
+				"Warn: SSIF new request with pending response");
+			complete_response(ssif_bmc);
+		}
+	} else if (unlikely(ssif_bmc->smbus_cmd != SSIF_IPMI_SINGLEPART_READ &&
+			    ssif_bmc->smbus_cmd != SSIF_IPMI_MULTIPART_WRITE_MIDDLE &&
+			    ssif_bmc->smbus_cmd != SSIF_IPMI_MULTIPART_WRITE_END &&
+			    ssif_bmc->smbus_cmd != SSIF_IPMI_MULTIPART_READ_START &&
+			    ssif_bmc->smbus_cmd != SSIF_IPMI_MULTIPART_READ_MIDDLE)) {
+		/* Unknown command, clear it */
+		dev_err(&ssif_bmc->client->dev, "Warn: Unknown SMBus command");
+		complete_response(ssif_bmc);
+	}
+}
+
+/*
+ * Callback function to handle I2C slave events
+ */
+static int ssif_bmc_cb(struct i2c_client *client, enum i2c_slave_event event, u8 *val)
+{
+	unsigned long flags;
+	struct ssif_bmc_ctx *ssif_bmc = i2c_get_clientdata(client);
+
+	spin_lock_irqsave(&ssif_bmc->lock, flags);
+
+	/* I2C Event Handler:
+	 *   I2C_SLAVE_READ_REQUESTED	0x0
+	 *   I2C_SLAVE_WRITE_REQUESTED	0x1
+	 *   I2C_SLAVE_READ_PROCESSED	0x2
+	 *   I2C_SLAVE_WRITE_RECEIVED	0x3
+	 *   I2C_SLAVE_STOP		0x4
+	 */
+	switch (event) {
+	case I2C_SLAVE_READ_REQUESTED:
+		ssif_bmc->msg_idx = 0;
+		if (ssif_bmc->is_singlepart_read)
+			*val = ssif_bmc->response.len;
+		else
+			set_multipart_response_buffer(ssif_bmc, val);
+		break;
+
+	case I2C_SLAVE_WRITE_REQUESTED:
+		ssif_bmc->msg_idx = 0;
+		break;
+
+	case I2C_SLAVE_READ_PROCESSED:
+		handle_read_processed(ssif_bmc, val);
+		break;
+
+	case I2C_SLAVE_WRITE_RECEIVED:
+		if (ssif_bmc->msg_idx)
+			handle_write_received(ssif_bmc, val);
+		else
+			initialize_transfer(ssif_bmc, val);
+
+		break;
+
+	case I2C_SLAVE_STOP:
+		if (ssif_bmc->last_event == I2C_SLAVE_WRITE_RECEIVED)
+			complete_write_received(ssif_bmc);
+		/* Reset message index */
+		ssif_bmc->msg_idx = 0;
+		break;
+
+	default:
+		break;
+	}
+	ssif_bmc->last_event = event;
+	spin_unlock_irqrestore(&ssif_bmc->lock, flags);
+
+	return 0;
+}
+
+struct ssif_bmc_ctx *ssif_bmc_alloc(struct i2c_client *client, int sizeof_priv)
+{
+	struct ssif_bmc_ctx *ssif_bmc;
+	int ret;
+
+	ssif_bmc = devm_kzalloc(&client->dev, sizeof(*ssif_bmc) + sizeof_priv, GFP_KERNEL);
+	if (!ssif_bmc)
+		return ERR_PTR(-ENOMEM);
+
+	spin_lock_init(&ssif_bmc->lock);
+
+	init_waitqueue_head(&ssif_bmc->wait_queue);
+	ssif_bmc->request_available = false;
+	ssif_bmc->response_in_progress = false;
+
+	/* Register misc device interface */
+	ssif_bmc->miscdev.minor = MISC_DYNAMIC_MINOR;
+	ssif_bmc->miscdev.name = DEVICE_NAME;
+	ssif_bmc->miscdev.fops = &ssif_bmc_fops;
+	ssif_bmc->miscdev.parent = &client->dev;
+	ret = misc_register(&ssif_bmc->miscdev);
+	if (ret)
+		goto out;
+
+	ssif_bmc->client = client;
+	ssif_bmc->client->flags |= I2C_CLIENT_SLAVE;
+
+	/* Register I2C slave */
+	i2c_set_clientdata(client, ssif_bmc);
+	ret = i2c_slave_register(client, ssif_bmc_cb);
+	if (ret) {
+		misc_deregister(&ssif_bmc->miscdev);
+		goto out;
+	}
+
+	return ssif_bmc;
+
+out:
+	devm_kfree(&client->dev, ssif_bmc);
+	return ERR_PTR(ret);
+}
+EXPORT_SYMBOL(ssif_bmc_alloc);
+
+MODULE_AUTHOR("Chuong Tran <[email protected]>");
+MODULE_AUTHOR("Quan Nguyen <[email protected]>");
+MODULE_DESCRIPTION("Linux device driver of the BMC IPMI SSIF interface.");
+MODULE_LICENSE("GPL v2");
drivers/char/ipmi/ssif_bmc.h new file +93 -0
new file mode 100644
index 000000000000..eab540061f14
@@ -0,0 +1,93 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * The driver for BMC side of SSIF interface
+ *
+ * Copyright (c) 2021, Ampere Computing LLC
+ *
+ * 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 <https://www.gnu.org/licenses/>.
+ */
+#ifndef __SSIF_BMC_H__
+#define __SSIF_BMC_H__
+
+#define DEVICE_NAME				"ipmi-ssif-host"
+
+#define GET_8BIT_ADDR(addr_7bit)		(((addr_7bit) << 1) & 0xff)
+
+#define MSG_PAYLOAD_LEN_MAX			252
+
+/* A standard SMBus Transaction is limited to 32 data bytes */
+#define MAX_PAYLOAD_PER_TRANSACTION		32
+
+#define MAX_IPMI_DATA_PER_START_TRANSACTION	30
+#define MAX_IPMI_DATA_PER_MIDDLE_TRANSACTION	31
+
+#define SSIF_IPMI_SINGLEPART_WRITE		0x2
+#define SSIF_IPMI_SINGLEPART_READ		0x3
+#define SSIF_IPMI_MULTIPART_WRITE_START		0x6
+#define SSIF_IPMI_MULTIPART_WRITE_MIDDLE	0x7
+#define SSIF_IPMI_MULTIPART_WRITE_END		0x8
+#define SSIF_IPMI_MULTIPART_READ_START		0x3
+#define SSIF_IPMI_MULTIPART_READ_MIDDLE		0x9
+
+struct ssif_msg {
+	u8 len;
+	u8 netfn_lun;
+	u8 cmd;
+	u8 payload[MSG_PAYLOAD_LEN_MAX];
+} __packed;
+
+static inline u32 ssif_msg_len(struct ssif_msg *ssif_msg)
+{
+	return ssif_msg->len + 1;
+}
+
+#define SSIF_BMC_BUSY   0x01
+#define SSIF_BMC_READY  0x02
+
+struct ssif_bmc_ctx {
+	struct i2c_client	*client;
+	struct miscdevice	miscdev;
+	u8			running;
+	u8			smbus_cmd;
+	struct ssif_msg		request;
+	bool			request_available;
+	struct ssif_msg		response;
+	bool			response_in_progress;
+	/* Response buffer for Multi-part Read Transaction */
+	u8			response_buf[MAX_PAYLOAD_PER_TRANSACTION];
+	/* Flag to identify a Multi-part Read Transaction */
+	bool			is_singlepart_read;
+	u8			nbytes_processed;
+	u8			remain_len;
+	u8			recv_len;
+	/* Block Number of a Multi-part Read Transaction */
+	u8			block_num;
+	size_t			msg_idx;
+	enum i2c_slave_event	last_event;
+	bool			pec_support;
+	/* spinlock */
+	spinlock_t		lock;
+	wait_queue_head_t	wait_queue;
+	void (*set_ssif_bmc_status)(struct ssif_bmc_ctx *ssif_bmc, unsigned int flags);
+	void			*priv;
+};
+
+static inline struct ssif_bmc_ctx *to_ssif_bmc(struct file *file)
+{
+	return container_of(file->private_data, struct ssif_bmc_ctx, miscdev);
+}
+
+struct ssif_bmc_ctx *ssif_bmc_alloc(struct i2c_client *client, int sizeof_priv);
+
+#endif /* __SSIF_BMC_H__ */
Signature · 1 line
2.28.0
#5 lore raw

Re: [PATCH v3 2/7] ipmi: ssif_bmc: Add SSIF BMC driver

Quoted 91 lines · Corey Minyard
Quoted 90 lines · Quan Nguyen

The SMBus system interface (SSIF) IPMI BMC driver can be used to perform in-band IPMI communication with their host in management (BMC) side.

---
v3:
+ Removed redundant license info [Joel]
+ Switched to use traditional if-else [Joel]
+ Removed unused ssif_bmc_ioctl() [Joel]
+ Made handle_request()/complete_response() to return void [Joel]
+ Refactored send_ssif_bmc_response()/receive_ssif_bmc_request() [Corey]
+ Removed mutex [Corey]
+ Use spin_lock/unlock_irqsave/restore in callback [Corey]
+ Removed the unnecessary memset [Corey]
+ Switch to use dev_err() [Corey]

v2:
+ Fixed compiling error with COMPILE_TEST for arc

  drivers/char/ipmi/Kconfig    |  11 +
  drivers/char/ipmi/Makefile   |   1 +
  drivers/char/ipmi/ssif_bmc.c | 605 +++++++++++++++++++++++++++++++++++
  drivers/char/ipmi/ssif_bmc.h |  93 ++++++
  4 files changed, 710 insertions(+)

create mode 100644 drivers/char/ipmi/ssif_bmc.c
create mode 100644 drivers/char/ipmi/ssif_bmc.h

drivers/char/ipmi/Kconfig +11 -0
index 07847d9a459a..ad5c5161bcd6 100644
@@ -133,6 +133,17 @@ config ASPEED_BT_IPMI_BMC
  	  found on Aspeed SOCs (AST2400 and AST2500). The driver
  	  implements the BMC side of the BT interface.
 
+config SSIF_IPMI_BMC
+	tristate "SSIF IPMI BMC driver"
+	select I2C
+	select I2C_SLAVE
+	help
+	  This enables the IPMI SMBus system interface (SSIF) at the
+	  management (BMC) side.
+
+	  The driver implements the BMC side of the SMBus system
+	  interface (SSIF).
+
  config IPMB_DEVICE_INTERFACE
  	tristate 'IPMB Interface handler'
  	depends on I2C
drivers/char/ipmi/Makefile +1 -0
index 0822adc2ec41..d04a214d74c4 100644
@@ -27,3 +27,4 @@ obj-$(CONFIG_ASPEED_BT_IPMI_BMC) += bt-bmc.o
  obj-$(CONFIG_ASPEED_KCS_IPMI_BMC) += kcs_bmc_aspeed.o
  obj-$(CONFIG_NPCM7XX_KCS_IPMI_BMC) += kcs_bmc_npcm7xx.o
  obj-$(CONFIG_IPMB_DEVICE_INTERFACE) += ipmb_dev_int.o
+obj-$(CONFIG_SSIF_IPMI_BMC) += ssif_bmc.o
drivers/char/ipmi/ssif_bmc.c new file +26 -0
new file mode 100644
index 000000000000..d0ea7059b1db
@@ -0,0 +1,605 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * The driver for BMC side of SSIF interface
+ *
+ * Copyright (c) 2021, Ampere Computing LLC
+ *
+ */
+
+#include <linux/i2c.h>
+#include <linux/miscdevice.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/platform_device.h>
+#include <linux/poll.h>
+#include <linux/sched.h>
+#include <linux/mutex.h>
+#include <linux/spinlock.h>
+
+#include "ssif_bmc.h"
+
+/* Handle SSIF message that will be sent to user */
+static ssize_t ssif_bmc_read(struct file *file, char __user *buf, size_t count, loff_t *ppos)
+{
+	struct ssif_bmc_ctx *ssif_bmc = to_ssif_bmc(file);
+	struct ssif_msg msg;
+	unsigned long flags;
Quoted 9 lines

drivers/char/ipmi/ssif_bmc.c line 31

Quoted 5 lines
drivers/char/ipmi/ssif_bmc.c new file +5 -0
+	ssize_t ret;
+
+	spin_lock_irqsave(&ssif_bmc->lock, flags);
+	while (!ssif_bmc->request_available) {
+		if (file->f_flags & O_NONBLOCK) {

There is no real need to check this under the lock. It would simplify the end of this function if you checked this outside the lock.

Thanks for your reviewing.

Will apply change like below in next version:

         spin_lock_irqsave(&ssif_bmc->lock, flags);
         while (!ssif_bmc->request_available) {

- if (file->f_flags & O_NONBLOCK) {
- ret = -EAGAIN;
- goto out_unlock;
- }

                 spin_unlock_irqrestore(&ssif_bmc->lock, flags);
+               if (file->f_flags & O_NONBLOCK)
+                       return -EAGAIN;
                 ret = wait_event_interruptible(ssif_bmc->wait_queue,

ssif_bmc->request_available);
if (ret)

Quoted 10 lines
Quoted 10 lines
drivers/char/ipmi/ssif_bmc.c new file +10 -0
+			ret = -EAGAIN;
+			goto out_unlock;
+		}
+		spin_unlock_irqrestore(&ssif_bmc->lock, flags);
+		ret = wait_event_interruptible(ssif_bmc->wait_queue,
+					       ssif_bmc->request_available);
+		if (ret)
+			return ret;
+		spin_lock_irqsave(&ssif_bmc->lock, flags);
+	}
Quoted 9 lines

drivers/char/ipmi/ssif_bmc.c line 44

Quoted 3 lines
drivers/char/ipmi/ssif_bmc.c new file +3 -0
+
+	count = min_t(ssize_t, ssif_msg_len(&ssif_bmc->request),
+		      min_t(ssize_t, sizeof(struct ssif_msg), count));

Just curious, what happens if userland supplies a buffer that is too small? Is data lost? In that case, I would think returning an error here would be the better thing, as opposed to silently truncating the message.

Will return -EINVAL if buffer is too small in next version

Quoted 29 lines
Quoted 29 lines
drivers/char/ipmi/ssif_bmc.c new file +29 -0
+	memcpy(&msg, &ssif_bmc->request, count);
+	ssif_bmc->request_available = false;
+	spin_unlock_irqrestore(&ssif_bmc->lock, flags);
+
+	ret = copy_to_user(buf, &msg, count);
+	goto out;
+
+out_unlock:
+	spin_unlock_irqrestore(&ssif_bmc->lock, flags);
+out:
+	return (ret < 0) ? ret : count;
+}
+
+/* Handle SSIF message that is written by user */
+static ssize_t ssif_bmc_write(struct file *file, const char __user *buf, size_t count,
+			      loff_t *ppos)
+{
+	struct ssif_bmc_ctx *ssif_bmc = to_ssif_bmc(file);
+	struct ssif_msg msg;
+	unsigned long flags;
+	ssize_t ret;
+
+	if (count > sizeof(struct ssif_msg))
+		return -EINVAL;
+
+	ret = copy_from_user(&msg, buf, count);
+	if (ret)
+		return ret;
+
Quoted 9 lines

drivers/char/ipmi/ssif_bmc.c line 79

Quoted 6 lines
drivers/char/ipmi/ssif_bmc.c new file +6 -0
+	if (!msg.len || count < ssif_msg_len(&msg))
+		return -EINVAL;
+
+	spin_lock_irqsave(&ssif_bmc->lock, flags);
+	while (ssif_bmc->response_in_progress) {
+		if (file->f_flags & O_NONBLOCK) {

Same comment as O_NONBLOCK before.

Will apply this follow the above in ssif_bmc_read()

Quoted 250 lines
Quoted 250 lines
drivers/char/ipmi/ssif_bmc.c new file +250 -0
+			ret = -EAGAIN;
+			goto out_unlock;
+		}
+		spin_unlock_irqrestore(&ssif_bmc->lock, flags);
+		ret = wait_event_interruptible(ssif_bmc->wait_queue,
+					       !ssif_bmc->response_in_progress);
+		if (ret)
+			goto out;
+		spin_lock_irqsave(&ssif_bmc->lock, flags);
+	}
+
+	memcpy(&ssif_bmc->response, &msg, count);
+	ssif_bmc->is_singlepart_read = (ssif_msg_len(&msg) <= MAX_PAYLOAD_PER_TRANSACTION + 1);
+	ssif_bmc->response_in_progress = true;
+	if (ssif_bmc->set_ssif_bmc_status)
+		ssif_bmc->set_ssif_bmc_status(ssif_bmc, SSIF_BMC_READY);
+
+out_unlock:
+	spin_unlock_irqrestore(&ssif_bmc->lock, flags);
+
+out:
+	return (ret < 0) ? ret : count;
+}
+
+static int ssif_bmc_open(struct inode *inode, struct file *file)
+{
+	struct ssif_bmc_ctx *ssif_bmc = to_ssif_bmc(file);
+	int ret = 0;
+
+	spin_lock_irq(&ssif_bmc->lock);
+	if (!ssif_bmc->running)
+		ssif_bmc->running = 1;
+	else
+		ret = -EBUSY;
+	spin_unlock_irq(&ssif_bmc->lock);
+
+	return ret;
+}
+
+static unsigned int ssif_bmc_poll(struct file *file, poll_table *wait)
+{
+	struct ssif_bmc_ctx *ssif_bmc = to_ssif_bmc(file);
+	unsigned int mask = 0;
+
+	poll_wait(file, &ssif_bmc->wait_queue, wait);
+
+	spin_lock_irq(&ssif_bmc->lock);
+	/*
+	 * The request message is now available so userspace application can
+	 * get the request
+	 */
+	if (ssif_bmc->request_available)
+		mask |= POLLIN;
+
+	spin_unlock_irq(&ssif_bmc->lock);
+
+	return mask;
+}
+
+static int ssif_bmc_release(struct inode *inode, struct file *file)
+{
+	struct ssif_bmc_ctx *ssif_bmc = to_ssif_bmc(file);
+
+	spin_lock_irq(&ssif_bmc->lock);
+	ssif_bmc->running = 0;
+	spin_unlock_irq(&ssif_bmc->lock);
+
+	return 0;
+}
+
+/*
+ * System calls to device interface for user apps
+ */
+static const struct file_operations ssif_bmc_fops = {
+	.owner		= THIS_MODULE,
+	.open		= ssif_bmc_open,
+	.read		= ssif_bmc_read,
+	.write		= ssif_bmc_write,
+	.release	= ssif_bmc_release,
+	.poll		= ssif_bmc_poll,
+};
+
+/* Called with ssif_bmc->lock held. */
+static void handle_request(struct ssif_bmc_ctx *ssif_bmc)
+{
+	if (ssif_bmc->set_ssif_bmc_status)
+		ssif_bmc->set_ssif_bmc_status(ssif_bmc, SSIF_BMC_BUSY);
+
+	/* Request message is available to process */
+	ssif_bmc->request_available = true;
+	/*
+	 * This is the new READ request.
+	 */
+	wake_up_all(&ssif_bmc->wait_queue);
+}
+
+/* Called with ssif_bmc->lock held. */
+static void complete_response(struct ssif_bmc_ctx *ssif_bmc)
+{
+	/* Invalidate response in buffer to denote it having been sent. */
+	ssif_bmc->response.len = 0;
+	ssif_bmc->response_in_progress = false;
+	ssif_bmc->nbytes_processed = 0;
+	ssif_bmc->remain_len = 0;
+	wake_up_all(&ssif_bmc->wait_queue);
+}
+
+static void set_multipart_response_buffer(struct ssif_bmc_ctx *ssif_bmc, u8 *val)
+{
+	u8 response_len = 0;
+	int idx = 0;
+	u8 data_len;
+
+	data_len = ssif_bmc->response.len;
+	switch (ssif_bmc->smbus_cmd) {
+	case SSIF_IPMI_MULTIPART_READ_START:
+		/*
+		 * Read Start length is 32 bytes.
+		 * Read Start transfer first 30 bytes of IPMI response
+		 * and 2 special code 0x00, 0x01.
+		 */
+		*val = MAX_PAYLOAD_PER_TRANSACTION;
+		ssif_bmc->remain_len = data_len - MAX_IPMI_DATA_PER_START_TRANSACTION;
+		ssif_bmc->block_num = 0;
+
+		ssif_bmc->response_buf[idx++] = 0x00; /* Start Flag */
+		ssif_bmc->response_buf[idx++] = 0x01; /* Start Flag */
+		ssif_bmc->response_buf[idx++] = ssif_bmc->response.netfn_lun;
+		ssif_bmc->response_buf[idx++] = ssif_bmc->response.cmd;
+		ssif_bmc->response_buf[idx++] = ssif_bmc->response.payload[0];
+
+		response_len = MAX_PAYLOAD_PER_TRANSACTION - idx;
+
+		memcpy(&ssif_bmc->response_buf[idx], &ssif_bmc->response.payload[1],
+		       response_len);
+		break;
+
+	case SSIF_IPMI_MULTIPART_READ_MIDDLE:
+		/*
+		 * IPMI READ Middle or READ End messages can carry up to 31 bytes
+		 * IPMI data plus block number byte.
+		 */
+		if (ssif_bmc->remain_len < MAX_IPMI_DATA_PER_MIDDLE_TRANSACTION) {
+			/*
+			 * This is READ End message
+			 *  Return length is the remaining response data length
+			 *  plus block number
+			 *  Block number 0xFF is to indicate this is last message
+			 *
+			 */
+			*val = ssif_bmc->remain_len + 1;
+			ssif_bmc->block_num = 0xFF;
+			ssif_bmc->response_buf[idx++] = ssif_bmc->block_num;
+			response_len = ssif_bmc->remain_len;
+			/* Clean the buffer */
+			memset(&ssif_bmc->response_buf[idx], 0, MAX_PAYLOAD_PER_TRANSACTION - idx);
+		} else {
+			/*
+			 * This is READ Middle message
+			 *  Response length is the maximum SMBUS transfer length
+			 *  Block number byte is incremented
+			 * Return length is maximum SMBUS transfer length
+			 */
+			*val = MAX_PAYLOAD_PER_TRANSACTION;
+			ssif_bmc->remain_len -= MAX_IPMI_DATA_PER_MIDDLE_TRANSACTION;
+			response_len = MAX_IPMI_DATA_PER_MIDDLE_TRANSACTION;
+			ssif_bmc->response_buf[idx++] = ssif_bmc->block_num;
+			ssif_bmc->block_num++;
+		}
+
+		memcpy(&ssif_bmc->response_buf[idx],
+		       ssif_bmc->response.payload + 1 + ssif_bmc->nbytes_processed,
+		       response_len);
+		break;
+
+	default:
+		/* Do not expect to go to this case */
+		dev_err(&ssif_bmc->client->dev,
+			"Error: Unexpected SMBus command received 0x%x\n", ssif_bmc->smbus_cmd);
+		break;
+	}
+
+	ssif_bmc->nbytes_processed += response_len;
+}
+
+/* Process the IPMI response that will be read by master */
+static void handle_read_processed(struct ssif_bmc_ctx *ssif_bmc, u8 *val)
+{
+	u8 *buf;
+	u8 pec_len, addr, len;
+	u8 pec = 0;
+
+	pec_len = ssif_bmc->pec_support ? 1 : 0;
+	/* PEC - Start Read Address */
+	addr = GET_8BIT_ADDR(ssif_bmc->client->addr);
+	pec = i2c_smbus_pec(pec, &addr, 1);
+	/* PEC - SSIF Command */
+	pec = i2c_smbus_pec(pec, &ssif_bmc->smbus_cmd, 1);
+	/* PEC - Restart Write Address */
+	addr = addr | 0x01;
+	pec = i2c_smbus_pec(pec, &addr, 1);
+
+	if (ssif_bmc->is_singlepart_read) {
+		/* Single-part Read processing */
+		buf = (u8 *)&ssif_bmc->response;
+
+		if (ssif_bmc->response.len && ssif_bmc->msg_idx < ssif_bmc->response.len) {
+			ssif_bmc->msg_idx++;
+			*val = buf[ssif_bmc->msg_idx];
+		} else if (ssif_bmc->response.len && ssif_bmc->msg_idx == ssif_bmc->response.len) {
+			ssif_bmc->msg_idx++;
+			*val = i2c_smbus_pec(pec, buf, ssif_msg_len(&ssif_bmc->response));
+		} else {
+			*val = 0;
+		}
+		/* Invalidate response buffer to denote it is sent */
+		if (ssif_bmc->msg_idx + 1 >= (ssif_msg_len(&ssif_bmc->response) + pec_len))
+			complete_response(ssif_bmc);
+	} else {
+		/* Multi-part Read processing */
+		switch (ssif_bmc->smbus_cmd) {
+		case SSIF_IPMI_MULTIPART_READ_START:
+		case SSIF_IPMI_MULTIPART_READ_MIDDLE:
+			buf = (u8 *)&ssif_bmc->response_buf;
+			*val = buf[ssif_bmc->msg_idx];
+			ssif_bmc->msg_idx++;
+			break;
+		default:
+			/* Do not expect to go to this case */
+			dev_err(&ssif_bmc->client->dev,
+				"Error: Unexpected SMBus command received 0x%x\n",
+				ssif_bmc->smbus_cmd);
+			break;
+		}
+		len = (ssif_bmc->block_num == 0xFF) ?
+		       ssif_bmc->remain_len + 1 : MAX_PAYLOAD_PER_TRANSACTION;
+		if (ssif_bmc->msg_idx == (len + 1)) {
+			pec = i2c_smbus_pec(pec, &len, 1);
+			*val = i2c_smbus_pec(pec, ssif_bmc->response_buf, len);
+		}
+		/* Invalidate response buffer to denote last response is sent */
+		if (ssif_bmc->block_num == 0xFF &&
+		    ssif_bmc->msg_idx > (ssif_bmc->remain_len + pec_len)) {
+			complete_response(ssif_bmc);
+		}
+	}
+}
+
+static void handle_write_received(struct ssif_bmc_ctx *ssif_bmc, u8 *val)
+{
Quoted 9 lines

drivers/char/ipmi/ssif_bmc.c line 334

Quoted 5 lines
drivers/char/ipmi/ssif_bmc.c new file +5 -0
+	u8 *buf;
+	u8 smbus_cmd;
+
+	buf = (u8 *)&ssif_bmc->request;
+	if (ssif_bmc->msg_idx >= sizeof(struct ssif_msg))

Shouldn't you abort the message if you get too much data? Otherwise you get silent truncation.

Will try to abort bad request in next version.

Quoted 5 lines
Quoted 5 lines
drivers/char/ipmi/ssif_bmc.c new file +5 -0
+		return;
+
+	smbus_cmd = ssif_bmc->smbus_cmd;
+	switch (smbus_cmd) {
+	case SSIF_IPMI_SINGLEPART_WRITE:
Quoted 9 lines

drivers/char/ipmi/ssif_bmc.c line 344

Quoted 5 lines
drivers/char/ipmi/ssif_bmc.c new file +5 -0
+		/* Single-part write */
+		buf[ssif_bmc->msg_idx - 1] = *val;
+		ssif_bmc->msg_idx++;
+
+		break;

Generally you put the space after the break, not before. You did this right before, but didn't from here down in the file.

Will fix them in next version.

Quoted 7 lines

drivers/char/ipmi/ssif_bmc.c line 348

Quoted 4 lines
drivers/char/ipmi/ssif_bmc.c new file +4 -0
+	case SSIF_IPMI_MULTIPART_WRITE_START:
+		/* Reset length to zero */
+		if (ssif_bmc->msg_idx == 1)
+			ssif_bmc->request.len = 0;

What happens if you get this and you are in the middle of a message?

As per specs, this message should be discarded. This part is still missing. I'll try to add this feature in next version.

Quoted 7 lines

drivers/char/ipmi/ssif_bmc.c line 352

Quoted 4 lines
drivers/char/ipmi/ssif_bmc.c new file +4 -0
+
+		fallthrough;
+	case SSIF_IPMI_MULTIPART_WRITE_MIDDLE:
+	case SSIF_IPMI_MULTIPART_WRITE_END:

What happens if you get these and you are not in a message?

Same as my comment above.

Quoted 355 lines
Quoted 355 lines
drivers/char/ipmi/ssif_bmc.c new file +253 -0
+		/* Multi-part write, 2nd byte received is length */
+		if (ssif_bmc->msg_idx == 1) {
+			ssif_bmc->request.len += *val;
+			ssif_bmc->recv_len = *val;
+		} else {
+			buf[ssif_bmc->msg_idx - 1 +
+			    ssif_bmc->request.len - ssif_bmc->recv_len]	= *val;
+		}
+
+		ssif_bmc->msg_idx++;
+
+		break;
+	default:
+		/* Do not expect to go to this case */
+		dev_err(&ssif_bmc->client->dev,
+			"Error: Unexpected SMBus command received 0x%x\n", ssif_bmc->smbus_cmd);
+		break;
+	}
+}
+
+static bool validate_pec(struct ssif_bmc_ctx *ssif_bmc)
+{
+	u8 rpec = 0, cpec = 0;
+	bool ret = true;
+	u8 addr, index;
+	u8 *buf;
+
+	buf = (u8 *)&ssif_bmc->request;
+	switch (ssif_bmc->smbus_cmd) {
+	case SSIF_IPMI_SINGLEPART_WRITE:
+		if ((ssif_bmc->msg_idx - 1) == ssif_msg_len(&ssif_bmc->request)) {
+			/* PEC is not included */
+			ssif_bmc->pec_support = false;
+			return true;
+		}
+
+		if ((ssif_bmc->msg_idx - 1) != (ssif_msg_len(&ssif_bmc->request) + 1))
+			goto error;
+
+		/* PEC is included */
+		ssif_bmc->pec_support = true;
+		rpec = buf[ssif_bmc->msg_idx - 2];
+		addr = GET_8BIT_ADDR(ssif_bmc->client->addr);
+		cpec = i2c_smbus_pec(cpec, &addr, 1);
+		cpec = i2c_smbus_pec(cpec, &ssif_bmc->smbus_cmd, 1);
+		cpec = i2c_smbus_pec(cpec, buf, ssif_msg_len(&ssif_bmc->request));
+		if (rpec != cpec) {
+			dev_err(&ssif_bmc->client->dev, "Bad PEC 0x%02x vs. 0x%02x\n", rpec, cpec);
+			ret = false;
+		}
+
+		break;
+	case SSIF_IPMI_MULTIPART_WRITE_START:
+	case SSIF_IPMI_MULTIPART_WRITE_MIDDLE:
+	case SSIF_IPMI_MULTIPART_WRITE_END:
+		index = ssif_bmc->request.len - ssif_bmc->recv_len;
+		if ((ssif_bmc->msg_idx - 1 + index) == ssif_msg_len(&ssif_bmc->request)) {
+			/* PEC is not included */
+			ssif_bmc->pec_support = false;
+			return true;
+		}
+
+		if ((ssif_bmc->msg_idx - 1 + index) != (ssif_msg_len(&ssif_bmc->request) + 1))
+			goto error;
+
+		/* PEC is included */
+		ssif_bmc->pec_support = true;
+		rpec = buf[ssif_bmc->msg_idx - 2 + index];
+		addr = GET_8BIT_ADDR(ssif_bmc->client->addr);
+		cpec = i2c_smbus_pec(cpec, &addr, 1);
+		cpec = i2c_smbus_pec(cpec, &ssif_bmc->smbus_cmd, 1);
+		cpec = i2c_smbus_pec(cpec, &ssif_bmc->recv_len, 1);
+		/* As SMBus specification does not allow the length
+		 * (byte count) in the Write-Block protocol to be zero.
+		 * Therefore, it is illegal to have the last Middle
+		 * transaction in the sequence carry 32-bytes and have
+		 * a length of ?0? in the End transaction.
+		 * But some users may try to use this way and we should
+		 * prevent ssif_bmc driver broken in this case.
+		 */
+		if (ssif_bmc->recv_len != 0)
+			cpec = i2c_smbus_pec(cpec, buf + 1 + index, ssif_bmc->recv_len);
+
+		if (rpec != cpec) {
+			dev_err(&ssif_bmc->client->dev, "Bad PEC 0x%02x vs. 0x%02x\n", rpec, cpec);
+			ret = false;
+		}
+
+		break;
+	default:
+		break;
+	}
+
+	return ret;
+error:
+	/* Do not expect to go to this case */
+	dev_err(&ssif_bmc->client->dev,
+		"Error: Unexpected length received %d\n", ssif_msg_len(&ssif_bmc->request));
+
+	return false;
+}
+
+static void complete_write_received(struct ssif_bmc_ctx *ssif_bmc)
+{
+	u8 cmd = ssif_bmc->smbus_cmd;
+
+	/* A BMC that receives an invalid PEC shall drop the data for the write
+	 * transaction and any further transactions (read or write) until
+	 * the next valid read or write Start transaction is received
+	 */
+	if (!validate_pec(ssif_bmc)) {
+		dev_err(&ssif_bmc->client->dev, "Received invalid PEC\n");
+		return;
+	}
+
+	if (cmd == SSIF_IPMI_SINGLEPART_WRITE || cmd == SSIF_IPMI_MULTIPART_WRITE_END)
+		handle_request(ssif_bmc);
+}
+
+static void initialize_transfer(struct ssif_bmc_ctx *ssif_bmc, u8 *val)
+{
+	/* SMBUS command can vary (single or multi-part) */
+	ssif_bmc->smbus_cmd = *val;
+	ssif_bmc->msg_idx++;
+
+	if (ssif_bmc->smbus_cmd == SSIF_IPMI_SINGLEPART_WRITE ||
+	    ssif_bmc->smbus_cmd == SSIF_IPMI_MULTIPART_WRITE_START) {
+		/*
+		 * The response can be delayed in BMC causing host SSIF driver
+		 * to timeout and send a new request once BMC slave is ready.
+		 * In that case check for pending response and clear it
+		 */
+		if (ssif_bmc->response_in_progress) {
+			dev_err(&ssif_bmc->client->dev,
+				"Warn: SSIF new request with pending response");
+			complete_response(ssif_bmc);
+		}
+	} else if (unlikely(ssif_bmc->smbus_cmd != SSIF_IPMI_SINGLEPART_READ &&
+			    ssif_bmc->smbus_cmd != SSIF_IPMI_MULTIPART_WRITE_MIDDLE &&
+			    ssif_bmc->smbus_cmd != SSIF_IPMI_MULTIPART_WRITE_END &&
+			    ssif_bmc->smbus_cmd != SSIF_IPMI_MULTIPART_READ_START &&
+			    ssif_bmc->smbus_cmd != SSIF_IPMI_MULTIPART_READ_MIDDLE)) {
+		/* Unknown command, clear it */
+		dev_err(&ssif_bmc->client->dev, "Warn: Unknown SMBus command");
+		complete_response(ssif_bmc);
+	}
+}
+
+/*
+ * Callback function to handle I2C slave events
+ */
+static int ssif_bmc_cb(struct i2c_client *client, enum i2c_slave_event event, u8 *val)
+{
+	unsigned long flags;
+	struct ssif_bmc_ctx *ssif_bmc = i2c_get_clientdata(client);
+
+	spin_lock_irqsave(&ssif_bmc->lock, flags);
+
+	/* I2C Event Handler:
+	 *   I2C_SLAVE_READ_REQUESTED	0x0
+	 *   I2C_SLAVE_WRITE_REQUESTED	0x1
+	 *   I2C_SLAVE_READ_PROCESSED	0x2
+	 *   I2C_SLAVE_WRITE_RECEIVED	0x3
+	 *   I2C_SLAVE_STOP		0x4
+	 */
+	switch (event) {
+	case I2C_SLAVE_READ_REQUESTED:
+		ssif_bmc->msg_idx = 0;
+		if (ssif_bmc->is_singlepart_read)
+			*val = ssif_bmc->response.len;
+		else
+			set_multipart_response_buffer(ssif_bmc, val);
+		break;
+
+	case I2C_SLAVE_WRITE_REQUESTED:
+		ssif_bmc->msg_idx = 0;
+		break;
+
+	case I2C_SLAVE_READ_PROCESSED:
+		handle_read_processed(ssif_bmc, val);
+		break;
+
+	case I2C_SLAVE_WRITE_RECEIVED:
+		if (ssif_bmc->msg_idx)
+			handle_write_received(ssif_bmc, val);
+		else
+			initialize_transfer(ssif_bmc, val);
+
+		break;
+
+	case I2C_SLAVE_STOP:
+		if (ssif_bmc->last_event == I2C_SLAVE_WRITE_RECEIVED)
+			complete_write_received(ssif_bmc);
+		/* Reset message index */
+		ssif_bmc->msg_idx = 0;
+		break;
+
+	default:
+		break;
+	}
+	ssif_bmc->last_event = event;
+	spin_unlock_irqrestore(&ssif_bmc->lock, flags);
+
+	return 0;
+}
+
+struct ssif_bmc_ctx *ssif_bmc_alloc(struct i2c_client *client, int sizeof_priv)
+{
+	struct ssif_bmc_ctx *ssif_bmc;
+	int ret;
+
+	ssif_bmc = devm_kzalloc(&client->dev, sizeof(*ssif_bmc) + sizeof_priv, GFP_KERNEL);
+	if (!ssif_bmc)
+		return ERR_PTR(-ENOMEM);
+
+	spin_lock_init(&ssif_bmc->lock);
+
+	init_waitqueue_head(&ssif_bmc->wait_queue);
+	ssif_bmc->request_available = false;
+	ssif_bmc->response_in_progress = false;
+
+	/* Register misc device interface */
+	ssif_bmc->miscdev.minor = MISC_DYNAMIC_MINOR;
+	ssif_bmc->miscdev.name = DEVICE_NAME;
+	ssif_bmc->miscdev.fops = &ssif_bmc_fops;
+	ssif_bmc->miscdev.parent = &client->dev;
+	ret = misc_register(&ssif_bmc->miscdev);
+	if (ret)
+		goto out;
+
+	ssif_bmc->client = client;
+	ssif_bmc->client->flags |= I2C_CLIENT_SLAVE;
+
+	/* Register I2C slave */
+	i2c_set_clientdata(client, ssif_bmc);
+	ret = i2c_slave_register(client, ssif_bmc_cb);
+	if (ret) {
+		misc_deregister(&ssif_bmc->miscdev);
+		goto out;
+	}
+
+	return ssif_bmc;
+
+out:
+	devm_kfree(&client->dev, ssif_bmc);
+	return ERR_PTR(ret);
+}
+EXPORT_SYMBOL(ssif_bmc_alloc);
+
+MODULE_AUTHOR("Chuong Tran <[email protected]>");
+MODULE_AUTHOR("Quan Nguyen <[email protected]>");
+MODULE_DESCRIPTION("Linux device driver of the BMC IPMI SSIF interface.");
+MODULE_LICENSE("GPL v2");
drivers/char/ipmi/ssif_bmc.h new file +93 -0
new file mode 100644
index 000000000000..eab540061f14
@@ -0,0 +1,93 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * The driver for BMC side of SSIF interface
+ *
+ * Copyright (c) 2021, Ampere Computing LLC
+ *
+ * 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 <https://www.gnu.org/licenses/>.
+ */
+#ifndef __SSIF_BMC_H__
+#define __SSIF_BMC_H__
+
+#define DEVICE_NAME				"ipmi-ssif-host"
+
+#define GET_8BIT_ADDR(addr_7bit)		(((addr_7bit) << 1) & 0xff)
+
+#define MSG_PAYLOAD_LEN_MAX			252
+
+/* A standard SMBus Transaction is limited to 32 data bytes */
+#define MAX_PAYLOAD_PER_TRANSACTION		32
+
+#define MAX_IPMI_DATA_PER_START_TRANSACTION	30
+#define MAX_IPMI_DATA_PER_MIDDLE_TRANSACTION	31
+
+#define SSIF_IPMI_SINGLEPART_WRITE		0x2
+#define SSIF_IPMI_SINGLEPART_READ		0x3
+#define SSIF_IPMI_MULTIPART_WRITE_START		0x6
+#define SSIF_IPMI_MULTIPART_WRITE_MIDDLE	0x7
+#define SSIF_IPMI_MULTIPART_WRITE_END		0x8
+#define SSIF_IPMI_MULTIPART_READ_START		0x3
+#define SSIF_IPMI_MULTIPART_READ_MIDDLE		0x9
+
+struct ssif_msg {
+	u8 len;
+	u8 netfn_lun;
+	u8 cmd;
+	u8 payload[MSG_PAYLOAD_LEN_MAX];
+} __packed;
+
+static inline u32 ssif_msg_len(struct ssif_msg *ssif_msg)
+{
+	return ssif_msg->len + 1;
+}
+
+#define SSIF_BMC_BUSY   0x01
+#define SSIF_BMC_READY  0x02
+
+struct ssif_bmc_ctx {
+	struct i2c_client	*client;
+	struct miscdevice	miscdev;
+	u8			running;
+	u8			smbus_cmd;
+	struct ssif_msg		request;
+	bool			request_available;
+	struct ssif_msg		response;
+	bool			response_in_progress;
+	/* Response buffer for Multi-part Read Transaction */
+	u8			response_buf[MAX_PAYLOAD_PER_TRANSACTION];
+	/* Flag to identify a Multi-part Read Transaction */
+	bool			is_singlepart_read;
+	u8			nbytes_processed;
+	u8			remain_len;
+	u8			recv_len;
+	/* Block Number of a Multi-part Read Transaction */
+	u8			block_num;
+	size_t			msg_idx;
+	enum i2c_slave_event	last_event;
+	bool			pec_support;
+	/* spinlock */
+	spinlock_t		lock;
+	wait_queue_head_t	wait_queue;
+	void (*set_ssif_bmc_status)(struct ssif_bmc_ctx *ssif_bmc, unsigned int flags);
+	void			*priv;
+};
+
+static inline struct ssif_bmc_ctx *to_ssif_bmc(struct file *file)
+{
+	return container_of(file->private_data, struct ssif_bmc_ctx, miscdev);
+}
+
+struct ssif_bmc_ctx *ssif_bmc_alloc(struct i2c_client *client, int sizeof_priv);
+
+#endif /* __SSIF_BMC_H__ */
Signature · 1 line
2.28.0
#6 lore raw

PATCH v3 3/7 i2c: aspeed: Fix unhandled Tx done with NAK

In reply to Quan Nguyen

It is observed that in normal condition, when the last byte sent by slave, the Tx Done with NAK irq will raise.
But it is also observed that sometimes master issues next transaction too quick while the slave irq handler is not yet invoked and Tx Done with NAK irq of last byte of previous READ PROCESSED was not ack'ed. This Tx Done with NAK irq is raised together with the Slave Match and Rx Done irq of the next coming transaction from master. Unfortunately, the current slave irq handler handles the Slave Match and Rx Done only in higher priority and ignore the Tx Done with NAK, causing the complain as below:
"aspeed-i2c-bus 1e78a040.i2c-bus: irq handled != irq. expected 0x00000086, but was 0x00000084"

This commit handles this case by emitting a Slave Stop event for the Tx Done with NAK before processing Slave Match and Rx Done for the coming transaction from master.

---
v3:
+ First introduce in v3 [Quan]

 drivers/i2c/busses/i2c-aspeed.c | 5 +++++
 1 file changed, 5 insertions(+)
drivers/i2c/busses/i2c-aspeed.c +5 -0
index 724bf30600d6..3fb37c3f23d4 100644
@@ -254,6 +254,11 @@ static u32 aspeed_i2c_slave_irq(struct aspeed_i2c_bus *bus, u32 irq_status)
 
 	/* Slave was requested, restart state machine. */
 	if (irq_status & ASPEED_I2CD_INTR_SLAVE_MATCH) {
+		if (irq_status & ASPEED_I2CD_INTR_TX_NAK &&
+		    bus->slave_state == ASPEED_I2C_SLAVE_READ_PROCESSED) {
+			irq_handled |= ASPEED_I2CD_INTR_TX_NAK;
+			i2c_slave_event(slave, I2C_SLAVE_STOP, &value);
+		}
 		irq_handled |= ASPEED_I2CD_INTR_SLAVE_MATCH;
 		bus->slave_state = ASPEED_I2C_SLAVE_START;
 	}
Signature · 1 line
2.28.0
maintainer
#7 lore raw

Re: [PATCH v3 3/7] i2c: aspeed: Fix unhandled Tx done with NAK

Ryan, can you please review this change?

Quoted 14 lines · Quan Nguyen

It is observed that in normal condition, when the last byte sent by slave, the Tx Done with NAK irq will raise.
But it is also observed that sometimes master issues next transaction too quick while the slave irq handler is not yet invoked and Tx Done with NAK irq of last byte of previous READ PROCESSED was not ack'ed. This Tx Done with NAK irq is raised together with the Slave Match and Rx Done irq of the next coming transaction from master. Unfortunately, the current slave irq handler handles the Slave Match and Rx Done only in higher priority and ignore the Tx Done with NAK, causing the complain as below:
"aspeed-i2c-bus 1e78a040.i2c-bus: irq handled != irq. expected 0x00000086, but was 0x00000084"

Quoted 3 lines

This commit handles this case by emitting a Slave Stop event for the Tx Done with NAK before processing Slave Match and Rx Done for the coming transaction from master.

It sounds like this patch is independent of the rest of the series, and can go in on it's own. Please send it separately to the i2c maintainers and add a suitable Fixes line, such as:

Fixes f9eb91350bb2 ("i2c: aspeed: added slave support for Aspeed I2C driver")
Quoted 14 lines

---
v3:
+ First introduce in v3 [Quan]

 drivers/i2c/busses/i2c-aspeed.c | 5 +++++
 1 file changed, 5 insertions(+)
drivers/i2c/busses/i2c-aspeed.c +0 -0
index 724bf30600d6..3fb37c3f23d4 100644

drivers/i2c/busses/i2c-aspeed.c static u32 aspeed_i2c_slave_irq(struct aspeed_i2c_bus *bus, u32 irq_status) · line 256

Quoted 3 lines
drivers/i2c/busses/i2c-aspeed.c +0 -0
@@ -254,6 +254,11 @@ static u32 aspeed_i2c_slave_irq(struct aspeed_i2c_bus *bus, u32 irq_status)
 
        /* Slave was requested, restart state machine. */
        if (irq_status & ASPEED_I2CD_INTR_SLAVE_MATCH) {

Can you explain why you need to do this handing inside the SLAVE_MATCH case?

Could you instead move the TX_NAK handling to be above the SLAVE_MATCH case?

drivers/i2c/busses/i2c-aspeed.c static u32 aspeed_i2c_slave_irq(struct aspeed_i2c_bus *bus, u32 irq_status) · line 258

Quoted 2 lines
drivers/i2c/busses/i2c-aspeed.c +2 -0
+               if (irq_status & ASPEED_I2CD_INTR_TX_NAK &&
+                   bus->slave_state == ASPEED_I2C_SLAVE_READ_PROCESSED) {

Either way, this needs a comment to explain what we're working around.

Quoted 9 lines
drivers/i2c/busses/i2c-aspeed.c +3 -1
+                       irq_handled |= ASPEED_I2CD_INTR_TX_NAK;
+                       i2c_slave_event(slave, I2C_SLAVE_STOP, &value);
+               }
                irq_handled |= ASPEED_I2CD_INTR_SLAVE_MATCH;
                bus->slave_state = ASPEED_I2C_SLAVE_START;
        }
--

2.28.0

maintainer
#8 lore raw

RE: [PATCH v3 3/7] i2c: aspeed: Fix unhandled Tx done with NAK

Quoted 79 lines

-----Original Message-----
From: Joel Stanley <[email protected]>
Sent: Thursday, May 20, 2021 7:29 AM
To: Quan Nguyen <[email protected]>; Ryan Chen <[email protected]>
Cc: Corey Minyard <[email protected]>; Rob Herring <[email protected]>; Andrew Jeffery <[email protected]>; Brendan Higgins <[email protected]>; Benjamin Herrenschmidt <[email protected]>; Wolfram Sang <[email protected]>; Philipp Zabel <[email protected]>; openipmi-developer at lists.sourceforge.net; devicetree <[email protected]>; Linux ARM <[email protected]>; linux-aspeed <[email protected]>; Linux Kernel Mailing List <[email protected]>; linux-i2c at vger.kernel.org; Open Source Submission <[email protected]>; Phong Vo <[email protected]>; Thang Q . Nguyen <[email protected]>; OpenBMC Maillist <[email protected]>
Subject: Re: [PATCH v3 3/7] i2c: aspeed: Fix unhandled Tx done with NAK

Ryan, can you please review this change?

Quoted 14 lines · Quan Nguyen

It is observed that in normal condition, when the last byte sent by slave, the Tx Done with NAK irq will raise.
But it is also observed that sometimes master issues next transaction too quick while the slave irq handler is not yet invoked and Tx Done with NAK irq of last byte of previous READ PROCESSED was not ack'ed. This Tx Done with NAK irq is raised together with the Slave Match and Rx Done irq of the next coming transaction from master. Unfortunately, the current slave irq handler handles the Slave Match and Rx Done only in higher priority and ignore the Tx Done with NAK, causing the complain as below:
"aspeed-i2c-bus 1e78a040.i2c-bus: irq handled != irq. expected 0x00000086, but was 0x00000084"

Quoted 3 lines

This commit handles this case by emitting a Slave Stop event for the Tx Done with NAK before processing Slave Match and Rx Done for the coming transaction from master.

It sounds like this patch is independent of the rest of the series, and can go in on it's own. Please send it separately to the i2c maintainers and add a suitable Fixes line, such as:

Fixes f9eb91350bb2 ("i2c: aspeed: added slave support for Aspeed I2C

driver")

Quoted 15 lines

---
v3:
+ First introduce in v3 [Quan]

 drivers/i2c/busses/i2c-aspeed.c | 5 +++++
 1 file changed, 5 insertions(+)
(quoted diff) +0 -0

b/drivers/i2c/busses/i2c-aspeed.c index 724bf30600d6..3fb37c3f23d4 100644

drivers/i2c/busses/i2c-aspeed.c static u32 aspeed_i2c_slave_irq(struct · line 257

Quoted 4 lines
drivers/i2c/busses/i2c-aspeed.c +0 -0
@@ -254,6 +254,11 @@ static u32 aspeed_i2c_slave_irq(struct
 aspeed_i2c_bus *bus, u32 irq_status)
 
        /* Slave was requested, restart state machine. */
        if (irq_status & ASPEED_I2CD_INTR_SLAVE_MATCH) {

Can you explain why you need to do this handing inside the SLAVE_MATCH case?

Could you instead move the TX_NAK handling to be above the SLAVE_MATCH case?

drivers/i2c/busses/i2c-aspeed.c static u32 aspeed_i2c_slave_irq(struct · line 260

Quoted 3 lines
drivers/i2c/busses/i2c-aspeed.c +3 -0
+               if (irq_status & ASPEED_I2CD_INTR_TX_NAK &&
+                   bus->slave_state ==
+ ASPEED_I2C_SLAVE_READ_PROCESSED) {
Quoted 5 lines

Either way, this needs a comment to explain what we're working around.

drivers/i2c/busses/i2c-aspeed.c static u32 aspeed_i2c_slave_irq(struct · line 262

Quoted 2 lines
drivers/i2c/busses/i2c-aspeed.c +2 -0
+                       irq_handled |= ASPEED_I2CD_INTR_TX_NAK;
+                       i2c_slave_event(slave, I2C_SLAVE_STOP,

&value);

According the patch assume slave receive TX_NAK will be go to SLAVE_STOP state?

Quoted 7 lines
Quoted 7 lines
drivers/i2c/busses/i2c-aspeed.c +1 -1
+               }
                irq_handled |= ASPEED_I2CD_INTR_SLAVE_MATCH;
                bus->slave_state = ASPEED_I2C_SLAVE_START;
        }
--

2.28.0

#9 lore raw

Re: [PATCH v3 3/7] i2c: aspeed: Fix unhandled Tx done with NAK

Quoted 79 lines · Ryan Chen
Quoted 79 lines

-----Original Message-----
From: Joel Stanley <[email protected]>
Sent: Thursday, May 20, 2021 7:29 AM
To: Quan Nguyen <[email protected]>; Ryan Chen <[email protected]>
Cc: Corey Minyard <[email protected]>; Rob Herring <[email protected]>; Andrew Jeffery <[email protected]>; Brendan Higgins <[email protected]>; Benjamin Herrenschmidt <[email protected]>; Wolfram Sang <[email protected]>; Philipp Zabel <[email protected]>; openipmi-developer at lists.sourceforge.net; devicetree <[email protected]>; Linux ARM <[email protected]>; linux-aspeed <[email protected]>; Linux Kernel Mailing List <[email protected]>; linux-i2c at vger.kernel.org; Open Source Submission <[email protected]>; Phong Vo <[email protected]>; Thang Q . Nguyen <[email protected]>; OpenBMC Maillist <[email protected]>
Subject: Re: [PATCH v3 3/7] i2c: aspeed: Fix unhandled Tx done with NAK

Ryan, can you please review this change?

Quoted 14 lines · Quan Nguyen

It is observed that in normal condition, when the last byte sent by slave, the Tx Done with NAK irq will raise.
But it is also observed that sometimes master issues next transaction too quick while the slave irq handler is not yet invoked and Tx Done with NAK irq of last byte of previous READ PROCESSED was not ack'ed. This Tx Done with NAK irq is raised together with the Slave Match and Rx Done irq of the next coming transaction from master. Unfortunately, the current slave irq handler handles the Slave Match and Rx Done only in higher priority and ignore the Tx Done with NAK, causing the complain as below:
"aspeed-i2c-bus 1e78a040.i2c-bus: irq handled != irq. expected 0x00000086, but was 0x00000084"

Quoted 3 lines

This commit handles this case by emitting a Slave Stop event for the Tx Done with NAK before processing Slave Match and Rx Done for the coming transaction from master.

It sounds like this patch is independent of the rest of the series, and can go in on it's own. Please send it separately to the i2c maintainers and add a suitable Fixes line, such as:

Fixes f9eb91350bb2 ("i2c: aspeed: added slave support for Aspeed I2C

driver")

Quoted 15 lines

---
v3:
+ First introduce in v3 [Quan]

  drivers/i2c/busses/i2c-aspeed.c | 5 +++++
  1 file changed, 5 insertions(+)
(quoted diff) +0 -0

b/drivers/i2c/busses/i2c-aspeed.c index 724bf30600d6..3fb37c3f23d4 100644

drivers/i2c/busses/i2c-aspeed.c static u32 aspeed_i2c_slave_irq(struct · line 257

Quoted 4 lines
drivers/i2c/busses/i2c-aspeed.c +0 -0
@@ -254,6 +254,11 @@ static u32 aspeed_i2c_slave_irq(struct
 aspeed_i2c_bus *bus, u32 irq_status)
 
         /* Slave was requested, restart state machine. */
         if (irq_status & ASPEED_I2CD_INTR_SLAVE_MATCH) {

Can you explain why you need to do this handing inside the SLAVE_MATCH case?

Could you instead move the TX_NAK handling to be above the SLAVE_MATCH case?

Quoted 3 lines
drivers/i2c/busses/i2c-aspeed.c +3 -0
+               if (irq_status & ASPEED_I2CD_INTR_TX_NAK &&
+                   bus->slave_state ==
+ ASPEED_I2C_SLAVE_READ_PROCESSED) {
Quoted 8 lines
Quoted 5 lines

Either way, this needs a comment to explain what we're working around.

drivers/i2c/busses/i2c-aspeed.c static u32 aspeed_i2c_slave_irq(struct · line 262

Quoted 2 lines
drivers/i2c/busses/i2c-aspeed.c +2 -0
+                       irq_handled |= ASPEED_I2CD_INTR_TX_NAK;
+                       i2c_slave_event(slave, I2C_SLAVE_STOP,

&value);

According the patch assume slave receive TX_NAK will be go to SLAVE_STOP state?

Hi Ryan,

As per my explain in other email, we need to emit one SLAVE_STOP event to complete the previous transaction before start processing for the next transaction.

- Quan

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
drivers/i2c/busses/i2c-aspeed.c +1 -1
+               }
                 irq_handled |= ASPEED_I2CD_INTR_SLAVE_MATCH;
                 bus->slave_state = ASPEED_I2C_SLAVE_START;
         }
--

2.28.0

#10 lore raw

Re: [PATCH v3 3/7] i2c: aspeed: Fix unhandled Tx done with NAK

In reply to Joel Stanley
Quoted 21 lines · Joel Stanley

Ryan, can you please review this change?

Quoted 14 lines · Quan Nguyen

It is observed that in normal condition, when the last byte sent by slave, the Tx Done with NAK irq will raise.
But it is also observed that sometimes master issues next transaction too quick while the slave irq handler is not yet invoked and Tx Done with NAK irq of last byte of previous READ PROCESSED was not ack'ed. This Tx Done with NAK irq is raised together with the Slave Match and Rx Done irq of the next coming transaction from master. Unfortunately, the current slave irq handler handles the Slave Match and Rx Done only in higher priority and ignore the Tx Done with NAK, causing the complain as below:
"aspeed-i2c-bus 1e78a040.i2c-bus: irq handled != irq. expected 0x00000086, but was 0x00000084"

Quoted 3 lines

This commit handles this case by emitting a Slave Stop event for the Tx Done with NAK before processing Slave Match and Rx Done for the coming transaction from master.

Quoted 6 lines

It sounds like this patch is independent of the rest of the series, and can go in on it's own. Please send it separately to the i2c maintainers and add a suitable Fixes line, such as:

Fixes f9eb91350bb2 ("i2c: aspeed: added slave support for Aspeed I2C driver")

Will separate this patch into independent series in next version.

Quoted 20 lines
Quoted 14 lines

---
v3:
+ First introduce in v3 [Quan]

  drivers/i2c/busses/i2c-aspeed.c | 5 +++++
  1 file changed, 5 insertions(+)
drivers/i2c/busses/i2c-aspeed.c +0 -0
index 724bf30600d6..3fb37c3f23d4 100644

drivers/i2c/busses/i2c-aspeed.c static u32 aspeed_i2c_slave_irq(struct aspeed_i2c_bus *bus, u32 irq_status) · line 256

Quoted 3 lines
drivers/i2c/busses/i2c-aspeed.c +0 -0
@@ -254,6 +254,11 @@ static u32 aspeed_i2c_slave_irq(struct aspeed_i2c_bus *bus, u32 irq_status)
 
         /* Slave was requested, restart state machine. */
         if (irq_status & ASPEED_I2CD_INTR_SLAVE_MATCH) {

Can you explain why you need to do this handing inside the SLAVE_MATCH case?

Quoted 7 lines

Could you instead move the TX_NAK handling to be above the SLAVE_MATCH case?

drivers/i2c/busses/i2c-aspeed.c static u32 aspeed_i2c_slave_irq(struct aspeed_i2c_bus *bus, u32 irq_status) · line 258

Quoted 2 lines
drivers/i2c/busses/i2c-aspeed.c +2 -0
+               if (irq_status & ASPEED_I2CD_INTR_TX_NAK &&
+                   bus->slave_state == ASPEED_I2C_SLAVE_READ_PROCESSED) {

Either way, this needs a comment to explain what we're working around.

Let me explain with the two examples below in normal case and the case where this patch is for:

In normal case:
The first transaction is Slave send (Master read):

    20(addr) 03(singlepart read) 03 1c 2e d5

Then the second Master write follow as below:

    20(addr) 02(singlepart write) 02 18 08 59

The irq will raise in sequence below:

irq data from-state to-state

00000084  20    INACTIVE        WRITE_RECEIVED
00000004  03    WRITE_RECEIVED  WRITE_RECEIVED <= RX_DONE
00000084  03    WRITE_RECEIVED  READ_PROCESSED
00000001  1c    READ_PROCESSED  READ_PROCESSED <= TX_ACK
00000001  2e    READ_PROCESSED  READ_PROCESSED
00000001  d5    READ_PROCESSED  READ_PROCESSED
00000002  xx    READ_PROCESSED  INACTIVE       <= TX_NAK
00000084  20    INACTIVE        WRITE_RECEIVED <= SLAVE_MATCH & RX_DONE
00000004  02    WRITE_RECEIVED  WRITE_RECEIVED
00000084  02    WRITE_RECEIVED  WRITE_RECEIVED
00000004  18    WRITE_RECEIVED  WRITE_RECEIVED
00000004  08    WRITE_RECEIVED  WRITE_RECEIVED
00000004  59    WRITE_RECEIVED  WRITE_RECEIVED
00000010  xx    WRITE_RECEIVED  INACTIVE

But sometimes:
The first transaction is Slave send (Master read):

    20(addr) 03(singlepart read) 03 1c 42 cc a5

Then the second Master write follow as below:

    20(addr) 02(singlepart write) 03 18 42 0c 63

The irq will raise in sequence below:

irq data from-state to-state

00000084  20    INACTIVE        WRITE_RECEIVED
00000004  03    WRITE_RECEIVED  WRITE_RECEIVED
00000084  03    WRITE_RECEIVED  READ_PROCESSED
00000001  1c    READ_PROCESSED  READ_PROCESSED
00000001  42    READ_PROCESSED  READ_PROCESSED
00000001  0c    READ_PROCESSED  READ_PROCESSED
00000001  63    READ_PROCESSED  READ_PROCESSED
00000086  20    READ_PROCESSED  WRITE_RECEIVED <= both 3 irqs raised
00000004  02    WRITE_RECEIVED  WRITE_RECEIVED
00000084  03    WRITE_RECEIVED  WRITE_RECEIVED
00000004  18    WRITE_RECEIVED  WRITE_RECEIVED
00000004  42    WRITE_RECEIVED  WRITE_RECEIVED
00000004  0c    WRITE_RECEIVED  WRITE_RECEIVED
00000004  63    WRITE_RECEIVED  WRITE_RECEIVED
00000010  xx    WRITE_RECEIVED  INACTIVE

This patch is to address this case where TX_NAK, SLAVE_MATCH and RX_DONE are raised together.

Quoted 9 lines
Quoted 9 lines
drivers/i2c/busses/i2c-aspeed.c +3 -1
+                       irq_handled |= ASPEED_I2CD_INTR_TX_NAK;
+                       i2c_slave_event(slave, I2C_SLAVE_STOP, &value);
+               }
                 irq_handled |= ASPEED_I2CD_INTR_SLAVE_MATCH;
                 bus->slave_state = ASPEED_I2C_SLAVE_START;
         }
--

2.28.0

#11 lore raw

PATCH v3 4/7 i2c: aspeed: Acknowledge Tx done w/wo ACK irq late

In reply to Quan Nguyen

With Tx done w/wo ACK are ack'ed early at beginning of irq handler, it is observed that, usually, the Tx done with Ack irq raises in the READ REQUESTED state. This is unexpected and complaint as below appear:
"Unexpected Ack on read request"

Assumed that Tx done should only be ack'ed once it was truly processed, switch to late ack'ed this two irqs and seen this issue go away through test with AST2500..

---
v3:
+ First introduce in v3 [Quan]

 drivers/i2c/busses/i2c-aspeed.c | 26 ++++++++++++++++++--------
 1 file changed, 18 insertions(+), 8 deletions(-)
drivers/i2c/busses/i2c-aspeed.c +18 -8
index 3fb37c3f23d4..b2e9c8f0ddf7 100644
@@ -606,8 +606,12 @@ static irqreturn_t aspeed_i2c_bus_irq(int irq, void *dev_id)
 
 	spin_lock(&bus->lock);
 	irq_received = readl(bus->base + ASPEED_I2C_INTR_STS_REG);
-	/* Ack all interrupts except for Rx done */
-	writel(irq_received & ~ASPEED_I2CD_INTR_RX_DONE,
+	/*
+	 * Ack all interrupts except for Rx done and
+	 * Tx done with/without ACK
+	 */
+	writel(irq_received &
+	       ~(ASPEED_I2CD_INTR_RX_DONE | ASPEED_I2CD_INTR_TX_ACK | ASPEED_I2CD_INTR_TX_NAK),
 	       bus->base + ASPEED_I2C_INTR_STS_REG);
 	readl(bus->base + ASPEED_I2C_INTR_STS_REG);
 	irq_received &= ASPEED_I2CD_INTR_RECV_MASK;
@@ -652,12 +656,18 @@ static irqreturn_t aspeed_i2c_bus_irq(int irq, void *dev_id)
 			"irq handled != irq. expected 0x%08x, but was 0x%08x\n",
 			irq_received, irq_handled);
 
-	/* Ack Rx done */
-	if (irq_received & ASPEED_I2CD_INTR_RX_DONE) {
-		writel(ASPEED_I2CD_INTR_RX_DONE,
-		       bus->base + ASPEED_I2C_INTR_STS_REG);
-		readl(bus->base + ASPEED_I2C_INTR_STS_REG);
-	}
+	/* Ack Rx done and Tx done with/without ACK */
+	/* Note: Re-use irq_handled variable */
+	irq_handled = 0;
+	if (irq_received & ASPEED_I2CD_INTR_RX_DONE)
+		irq_handled |= ASPEED_I2CD_INTR_RX_DONE;
+	if (irq_received & ASPEED_I2CD_INTR_TX_ACK)
+		irq_handled |= ASPEED_I2CD_INTR_TX_ACK;
+	if (irq_received & ASPEED_I2CD_INTR_TX_NAK)
+		irq_handled |= ASPEED_I2CD_INTR_TX_NAK;
+	writel(irq_handled, bus->base + ASPEED_I2C_INTR_STS_REG);
+	readl(bus->base + ASPEED_I2C_INTR_STS_REG);
+
 	spin_unlock(&bus->lock);
 	return irq_remaining ? IRQ_NONE : IRQ_HANDLED;
 }
Signature · 1 line
2.28.0
maintainer
#12 lore raw

Re: [PATCH v3 4/7] i2c: aspeed: Acknowledge Tx done w/wo ACK irq late

Quoted 2 lines · Quan Nguyen

With Tx done w/wo ACK are ack'ed early at beginning of irq handler,

Is w/wo a typo? If not, please write the full words ("with and without")

Quoted 7 lines

it is observed that, usually, the Tx done with Ack irq raises in the READ REQUESTED state. This is unexpected and complaint as below appear:
"Unexpected Ack on read request"

Assumed that Tx done should only be ack'ed once it was truly processed, switch to late ack'ed this two irqs and seen this issue go away through test with AST2500..

Please read Guneter's commit message 2be6b47211e17e6c90ead40d24d2a5cc815f2d5c to confirm that your changes do not invalidate the fix that they made. Add them to CC for review.

Again, this is a fix that is independent of the ssif work. Please send it separately with a Fixes line.

Quoted 14 lines

---
v3:
+ First introduce in v3 [Quan]

 drivers/i2c/busses/i2c-aspeed.c | 26 ++++++++++++++++++--------
 1 file changed, 18 insertions(+), 8 deletions(-)
drivers/i2c/busses/i2c-aspeed.c +0 -0
index 3fb37c3f23d4..b2e9c8f0ddf7 100644

drivers/i2c/busses/i2c-aspeed.c static irqreturn_t aspeed_i2c_bus_irq(int irq, void *dev_id) · line 611

Quoted 8 lines
drivers/i2c/busses/i2c-aspeed.c +3 -2
@@ -606,8 +606,12 @@ static irqreturn_t aspeed_i2c_bus_irq(int irq, void *dev_id)
 
        spin_lock(&bus->lock);
        irq_received = readl(bus->base + ASPEED_I2C_INTR_STS_REG);
-       /* Ack all interrupts except for Rx done */
-       writel(irq_received & ~ASPEED_I2CD_INTR_RX_DONE,
+       /*
+        * Ack all interrupts except for Rx done and
+        * Tx done with/without ACK

Nit: this comment can be on one line.

Quoted 8 lines
drivers/i2c/busses/i2c-aspeed.c +3 -0
+        */
+       writel(irq_received &
+              ~(ASPEED_I2CD_INTR_RX_DONE | ASPEED_I2CD_INTR_TX_ACK | ASPEED_I2CD_INTR_TX_NAK),
               bus->base + ASPEED_I2C_INTR_STS_REG);
        readl(bus->base + ASPEED_I2C_INTR_STS_REG);
        irq_received &= ASPEED_I2CD_INTR_RECV_MASK;
@@ -652,12 +656,18 @@ static irqreturn_t aspeed_i2c_bus_irq(int irq, void *dev_id)
                        "irq handled != irq. expected 0x%08x, but was 0x%08x\n",

drivers/i2c/busses/i2c-aspeed.c static irqreturn_t aspeed_i2c_bus_irq(int irq, void *dev_id) · line 660

Quoted 10 lines
drivers/i2c/busses/i2c-aspeed.c +2 -6
                        irq_received, irq_handled);
 
-       /* Ack Rx done */
-       if (irq_received & ASPEED_I2CD_INTR_RX_DONE) {
-               writel(ASPEED_I2CD_INTR_RX_DONE,
-                      bus->base + ASPEED_I2C_INTR_STS_REG);
-               readl(bus->base + ASPEED_I2C_INTR_STS_REG);
-       }
+       /* Ack Rx done and Tx done with/without ACK */
+       /* Note: Re-use irq_handled variable */

I'm not sure what this note means.

drivers/i2c/busses/i2c-aspeed.c static irqreturn_t aspeed_i2c_bus_irq(int irq, void *dev_id) · line 668

Quoted 8 lines
drivers/i2c/busses/i2c-aspeed.c +8 -0
+       irq_handled = 0;
+       if (irq_received & ASPEED_I2CD_INTR_RX_DONE)
+               irq_handled |= ASPEED_I2CD_INTR_RX_DONE;
+       if (irq_received & ASPEED_I2CD_INTR_TX_ACK)
+               irq_handled |= ASPEED_I2CD_INTR_TX_ACK;
+       if (irq_received & ASPEED_I2CD_INTR_TX_NAK)
+               irq_handled |= ASPEED_I2CD_INTR_TX_NAK;
+       writel(irq_handled, bus->base + ASPEED_I2C_INTR_STS_REG);

Are you intentionally only acking the bits that are set when we read from STS_REG at the start of the handler? If not, we could write this instead:

writel(ASPEED_I2CD_INTR_RX_DONE | ASPEED_I2CD_INTR_TX_ACK | ASPEED_I2CD_INTR_TX_NAK,

        bus->base + ASPEED_I2C_INTR_STS_REG);

If you only want to ack the bits that are set, then do this:

writel(irq_received &
(ASPEED_I2CD_INTR_RX_DONE | ASPEED_I2CD_INTR_TX_ACK |
ASPEED_I2CD_INTR_TX_NAK),

         bus->base + ASPEED_I2C_INTR_STS_REG);

That way, you can avoid all of the tests.

drivers/i2c/busses/i2c-aspeed.c static irqreturn_t aspeed_i2c_bus_irq(int irq, void *dev_id) · line 669

Quoted 1 line
drivers/i2c/busses/i2c-aspeed.c +1 -0
+       readl(bus->base + ASPEED_I2C_INTR_STS_REG);

When you move this, please add a comment that reminds us why we do a write-then-read (see commit c926c87b8e36dcc0ea5c2a0a0227ed4f32d0516a).

Quoted 7 lines
drivers/i2c/busses/i2c-aspeed.c +1 -1
+
        spin_unlock(&bus->lock);
        return irq_remaining ? IRQ_NONE : IRQ_HANDLED;
 }
--

2.28.0

#13 lore raw

Re: [PATCH v3 4/7] i2c: aspeed: Acknowledge Tx done w/wo ACK irq late

Quoted 10 lines · Joel Stanley
Quoted 2 lines · Quan Nguyen

With Tx done w/wo ACK are ack'ed early at beginning of irq handler,

Is w/wo a typo? If not, please write the full words ("with and without")

Quoted 4 lines

it is observed that, usually, the Tx done with Ack irq raises in the READ REQUESTED state. This is unexpected and complaint as below appear:
"Unexpected Ack on read request"

Quoted 8 lines
Quoted 3 lines

Assumed that Tx done should only be ack'ed once it was truly processed, switch to late ack'ed this two irqs and seen this issue go away through test with AST2500..

Please read Guneter's commit message 2be6b47211e17e6c90ead40d24d2a5cc815f2d5c to confirm that your changes do not invalidate the fix that they made. Add them to CC for review.

This might re-introduce a race condition if the code that is handling Tx done sends another byte without acknowledging the original interrupt, and another Tx done (or Tx nack) interrupt is received before the interrupt handler returns. If that happens, the second Tx done interrupt would be acknowledged but not be handled, and transmit would stall. That may well be what I had observed at the time but it is too long ago to remember, sorry.

Quoted 59 lines

Again, this is a fix that is independent of the ssif work. Please send it separately with a Fixes line.

Quoted 14 lines

---
v3:
+ First introduce in v3 [Quan]

  drivers/i2c/busses/i2c-aspeed.c | 26 ++++++++++++++++++--------
  1 file changed, 18 insertions(+), 8 deletions(-)
drivers/i2c/busses/i2c-aspeed.c +0 -0
index 3fb37c3f23d4..b2e9c8f0ddf7 100644

drivers/i2c/busses/i2c-aspeed.c static irqreturn_t aspeed_i2c_bus_irq(int irq, void *dev_id) · line 611

Quoted 8 lines
drivers/i2c/busses/i2c-aspeed.c +3 -2
@@ -606,8 +606,12 @@ static irqreturn_t aspeed_i2c_bus_irq(int irq, void *dev_id)
 
         spin_lock(&bus->lock);
         irq_received = readl(bus->base + ASPEED_I2C_INTR_STS_REG);
-       /* Ack all interrupts except for Rx done */
-       writel(irq_received & ~ASPEED_I2CD_INTR_RX_DONE,
+       /*
+        * Ack all interrupts except for Rx done and
+        * Tx done with/without ACK

Nit: this comment can be on one line.

Quoted 8 lines
drivers/i2c/busses/i2c-aspeed.c +3 -0
+        */
+       writel(irq_received &
+              ~(ASPEED_I2CD_INTR_RX_DONE | ASPEED_I2CD_INTR_TX_ACK | ASPEED_I2CD_INTR_TX_NAK),
                bus->base + ASPEED_I2C_INTR_STS_REG);
         readl(bus->base + ASPEED_I2C_INTR_STS_REG);
         irq_received &= ASPEED_I2CD_INTR_RECV_MASK;
@@ -652,12 +656,18 @@ static irqreturn_t aspeed_i2c_bus_irq(int irq, void *dev_id)
                         "irq handled != irq. expected 0x%08x, but was 0x%08x\n",

drivers/i2c/busses/i2c-aspeed.c static irqreturn_t aspeed_i2c_bus_irq(int irq, void *dev_id) · line 660

Quoted 10 lines
drivers/i2c/busses/i2c-aspeed.c +2 -6
                         irq_received, irq_handled);
 
-       /* Ack Rx done */
-       if (irq_received & ASPEED_I2CD_INTR_RX_DONE) {
-               writel(ASPEED_I2CD_INTR_RX_DONE,
-                      bus->base + ASPEED_I2C_INTR_STS_REG);
-               readl(bus->base + ASPEED_I2C_INTR_STS_REG);
-       }
+       /* Ack Rx done and Tx done with/without ACK */
+       /* Note: Re-use irq_handled variable */

I'm not sure what this note means.

drivers/i2c/busses/i2c-aspeed.c static irqreturn_t aspeed_i2c_bus_irq(int irq, void *dev_id) · line 668

Quoted 8 lines
drivers/i2c/busses/i2c-aspeed.c +8 -0
+       irq_handled = 0;
+       if (irq_received & ASPEED_I2CD_INTR_RX_DONE)
+               irq_handled |= ASPEED_I2CD_INTR_RX_DONE;
+       if (irq_received & ASPEED_I2CD_INTR_TX_ACK)
+               irq_handled |= ASPEED_I2CD_INTR_TX_ACK;
+       if (irq_received & ASPEED_I2CD_INTR_TX_NAK)
+               irq_handled |= ASPEED_I2CD_INTR_TX_NAK;
+       writel(irq_handled, bus->base + ASPEED_I2C_INTR_STS_REG);
Quoted 8 lines

Are you intentionally only acking the bits that are set when we read from STS_REG at the start of the handler? If not, we could write this instead:

writel(ASPEED_I2CD_INTR_RX_DONE | ASPEED_I2CD_INTR_TX_ACK | ASPEED_I2CD_INTR_TX_NAK,

         bus->base + ASPEED_I2C_INTR_STS_REG);

This would clear those bits unconditionally even if they were not handled.

Quoted 9 lines

If you only want to ack the bits that are set, then do this:

writel(irq_received &
(ASPEED_I2CD_INTR_RX_DONE | ASPEED_I2CD_INTR_TX_ACK |
ASPEED_I2CD_INTR_TX_NAK),

          bus->base + ASPEED_I2C_INTR_STS_REG);

That way, you can avoid all of the tests.

Or

	irq_handled = irq_received &
		(ASPEED_I2CD_INTR_RX_DONE | ASPEED_I2CD_INTR_TX_ACK | ASPEED_I2CD_INTR_TX_NAK);
	writel(irq_handled, bus->base + ASPEED_I2C_INTR_STS_REG);

if the idea was to avoid the long statement.

Guenter

Quoted 12 lines

drivers/i2c/busses/i2c-aspeed.c static irqreturn_t aspeed_i2c_bus_irq(int irq, void *dev_id) · line 669

Quoted 1 line
drivers/i2c/busses/i2c-aspeed.c +1 -0
+       readl(bus->base + ASPEED_I2C_INTR_STS_REG);

When you move this, please add a comment that reminds us why we do a write-then-read (see commit c926c87b8e36dcc0ea5c2a0a0227ed4f32d0516a).

Quoted 7 lines
drivers/i2c/busses/i2c-aspeed.c +1 -1
+
         spin_unlock(&bus->lock);
         return irq_remaining ? IRQ_NONE : IRQ_HANDLED;
  }
--

2.28.0

#14 lore raw

Re: [PATCH v3 4/7] i2c: aspeed: Acknowledge Tx done w/wo ACK irq late

Quoted 21 lines · Guenter Roeck
Quoted 11 lines · Joel Stanley
Quoted 2 lines · Quan Nguyen

With Tx done w/wo ACK are ack'ed early at beginning of irq handler,

Is w/wo a typo? If not, please write the full words ("with and without")

Quoted 4 lines

it is observed that, usually, the Tx done with Ack irq raises in the READ REQUESTED state. This is unexpected and complaint as below appear:
"Unexpected Ack on read request"

Quoted 8 lines
Quoted 3 lines

Assumed that Tx done should only be ack'ed once it was truly processed, switch to late ack'ed this two irqs and seen this issue go away through test with AST2500..

Please read Guneter's commit message 2be6b47211e17e6c90ead40d24d2a5cc815f2d5c to confirm that your changes do not invalidate the fix that they made.? Add them to CC for review.

Quoted 7 lines

This might re-introduce a race condition if the code that is handling Tx done sends another byte without acknowledging the original interrupt, and another Tx done (or Tx nack) interrupt is received before the interrupt handler returns. If that happens, the second Tx done interrupt would be acknowledged but not be handled, and transmit would stall. That may well be what I had observed at the time but it is too long ago to remember, sorry.

My assumption is that HW will start sending another byte as soon as the Tx done of the previous byte is ack'ed. So if it was ack'ed early before the slave irq handler to actually prepare the next byte to send, garbage value might be sent out instead.
I'm not sure if my assumption is correct but this patch do help in this case.
Please help with your comment.

Quoted 90 lines
Quoted 64 lines

Again, this is a fix that is independent of the ssif work. Please send it separately with a Fixes line.

Quoted 15 lines

---
v3:
?? + First introduce in v3 [Quan]

? drivers/i2c/busses/i2c-aspeed.c | 26 ++++++++++++++++++-------- ? 1 file changed, 18 insertions(+), 8 deletions(-)

drivers/i2c/busses/i2c-aspeed.c +0 -0
index 3fb37c3f23d4..b2e9c8f0ddf7 100644

drivers/i2c/busses/i2c-aspeed.c static irqreturn_t aspeed_i2c_bus_irq(int irq, · line 612

Quoted 9 lines
drivers/i2c/busses/i2c-aspeed.c +3 -2
@@ -606,8 +606,12 @@ static irqreturn_t aspeed_i2c_bus_irq(int irq, 
 void *dev_id)
 
 ???????? spin_lock(&bus->lock);
 ???????? irq_received = readl(bus->base + ASPEED_I2C_INTR_STS_REG);
-?????? /* Ack all interrupts except for Rx done */
-?????? writel(irq_received & ~ASPEED_I2CD_INTR_RX_DONE,
+?????? /*
+??????? * Ack all interrupts except for Rx done and
+??????? * Tx done with/without ACK

Nit: this comment can be on one line.

Quoted 13 lines
drivers/i2c/busses/i2c-aspeed.c +3 -0
+??????? */
+?????? writel(irq_received &
+????????????? ~(ASPEED_I2CD_INTR_RX_DONE | ASPEED_I2CD_INTR_TX_ACK | 
 ASPEED_I2CD_INTR_TX_NAK),
 ??????????????? bus->base + ASPEED_I2C_INTR_STS_REG);

???????? readl(bus->base + ASPEED_I2C_INTR_STS_REG); ???????? irq_received &= ASPEED_I2CD_INTR_RECV_MASK;

drivers/i2c/busses/i2c-aspeed.c +0 -0
@@ -652,12 +656,18 @@ static irqreturn_t aspeed_i2c_bus_irq(int irq, 
 void *dev_id)

???????????????????????? "irq handled != irq. expected 0x%08x, but was 0x%08x\n",
???????????????????????? irq_received, irq_handled);

Quoted 8 lines

-?????? /* Ack Rx done */
-?????? if (irq_received & ASPEED_I2CD_INTR_RX_DONE) { -?????????????? writel(ASPEED_I2CD_INTR_RX_DONE, -????????????????????? bus->base + ASPEED_I2C_INTR_STS_REG); -?????????????? readl(bus->base + ASPEED_I2C_INTR_STS_REG); -?????? }
+?????? /* Ack Rx done and Tx done with/without ACK */ +?????? /* Note: Re-use irq_handled variable */

I'm not sure what this note means.

Quoted 8 lines

+?????? irq_handled = 0;
+?????? if (irq_received & ASPEED_I2CD_INTR_RX_DONE) +?????????????? irq_handled |= ASPEED_I2CD_INTR_RX_DONE; +?????? if (irq_received & ASPEED_I2CD_INTR_TX_ACK) +?????????????? irq_handled |= ASPEED_I2CD_INTR_TX_ACK; +?????? if (irq_received & ASPEED_I2CD_INTR_TX_NAK) +?????????????? irq_handled |= ASPEED_I2CD_INTR_TX_NAK; +?????? writel(irq_handled, bus->base + ASPEED_I2C_INTR_STS_REG);

Quoted 8 lines

Are you intentionally only acking the bits that are set when we read from STS_REG at the start of the handler? If not, we could write this instead:

writel(ASPEED_I2CD_INTR_RX_DONE | ASPEED_I2CD_INTR_TX_ACK | ASPEED_I2CD_INTR_TX_NAK,
???????? bus->base + ASPEED_I2C_INTR_STS_REG);

This would clear those bits unconditionally even if they were not handled.

Quoted 9 lines

If you only want to ack the bits that are set, then do this:

?? writel(irq_received &
???????????? (ASPEED_I2CD_INTR_RX_DONE | ASPEED_I2CD_INTR_TX_ACK | ASPEED_I2CD_INTR_TX_NAK),
????????? bus->base + ASPEED_I2C_INTR_STS_REG);

That way, you can avoid all of the tests.

Or
????irq_handled = irq_received &
??????? (ASPEED_I2CD_INTR_RX_DONE | ASPEED_I2CD_INTR_TX_ACK |
ASPEED_I2CD_INTR_TX_NAK);
????writel(irq_handled, bus->base + ASPEED_I2C_INTR_STS_REG);

Quoted 4 lines

if the idea was to avoid the long statement.

Guenter

Thanks Guenter,
Will apply in next version.

Quoted 13 lines
Quoted 12 lines
Quoted 1 line

+?????? readl(bus->base + ASPEED_I2C_INTR_STS_REG);

When you move this, please add a comment that reminds us why we do a write-then-read (see commit c926c87b8e36dcc0ea5c2a0a0227ed4f32d0516a).

Quoted 7 lines

+
???????? spin_unlock(&bus->lock);
???????? return irq_remaining ? IRQ_NONE : IRQ_HANDLED; ? }

Signature · 1 line
2.28.0
#15 lore raw

Re: [PATCH v3 4/7] i2c: aspeed: Acknowledge Tx done w/wo ACK irq late

In reply to Joel Stanley
Quoted 6 lines · Joel Stanley
Quoted 2 lines · Quan Nguyen

With Tx done w/wo ACK are ack'ed early at beginning of irq handler,

Is w/wo a typo? If not, please write the full words ("with and without")

It is "with and without", will fix in next version

Quoted 8 lines
Quoted 7 lines

it is observed that, usually, the Tx done with Ack irq raises in the READ REQUESTED state. This is unexpected and complaint as below appear:
"Unexpected Ack on read request"

Assumed that Tx done should only be ack'ed once it was truly processed, switch to late ack'ed this two irqs and seen this issue go away through test with AST2500..

Quoted 7 lines

Please read Guneter's commit message 2be6b47211e17e6c90ead40d24d2a5cc815f2d5c to confirm that your changes do not invalidate the fix that they made. Add them to CC for review.

Again, this is a fix that is independent of the ssif work. Please send it separately with a Fixes line.

Will do and separate this patch into other series in next version.

Quoted 15 lines
Quoted 15 lines

---
v3:
+ First introduce in v3 [Quan]

  drivers/i2c/busses/i2c-aspeed.c | 26 ++++++++++++++++++--------
  1 file changed, 18 insertions(+), 8 deletions(-)
drivers/i2c/busses/i2c-aspeed.c +0 -0
index 3fb37c3f23d4..b2e9c8f0ddf7 100644
@@ -606,8 +606,12 @@ static irqreturn_t aspeed_i2c_bus_irq(int irq, void *dev_id)
 
Quoted 10 lines

drivers/i2c/busses/i2c-aspeed.c static irqreturn_t aspeed_i2c_bus_irq(int irq, void *dev_id) · line 611

Quoted 7 lines
drivers/i2c/busses/i2c-aspeed.c +3 -2
         spin_lock(&bus->lock);
         irq_received = readl(bus->base + ASPEED_I2C_INTR_STS_REG);
-       /* Ack all interrupts except for Rx done */
-       writel(irq_received & ~ASPEED_I2CD_INTR_RX_DONE,
+       /*
+        * Ack all interrupts except for Rx done and
+        * Tx done with/without ACK

Nit: this comment can be on one line.

Thanks, will fix.

Quoted 39 lines
Quoted 9 lines
drivers/i2c/busses/i2c-aspeed.c +3 -0
+        */
+       writel(irq_received &
+              ~(ASPEED_I2CD_INTR_RX_DONE | ASPEED_I2CD_INTR_TX_ACK | ASPEED_I2CD_INTR_TX_NAK),
                bus->base + ASPEED_I2C_INTR_STS_REG);
         readl(bus->base + ASPEED_I2C_INTR_STS_REG);
         irq_received &= ASPEED_I2CD_INTR_RECV_MASK;
@@ -652,12 +656,18 @@ static irqreturn_t aspeed_i2c_bus_irq(int irq, void *dev_id)
                         "irq handled != irq. expected 0x%08x, but was 0x%08x\n",

drivers/i2c/busses/i2c-aspeed.c static irqreturn_t aspeed_i2c_bus_irq(int irq, void *dev_id) · line 660

Quoted 10 lines
drivers/i2c/busses/i2c-aspeed.c +2 -6
                         irq_received, irq_handled);
 
-       /* Ack Rx done */
-       if (irq_received & ASPEED_I2CD_INTR_RX_DONE) {
-               writel(ASPEED_I2CD_INTR_RX_DONE,
-                      bus->base + ASPEED_I2C_INTR_STS_REG);
-               readl(bus->base + ASPEED_I2C_INTR_STS_REG);
-       }
+       /* Ack Rx done and Tx done with/without ACK */
+       /* Note: Re-use irq_handled variable */

I'm not sure what this note means.

drivers/i2c/busses/i2c-aspeed.c static irqreturn_t aspeed_i2c_bus_irq(int irq, void *dev_id) · line 668

Quoted 8 lines
drivers/i2c/busses/i2c-aspeed.c +8 -0
+       irq_handled = 0;
+       if (irq_received & ASPEED_I2CD_INTR_RX_DONE)
+               irq_handled |= ASPEED_I2CD_INTR_RX_DONE;
+       if (irq_received & ASPEED_I2CD_INTR_TX_ACK)
+               irq_handled |= ASPEED_I2CD_INTR_TX_ACK;
+       if (irq_received & ASPEED_I2CD_INTR_TX_NAK)
+               irq_handled |= ASPEED_I2CD_INTR_TX_NAK;
+       writel(irq_handled, bus->base + ASPEED_I2C_INTR_STS_REG);

Are you intentionally only acking the bits that are set when we read from STS_REG at the start of the handler? If not, we could write this instead:

writel(ASPEED_I2CD_INTR_RX_DONE | ASPEED_I2CD_INTR_TX_ACK | ASPEED_I2CD_INTR_TX_NAK,

         bus->base + ASPEED_I2C_INTR_STS_REG);
Quoted 9 lines

If you only want to ack the bits that are set, then do this:

writel(irq_received &
(ASPEED_I2CD_INTR_RX_DONE | ASPEED_I2CD_INTR_TX_ACK |
ASPEED_I2CD_INTR_TX_NAK),

          bus->base + ASPEED_I2C_INTR_STS_REG);

That way, you can avoid all of the tests.

Thanks, will fix this in next version.

Quoted 5 lines

drivers/i2c/busses/i2c-aspeed.c static irqreturn_t aspeed_i2c_bus_irq(int irq, void *dev_id) · line 669

Quoted 1 line
drivers/i2c/busses/i2c-aspeed.c +1 -0
+       readl(bus->base + ASPEED_I2C_INTR_STS_REG);

When you move this, please add a comment that reminds us why we do a write-then-read (see commit c926c87b8e36dcc0ea5c2a0a0227ed4f32d0516a).

Will fix in next version.

Quoted 7 lines
Quoted 7 lines
drivers/i2c/busses/i2c-aspeed.c +1 -1
+
         spin_unlock(&bus->lock);
         return irq_remaining ? IRQ_NONE : IRQ_HANDLED;
  }
--

2.28.0

#16 lore raw

PATCH v3 5/7 i2c: aspeed: Add aspeed_set_slave_busy()

In reply to Quan Nguyen

Slave i2c device on AST2500 received a lot of slave irq while it is busy processing the response. To handle this case, adds and exports aspeed_set_slave_busy() for controller to temporary stop slave irq while slave is handling the response, and re-enable them again when the response is ready.

---
v3:
+ First introduce in v3 [Quan]

 drivers/i2c/busses/i2c-aspeed.c | 20 ++++++++++++++++++++
 1 file changed, 20 insertions(+)
drivers/i2c/busses/i2c-aspeed.c +20 -0
index b2e9c8f0ddf7..9926d04831a2 100644
@@ -944,6 +944,26 @@ static int aspeed_i2c_init(struct aspeed_i2c_bus *bus,
 	return 0;
 }
 
+#if IS_ENABLED(CONFIG_I2C_SLAVE)
+void aspeed_set_slave_busy(struct i2c_adapter *adap, bool busy)
+{
+	struct aspeed_i2c_bus *bus = i2c_get_adapdata(adap);
+	unsigned long current_mask, flags;
+
+	spin_lock_irqsave(&bus->lock, flags);
+
+	current_mask = readl(bus->base + ASPEED_I2C_INTR_CTRL_REG);
+	if (busy)
+		current_mask &= ~(ASPEED_I2CD_INTR_RX_DONE | ASPEED_I2CD_INTR_SLAVE_MATCH);
+	else
+		current_mask |= ASPEED_I2CD_INTR_RX_DONE | ASPEED_I2CD_INTR_SLAVE_MATCH;
+	writel(current_mask, bus->base + ASPEED_I2C_INTR_CTRL_REG);
+
+	spin_unlock_irqrestore(&bus->lock, flags);
+}
+EXPORT_SYMBOL_GPL(aspeed_set_slave_busy);
+#endif
+
 static int aspeed_i2c_reset(struct aspeed_i2c_bus *bus)
 {
 	struct platform_device *pdev = to_platform_device(bus->dev);
Signature · 1 line
2.28.0
maintainer
#17 lore raw

RE: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

Quoted 63 lines

-----Original Message-----
From: openbmc <[email protected]> On Behalf Of Quan Nguyen
Sent: Wednesday, May 19, 2021 3:50 PM
To: Corey Minyard <[email protected]>; Rob Herring <[email protected]>; Joel Stanley <[email protected]>; Andrew Jeffery <[email protected]>; Brendan Higgins <[email protected]>; Benjamin Herrenschmidt <[email protected]>; Wolfram Sang <[email protected]>; Philipp Zabel <[email protected]>; openipmi-developer at lists.sourceforge.net; devicetree at vger.kernel.org; linux-arm-kernel at lists.infradead.org; linux-aspeed at lists.ozlabs.org; linux-kernel at vger.kernel.org; linux-i2c at vger.kernel.org
Cc: Open Source Submission <[email protected]>; Thang Q . Nguyen <[email protected]>; Phong Vo <[email protected]>; openbmc at lists.ozlabs.org
Subject: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

Slave i2c device on AST2500 received a lot of slave irq while it is busy processing the response. To handle this case, adds and exports aspeed_set_slave_busy() for controller to temporary stop slave irq while slave is handling the response, and re-enable them again when the response is ready.

---
v3:
+ First introduce in v3 [Quan]

 drivers/i2c/busses/i2c-aspeed.c | 20 ++++++++++++++++++++
 1 file changed, 20 insertions(+)
drivers/i2c/busses/i2c-aspeed.c +18 -1
index b2e9c8f0ddf7..9926d04831a2 100644
@@ -944,6 +944,26 @@ static int aspeed_i2c_init(struct aspeed_i2c_bus
 *bus,
 	return 0;
 }
 
+#if IS_ENABLED(CONFIG_I2C_SLAVE)
+void aspeed_set_slave_busy(struct i2c_adapter *adap, bool busy) {
+	struct aspeed_i2c_bus *bus = i2c_get_adapdata(adap);
+	unsigned long current_mask, flags;
+
+	spin_lock_irqsave(&bus->lock, flags);
+
+	current_mask = readl(bus->base + ASPEED_I2C_INTR_CTRL_REG);
+	if (busy)
+		current_mask &= ~(ASPEED_I2CD_INTR_RX_DONE |
 ASPEED_I2CD_INTR_SLAVE_MATCH);
+	else
+		current_mask |= ASPEED_I2CD_INTR_RX_DONE |
 ASPEED_I2CD_INTR_SLAVE_MATCH;
+	writel(current_mask, bus->base + ASPEED_I2C_INTR_CTRL_REG);
+
+	spin_unlock_irqrestore(&bus->lock, flags); }
+EXPORT_SYMBOL_GPL(aspeed_set_slave_busy);
+#endif
+
 static int aspeed_i2c_reset(struct aspeed_i2c_bus *bus)  {
 	struct platform_device *pdev = to_platform_device(bus->dev);
--
Quoted 1 line

2.28.0

Hello,
The better idea is use disable i2c slave mode.
Due to if i2c controller running in slave will get slave match, and latch the SCL. Until cpu clear interrupt status.
Ryan

#18 lore raw

Re: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

Quoted 62 lines · Ryan Chen
Quoted 62 lines

-----Original Message-----
From: openbmc <[email protected]> On Behalf Of Quan Nguyen
Sent: Wednesday, May 19, 2021 3:50 PM
To: Corey Minyard <[email protected]>; Rob Herring <[email protected]>; Joel Stanley <[email protected]>; Andrew Jeffery <[email protected]>; Brendan Higgins <[email protected]>; Benjamin Herrenschmidt <[email protected]>; Wolfram Sang <[email protected]>; Philipp Zabel <[email protected]>; openipmi-developer at lists.sourceforge.net; devicetree at vger.kernel.org; linux-arm-kernel at lists.infradead.org; linux-aspeed at lists.ozlabs.org; linux-kernel at vger.kernel.org; linux-i2c at vger.kernel.org
Cc: Open Source Submission <[email protected]>; Thang Q . Nguyen <[email protected]>; Phong Vo <[email protected]>; openbmc at lists.ozlabs.org
Subject: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

Slave i2c device on AST2500 received a lot of slave irq while it is busy processing the response. To handle this case, adds and exports aspeed_set_slave_busy() for controller to temporary stop slave irq while slave is handling the response, and re-enable them again when the response is ready.

---
v3:
+ First introduce in v3 [Quan]

  drivers/i2c/busses/i2c-aspeed.c | 20 ++++++++++++++++++++
  1 file changed, 20 insertions(+)
drivers/i2c/busses/i2c-aspeed.c +18 -0
index b2e9c8f0ddf7..9926d04831a2 100644
@@ -944,6 +944,26 @@ static int aspeed_i2c_init(struct aspeed_i2c_bus
 *bus,
  	return 0;
  }
 
+#if IS_ENABLED(CONFIG_I2C_SLAVE)
+void aspeed_set_slave_busy(struct i2c_adapter *adap, bool busy) {
+	struct aspeed_i2c_bus *bus = i2c_get_adapdata(adap);
+	unsigned long current_mask, flags;
+
+	spin_lock_irqsave(&bus->lock, flags);
+
+	current_mask = readl(bus->base + ASPEED_I2C_INTR_CTRL_REG);
+	if (busy)
+		current_mask &= ~(ASPEED_I2CD_INTR_RX_DONE |
 ASPEED_I2CD_INTR_SLAVE_MATCH);
+	else
+		current_mask |= ASPEED_I2CD_INTR_RX_DONE |
 ASPEED_I2CD_INTR_SLAVE_MATCH;
+	writel(current_mask, bus->base + ASPEED_I2C_INTR_CTRL_REG);
+
+	spin_unlock_irqrestore(&bus->lock, flags); }
+EXPORT_SYMBOL_GPL(aspeed_set_slave_busy);
+#endif
+
  static int aspeed_i2c_reset(struct aspeed_i2c_bus *bus)  {
  	struct platform_device *pdev = to_platform_device(bus->dev);
Quoted 9 lines
Quoted 2 lines

--
2.28.0

Hello,
The better idea is use disable i2c slave mode.
Due to if i2c controller running in slave will get slave match, and latch the SCL. Until cpu clear interrupt status.
Ryan

Thanks Ryan,

Do you mean to enable/disable slave function as per example code below ?

         /* Turn on slave mode. */
         func_ctrl_reg_val = readl(bus->base + ASPEED_I2C_FUN_CTRL_REG);
         func_ctrl_reg_val |= ASPEED_I2CD_SLAVE_EN;
         writel(func_ctrl_reg_val, bus->base + ASPEED_I2C_FUN_CTRL_REG);

Will try this idea.
- Quan

maintainer
#19 lore raw

RE: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

Quoted 90 lines

-----Original Message-----
From: Quan Nguyen <[email protected]>
Sent: Thursday, May 20, 2021 10:10 PM
To: Ryan Chen <[email protected]>; Corey Minyard <[email protected]>; Rob Herring <[email protected]>; Joel Stanley <[email protected]>; Andrew Jeffery <[email protected]>; Brendan Higgins <[email protected]>; Benjamin Herrenschmidt <[email protected]>; Wolfram Sang <[email protected]>; Philipp Zabel <[email protected]>; openipmi-developer at lists.sourceforge.net; devicetree at vger.kernel.org; linux-arm-kernel at lists.infradead.org; linux-aspeed at lists.ozlabs.org; linux-kernel at vger.kernel.org; linux-i2c at vger.kernel.org
Cc: Open Source Submission <[email protected]>; Thang Q . Nguyen <[email protected]>; Phong Vo <[email protected]>; openbmc at lists.ozlabs.org
Subject: Re: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

Quoted 3 lines · Ryan Chen
Quoted 3 lines

-----Original Message-----
From: openbmc <[email protected]> On

Behalf

Quoted 16 lines
Quoted 16 lines

Of Quan Nguyen
Sent: Wednesday, May 19, 2021 3:50 PM
To: Corey Minyard <[email protected]>; Rob Herring <[email protected]>; Joel Stanley <[email protected]>; Andrew Jeffery <[email protected]>; Brendan Higgins <[email protected]>; Benjamin Herrenschmidt <[email protected]>; Wolfram Sang <[email protected]>; Philipp Zabel <[email protected]>; openipmi-developer at lists.sourceforge.net;
devicetree at vger.kernel.org; linux-arm-kernel at lists.infradead.org; linux-aspeed at lists.ozlabs.org; linux-kernel at vger.kernel.org; linux-i2c at vger.kernel.org
Cc: Open Source Submission <[email protected]>; Thang Q . Nguyen <[email protected]>; Phong Vo <[email protected]>; openbmc at lists.ozlabs.org
Subject: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

Quoted 4 lines
Quoted 4 lines

Slave i2c device on AST2500 received a lot of slave irq while it is busy processing the response. To handle this case, adds and exports aspeed_set_slave_busy() for controller to temporary stop slave irq while slave is handling the response, and re-enable them again when the

response is ready.

Quoted 39 lines
Quoted 39 lines

---
v3:
+ First introduce in v3 [Quan]

  drivers/i2c/busses/i2c-aspeed.c | 20 ++++++++++++++++++++
  1 file changed, 20 insertions(+)
(quoted diff) +0 -0

b/drivers/i2c/busses/i2c-aspeed.c index b2e9c8f0ddf7..9926d04831a2 100644

drivers/i2c/busses/i2c-aspeed.c +18 -0
@@ -944,6 +944,26 @@ static int aspeed_i2c_init(struct aspeed_i2c_bus
 *bus,
  	return 0;
  }
 
+#if IS_ENABLED(CONFIG_I2C_SLAVE)
+void aspeed_set_slave_busy(struct i2c_adapter *adap, bool busy) {
+	struct aspeed_i2c_bus *bus = i2c_get_adapdata(adap);
+	unsigned long current_mask, flags;
+
+	spin_lock_irqsave(&bus->lock, flags);
+
+	current_mask = readl(bus->base + ASPEED_I2C_INTR_CTRL_REG);
+	if (busy)
+		current_mask &= ~(ASPEED_I2CD_INTR_RX_DONE |
 ASPEED_I2CD_INTR_SLAVE_MATCH);
+	else
+		current_mask |= ASPEED_I2CD_INTR_RX_DONE |
 ASPEED_I2CD_INTR_SLAVE_MATCH;
+	writel(current_mask, bus->base + ASPEED_I2C_INTR_CTRL_REG);
+
+	spin_unlock_irqrestore(&bus->lock, flags); }
+EXPORT_SYMBOL_GPL(aspeed_set_slave_busy);
+#endif
+
Quoted 8 lines
Quoted 4 lines

static int aspeed_i2c_reset(struct aspeed_i2c_bus *bus) {
struct platform_device *pdev = to_platform_device(bus->dev);
--
2.28.0

Hello,
The better idea is use disable i2c slave mode.
Due to if i2c controller running in slave will get slave match, and latch the

Quoted 7 lines

SCL.

Quoted 3 lines

Until cpu clear interrupt status.
Ryan

Thanks Ryan,

Do you mean to enable/disable slave function as per example code below ?

Yes. it is.

Quoted 10 lines
         /* Turn on slave mode. */
         func_ctrl_reg_val = readl(bus->base +
ASPEED_I2C_FUN_CTRL_REG);
         func_ctrl_reg_val |= ASPEED_I2CD_SLAVE_EN;
         writel(func_ctrl_reg_val, bus->base +

ASPEED_I2C_FUN_CTRL_REG);

Will try this idea.
- Quan

#20 lore raw

Re: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

Quoted 90 lines · Ryan Chen
Quoted 90 lines

-----Original Message-----
From: Quan Nguyen <[email protected]>
Sent: Thursday, May 20, 2021 10:10 PM
To: Ryan Chen <[email protected]>; Corey Minyard <[email protected]>; Rob Herring <[email protected]>; Joel Stanley <[email protected]>; Andrew Jeffery <[email protected]>; Brendan Higgins <[email protected]>; Benjamin Herrenschmidt <[email protected]>; Wolfram Sang <[email protected]>; Philipp Zabel <[email protected]>; openipmi-developer at lists.sourceforge.net; devicetree at vger.kernel.org; linux-arm-kernel at lists.infradead.org; linux-aspeed at lists.ozlabs.org; linux-kernel at vger.kernel.org; linux-i2c at vger.kernel.org
Cc: Open Source Submission <[email protected]>; Thang Q . Nguyen <[email protected]>; Phong Vo <[email protected]>; openbmc at lists.ozlabs.org
Subject: Re: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

Quoted 3 lines · Ryan Chen
Quoted 3 lines

-----Original Message-----
From: openbmc <[email protected]> On

Behalf

Quoted 16 lines
Quoted 16 lines

Of Quan Nguyen
Sent: Wednesday, May 19, 2021 3:50 PM
To: Corey Minyard <[email protected]>; Rob Herring <[email protected]>; Joel Stanley <[email protected]>; Andrew Jeffery <[email protected]>; Brendan Higgins <[email protected]>; Benjamin Herrenschmidt <[email protected]>; Wolfram Sang <[email protected]>; Philipp Zabel <[email protected]>; openipmi-developer at lists.sourceforge.net;
devicetree at vger.kernel.org; linux-arm-kernel at lists.infradead.org; linux-aspeed at lists.ozlabs.org; linux-kernel at vger.kernel.org; linux-i2c at vger.kernel.org
Cc: Open Source Submission <[email protected]>; Thang Q . Nguyen <[email protected]>; Phong Vo <[email protected]>; openbmc at lists.ozlabs.org
Subject: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

Quoted 4 lines
Quoted 4 lines

Slave i2c device on AST2500 received a lot of slave irq while it is busy processing the response. To handle this case, adds and exports aspeed_set_slave_busy() for controller to temporary stop slave irq while slave is handling the response, and re-enable them again when the

response is ready.

Quoted 47 lines
Quoted 42 lines

---
v3:
+ First introduce in v3 [Quan]

   drivers/i2c/busses/i2c-aspeed.c | 20 ++++++++++++++++++++
   1 file changed, 20 insertions(+)
(quoted diff) +0 -0

b/drivers/i2c/busses/i2c-aspeed.c index b2e9c8f0ddf7..9926d04831a2 100644

drivers/i2c/busses/i2c-aspeed.c +18 -1
@@ -944,6 +944,26 @@ static int aspeed_i2c_init(struct aspeed_i2c_bus
 *bus,
   	return 0;
   }
 
+#if IS_ENABLED(CONFIG_I2C_SLAVE)
+void aspeed_set_slave_busy(struct i2c_adapter *adap, bool busy) {
+	struct aspeed_i2c_bus *bus = i2c_get_adapdata(adap);
+	unsigned long current_mask, flags;
+
+	spin_lock_irqsave(&bus->lock, flags);
+
+	current_mask = readl(bus->base + ASPEED_I2C_INTR_CTRL_REG);
+	if (busy)
+		current_mask &= ~(ASPEED_I2CD_INTR_RX_DONE |
 ASPEED_I2CD_INTR_SLAVE_MATCH);
+	else
+		current_mask |= ASPEED_I2CD_INTR_RX_DONE |
 ASPEED_I2CD_INTR_SLAVE_MATCH;
+	writel(current_mask, bus->base + ASPEED_I2C_INTR_CTRL_REG);
+
+	spin_unlock_irqrestore(&bus->lock, flags); }
+EXPORT_SYMBOL_GPL(aspeed_set_slave_busy);
+#endif
+
   static int aspeed_i2c_reset(struct aspeed_i2c_bus *bus)  {
   	struct platform_device *pdev = to_platform_device(bus->dev);
--
Quoted 1 line

2.28.0

Hello,
The better idea is use disable i2c slave mode.
Due to if i2c controller running in slave will get slave match, and latch the

Quoted 8 lines
Quoted 7 lines

SCL.

Quoted 3 lines

Until cpu clear interrupt status.
Ryan

Thanks Ryan,

Do you mean to enable/disable slave function as per example code below ?

Yes. it is.

Dear Ryan,

This solution looks good. I'll switch to use this way in next version. Thanks for the suggestion.

- Quan

Quoted 10 lines
Quoted 10 lines
          /* Turn on slave mode. */
          func_ctrl_reg_val = readl(bus->base +
ASPEED_I2C_FUN_CTRL_REG);
          func_ctrl_reg_val |= ASPEED_I2CD_SLAVE_EN;
          writel(func_ctrl_reg_val, bus->base +

ASPEED_I2C_FUN_CTRL_REG);

Will try this idea.
- Quan

maintainer
#21 lore raw

RE: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

In reply to Quan Nguyen
Quoted 38 lines

-----Original Message-----
From: openbmc <[email protected]> On Behalf Of Quan Nguyen
Sent: Wednesday, May 19, 2021 3:50 PM
To: Corey Minyard <[email protected]>; Rob Herring <[email protected]>; Joel Stanley <[email protected]>; Andrew Jeffery <[email protected]>; Brendan Higgins <[email protected]>; Benjamin Herrenschmidt <[email protected]>; Wolfram Sang <[email protected]>; Philipp Zabel <[email protected]>; openipmi-developer at lists.sourceforge.net; devicetree at vger.kernel.org; linux-arm-kernel at lists.infradead.org; linux-aspeed at lists.ozlabs.org; linux-kernel at vger.kernel.org; linux-i2c at vger.kernel.org
Cc: Open Source Submission <[email protected]>; Thang Q . Nguyen <[email protected]>; Phong Vo <[email protected]>; openbmc at lists.ozlabs.org
Subject: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

Slave i2c device on AST2500 received a lot of slave irq while it is busy processing the response. To handle this case, adds and exports aspeed_set_slave_busy() for controller to temporary stop slave irq while slave is handling the response, and re-enable them again when the response is ready.

---
v3:
+ First introduce in v3 [Quan]

 drivers/i2c/busses/i2c-aspeed.c | 20 ++++++++++++++++++++
 1 file changed, 20 insertions(+)
drivers/i2c/busses/i2c-aspeed.c +0 -0
index b2e9c8f0ddf7..9926d04831a2 100644
@@ -944,6 +944,26 @@ static int aspeed_i2c_init(struct aspeed_i2c_bus
 *bus,
 	return 0;

drivers/i2c/busses/i2c-aspeed.c static int aspeed_i2c_init(struct aspeed_i2c_bus · line 955

Quoted 10 lines
drivers/i2c/busses/i2c-aspeed.c +8 -0
 }
 
+#if IS_ENABLED(CONFIG_I2C_SLAVE)
+void aspeed_set_slave_busy(struct i2c_adapter *adap, bool busy) {
+	struct aspeed_i2c_bus *bus = i2c_get_adapdata(adap);
+	unsigned long current_mask, flags;
+
+	spin_lock_irqsave(&bus->lock, flags);
+
+	current_mask = readl(bus->base + ASPEED_I2C_INTR_CTRL_REG);

Hello

	Where the bus->base to be remap?
Quoted 16 lines
drivers/i2c/busses/i2c-aspeed.c +10 -1
+	if (busy)
+		current_mask &= ~(ASPEED_I2CD_INTR_RX_DONE |
 ASPEED_I2CD_INTR_SLAVE_MATCH);
+	else
+		current_mask |= ASPEED_I2CD_INTR_RX_DONE |
 ASPEED_I2CD_INTR_SLAVE_MATCH;
+	writel(current_mask, bus->base + ASPEED_I2C_INTR_CTRL_REG);
+
+	spin_unlock_irqrestore(&bus->lock, flags); }
+EXPORT_SYMBOL_GPL(aspeed_set_slave_busy);
+#endif
+
 static int aspeed_i2c_reset(struct aspeed_i2c_bus *bus)  {
 	struct platform_device *pdev = to_platform_device(bus->dev);
--

2.28.0

#22 lore raw

Re: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

Quoted 49 lines · Ryan Chen
Quoted 38 lines

-----Original Message-----
From: openbmc <[email protected]> On Behalf Of Quan Nguyen
Sent: Wednesday, May 19, 2021 3:50 PM
To: Corey Minyard <[email protected]>; Rob Herring <[email protected]>; Joel Stanley <[email protected]>; Andrew Jeffery <[email protected]>; Brendan Higgins <[email protected]>; Benjamin Herrenschmidt <[email protected]>; Wolfram Sang <[email protected]>; Philipp Zabel <[email protected]>; openipmi-developer at lists.sourceforge.net; devicetree at vger.kernel.org; linux-arm-kernel at lists.infradead.org; linux-aspeed at lists.ozlabs.org; linux-kernel at vger.kernel.org; linux-i2c at vger.kernel.org
Cc: Open Source Submission <[email protected]>; Thang Q . Nguyen <[email protected]>; Phong Vo <[email protected]>; openbmc at lists.ozlabs.org
Subject: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

Slave i2c device on AST2500 received a lot of slave irq while it is busy processing the response. To handle this case, adds and exports aspeed_set_slave_busy() for controller to temporary stop slave irq while slave is handling the response, and re-enable them again when the response is ready.

---
v3:
+ First introduce in v3 [Quan]

  drivers/i2c/busses/i2c-aspeed.c | 20 ++++++++++++++++++++
  1 file changed, 20 insertions(+)
drivers/i2c/busses/i2c-aspeed.c +0 -0
index b2e9c8f0ddf7..9926d04831a2 100644
@@ -944,6 +944,26 @@ static int aspeed_i2c_init(struct aspeed_i2c_bus
 *bus,
  	return 0;

drivers/i2c/busses/i2c-aspeed.c static int aspeed_i2c_init(struct aspeed_i2c_bus · line 955

Quoted 10 lines
drivers/i2c/busses/i2c-aspeed.c +8 -0
  }
 
+#if IS_ENABLED(CONFIG_I2C_SLAVE)
+void aspeed_set_slave_busy(struct i2c_adapter *adap, bool busy) {
+	struct aspeed_i2c_bus *bus = i2c_get_adapdata(adap);
+	unsigned long current_mask, flags;
+
+	spin_lock_irqsave(&bus->lock, flags);
+
+	current_mask = readl(bus->base + ASPEED_I2C_INTR_CTRL_REG);

Hello

Quoted 2 lines
	Where the bus->base to be remap?

Hi Ryan,

In "[PATCH v3 6/7] ipmi: ssif_bmc: Add Aspeed SSIF BMC driver", the ->priv is retrieved by calling i2c_get_adapdata(client->adapter). And in aspeed_set_ssif_bmc_status(), call the exported aspeed_set_slave_busy() using ->priv pointer as code below.

+extern void aspeed_set_slave_busy(struct i2c_adapter *adap, bool busy); +static void aspeed_set_ssif_bmc_status(struct ssif_bmc_ctx *ssif_bmc, unsigned int status)
+{
+ if (status & SSIF_BMC_BUSY)
+ aspeed_set_slave_busy((struct i2c_adapter *)ssif_bmc->priv, true);
+ else if (status & SSIF_BMC_READY)
+ aspeed_set_slave_busy((struct i2c_adapter *)ssif_bmc->priv, false);
+}
+
+static int ssif_bmc_probe(struct i2c_client *client, const struct i2c_device_id *id)
+{
+ struct ssif_bmc_ctx *ssif_bmc;
+
+ ssif_bmc = ssif_bmc_alloc(client, 0);
+ if (IS_ERR(ssif_bmc))
+ return PTR_ERR(ssif_bmc);
+
+ ssif_bmc->priv = i2c_get_adapdata(client->adapter);
+ ssif_bmc->set_ssif_bmc_status = aspeed_set_ssif_bmc_status;
+
+ return 0;
+}

- Quan

maintainer
#23 lore raw

RE: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

Quoted 74 lines

-----Original Message-----
From: Quan Nguyen <[email protected]>
Sent: Monday, May 24, 2021 6:20 PM
To: Ryan Chen <[email protected]>; Corey Minyard <[email protected]>; Rob Herring <[email protected]>; Joel Stanley <[email protected]>; Andrew Jeffery <[email protected]>; Brendan Higgins <[email protected]>; Benjamin Herrenschmidt <[email protected]>; Wolfram Sang <[email protected]>; Philipp Zabel <[email protected]>; openipmi-developer at lists.sourceforge.net; devicetree at vger.kernel.org; linux-arm-kernel at lists.infradead.org; linux-aspeed at lists.ozlabs.org; linux-kernel at vger.kernel.org; linux-i2c at vger.kernel.org
Cc: Open Source Submission <[email protected]>; Thang Q . Nguyen <[email protected]>; Phong Vo <[email protected]>; openbmc at lists.ozlabs.org
Subject: Re: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

Quoted 3 lines · Ryan Chen
Quoted 3 lines

-----Original Message-----
From: openbmc <[email protected]> On

Behalf

Quoted 16 lines
Quoted 16 lines

Of Quan Nguyen
Sent: Wednesday, May 19, 2021 3:50 PM
To: Corey Minyard <[email protected]>; Rob Herring <[email protected]>; Joel Stanley <[email protected]>; Andrew Jeffery <[email protected]>; Brendan Higgins <[email protected]>; Benjamin Herrenschmidt <[email protected]>; Wolfram Sang <[email protected]>; Philipp Zabel <[email protected]>; openipmi-developer at lists.sourceforge.net;
devicetree at vger.kernel.org; linux-arm-kernel at lists.infradead.org; linux-aspeed at lists.ozlabs.org; linux-kernel at vger.kernel.org; linux-i2c at vger.kernel.org
Cc: Open Source Submission <[email protected]>; Thang Q . Nguyen <[email protected]>; Phong Vo <[email protected]>; openbmc at lists.ozlabs.org
Subject: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

Quoted 4 lines
Quoted 4 lines

Slave i2c device on AST2500 received a lot of slave irq while it is busy processing the response. To handle this case, adds and exports aspeed_set_slave_busy() for controller to temporary stop slave irq while slave is handling the response, and re-enable them again when the

response is ready.

Quoted 28 lines
Quoted 17 lines

---
v3:
+ First introduce in v3 [Quan]

  drivers/i2c/busses/i2c-aspeed.c | 20 ++++++++++++++++++++
  1 file changed, 20 insertions(+)
(quoted diff) +0 -0

b/drivers/i2c/busses/i2c-aspeed.c index b2e9c8f0ddf7..9926d04831a2 100644

drivers/i2c/busses/i2c-aspeed.c +0 -0
@@ -944,6 +944,26 @@ static int aspeed_i2c_init(struct aspeed_i2c_bus
 *bus,
  	return 0;

drivers/i2c/busses/i2c-aspeed.c static int aspeed_i2c_init(struct aspeed_i2c_bus · line 955

Quoted 10 lines
drivers/i2c/busses/i2c-aspeed.c +8 -0
  }
 
+#if IS_ENABLED(CONFIG_I2C_SLAVE)
+void aspeed_set_slave_busy(struct i2c_adapter *adap, bool busy) {
+	struct aspeed_i2c_bus *bus = i2c_get_adapdata(adap);
+	unsigned long current_mask, flags;
+
+	spin_lock_irqsave(&bus->lock, flags);
+
+	current_mask = readl(bus->base + ASPEED_I2C_INTR_CTRL_REG);

Hello

Quoted 2 lines
	Where the bus->base to be remap?
Quoted 7 lines

Hi Ryan,

In "[PATCH v3 6/7] ipmi: ssif_bmc: Add Aspeed SSIF BMC driver", the ->priv is retrieved by calling i2c_get_adapdata(client->adapter). And in aspeed_set_ssif_bmc_status(), call the exported aspeed_set_slave_busy() using ->priv pointer as code below.

Yes, I see the probe function " ssif_bmc->priv = i2c_get_adapdata(client->adapter);" to get priv. But my question I don?t see the bus->base address be assigned.

Quoted 27 lines
(quoted diff) +2 -0
+extern void aspeed_set_slave_busy(struct i2c_adapter *adap, bool busy);
+static void aspeed_set_ssif_bmc_status(struct ssif_bmc_ctx *ssif_bmc,

unsigned int status)

(quoted diff) +7 -0
+{
+	if (status & SSIF_BMC_BUSY)
+		aspeed_set_slave_busy((struct i2c_adapter *)ssif_bmc->priv, true);
+	else if (status & SSIF_BMC_READY)
+		aspeed_set_slave_busy((struct i2c_adapter *)ssif_bmc->priv, false); }
+
+static int ssif_bmc_probe(struct i2c_client *client, const struct

i2c_device_id *id)

(quoted diff) +12 -1
+{
+	struct ssif_bmc_ctx *ssif_bmc;
+
+	ssif_bmc = ssif_bmc_alloc(client, 0);
+	if (IS_ERR(ssif_bmc))
+		return PTR_ERR(ssif_bmc);
+
+	ssif_bmc->priv = i2c_get_adapdata(client->adapter);
+	ssif_bmc->set_ssif_bmc_status = aspeed_set_ssif_bmc_status;
+
+	return 0;
+}
 
- Quan
#24 lore raw

Re: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

Quoted 76 lines · Ryan Chen
Quoted 76 lines

-----Original Message-----
From: Quan Nguyen <[email protected]>
Sent: Monday, May 24, 2021 6:20 PM
To: Ryan Chen <[email protected]>; Corey Minyard <[email protected]>; Rob Herring <[email protected]>; Joel Stanley <[email protected]>; Andrew Jeffery <[email protected]>; Brendan Higgins <[email protected]>; Benjamin Herrenschmidt <[email protected]>; Wolfram Sang <[email protected]>; Philipp Zabel <[email protected]>; openipmi-developer at lists.sourceforge.net; devicetree at vger.kernel.org; linux-arm-kernel at lists.infradead.org; linux-aspeed at lists.ozlabs.org; linux-kernel at vger.kernel.org; linux-i2c at vger.kernel.org
Cc: Open Source Submission <[email protected]>; Thang Q . Nguyen <[email protected]>; Phong Vo <[email protected]>; openbmc at lists.ozlabs.org
Subject: Re: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

Quoted 3 lines · Ryan Chen
Quoted 3 lines

-----Original Message-----
From: openbmc <[email protected]> On

Behalf

Quoted 16 lines
Quoted 16 lines

Of Quan Nguyen
Sent: Wednesday, May 19, 2021 3:50 PM
To: Corey Minyard <[email protected]>; Rob Herring <[email protected]>; Joel Stanley <[email protected]>; Andrew Jeffery <[email protected]>; Brendan Higgins <[email protected]>; Benjamin Herrenschmidt <[email protected]>; Wolfram Sang <[email protected]>; Philipp Zabel <[email protected]>; openipmi-developer at lists.sourceforge.net;
devicetree at vger.kernel.org; linux-arm-kernel at lists.infradead.org; linux-aspeed at lists.ozlabs.org; linux-kernel at vger.kernel.org; linux-i2c at vger.kernel.org
Cc: Open Source Submission <[email protected]>; Thang Q . Nguyen <[email protected]>; Phong Vo <[email protected]>; openbmc at lists.ozlabs.org
Subject: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

Quoted 4 lines
Quoted 4 lines

Slave i2c device on AST2500 received a lot of slave irq while it is busy processing the response. To handle this case, adds and exports aspeed_set_slave_busy() for controller to temporary stop slave irq while slave is handling the response, and re-enable them again when the

response is ready.

Quoted 28 lines
Quoted 17 lines

---
v3:
+ First introduce in v3 [Quan]

   drivers/i2c/busses/i2c-aspeed.c | 20 ++++++++++++++++++++
   1 file changed, 20 insertions(+)
(quoted diff) +0 -0

b/drivers/i2c/busses/i2c-aspeed.c index b2e9c8f0ddf7..9926d04831a2 100644

drivers/i2c/busses/i2c-aspeed.c +0 -0
@@ -944,6 +944,26 @@ static int aspeed_i2c_init(struct aspeed_i2c_bus
 *bus,
   	return 0;

drivers/i2c/busses/i2c-aspeed.c static int aspeed_i2c_init(struct aspeed_i2c_bus · line 955

Quoted 10 lines
drivers/i2c/busses/i2c-aspeed.c +8 -0
   }
 
+#if IS_ENABLED(CONFIG_I2C_SLAVE)
+void aspeed_set_slave_busy(struct i2c_adapter *adap, bool busy) {
+	struct aspeed_i2c_bus *bus = i2c_get_adapdata(adap);
+	unsigned long current_mask, flags;
+
+	spin_lock_irqsave(&bus->lock, flags);
+
+	current_mask = readl(bus->base + ASPEED_I2C_INTR_CTRL_REG);

Hello

Quoted 2 lines
	Where the bus->base to be remap?

Hi Ryan,

Quoted 8 lines
Quoted 5 lines

In "[PATCH v3 6/7] ipmi: ssif_bmc: Add Aspeed SSIF BMC driver", the ->priv is retrieved by calling i2c_get_adapdata(client->adapter). And in aspeed_set_ssif_bmc_status(), call the exported aspeed_set_slave_busy() using ->priv pointer as code below.

Yes, I see the probe function " ssif_bmc->priv = i2c_get_adapdata(client->adapter);" to get priv. But my question I don?t see the bus->base address be assigned.

Hi Ryan,

In drivers/i2c/busses/i2c-aspeed.c:

struct aspeed_i2c_bus {
         struct i2c_adapter              adap;
         struct device                   *dev;
         void __iomem                    *base;
         struct reset_control            *rst;
         /* Synchronizes I/O mem access to base. */
         spinlock_t                      lock;

So when "struct aspeed_i2c_bus *bus = i2c_get_adapdata(adap);", the bus->base should point to the base of the aspeed_i2c_bus, which is already initialized by the aspeed i2c bus driver.

Do I miss something?

- Quan

Quoted 28 lines
Quoted 27 lines
(quoted diff) +2 -0
+extern void aspeed_set_slave_busy(struct i2c_adapter *adap, bool busy);
+static void aspeed_set_ssif_bmc_status(struct ssif_bmc_ctx *ssif_bmc,

unsigned int status)

(quoted diff) +7 -0
+{
+	if (status & SSIF_BMC_BUSY)
+		aspeed_set_slave_busy((struct i2c_adapter *)ssif_bmc->priv, true);
+	else if (status & SSIF_BMC_READY)
+		aspeed_set_slave_busy((struct i2c_adapter *)ssif_bmc->priv, false); }
+
+static int ssif_bmc_probe(struct i2c_client *client, const struct

i2c_device_id *id)

(quoted diff) +12 -1
+{
+	struct ssif_bmc_ctx *ssif_bmc;
+
+	ssif_bmc = ssif_bmc_alloc(client, 0);
+	if (IS_ERR(ssif_bmc))
+		return PTR_ERR(ssif_bmc);
+
+	ssif_bmc->priv = i2c_get_adapdata(client->adapter);
+	ssif_bmc->set_ssif_bmc_status = aspeed_set_ssif_bmc_status;
+
+	return 0;
+}
 
- Quan
maintainer
#25 lore raw

RE: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

Quoted 118 lines

-----Original Message-----
From: Quan Nguyen <[email protected]>
Sent: Monday, May 24, 2021 6:49 PM
To: Ryan Chen <[email protected]>; Corey Minyard <[email protected]>; Rob Herring <[email protected]>; Joel Stanley <[email protected]>; Andrew Jeffery <[email protected]>; Brendan Higgins <[email protected]>; Benjamin Herrenschmidt <[email protected]>; Wolfram Sang <[email protected]>; Philipp Zabel <[email protected]>; openipmi-developer at lists.sourceforge.net; devicetree at vger.kernel.org; linux-arm-kernel at lists.infradead.org; linux-aspeed at lists.ozlabs.org; linux-kernel at vger.kernel.org; linux-i2c at vger.kernel.org
Cc: Open Source Submission <[email protected]>; Thang Q . Nguyen <[email protected]>; Phong Vo <[email protected]>; openbmc at lists.ozlabs.org
Subject: Re: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

Quoted 23 lines · Ryan Chen
Quoted 23 lines

-----Original Message-----
From: Quan Nguyen <[email protected]>
Sent: Monday, May 24, 2021 6:20 PM
To: Ryan Chen <[email protected]>; Corey Minyard <[email protected]>; Rob Herring <[email protected]>; Joel Stanley <[email protected]>; Andrew Jeffery <[email protected]>; Brendan Higgins <[email protected]>; Benjamin Herrenschmidt <[email protected]>; Wolfram Sang <[email protected]>; Philipp Zabel <[email protected]>;
openipmi-developer at lists.sourceforge.net;
devicetree at vger.kernel.org; linux-arm-kernel at lists.infradead.org; linux-aspeed at lists.ozlabs.org; linux-kernel at vger.kernel.org; linux-i2c at vger.kernel.org
Cc: Open Source Submission <[email protected]>; Thang Q . Nguyen <[email protected]>; Phong Vo <[email protected]>; openbmc at lists.ozlabs.org
Subject: Re: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

Quoted 3 lines · Ryan Chen
Quoted 3 lines

-----Original Message-----
From: openbmc <[email protected]> On

Behalf

Quoted 12 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

Of Quan Nguyen
Sent: Wednesday, May 19, 2021 3:50 PM
To: Corey Minyard <[email protected]>; Rob Herring <[email protected]>; Joel Stanley <[email protected]>; Andrew Jeffery

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

<[email protected]>; Brendan Higgins <[email protected]>; Benjamin Herrenschmidt <[email protected]>; Wolfram Sang <[email protected]>; Philipp Zabel <[email protected]>; openipmi-developer at lists.sourceforge.net;
devicetree at vger.kernel.org; linux-arm-kernel at lists.infradead.org; linux-aspeed at lists.ozlabs.org; linux-kernel at vger.kernel.org; linux-i2c at vger.kernel.org
Cc: Open Source Submission <[email protected]>; Thang

Q .

Quoted 43 lines
Quoted 43 lines
Quoted 9 lines
Quoted 9 lines

Nguyen <[email protected]>; Phong Vo <[email protected]>; openbmc at lists.ozlabs.org
Subject: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

Slave i2c device on AST2500 received a lot of slave irq while it is busy processing the response. To handle this case, adds and exports aspeed_set_slave_busy() for controller to temporary stop slave irq while slave is handling the response, and re-enable them again when the

response is ready.

Quoted 28 lines
Quoted 17 lines

---
v3:
+ First introduce in v3 [Quan]

   drivers/i2c/busses/i2c-aspeed.c | 20 ++++++++++++++++++++
   1 file changed, 20 insertions(+)
(quoted diff) +0 -0

b/drivers/i2c/busses/i2c-aspeed.c index b2e9c8f0ddf7..9926d04831a2 100644

drivers/i2c/busses/i2c-aspeed.c +0 -0
@@ -944,6 +944,26 @@ static int aspeed_i2c_init(struct
 aspeed_i2c_bus *bus,
   	return 0;

drivers/i2c/busses/i2c-aspeed.c static int aspeed_i2c_init(struct · line 955

Quoted 10 lines
drivers/i2c/busses/i2c-aspeed.c +8 -0
   }
 
+#if IS_ENABLED(CONFIG_I2C_SLAVE)
+void aspeed_set_slave_busy(struct i2c_adapter *adap, bool busy) {
+	struct aspeed_i2c_bus *bus = i2c_get_adapdata(adap);
+	unsigned long current_mask, flags;
+
+	spin_lock_irqsave(&bus->lock, flags);
+
+	current_mask = readl(bus->base + ASPEED_I2C_INTR_CTRL_REG);

Hello

Quoted 2 lines
	Where the bus->base to be remap?

Hi Ryan,

Quoted 7 lines
Quoted 6 lines

In "[PATCH v3 6/7] ipmi: ssif_bmc: Add Aspeed SSIF BMC driver", the ->priv is retrieved by calling i2c_get_adapdata(client->adapter). And ->in
aspeed_set_ssif_bmc_status(), call the exported aspeed_set_slave_busy() using ->priv pointer as code below.

Yes, I see the probe function " ssif_bmc->priv =

i2c_get_adapdata(client->adapter);" to get priv.

Quoted 2 lines

But my question I don?t see the bus->base address be assigned.

Hi Ryan,

In drivers/i2c/busses/i2c-aspeed.c:

struct aspeed_i2c_bus {
         struct i2c_adapter              adap;
         struct device                   *dev;
         void __iomem                    *base;
         struct reset_control            *rst;
         /* Synchronizes I/O mem access to base. */
         spinlock_t                      lock;
Quoted 5 lines

So when "struct aspeed_i2c_bus *bus = i2c_get_adapdata(adap);", the bus->base should point to the base of the aspeed_i2c_bus, which is already initialized by the aspeed i2c bus driver.

Do I miss something?

Hello Quan,
After study. I think the ssif_bmc_aspeed.c assume the priv data is the same struct. That is trick.
Do we have a better for slave enable/disable call back to implement this? If add call back in struct i2c_algorithm; is it workable?

Quoted 29 lines
Quoted 29 lines
Quoted 28 lines
(quoted diff) +2 -0
+extern void aspeed_set_slave_busy(struct i2c_adapter *adap, bool busy);
+static void aspeed_set_ssif_bmc_status(struct ssif_bmc_ctx *ssif_bmc,

unsigned int status)

(quoted diff) +7 -0
+{
+	if (status & SSIF_BMC_BUSY)
+		aspeed_set_slave_busy((struct i2c_adapter *)ssif_bmc->priv, true);
+	else if (status & SSIF_BMC_READY)
+		aspeed_set_slave_busy((struct i2c_adapter *)ssif_bmc->priv, false); }
+
+static int ssif_bmc_probe(struct i2c_client *client, const struct

i2c_device_id *id)

(quoted diff) +12 -1
+{
+	struct ssif_bmc_ctx *ssif_bmc;
+
+	ssif_bmc = ssif_bmc_alloc(client, 0);
+	if (IS_ERR(ssif_bmc))
+		return PTR_ERR(ssif_bmc);
+
+	ssif_bmc->priv = i2c_get_adapdata(client->adapter);
+	ssif_bmc->set_ssif_bmc_status = aspeed_set_ssif_bmc_status;
+
+	return 0;
+}
 
- Quan
#26 lore raw

Re: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

Quoted 122 lines · Ryan Chen
Quoted 122 lines

-----Original Message-----
From: Quan Nguyen <[email protected]>
Sent: Monday, May 24, 2021 6:49 PM
To: Ryan Chen <[email protected]>; Corey Minyard <[email protected]>; Rob Herring <[email protected]>; Joel Stanley <[email protected]>; Andrew Jeffery <[email protected]>; Brendan Higgins <[email protected]>; Benjamin Herrenschmidt <[email protected]>; Wolfram Sang <[email protected]>; Philipp Zabel <[email protected]>; openipmi-developer at lists.sourceforge.net; devicetree at vger.kernel.org; linux-arm-kernel at lists.infradead.org; linux-aspeed at lists.ozlabs.org; linux-kernel at vger.kernel.org; linux-i2c at vger.kernel.org
Cc: Open Source Submission <[email protected]>; Thang Q . Nguyen <[email protected]>; Phong Vo <[email protected]>; openbmc at lists.ozlabs.org
Subject: Re: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

Quoted 23 lines · Ryan Chen
Quoted 23 lines

-----Original Message-----
From: Quan Nguyen <[email protected]>
Sent: Monday, May 24, 2021 6:20 PM
To: Ryan Chen <[email protected]>; Corey Minyard <[email protected]>; Rob Herring <[email protected]>; Joel Stanley <[email protected]>; Andrew Jeffery <[email protected]>; Brendan Higgins <[email protected]>; Benjamin Herrenschmidt <[email protected]>; Wolfram Sang <[email protected]>; Philipp Zabel <[email protected]>;
openipmi-developer at lists.sourceforge.net;
devicetree at vger.kernel.org; linux-arm-kernel at lists.infradead.org; linux-aspeed at lists.ozlabs.org; linux-kernel at vger.kernel.org; linux-i2c at vger.kernel.org
Cc: Open Source Submission <[email protected]>; Thang Q . Nguyen <[email protected]>; Phong Vo <[email protected]>; openbmc at lists.ozlabs.org
Subject: Re: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

Quoted 3 lines · Ryan Chen
Quoted 3 lines

-----Original Message-----
From: openbmc <[email protected]> On

Behalf

Quoted 12 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

Of Quan Nguyen
Sent: Wednesday, May 19, 2021 3:50 PM
To: Corey Minyard <[email protected]>; Rob Herring <[email protected]>; Joel Stanley <[email protected]>; Andrew Jeffery

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

<[email protected]>; Brendan Higgins <[email protected]>; Benjamin Herrenschmidt <[email protected]>; Wolfram Sang <[email protected]>; Philipp Zabel <[email protected]>; openipmi-developer at lists.sourceforge.net;
devicetree at vger.kernel.org; linux-arm-kernel at lists.infradead.org; linux-aspeed at lists.ozlabs.org; linux-kernel at vger.kernel.org; linux-i2c at vger.kernel.org
Cc: Open Source Submission <[email protected]>; Thang

Q .

Quoted 43 lines
Quoted 43 lines
Quoted 9 lines
Quoted 9 lines

Nguyen <[email protected]>; Phong Vo <[email protected]>; openbmc at lists.ozlabs.org
Subject: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

Slave i2c device on AST2500 received a lot of slave irq while it is busy processing the response. To handle this case, adds and exports aspeed_set_slave_busy() for controller to temporary stop slave irq while slave is handling the response, and re-enable them again when the

response is ready.

Quoted 28 lines
Quoted 17 lines

---
v3:
+ First introduce in v3 [Quan]

drivers/i2c/busses/i2c-aspeed.c | 20 ++++++++++++++++++++ 1 file changed, 20 insertions(+)

(quoted diff) +0 -0

b/drivers/i2c/busses/i2c-aspeed.c index b2e9c8f0ddf7..9926d04831a2 100644

drivers/i2c/busses/i2c-aspeed.c +0 -0
@@ -944,6 +944,26 @@ static int aspeed_i2c_init(struct
 aspeed_i2c_bus *bus,
    	return 0;

drivers/i2c/busses/i2c-aspeed.c static int aspeed_i2c_init(struct · line 955

Quoted 10 lines
drivers/i2c/busses/i2c-aspeed.c +8 -0
    }
 
+#if IS_ENABLED(CONFIG_I2C_SLAVE)
+void aspeed_set_slave_busy(struct i2c_adapter *adap, bool busy) {
+	struct aspeed_i2c_bus *bus = i2c_get_adapdata(adap);
+	unsigned long current_mask, flags;
+
+	spin_lock_irqsave(&bus->lock, flags);
+
+	current_mask = readl(bus->base + ASPEED_I2C_INTR_CTRL_REG);

Hello

Quoted 2 lines
	Where the bus->base to be remap?

Hi Ryan,

Quoted 7 lines
Quoted 6 lines

In "[PATCH v3 6/7] ipmi: ssif_bmc: Add Aspeed SSIF BMC driver", the ->priv is retrieved by calling i2c_get_adapdata(client->adapter). And ->in
aspeed_set_ssif_bmc_status(), call the exported aspeed_set_slave_busy() using ->priv pointer as code below.

Yes, I see the probe function " ssif_bmc->priv =

i2c_get_adapdata(client->adapter);" to get priv.

Quoted 2 lines

But my question I don?t see the bus->base address be assigned.

Hi Ryan,

In drivers/i2c/busses/i2c-aspeed.c:

struct aspeed_i2c_bus {
          struct i2c_adapter              adap;
          struct device                   *dev;
          void __iomem                    *base;
          struct reset_control            *rst;
          /* Synchronizes I/O mem access to base. */
          spinlock_t                      lock;

So when "struct aspeed_i2c_bus *bus = i2c_get_adapdata(adap);", the bus->base should point to the base of the aspeed_i2c_bus, which is already initialized by the aspeed i2c bus driver.

Quoted 7 lines
Quoted 1 line

Do I miss something?

Hello Quan,
After study. I think the ssif_bmc_aspeed.c assume the priv data is the same struct. That is trick.
Do we have a better for slave enable/disable call back to implement this? If add call back in struct i2c_algorithm; is it workable?

Quoted 1 line

Hi Ryan,

I dont know which is better, ie: adding callback to struct i2c_algorithm or to struct i2c_adapter.
I have tried to add generic callback to struct i2c_adapter as below and it works.

include/linux/i2c.h +4 -0
index 4e7714c88f95..6e9abf2d6abb 100644
@@ -713,6 +713,10 @@ struct i2c_adapter {
  	const struct i2c_adapter_quirks *quirks;
 
  	struct irq_domain *host_notify_domain;
+#if IS_ENABLED(CONFIG_I2C_SLAVE)
+	int (*slave_enable)(struct i2c_adapter *adap);
+	int (*slave_disable)(struct i2c_adapter *adap);
+#endif
  };
  #define to_i2c_adapter(d) container_of(d, struct i2c_adapter, dev)
 
Quoted 29 lines
Quoted 28 lines
Quoted 28 lines
Quoted 27 lines
(quoted diff) +2 -0
+extern void aspeed_set_slave_busy(struct i2c_adapter *adap, bool busy);
+static void aspeed_set_ssif_bmc_status(struct ssif_bmc_ctx *ssif_bmc,

unsigned int status)

(quoted diff) +7 -0
+{
+	if (status & SSIF_BMC_BUSY)
+		aspeed_set_slave_busy((struct i2c_adapter *)ssif_bmc->priv, true);
+	else if (status & SSIF_BMC_READY)
+		aspeed_set_slave_busy((struct i2c_adapter *)ssif_bmc->priv, false); }
+
+static int ssif_bmc_probe(struct i2c_client *client, const struct

i2c_device_id *id)

(quoted diff) +12 -1
+{
+	struct ssif_bmc_ctx *ssif_bmc;
+
+	ssif_bmc = ssif_bmc_alloc(client, 0);
+	if (IS_ERR(ssif_bmc))
+		return PTR_ERR(ssif_bmc);
+
+	ssif_bmc->priv = i2c_get_adapdata(client->adapter);
+	ssif_bmc->set_ssif_bmc_status = aspeed_set_ssif_bmc_status;
+
+	return 0;
+}
 
- Quan
maintainer
#27 lore raw

RE: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

Quoted 156 lines

-----Original Message-----
From: Quan Nguyen <[email protected]>
Sent: Friday, May 28, 2021 8:53 AM
To: Ryan Chen <[email protected]>; Corey Minyard <[email protected]>; Rob Herring <[email protected]>; Joel Stanley <[email protected]>; Andrew Jeffery <[email protected]>; Brendan Higgins <[email protected]>; Benjamin Herrenschmidt <[email protected]>; Wolfram Sang <[email protected]>; Philipp Zabel <[email protected]>; openipmi-developer at lists.sourceforge.net; devicetree at vger.kernel.org; linux-arm-kernel at lists.infradead.org; linux-aspeed at lists.ozlabs.org; linux-kernel at vger.kernel.org; linux-i2c at vger.kernel.org
Cc: Open Source Submission <[email protected]>; Thang Q . Nguyen <[email protected]>; Phong Vo <[email protected]>; openbmc at lists.ozlabs.org
Subject: Re: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

Quoted 19 lines · Ryan Chen
Quoted 19 lines

-----Original Message-----
From: Quan Nguyen <[email protected]>
Sent: Monday, May 24, 2021 6:49 PM
To: Ryan Chen <[email protected]>; Corey Minyard <[email protected]>; Rob Herring <[email protected]>; Joel Stanley <[email protected]>; Andrew Jeffery <[email protected]>; Brendan Higgins <[email protected]>; Benjamin Herrenschmidt <[email protected]>; Wolfram Sang <[email protected]>; Philipp Zabel <[email protected]>;
openipmi-developer at lists.sourceforge.net;
devicetree at vger.kernel.org; linux-arm-kernel at lists.infradead.org; linux-aspeed at lists.ozlabs.org; linux-kernel at vger.kernel.org; linux-i2c at vger.kernel.org
Cc: Open Source Submission <[email protected]>; Thang Q . Nguyen <[email protected]>; Phong Vo <[email protected]>; openbmc at lists.ozlabs.org
Subject: Re: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

On 24/05/2021 17:36, Ryan Chen wrote:

Quoted 14 lines
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines

-----Original Message-----
From: Quan Nguyen <[email protected]>
Sent: Monday, May 24, 2021 6:20 PM
To: Ryan Chen <[email protected]>; Corey Minyard <[email protected]>; Rob Herring <[email protected]>; Joel Stanley <[email protected]>; Andrew Jeffery <[email protected]>; Brendan Higgins

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

<[email protected]>; Benjamin Herrenschmidt <[email protected]>; Wolfram Sang <[email protected]>; Philipp Zabel <[email protected]>;
openipmi-developer at lists.sourceforge.net;
devicetree at vger.kernel.org; linux-arm-kernel at lists.infradead.org; linux-aspeed at lists.ozlabs.org; linux-kernel at vger.kernel.org; linux-i2c at vger.kernel.org
Cc: Open Source Submission <[email protected]>; Thang

Q .

Quoted 10 lines
Quoted 10 lines
Quoted 10 lines
Quoted 10 lines

Nguyen <[email protected]>; Phong Vo <[email protected]>; openbmc at lists.ozlabs.org
Subject: Re: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

Quoted 3 lines · Ryan Chen
Quoted 3 lines

-----Original Message-----
From: openbmc <[email protected]> On

Behalf

Quoted 13 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines

Of Quan Nguyen
Sent: Wednesday, May 19, 2021 3:50 PM
To: Corey Minyard <[email protected]>; Rob Herring <[email protected]>; Joel Stanley <[email protected]>; Andrew Jeffery <[email protected]>; Brendan Higgins

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

<[email protected]>; Benjamin Herrenschmidt <[email protected]>; Wolfram Sang <[email protected]>; Philipp Zabel <[email protected]>;
openipmi-developer at lists.sourceforge.net;
devicetree at vger.kernel.org; linux-arm-kernel at lists.infradead.org; linux-aspeed at lists.ozlabs.org; linux-kernel at vger.kernel.org; linux-i2c at vger.kernel.org
Cc: Open Source Submission <[email protected]>;

Thang

Quoted 67 lines
Quoted 67 lines

Q .

Quoted 45 lines
Quoted 45 lines
Quoted 10 lines
Quoted 10 lines

Nguyen <[email protected]>; Phong Vo <[email protected]>; openbmc at lists.ozlabs.org
Subject: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

Slave i2c device on AST2500 received a lot of slave irq while it is busy processing the response. To handle this case, adds and exports
aspeed_set_slave_busy() for controller to temporary stop slave irq while slave is handling the response, and re-enable them again when the

response is ready.

Quoted 29 lines
Quoted 18 lines

---
v3:
+ First introduce in v3 [Quan]

drivers/i2c/busses/i2c-aspeed.c | 20 ++++++++++++++++++++ 1 file changed, 20 insertions(+)

(quoted diff) +0 -0

b/drivers/i2c/busses/i2c-aspeed.c index
b2e9c8f0ddf7..9926d04831a2
100644

drivers/i2c/busses/i2c-aspeed.c +0 -0
@@ -944,6 +944,26 @@ static int aspeed_i2c_init(struct
 aspeed_i2c_bus *bus,
    	return 0;

drivers/i2c/busses/i2c-aspeed.c static int aspeed_i2c_init(struct · line 955

Quoted 10 lines
drivers/i2c/busses/i2c-aspeed.c +8 -0
    }
 
+#if IS_ENABLED(CONFIG_I2C_SLAVE) void
+aspeed_set_slave_busy(struct i2c_adapter *adap, bool busy) {
+	struct aspeed_i2c_bus *bus = i2c_get_adapdata(adap);
+	unsigned long current_mask, flags;
+
+	spin_lock_irqsave(&bus->lock, flags);
+
+	current_mask = readl(bus->base + ASPEED_I2C_INTR_CTRL_REG);

Hello

Quoted 2 lines
	Where the bus->base to be remap?

Hi Ryan,

Quoted 7 lines
Quoted 6 lines

In "[PATCH v3 6/7] ipmi: ssif_bmc: Add Aspeed SSIF BMC driver", the ->priv is retrieved by calling i2c_get_adapdata(client->adapter). ->And in
aspeed_set_ssif_bmc_status(), call the exported aspeed_set_slave_busy() using ->priv pointer as code below.

Yes, I see the probe function " ssif_bmc->priv =

i2c_get_adapdata(client->adapter);" to get priv.

Quoted 2 lines

But my question I don?t see the bus->base address be assigned.

Hi Ryan,

In drivers/i2c/busses/i2c-aspeed.c:

struct aspeed_i2c_bus {
          struct i2c_adapter              adap;
          struct device                   *dev;
          void __iomem                    *base;
          struct reset_control            *rst;
          /* Synchronizes I/O mem access to base. */
          spinlock_t                      lock;
Quoted 7 lines
Quoted 5 lines

So when "struct aspeed_i2c_bus *bus = i2c_get_adapdata(adap);", the bus->base should point to the base of the aspeed_i2c_bus, which is already initialized by the aspeed i2c bus driver.

Do I miss something?

Hello Quan,
After study. I think the ssif_bmc_aspeed.c assume the priv data is the

same struct.

Quoted 4 lines

That is trick.
Do we have a better for slave enable/disable call back to implement this? If add call back in struct i2c_algorithm; is it workable?

Quoted 1 line
Quoted 7 lines

Hi Ryan,

I dont know which is better, ie: adding callback to struct i2c_algorithm or to struct i2c_adapter.
I have tried to add generic callback to struct i2c_adapter as below and it works.

Hello Quan,
Thanks your feedback.
Because, if we apply it in callback. It can unify it in ssif_bmc.c to do callback. Not have ssif_bmc_aspeed.c, ssif_bmc_xxx.c .....
Because the priv struct is different in different i2c driver implement.
Ryan

Quoted 46 lines
include/linux/i2c.h index renamed from include/linux/i2c.h +0 -0

4e7714c88f95..6e9abf2d6abb 100644

include/linux/i2c.h +3 -0
@@ -713,6 +713,10 @@ struct i2c_adapter {
  	const struct i2c_adapter_quirks *quirks;
 
  	struct irq_domain *host_notify_domain;
+#if IS_ENABLED(CONFIG_I2C_SLAVE)
+	int (*slave_enable)(struct i2c_adapter *adap);
+	int (*slave_disable)(struct i2c_adapter *adap); #endif
  };
  #define to_i2c_adapter(d) container_of(d, struct i2c_adapter, dev)
 
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines

Review

Quoted 7 lines
(quoted diff) +3 -0
+extern void aspeed_set_slave_busy(struct i2c_adapter *adap, bool
+busy); static void aspeed_set_ssif_bmc_status(struct ssif_bmc_ctx
+*ssif_bmc,

unsigned int status)

(quoted diff) +3 -0
+{
+	if (status & SSIF_BMC_BUSY)
+		aspeed_set_slave_busy((struct i2c_adapter *)ssif_bmc->priv,

true);

Quoted 24 lines
Quoted 23 lines
Quoted 23 lines
Quoted 22 lines
(quoted diff) +5 -0
+	else if (status & SSIF_BMC_READY)
+		aspeed_set_slave_busy((struct i2c_adapter *)ssif_bmc->priv,
+false); }
+
+static int ssif_bmc_probe(struct i2c_client *client, const struct

i2c_device_id *id)

(quoted diff) +12 -1
+{
+	struct ssif_bmc_ctx *ssif_bmc;
+
+	ssif_bmc = ssif_bmc_alloc(client, 0);
+	if (IS_ERR(ssif_bmc))
+		return PTR_ERR(ssif_bmc);
+
+	ssif_bmc->priv = i2c_get_adapdata(client->adapter);
+	ssif_bmc->set_ssif_bmc_status = aspeed_set_ssif_bmc_status;
+
+	return 0;
+}
 
- Quan
#28 lore raw

Re: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()

In reply to Quan Nguyen
Quoted 19 lines · Quan Nguyen

Slave i2c device on AST2500 received a lot of slave irq while it is busy processing the response. To handle this case, adds and exports aspeed_set_slave_busy() for controller to temporary stop slave irq while slave is handling the response, and re-enable them again when the response is ready.

---
v3:
+ First introduce in v3 [Quan]

 drivers/i2c/busses/i2c-aspeed.c | 20 ++++++++++++++++++++
 1 file changed, 20 insertions(+)
drivers/i2c/busses/i2c-aspeed.c +0 -0
index b2e9c8f0ddf7..9926d04831a2 100644

drivers/i2c/busses/i2c-aspeed.c static int aspeed_i2c_init(struct aspeed_i2c_bus *bus, · line 953

Quoted 10 lines
drivers/i2c/busses/i2c-aspeed.c +7 -0
@@ -944,6 +944,26 @@ static int aspeed_i2c_init(struct aspeed_i2c_bus *bus,
 	return 0;
 }
 
+#if IS_ENABLED(CONFIG_I2C_SLAVE)
+void aspeed_set_slave_busy(struct i2c_adapter *adap, bool busy)
+{
+	struct aspeed_i2c_bus *bus = i2c_get_adapdata(adap);
+	unsigned long current_mask, flags;
+
+	spin_lock_irqsave(&bus->lock, flags);

This as far as I can see is still a recursive spinlock, and the spinlock debugger seems to agree with me.

Graeme

Quoted 19 lines
drivers/i2c/busses/i2c-aspeed.c +13 -0
+
+	current_mask = readl(bus->base + ASPEED_I2C_INTR_CTRL_REG);
+	if (busy)
+		current_mask &= ~(ASPEED_I2CD_INTR_RX_DONE | ASPEED_I2CD_INTR_SLAVE_MATCH);
+	else
+		current_mask |= ASPEED_I2CD_INTR_RX_DONE | ASPEED_I2CD_INTR_SLAVE_MATCH;
+	writel(current_mask, bus->base + ASPEED_I2C_INTR_CTRL_REG);
+
+	spin_unlock_irqrestore(&bus->lock, flags);
+}
+EXPORT_SYMBOL_GPL(aspeed_set_slave_busy);
+#endif
+
 static int aspeed_i2c_reset(struct aspeed_i2c_bus *bus)
 {
 	struct platform_device *pdev = to_platform_device(bus->dev);
Signature · 1 line
2.28.0
#29 lore raw

PATCH v3 6/7 ipmi: ssif_bmc: Add Aspeed SSIF BMC driver

In reply to Quan Nguyen

This commits adds SSIF BMC driver specifically for Aspeed AST2500 which commonly used as Board Management Controllers.

---
v3:
+ Splited into separate commit [Corey, Joel]
+ Invoked aspeed-specific aspeed_set_slave_busy() when busy
to address deadlock from Graeme and comment from Philipp
+ Combined two interrupts to mask/unmask together [Corey]

 drivers/char/ipmi/Kconfig           | 11 +++++
 drivers/char/ipmi/Makefile          |  1 +
 drivers/char/ipmi/ssif_bmc_aspeed.c | 75 +++++++++++++++++++++++++++++
 3 files changed, 87 insertions(+)
 create mode 100644 drivers/char/ipmi/ssif_bmc_aspeed.c
drivers/char/ipmi/Kconfig +11 -0
index ad5c5161bcd6..45be57023577 100644
@@ -144,6 +144,17 @@ config SSIF_IPMI_BMC
 	  The driver implements the BMC side of the SMBus system
 	  interface (SSIF).
 
+config ASPEED_SSIF_IPMI_BMC
+	depends on ARCH_ASPEED || COMPILE_TEST
+	select SSIF_IPMI_BMC
+	tristate "Aspeed SSIF IPMI BMC driver"
+	help
+	  Provides a driver for the SSIF IPMI interface found on
+	  Aspeed AST2500 SoC.
+
+	  The driver implements the BMC side of the SMBus system
+	  interface (SSIF), specific for Aspeed AST2500 SoC.
+
 config IPMB_DEVICE_INTERFACE
 	tristate 'IPMB Interface handler'
 	depends on I2C
drivers/char/ipmi/Makefile +1 -0
index d04a214d74c4..05b993f7335b 100644
@@ -28,3 +28,4 @@ obj-$(CONFIG_ASPEED_KCS_IPMI_BMC) += kcs_bmc_aspeed.o
 obj-$(CONFIG_NPCM7XX_KCS_IPMI_BMC) += kcs_bmc_npcm7xx.o
 obj-$(CONFIG_IPMB_DEVICE_INTERFACE) += ipmb_dev_int.o
 obj-$(CONFIG_SSIF_IPMI_BMC) += ssif_bmc.o
+obj-$(CONFIG_ASPEED_SSIF_IPMI_BMC) += ssif_bmc_aspeed.o
drivers/char/ipmi/ssif_bmc_aspeed.c new file +75 -0

75 lines not on this page · raw message

Signature · 1 line
2.28.0
#30 lore raw

PATCH v3 7/7 bindings: ipmi: Add binding for Aspeed SSIF BMC driver

In reply to Quan Nguyen

Add device tree binding document for the Aspeed SSIF BMC driver.

---
v3:
+ Switched to use DT schema format [Rob]

 .../bindings/ipmi/aspeed-ssif-bmc.yaml        | 33 +++++++++++++++++++
 1 file changed, 33 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/ipmi/aspeed-ssif-bmc.yaml
Documentation/devicetree/bindings/ipmi/aspeed-ssif-bmc.yaml new file +33 -0
new file mode 100644
index 000000000000..8608041fbf7e
@@ -0,0 +1,33 @@
+# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
+%YAML 1.2
+---
+$id: http://devicetree.org/schemas/ipmi/aspeed-ssif-bmc.yaml#
+$schema: http://devicetree.org/meta-schemas/core.yaml#
+
+title: Aspeed SSIF IPMI interface
+
+description: SSIF IPMI device bindings for Aspeed
+
+maintainers:
+  - Thang Nguyen <[email protected]>
+
+properties:
+  compatible:
+    enum:
+      - aspeed,ast2500-ssif-bmc
+
+  reg:
+    maxItems: 1
+
+required:
+  - compatible
+  - reg
+
+additionalProperties: false
+
+examples:
+  - |
+    ssif-bmc at 10 {
+        compatible = "aspeed,ast2500-ssif-bmc";
+        reg = <0x10>;
+    };
Signature · 1 line
2.28.0
maintainer
#31 lore raw

Re: [PATCH v3 7/7] bindings: ipmi: Add binding for Aspeed SSIF BMC driver

Quoted 11 lines · Quan Nguyen

Add device tree binding document for the Aspeed SSIF BMC driver.

---
v3:
+ Switched to use DT schema format [Rob]

 .../bindings/ipmi/aspeed-ssif-bmc.yaml        | 33 +++++++++++++++++++
 1 file changed, 33 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/ipmi/aspeed-ssif-bmc.yaml

My bot found errors running 'make DT_CHECKER_FLAGS=-m dt_binding_check' on your patch (DT_CHECKER_FLAGS is new in v5.13):

yamllint warnings/errors:

dtschema/dtc warnings/errors:
Documentation/devicetree/bindings/ipmi/aspeed-ssif-bmc.example.dts:21.13-26: Warning (reg_format): /example-0/ssif-bmc at 10:reg: property has invalid length (4 bytes) (#address-cells == 1, #size-cells == 1) Documentation/devicetree/bindings/ipmi/aspeed-ssif-bmc.example.dt.yaml: Warning (pci_device_reg): Failed prerequisite 'reg_format' Documentation/devicetree/bindings/ipmi/aspeed-ssif-bmc.example.dt.yaml: Warning (pci_device_bus_num): Failed prerequisite 'reg_format' Documentation/devicetree/bindings/ipmi/aspeed-ssif-bmc.example.dt.yaml: Warning (simple_bus_reg): Failed prerequisite 'reg_format' Documentation/devicetree/bindings/ipmi/aspeed-ssif-bmc.example.dt.yaml: Warning (i2c_bus_reg): Failed prerequisite 'reg_format' Documentation/devicetree/bindings/ipmi/aspeed-ssif-bmc.example.dt.yaml: Warning (spi_bus_reg): Failed prerequisite 'reg_format' /builds/robherring/linux-dt-review/Documentation/devicetree/bindings/ipmi/aspeed-ssif-bmc.example.dt.yaml: example-0: ssif-bmc at 10:reg:0: [16] is too short
From schema: /usr/local/lib/python3.8/dist-packages/dtschema/schemas/reg.yaml

See https://patchwork.ozlabs.org/patch/1480727

This check can fail if there are any dependencies. The base for a patch series is generally the most recent rc1.

If you already ran 'make dt_binding_check' and didn't see the above error(s), then make sure 'yamllint' is installed and dt-schema is up to date:

pip3 install dtschema --upgrade

Please check and re-submit.

#32 lore raw

Re: [PATCH v3 7/7] bindings: ipmi: Add binding for Aspeed SSIF BMC driver

Quoted 33 lines · Rob Herring
Quoted 11 lines · Quan Nguyen

Add device tree binding document for the Aspeed SSIF BMC driver.

---
v3:
+ Switched to use DT schema format [Rob]

  .../bindings/ipmi/aspeed-ssif-bmc.yaml        | 33 +++++++++++++++++++
  1 file changed, 33 insertions(+)

create mode 100644 Documentation/devicetree/bindings/ipmi/aspeed-ssif-bmc.yaml

My bot found errors running 'make DT_CHECKER_FLAGS=-m dt_binding_check' on your patch (DT_CHECKER_FLAGS is new in v5.13):

yamllint warnings/errors:

dtschema/dtc warnings/errors:
Documentation/devicetree/bindings/ipmi/aspeed-ssif-bmc.example.dts:21.13-26: Warning (reg_format): /example-0/ssif-bmc at 10:reg: property has invalid length (4 bytes) (#address-cells == 1, #size-cells == 1) Documentation/devicetree/bindings/ipmi/aspeed-ssif-bmc.example.dt.yaml: Warning (pci_device_reg): Failed prerequisite 'reg_format' Documentation/devicetree/bindings/ipmi/aspeed-ssif-bmc.example.dt.yaml: Warning (pci_device_bus_num): Failed prerequisite 'reg_format' Documentation/devicetree/bindings/ipmi/aspeed-ssif-bmc.example.dt.yaml: Warning (simple_bus_reg): Failed prerequisite 'reg_format' Documentation/devicetree/bindings/ipmi/aspeed-ssif-bmc.example.dt.yaml: Warning (i2c_bus_reg): Failed prerequisite 'reg_format' Documentation/devicetree/bindings/ipmi/aspeed-ssif-bmc.example.dt.yaml: Warning (spi_bus_reg): Failed prerequisite 'reg_format' /builds/robherring/linux-dt-review/Documentation/devicetree/bindings/ipmi/aspeed-ssif-bmc.example.dt.yaml: example-0: ssif-bmc at 10:reg:0: [16] is too short
From schema: /usr/local/lib/python3.8/dist-packages/dtschema/schemas/reg.yaml

See https://patchwork.ozlabs.org/patch/1480727

This check can fail if there are any dependencies. The base for a patch series is generally the most recent rc1.

Quoted 8 lines

If you already ran 'make dt_binding_check' and didn't see the above error(s), then make sure 'yamllint' is installed and dt-schema is up to date:

pip3 install dtschema --upgrade

Please check and re-submit.

Thanks,
Will fix in next version.

- Quan

#33 lore raw
In reply to Quan Nguyen
Quoted 24 lines · Quan Nguyen

This series add support for the Aspeed specific SSIF BMC driver which is to perform in-band IPMI communication with the host in management (BMC) side.

v3:
+ Switched binding doc to use DT schema format [Rob]
+ Splited into generic ssif_bmc and aspeed-specific [Corey, Joel]
+ Removed redundant license info [Joel]
+ Switched to use traditional if-else [Joel]
+ Removed unused ssif_bmc_ioctl() [Joel]
+ Made handle_request()/complete_response() to return void [Joel]
+ Refactored send_ssif_bmc_response()/receive_ssif_bmc_request() [Corey]
+ Remove mutex [Corey]
+ Use spin_lock/unlock_irqsave/restore in callback [Corey]
+ Removed the unnecessary memset [Corey]
+ Switch to use dev_err() [Corey]
+ Combine mask/unmask two interrupts together [Corey]
+ Fixed unhandled Tx done with NAK [Quan]
+ Late ack'ed Tx done w/wo Ack irq [Quan]
+ Use aspeed-specific exported aspeed_set_slave_busy() when slave busy
to fix the deadlock [Graeme, Philipp, Quan]
+ Clean buffer for last multipart read [Quan]
+ Handle unknown incoming command [Quan]

Quoted 5 lines

v2:
+ Fixed compiling error with COMPILE_TEST for arc

Quan Nguyen (7):
i2c: i2c-core-smbus: Expose PEC calculate function for generic use

Note that for this I2C-specific change, I will need acks from the I2C maintainers to be able to include this in my tree.

Quoted 2 lines

ipmi: ssif_bmc: Add SSIF BMC driver
i2c: aspeed: Fix unhandled Tx done with NAK

For the aspeed changes, they don't really belong here, they belong in the aspeed tree. I see that you need them for the device to work correctly, though. I'll need acks from maintainers to include them.

Quoted 23 lines

i2c: aspeed: Acknowledge Tx done w/wo ACK irq late
i2c: aspeed: Add aspeed_set_slave_busy()
ipmi: ssif_bmc: Add Aspeed SSIF BMC driver
bindings: ipmi: Add binding for Aspeed SSIF BMC driver

 .../bindings/ipmi/aspeed-ssif-bmc.yaml        |  33 +
 drivers/char/ipmi/Kconfig                     |  22 +
 drivers/char/ipmi/Makefile                    |   2 +
 drivers/char/ipmi/ssif_bmc.c                  | 605 ++++++++++++++++++
 drivers/char/ipmi/ssif_bmc.h                  |  93 +++
 drivers/char/ipmi/ssif_bmc_aspeed.c           |  75 +++
 drivers/i2c/busses/i2c-aspeed.c               |  51 +-
 drivers/i2c/i2c-core-smbus.c                  |  12 +-
 include/linux/i2c.h                           |   1 +
 9 files changed, 884 insertions(+), 10 deletions(-)
 create mode 100644 Documentation/devicetree/bindings/ipmi/aspeed-ssif-bmc.yaml
 create mode 100644 drivers/char/ipmi/ssif_bmc.c
 create mode 100644 drivers/char/ipmi/ssif_bmc.h
 create mode 100644 drivers/char/ipmi/ssif_bmc_aspeed.c
Signature · 1 line
2.28.0
#34 lore raw
Quoted 31 lines · Corey Minyard
Quoted 24 lines · Quan Nguyen

This series add support for the Aspeed specific SSIF BMC driver which is to perform in-band IPMI communication with the host in management (BMC) side.

v3:
+ Switched binding doc to use DT schema format [Rob]
+ Splited into generic ssif_bmc and aspeed-specific [Corey, Joel]
+ Removed redundant license info [Joel]
+ Switched to use traditional if-else [Joel]
+ Removed unused ssif_bmc_ioctl() [Joel]
+ Made handle_request()/complete_response() to return void [Joel]
+ Refactored send_ssif_bmc_response()/receive_ssif_bmc_request() [Corey]
+ Remove mutex [Corey]
+ Use spin_lock/unlock_irqsave/restore in callback [Corey]
+ Removed the unnecessary memset [Corey]
+ Switch to use dev_err() [Corey]
+ Combine mask/unmask two interrupts together [Corey]
+ Fixed unhandled Tx done with NAK [Quan]
+ Late ack'ed Tx done w/wo Ack irq [Quan]
+ Use aspeed-specific exported aspeed_set_slave_busy() when slave busy
to fix the deadlock [Graeme, Philipp, Quan]
+ Clean buffer for last multipart read [Quan]
+ Handle unknown incoming command [Quan]

Quoted 5 lines

v2:
+ Fixed compiling error with COMPILE_TEST for arc

Quan Nguyen (7):
i2c: i2c-core-smbus: Expose PEC calculate function for generic use

Quoted 9 lines

Note that for this I2C-specific change, I will need acks from the I2C maintainers to be able to include this in my tree.

Quoted 2 lines
Quoted 1 line

ipmi: ssif_bmc: Add SSIF BMC driver

i2c: aspeed: Fix unhandled Tx done with NAK

For the aspeed changes, they don't really belong here, they belong in the aspeed tree. I see that you need them for the device to work correctly, though. I'll need acks from maintainers to include them.

Thanks for you information.
Will separate these i2c-aspeed's patches into other independent series in next version.

Quoted 23 lines
Quoted 23 lines

i2c: aspeed: Acknowledge Tx done w/wo ACK irq late
i2c: aspeed: Add aspeed_set_slave_busy()
ipmi: ssif_bmc: Add Aspeed SSIF BMC driver
bindings: ipmi: Add binding for Aspeed SSIF BMC driver

  .../bindings/ipmi/aspeed-ssif-bmc.yaml        |  33 +
  drivers/char/ipmi/Kconfig                     |  22 +
  drivers/char/ipmi/Makefile                    |   2 +
  drivers/char/ipmi/ssif_bmc.c                  | 605 ++++++++++++++++++
  drivers/char/ipmi/ssif_bmc.h                  |  93 +++
  drivers/char/ipmi/ssif_bmc_aspeed.c           |  75 +++
  drivers/i2c/busses/i2c-aspeed.c               |  51 +-
  drivers/i2c/i2c-core-smbus.c                  |  12 +-
  include/linux/i2c.h                           |   1 +
  9 files changed, 884 insertions(+), 10 deletions(-)

create mode 100644 Documentation/devicetree/bindings/ipmi/aspeed-ssif-bmc.yaml create mode 100644 drivers/char/ipmi/ssif_bmc.c
create mode 100644 drivers/char/ipmi/ssif_bmc.h
create mode 100644 drivers/char/ipmi/ssif_bmc_aspeed.c

Signature · 1 line
2.28.0