Add Aspeed SSIF BMC driver
Subsystems (6)
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
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
@@ -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
@@ -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
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.Signed-off-by Quan Nguyen <[email protected]>
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>
PATCH v3 2/7 ipmi: ssif_bmc: Add SSIF BMC driver
In reply to Quan NguyenThe 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
@@ -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
@@ -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-$(CONFIG_SSIF_IPMI_BMC) += ssif_bmc.o
drivers/char/ipmi/ssif_bmc.c new file +605 -0
605 lines not on this page · raw message
drivers/char/ipmi/ssif_bmc.h new file +93 -0
93 lines not on this page · raw message
Signature · 1 line
2.28.0
Re: [PATCH v3 2/7] ipmi: ssif_bmc: Add SSIF BMC driver
Quoted 85 lines · Quan Nguyen
The SMBus system interface (SSIF) IPMI BMC driver can be used to perform in-band IPMI communication with their host in management (BMC) side.
Signed-off-by Quan Nguyen <[email protected]>---
v3:
+ Removed redundant license info [Joel]
+ Switched to use traditional if-else [Joel]
+ Removed unusedssif_bmc_ioctl()[Joel]
+ Made handle_request()/complete_response() to return void [Joel]
+ Refactored send_ssif_bmc_response()/receive_ssif_bmc_request() [Corey]
+ Removed mutex [Corey]
+ Use spin_lock/unlock_irqsave/restore in callback [Corey]
+ Removed the unnecessary memset [Corey]
+ Switch to usedev_err()[Corey]v2:
+ Fixed compiling error withCOMPILE_TESTfor arcdrivers/char/ipmi/Kconfig | 11 + drivers/char/ipmi/Makefile | 1 + drivers/char/ipmi/ssif_bmc.c | 605 +++++++++++++++++++++++++++++++++++ drivers/char/ipmi/ssif_bmc.h | 93 ++++++ 4 files changed, 710 insertions(+) create mode 100644 drivers/char/ipmi/ssif_bmc.c create mode 100644 drivers/char/ipmi/ssif_bmc.hdrivers/char/ipmi/Kconfig +11
-0@@ -133,6 +133,17 @@ config ASPEED_BT_IPMI_BMC found on Aspeed SOCs (AST2400 and AST2500). The driver implements the BMC side of the BT interface. +config SSIF_IPMI_BMC + tristate "SSIF IPMI BMC driver" + select I2C + select I2C_SLAVE + help + This enables the IPMI SMBus system interface (SSIF) at the + management (BMC) side. + + The driver implements the BMC side of the SMBus system + interface (SSIF). + config IPMB_DEVICE_INTERFACE tristate 'IPMB Interface handler' depends on I2Cdrivers/char/ipmi/Makefile +1
-0@@ -27,3 +27,4 @@ obj-$(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.odrivers/char/ipmi/ssif_bmc.c new file +21
-0@@ -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@@ -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
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.
Signed-off-by Quan Nguyen <[email protected]>---
v3:
+ Removed redundant license info [Joel]
+ Switched to use traditional if-else [Joel]
+ Removed unusedssif_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 usedev_err()[Corey]v2:
+ Fixed compiling error withCOMPILE_TESTfor arcdrivers/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 100644drivers/char/ipmi/ssif_bmc.hdrivers/char/ipmi/Kconfig +11
-0@@ -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 I2Cdrivers/char/ipmi/Makefile +1
-0@@ -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.odrivers/char/ipmi/ssif_bmc.c new file +26
-0@@ -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_NONBLOCKbefore.
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@@ -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
PATCH v3 3/7 i2c: aspeed: Fix unhandled Tx done with NAK
In reply to Quan NguyenIt 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
@@ -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
Joel Stanley
maintainerRe: [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. expected0x00000086, but was0x00000084"
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:
Quoted 14 lines
Signed-off-by Quan Nguyen <[email protected]>---
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
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
Ryan Chen
maintainerRE: [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 NAKRyan, 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. expected0x00000086, but was0x00000084"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 I2Cdriver")
Quoted 15 lines
Signed-off-by Quan Nguyen <[email protected]>---
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.cindex724bf30600d6..3fb37c3f23d4100644drivers/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_MATCHcase?Could you instead move the
TX_NAKhandling to be above theSLAVE_MATCHcase?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
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 NAKRyan, 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. expected0x00000086, but was0x00000084"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 I2Cdriver")
Quoted 15 lines
Signed-off-by Quan Nguyen <[email protected]>---
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.cindex724bf30600d6..3fb37c3f23d4100644drivers/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_MATCHcase?Could you instead move the
TX_NAKhandling to be above theSLAVE_MATCHcase?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_NAKwill be go toSLAVE_STOPstate?
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
Re: [PATCH v3 3/7] i2c: aspeed: Fix unhandled Tx done with NAK
In reply to Joel StanleyQuoted 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. expected0x00000086, but was0x00000084"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
Signed-off-by Quan Nguyen <[email protected]>---
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
-0drivers/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_MATCHcase?
Quoted 7 lines
Could you instead move the
TX_NAKhandling to be above theSLAVE_MATCHcase?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
PATCH v3 4/7 i2c: aspeed: Acknowledge Tx done w/wo ACK irq late
In reply to Quan NguyenWith 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
@@ -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
Joel Stanley
maintainerRe: [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
Signed-off-by Quan Nguyen <[email protected]>---
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
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
-6irq_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
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
2be6b47211e17e6c90ead40d24d2a5cc815f2d5cto 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
Signed-off-by Quan Nguyen <[email protected]>---
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
-0drivers/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
-6irq_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_REGat 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
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
2be6b47211e17e6c90ead40d24d2a5cc815f2d5cto 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
Signed-off-by Quan Nguyen <[email protected]>---
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
-0drivers/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-useirq_handledvariable */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_REGat 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);
???????? returnirq_remaining?IRQ_NONE:IRQ_HANDLED; ? }Signature · 1 line
2.28.0
Re: [PATCH v3 4/7] i2c: aspeed: Acknowledge Tx done w/wo ACK irq late
In reply to Joel StanleyQuoted 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
2be6b47211e17e6c90ead40d24d2a5cc815f2d5cto 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
Signed-off-by Quan Nguyen <[email protected]>---
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@@ -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
-2spin_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
-6irq_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_REGat 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
PATCH v3 5/7 i2c: aspeed: Add aspeed_set_slave_busy()
In reply to Quan NguyenSlave 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
@@ -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
Ryan Chen
maintainerRE: [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: Addaspeed_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.Signed-off-by Quan Nguyen <[email protected]>---
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@@ -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
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: Addaspeed_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.Signed-off-by Quan Nguyen <[email protected]>---
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@@ -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
--
2.28.0Hello,
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
Ryan Chen
maintainerRE: [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: Addaspeed_set_slave_busy()Quoted 3 lines · Ryan Chen
Quoted 3 lines
-----Original Message-----
From: openbmc <[email protected]> OnBehalf
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: Addaspeed_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 theresponse is ready.
Quoted 39 lines
Quoted 39 lines
Signed-off-by Quan Nguyen <[email protected]>---
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.cindexb2e9c8f0ddf7..9926d04831a2100644drivers/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){
structplatform_device*pdev =to_platform_device(bus->dev);
--
2.28.0Hello,
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.
RyanThanks 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
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: Addaspeed_set_slave_busy()Quoted 3 lines · Ryan Chen
Quoted 3 lines
-----Original Message-----
From: openbmc <[email protected]> OnBehalf
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: Addaspeed_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 theresponse is ready.
Quoted 47 lines
Quoted 42 lines
Signed-off-by Quan Nguyen <[email protected]>---
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.cindexb2e9c8f0ddf7..9926d04831a2100644drivers/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.
RyanThanks 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
Ryan Chen
maintainerRE: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()
In reply to Quan NguyenQuoted 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: Addaspeed_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.Signed-off-by Quan Nguyen <[email protected]>---
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@@ -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
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: Addaspeed_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.Signed-off-by Quan Nguyen <[email protected]>---
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@@ -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
Ryan Chen
maintainerRE: [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: Addaspeed_set_slave_busy()Quoted 3 lines · Ryan Chen
Quoted 3 lines
-----Original Message-----
From: openbmc <[email protected]> OnBehalf
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: Addaspeed_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 theresponse is ready.
Quoted 28 lines
Quoted 17 lines
Signed-off-by Quan Nguyen <[email protected]>---
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.cindexb2e9c8f0ddf7..9926d04831a2100644drivers/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 callingi2c_get_adapdata(client->adapter). And inaspeed_set_ssif_bmc_status(), call the exportedaspeed_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
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: Addaspeed_set_slave_busy()Quoted 3 lines · Ryan Chen
Quoted 3 lines
-----Original Message-----
From: openbmc <[email protected]> OnBehalf
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: Addaspeed_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 theresponse is ready.
Quoted 28 lines
Quoted 17 lines
Signed-off-by Quan Nguyen <[email protected]>---
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.cindexb2e9c8f0ddf7..9926d04831a2100644drivers/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 callingi2c_get_adapdata(client->adapter). And inaspeed_set_ssif_bmc_status(), call the exportedaspeed_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 thebus->baseaddress 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
Ryan Chen
maintainerRE: [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: Addaspeed_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: Addaspeed_set_slave_busy()Quoted 3 lines · Ryan Chen
Quoted 3 lines
-----Original Message-----
From: openbmc <[email protected]> OnBehalf
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 JefferyQuoted 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]>; ThangQ .
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: Addaspeed_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 theresponse is ready.
Quoted 28 lines
Quoted 17 lines
Signed-off-by Quan Nguyen <[email protected]>---
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.cindexb2e9c8f0ddf7..9926d04831a2100644drivers/i2c/busses/i2c-aspeed.c +0
-0@@ -944,6 +944,26 @@ static int aspeed_i2c_init(structaspeed_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 callingi2c_get_adapdata(client->adapter). And ->inaspeed_set_ssif_bmc_status(), call the exportedaspeed_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->baseaddress 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);", thebus->baseshould point to the base of theaspeed_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
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: Addaspeed_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: Addaspeed_set_slave_busy()Quoted 3 lines · Ryan Chen
Quoted 3 lines
-----Original Message-----
From: openbmc <[email protected]> OnBehalf
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 JefferyQuoted 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]>; ThangQ .
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: Addaspeed_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 theresponse is ready.
Quoted 28 lines
Quoted 17 lines
Signed-off-by Quan Nguyen <[email protected]>---
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.cindexb2e9c8f0ddf7..9926d04831a2100644drivers/i2c/busses/i2c-aspeed.c +0
-0@@ -944,6 +944,26 @@ static int aspeed_i2c_init(structaspeed_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 callingi2c_get_adapdata(client->adapter). And ->inaspeed_set_ssif_bmc_status(), call the exportedaspeed_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->baseaddress 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);", thebus->baseshould point to the base of theaspeed_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 thessif_bmc_aspeed.cassume 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 structi2c_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
@@ -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
Ryan Chen
maintainerRE: [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: Addaspeed_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: Addaspeed_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 HigginsQuoted 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]>; ThangQ .
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: Addaspeed_set_slave_busy()Quoted 3 lines · Ryan Chen
Quoted 3 lines
-----Original Message-----
From: openbmc <[email protected]> OnBehalf
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 HigginsQuoted 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: Addaspeed_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 theresponse is ready.
Quoted 29 lines
Quoted 18 lines
Signed-off-by Quan Nguyen <[email protected]>---
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.cindexb2e9c8f0ddf7..9926d04831a2
100644drivers/i2c/busses/i2c-aspeed.c +0
-0@@ -944,6 +944,26 @@ static int aspeed_i2c_init(structaspeed_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 callingi2c_get_adapdata(client->adapter). ->And inaspeed_set_ssif_bmc_status(), call the exportedaspeed_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->baseaddress 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);", thebus->baseshould point to the base of theaspeed_i2c_bus, which is already initialized by the aspeed i2c bus driver.Do I miss something?
Hello Quan,
After study. I think thessif_bmc_aspeed.cassume the priv data is thesame 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 structi2c_algorithm; is it workable?Quoted 1 line
Quoted 7 lines
Hi Ryan,
I dont know which is better, ie: adding callback to struct
i2c_algorithmor to structi2c_adapter.
I have tried to add generic callback to structi2c_adapteras 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..6e9abf2d6abb100644include/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
Re: [PATCH v3 5/7] i2c: aspeed: Add aspeed_set_slave_busy()
In reply to Quan NguyenQuoted 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.Signed-off-by Quan Nguyen <[email protected]>---
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
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
PATCH v3 6/7 ipmi: ssif_bmc: Add Aspeed SSIF BMC driver
In reply to Quan NguyenThis 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
@@ -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
@@ -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
PATCH v3 7/7 bindings: ipmi: Add binding for Aspeed SSIF BMC driver
In reply to Quan NguyenAdd 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
@@ -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
Rob Herring
maintainerRe: [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.
Signed-off-by Quan Nguyen <[email protected]>---
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.
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.
Signed-off-by Quan Nguyen <[email protected]>---
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.yamlMy bot found errors running 'make DT_CHECKER_FLAGS=-m
dt_binding_check' on your patch (DT_CHECKER_FLAGSis 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.yamlSee 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
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 genericssif_bmcand aspeed-specific [Corey, Joel]
+ Removed redundant license info [Joel]
+ Switched to use traditional if-else [Joel]
+ Removed unusedssif_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 usedev_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 exportedaspeed_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 withCOMPILE_TESTfor arcQuan 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: Addaspeed_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.cSignature · 1 line
2.28.0
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 genericssif_bmcand aspeed-specific [Corey, Joel]
+ Removed redundant license info [Joel]
+ Switched to use traditional if-else [Joel]
+ Removed unusedssif_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 usedev_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 exportedaspeed_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 withCOMPILE_TESTfor arcQuan 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 driveri2c: 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: Addaspeed_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.yamlcreate mode 100644drivers/char/ipmi/ssif_bmc.c
create mode 100644drivers/char/ipmi/ssif_bmc.h
create mode 100644drivers/char/ipmi/ssif_bmc_aspeed.cSignature · 1 line
2.28.0