lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <de2698d9-7c74-2abb-ffbb-dff408d8415b@grandegger.com>
Date:   Tue, 22 Jan 2019 15:34:02 +0100
From:   Wolfgang Grandegger <wg@...ndegger.com>
To:     Dan Murphy <dmurphy@...com>, mkl@...gutronix.de,
        davem@...emloft.net, b29396@...escale.com
Cc:     linux-can@...r.kernel.org, netdev@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH v4 2/4] can: m_can: Migrate the m_can code to use the
 framework

Hello Dan,

Am 22.01.19 um 14:37 schrieb Dan Murphy:
> Wolfgang
> 
> On 1/22/19 3:35 AM, Wolfgang Grandegger wrote:
>> Hello,
>>
>> Am 17.01.19 um 21:05 schrieb Dan Murphy:
>>> Migrate the m_can code to use the m_can_platform framework
>>> code.
>>>
>>> Signed-off-by: Dan Murphy <dmurphy@...com>
>>> ---
>>>  drivers/net/can/m_can/Kconfig          |  12 +
>>>  drivers/net/can/m_can/Makefile         |   4 +-
>>>  drivers/net/can/m_can/m_can.c          | 764 ++++++++++++-------------
>>>  drivers/net/can/m_can/m_can_platform.h |   2 +-
>>>  4 files changed, 374 insertions(+), 408 deletions(-)
>>>
>>> diff --git a/drivers/net/can/m_can/Kconfig b/drivers/net/can/m_can/Kconfig
>>> index 04f20dd39007..b1a9358b7660 100644
>>> --- a/drivers/net/can/m_can/Kconfig
>>> +++ b/drivers/net/can/m_can/Kconfig
>>> @@ -1,5 +1,17 @@
>>>  config CAN_M_CAN
>>> +	tristate "Bosch M_CAN support"
>>> +	---help---
>>> +	  Say Y here if you want to support for Bosch M_CAN controller.
>>> +
>>> +config CAN_M_CAN_CORE
>>> +	depends on CAN_M_CAN
>>> +	tristate "Bosch M_CAN Core support"
>>> +	---help---
>>> +	  Say Y here if you want to support for Bosch M_CAN controller.
>>
>> Do you need that extra config? I think "CAN_M_CAN" is just fine.
>>
> 
> OK I can remove the CORE.  That should be built if CAN_M_CAN is selected then
> 
>>> +config CAN_M_CAN_PLATFORM
>>>  	depends on HAS_IOMEM
>>> +	depends on CAN_M_CAN_CORE
>>>  	tristate "Bosch M_CAN devices"
>>>  	---help---
>>>  	  Say Y here if you want to support for Bosch M_CAN controller.
>>> diff --git a/drivers/net/can/m_can/Makefile b/drivers/net/can/m_can/Makefile
>>> index 8bbd7f24f5be..04f36947ac3b 100644
>>> --- a/drivers/net/can/m_can/Makefile
>>> +++ b/drivers/net/can/m_can/Makefile
>>> @@ -2,4 +2,6 @@
>>>  #  Makefile for the Bosch M_CAN controller driver.
>>>  #
>>>  
>>> -obj-$(CONFIG_CAN_M_CAN) += m_can.o
>>> +obj-$(CONFIG_CAN_M_CAN_CORE) += m_can.o
>>> +obj-$(CONFIG_CAN_M_CAN_PLATFORM) += m_can_platform.o
>>> +obj-$(CONFIG_CAN_M_CAN_TCAN4X5X) += tcan4x5x.o
>>
>> This file is provided in a sub-sequent patched! The code *must* compile
>> for every single patch applied for bisect'ing. Looking to your first
>> patch, I just realize that this is also not the case!
>>
>> I think it makes sense to squash patch 1 and 2.
>>
> 
> Well.  Patch 1 should compile on its own I will check that.
> 
> Squashing patch 1 and patch 2 made the review even more impossible.
> So I made patch 1 to create the io-mapped code (which probably will not compile when enabled) and then
> patch 2 was the changes to the CORE to stitch in the io-mapped code.

Patch 1 just changes two small locations of "m_can.c". It does not use
the new interface or even include "m_can_platform.h". Patch 2 is
well readable, especially the common code. It would be even better if
"struct can_priv" would be kept and all the functions in the original
order. That's why I suggested to do so.

> Patch 4 was the addition of the peripherial.
> 
> I did this because there was a comment from the RFCs I sent in that there was to much change to review for a single patch
> conversion.  I can squash patch 1 and patch 2 after we finish the review process prior to merging it into the tree.
> 
> That was also the comment I received from an internal review.

Sorry, if I confused you! My concern was about the common code in m_can.c.

>>> diff --git a/drivers/net/can/m_can/m_can.c b/drivers/net/can/m_can/m_can.c
>>> index f817b28582e9..6da0ae26138e 100644
>>> --- a/drivers/net/can/m_can/m_can.c
>>> +++ b/drivers/net/can/m_can/m_can.c
>>> @@ -28,87 +28,14 @@
>>>  #include <linux/can/dev.h>
>>>  #include <linux/pinctrl/consumer.h>
>>>  
>>> +#include "m_can_platform.h"
>>> +
>>>  /* napi related */
>>>  #define M_CAN_NAPI_WEIGHT	64
>>>  
>>>  /* message ram configuration data length */
>>>  #define MRAM_CFG_LEN	8
>>>  
>>> -/* registers definition */
>>> -enum m_can_reg {
>>> -	M_CAN_CREL	= 0x0,
>>> -	M_CAN_ENDN	= 0x4,
>>> -	M_CAN_CUST	= 0x8,
>>> -	M_CAN_DBTP	= 0xc,
>>> -	M_CAN_TEST	= 0x10,
>>> -	M_CAN_RWD	= 0x14,
>>> -	M_CAN_CCCR	= 0x18,
>>> -	M_CAN_NBTP	= 0x1c,
>>> -	M_CAN_TSCC	= 0x20,
>>> -	M_CAN_TSCV	= 0x24,
>>> -	M_CAN_TOCC	= 0x28,
>>> -	M_CAN_TOCV	= 0x2c,
>>> -	M_CAN_ECR	= 0x40,
>>> -	M_CAN_PSR	= 0x44,
>>> -/* TDCR Register only available for version >=3.1.x */
>>> -	M_CAN_TDCR	= 0x48,
>>> -	M_CAN_IR	= 0x50,
>>> -	M_CAN_IE	= 0x54,
>>> -	M_CAN_ILS	= 0x58,
>>> -	M_CAN_ILE	= 0x5c,
>>> -	M_CAN_GFC	= 0x80,
>>> -	M_CAN_SIDFC	= 0x84,
>>> -	M_CAN_XIDFC	= 0x88,
>>> -	M_CAN_XIDAM	= 0x90,
>>> -	M_CAN_HPMS	= 0x94,
>>> -	M_CAN_NDAT1	= 0x98,
>>> -	M_CAN_NDAT2	= 0x9c,
>>> -	M_CAN_RXF0C	= 0xa0,
>>> -	M_CAN_RXF0S	= 0xa4,
>>> -	M_CAN_RXF0A	= 0xa8,
>>> -	M_CAN_RXBC	= 0xac,
>>> -	M_CAN_RXF1C	= 0xb0,
>>> -	M_CAN_RXF1S	= 0xb4,
>>> -	M_CAN_RXF1A	= 0xb8,
>>> -	M_CAN_RXESC	= 0xbc,
>>> -	M_CAN_TXBC	= 0xc0,
>>> -	M_CAN_TXFQS	= 0xc4,
>>> -	M_CAN_TXESC	= 0xc8,
>>> -	M_CAN_TXBRP	= 0xcc,
>>> -	M_CAN_TXBAR	= 0xd0,
>>> -	M_CAN_TXBCR	= 0xd4,
>>> -	M_CAN_TXBTO	= 0xd8,
>>> -	M_CAN_TXBCF	= 0xdc,
>>> -	M_CAN_TXBTIE	= 0xe0,
>>> -	M_CAN_TXBCIE	= 0xe4,
>>> -	M_CAN_TXEFC	= 0xf0,
>>> -	M_CAN_TXEFS	= 0xf4,
>>> -	M_CAN_TXEFA	= 0xf8,
>>> -};
>>> -
>>> -/* m_can lec values */
>>> -enum m_can_lec_type {
>>> -	LEC_NO_ERROR = 0,
>>> -	LEC_STUFF_ERROR,
>>> -	LEC_FORM_ERROR,
>>> -	LEC_ACK_ERROR,
>>> -	LEC_BIT1_ERROR,
>>> -	LEC_BIT0_ERROR,
>>> -	LEC_CRC_ERROR,
>>> -	LEC_UNUSED,
>>> -};
>>> -
>>> -enum m_can_mram_cfg {
>>> -	MRAM_SIDF = 0,
>>> -	MRAM_XIDF,
>>> -	MRAM_RXF0,
>>> -	MRAM_RXF1,
>>> -	MRAM_RXB,
>>> -	MRAM_TXE,
>>> -	MRAM_TXB,
>>> -	MRAM_CFG_NUM,
>>> -};
>>
>> Patch 1 should have already done that!
>>
> 
> Will look at it.

As I said above. Patch 1 did *not* touch "m_can.c". Therefore the
definitions above are in both files, m_can.c *and* m_can_platform.h.

> 
>>>  /* Core Release Register (CREL) */
>>>  #define CREL_REL_SHIFT		28
>>>  #define CREL_REL_MASK		(0xF << CREL_REL_SHIFT)
>>> @@ -343,72 +270,81 @@ enum m_can_mram_cfg {
>>>  #define TX_BUF_MM_MASK		(0xff << TX_BUF_MM_SHIFT)
>>>  
>>>  /* Tx event FIFO Element */
>>> -/* E1 */
>>>  #define TX_EVENT_MM_SHIFT	TX_BUF_MM_SHIFT
>>>  #define TX_EVENT_MM_MASK	(0xff << TX_EVENT_MM_SHIFT)
>>>  
>>> -/* address offset and element number for each FIFO/Buffer in the Message RAM */
>>> -struct mram_cfg {
>>> -	u16 off;
>>> -	u8  num;
>>> -};
>>> +static u32 m_can_read(struct m_can_classdev *priv, enum m_can_reg reg)
>>> +{
>>> +	u32 ret = -EINVAL;
>>>  
>>> -/* m_can private data structure */
>>> -struct m_can_priv {
>>> -	struct can_priv can;	/* must be the first member */
>>> -	struct napi_struct napi;
>>> -	struct net_device *dev;
>>> -	struct device *device;
>>> -	struct clk *hclk;
>>> -	struct clk *cclk;
>>> -	void __iomem *base;
>>> -	u32 irqstatus;
>>> -	int version;
>>> -
>>> -	/* message ram configuration */
>>> -	void __iomem *mram_base;
>>> -	struct mram_cfg mcfg[MRAM_CFG_NUM];
>>> -};
>>> +	if (priv->read_reg)
>>> +		ret = priv->read_reg(priv, reg);
>>>  
>>> -static inline u32 m_can_read(const struct m_can_priv *priv, enum m_can_reg reg)
>>> +	return ret;
>>> +}
>>> +
>>> +static int m_can_write(struct m_can_classdev *priv, enum m_can_reg reg, u32 val)
>>>  {
>>> -	return readl(priv->base + reg);
>>> +	int ret = -EINVAL;
>>> +
>>> +	if (priv->write_reg)
>>> +		ret = priv->write_reg(priv, reg, val);
>>> +
>>> +	return ret;
>>>  }
>>>  
>>> -static inline void m_can_write(const struct m_can_priv *priv,
>>> -			       enum m_can_reg reg, u32 val)
>>> +static u32 m_can_fifo_read(struct m_can_classdev *priv,
>>> +			   u32 fgi, unsigned int offset)
>>>  {
>>> -	writel(val, priv->base + reg);
>>> +	u32 addr_offset = priv->mcfg[MRAM_RXF0].off + fgi * RXF0_ELEMENT_SIZE + offset;
>>> +	u32 ret = -EINVAL;
>>> +
>>> +	if (priv->read_fifo)
>>> +		ret = priv->read_fifo(priv, addr_offset);
>>> +
>>> +	return ret;
>>>  }
>>>  
>>> -static inline u32 m_can_fifo_read(const struct m_can_priv *priv,
>>> -				  u32 fgi, unsigned int offset)
>>> +static u32 m_can_fifo_write(struct m_can_classdev *priv,
>>> +			    u32 fpi, unsigned int offset, u32 val)
>>>  {
>>> -	return readl(priv->mram_base + priv->mcfg[MRAM_RXF0].off +
>>> -		     fgi * RXF0_ELEMENT_SIZE + offset);
>>> +	u32 addr_offset =  priv->mcfg[MRAM_TXB].off + fpi * TXB_ELEMENT_SIZE + offset;
>>> +	u32 ret = -EINVAL;
>>> +
>>> +	if (priv->write_fifo)
>>> +		ret = priv->write_fifo(priv, addr_offset, val);
>>> +
>>> +	return ret;
>>
>> Why not just:
>>
>> 	if (priv->write_fifo)
>> 		return 	priv->write_fifo(priv, addr_offset, val);
>> 	else
>> 		return -EINVAL;
>>
>> Here and below...
>>
> 
> Ack.
> 
>>>  }
>>>  
>>> -static inline void m_can_fifo_write(const struct m_can_priv *priv,
>>> -				    u32 fpi, unsigned int offset, u32 val)
>>> +static u32 m_can_fifo_write_no_off(struct m_can_classdev *priv,
>>> +				   u32 fpi, u32 val)
>>>  {
>>> -	writel(val, priv->mram_base + priv->mcfg[MRAM_TXB].off +
>>> -	       fpi * TXB_ELEMENT_SIZE + offset);
>>> +	u32 ret = 0;
>>> +
>>> +	if (priv->write_fifo)
>>> +		ret = priv->write_fifo(priv, fpi, val);
>>> +
>>> +	return ret;
>>>  }
>>>  
>>> -static inline u32 m_can_txe_fifo_read(const struct m_can_priv *priv,
>>> -				      u32 fgi,
>>> -				      u32 offset) {
>>> -	return readl(priv->mram_base + priv->mcfg[MRAM_TXE].off +
>>> -			fgi * TXE_ELEMENT_SIZE + offset);
>>> +static u32 m_can_txe_fifo_read(struct m_can_classdev *priv, u32 fgi, u32 offset)
>>> +{
>>> +	u32 addr_offset = priv->mcfg[MRAM_TXE].off + fgi * TXE_ELEMENT_SIZE + offset;
>>> +	u32 ret = -EINVAL;
>>> +
>>> +	if (priv->read_fifo)
>>> +		ret = priv->read_fifo(priv, addr_offset);
>>> +
>>> +	return ret;
>>>  }
>>>  
>>> -static inline bool m_can_tx_fifo_full(const struct m_can_priv *priv)
>>> +static inline bool m_can_tx_fifo_full(struct m_can_classdev *priv)
>>>  {
>>> -		return !!(m_can_read(priv, M_CAN_TXFQS) & TXFQS_TFQF);
>>> +	return !!(m_can_read(priv, M_CAN_TXFQS) & TXFQS_TFQF);
>>>  }
>>>  
>>> -static inline void m_can_config_endisable(const struct m_can_priv *priv,
>>> -					  bool enable)
>>> +void m_can_config_endisable(struct m_can_classdev *priv, bool enable)
>>>  {
>>>  	u32 cccr = m_can_read(priv, M_CAN_CCCR);
>>>  	u32 timeout = 10;
>>> @@ -433,7 +369,7 @@ static inline void m_can_config_endisable(const struct m_can_priv *priv,
>>>  
>>>  	while ((m_can_read(priv, M_CAN_CCCR) & (CCCR_INIT | CCCR_CCE)) != val) {
>>>  		if (timeout == 0) {
>>> -			netdev_warn(priv->dev, "Failed to init module\n");
>>> +			netdev_warn(priv->net, "Failed to init module\n");
>>>  			return;
>>>  		}
>>>  		timeout--;
>>> @@ -441,13 +377,13 @@ static inline void m_can_config_endisable(const struct m_can_priv *priv,
>>>  	}
>>>  }
>>>  
>>> -static inline void m_can_enable_all_interrupts(const struct m_can_priv *priv)
>>> +static inline void m_can_enable_all_interrupts(struct m_can_classdev *priv)
>>>  {
>>>  	/* Only interrupt line 0 is used in this driver */
>>>  	m_can_write(priv, M_CAN_ILE, ILE_EINT0);
>>>  }
>>>  
>>> -static inline void m_can_disable_all_interrupts(const struct m_can_priv *priv)
>>> +static inline void m_can_disable_all_interrupts(struct m_can_classdev *priv)
>>>  {
>>>  	m_can_write(priv, M_CAN_ILE, 0x0);
>>>  }
>>> @@ -455,7 +391,7 @@ static inline void m_can_disable_all_interrupts(const struct m_can_priv *priv)
>>>  static void m_can_read_fifo(struct net_device *dev, u32 rxfs)
>>>  {
>>>  	struct net_device_stats *stats = &dev->stats;
>>> -	struct m_can_priv *priv = netdev_priv(dev);
>>> +	struct m_can_classdev *priv = netdev_priv(dev);
>>>  	struct canfd_frame *cf;
>>>  	struct sk_buff *skb;
>>>  	u32 id, fgi, dlc;
>>> @@ -512,7 +448,7 @@ static void m_can_read_fifo(struct net_device *dev, u32 rxfs)
>>>  
>>>  static int m_can_do_rx_poll(struct net_device *dev, int quota)
>>>  {
>>> -	struct m_can_priv *priv = netdev_priv(dev);
>>> +	struct m_can_classdev *priv = netdev_priv(dev);
>>>  	u32 pkts = 0;
>>>  	u32 rxfs;
>>>  
>>> @@ -565,7 +501,7 @@ static int m_can_handle_lost_msg(struct net_device *dev)
>>>  static int m_can_handle_lec_err(struct net_device *dev,
>>>  				enum m_can_lec_type lec_type)
>>>  {
>>> -	struct m_can_priv *priv = netdev_priv(dev);
>>> +	struct m_can_classdev *priv = netdev_priv(dev);
>>>  	struct net_device_stats *stats = &dev->stats;
>>>  	struct can_frame *cf;
>>>  	struct sk_buff *skb;
>>> @@ -622,7 +558,7 @@ static int m_can_handle_lec_err(struct net_device *dev,
>>>  static int __m_can_get_berr_counter(const struct net_device *dev,
>>>  				    struct can_berr_counter *bec)
>>>  {
>>> -	struct m_can_priv *priv = netdev_priv(dev);
>>> +	struct m_can_classdev *priv = netdev_priv(dev);
>>
>> Back to the naming: You use "priv" here because I only wanted to see
>> minimal changes to the m_can code. Why not also re-using "struct
>> m_can_priv" for the moment and use:
>>
>>   m_can_register(struct m_can_priv priv);
>>
>> Later-on, when the development has settled, we could introduce betetr
>> names. What do you think?
>>
> 
> m_can_classdev is the common struct for the devices that register.
> m_can_priv is the private struct for the device itself.  If you look at the m_can_priv
> struct and the tcan_priv struct they contain very different variables.
> 
> This is why I changed this to the common class struct so that the m_can common code has a common
> struct.

I know and I agree that the old "stuct m_can_priv" is not a good name.
The only reason why I suggest to keep it for the moment is to make the
changes to the common code minimal, esepcially to get rid of:

-       struct m_can_priv *priv = netdev_priv(dev);
+       struct m_can_classdev *priv = netdev_priv(dev);

Well, just leave it as it is now! I can follow the changes.

>>>  	unsigned int ecr;
>>>  
>>>  	ecr = m_can_read(priv, M_CAN_ECR);
>>> @@ -632,28 +568,32 @@ static int __m_can_get_berr_counter(const struct net_device *dev,
>>>  	return 0;
>>>  }
>>>  
>>> -static int m_can_clk_start(struct m_can_priv *priv)
>>> +static int m_can_clk_start(struct m_can_classdev *priv)
>>>  {
>>>  	int err;
>>>  
>>> -	err = pm_runtime_get_sync(priv->device);
>>> +	if (priv->pm_clock_support == 0)
>>> +		return 0;
>>> +
>>> +	err = pm_runtime_get_sync(priv->dev);
>>>  	if (err < 0) {
>>> -		pm_runtime_put_noidle(priv->device);
>>> +		pm_runtime_put_noidle(priv->dev);
>>>  		return err;
>>>  	}
>>>  
>>>  	return 0;
>>>  }
>>>  
>>> -static void m_can_clk_stop(struct m_can_priv *priv)
>>> +static void m_can_clk_stop(struct m_can_classdev *priv)
>>>  {
>>> -	pm_runtime_put_sync(priv->device);
>>> +	if (priv->pm_clock_support)
>>> +		pm_runtime_put_sync(priv->dev);
>>>  }
>>>  
>>>  static int m_can_get_berr_counter(const struct net_device *dev,
>>>  				  struct can_berr_counter *bec)
>>>  {
>>> -	struct m_can_priv *priv = netdev_priv(dev);
>>> +	struct m_can_classdev *priv = netdev_priv(dev);
>>>  	int err;
>>>  
>>>  	err = m_can_clk_start(priv);
>>> @@ -670,7 +610,7 @@ static int m_can_get_berr_counter(const struct net_device *dev,
>>>  static int m_can_handle_state_change(struct net_device *dev,
>>>  				     enum can_state new_state)
>>>  {
>>> -	struct m_can_priv *priv = netdev_priv(dev);
>>> +	struct m_can_classdev *priv = netdev_priv(dev);
>>>  	struct net_device_stats *stats = &dev->stats;
>>>  	struct can_frame *cf;
>>>  	struct sk_buff *skb;
>>> @@ -744,25 +684,22 @@ static int m_can_handle_state_change(struct net_device *dev,
>>>  
>>>  static int m_can_handle_state_errors(struct net_device *dev, u32 psr)
>>>  {
>>> -	struct m_can_priv *priv = netdev_priv(dev);
>>> +	struct m_can_classdev *priv = netdev_priv(dev);
>>>  	int work_done = 0;
>>>  
>>> -	if ((psr & PSR_EW) &&
>>> -	    (priv->can.state != CAN_STATE_ERROR_WARNING)) {
>>> +	if ((psr & PSR_EW) && priv->can.state != CAN_STATE_ERROR_WARNING) {
>>
>> This is an unrelated cosmetic change. We should avoid them here.
>> Could be done later-on with an extra patch.
>>
> 
> Ack
> 
>>>  		netdev_dbg(dev, "entered error warning state\n");
>>>  		work_done += m_can_handle_state_change(dev,
>>>  						       CAN_STATE_ERROR_WARNING);
>>>  	}
>>>  
>>> -	if ((psr & PSR_EP) &&
>>> -	    (priv->can.state != CAN_STATE_ERROR_PASSIVE)) {
>>> +	if ((psr & PSR_EP) && priv->can.state != CAN_STATE_ERROR_PASSIVE) {
>>>  		netdev_dbg(dev, "entered error passive state\n");
>>>  		work_done += m_can_handle_state_change(dev,
>>>  						       CAN_STATE_ERROR_PASSIVE);
>>>  	}
>>>  
>>> -	if ((psr & PSR_BO) &&
>>> -	    (priv->can.state != CAN_STATE_BUS_OFF)) {
>>> +	if ((psr & PSR_BO) && priv->can.state != CAN_STATE_BUS_OFF) {
>>>  		netdev_dbg(dev, "entered error bus off state\n");
>>>  		work_done += m_can_handle_state_change(dev,
>>>  						       CAN_STATE_BUS_OFF);
>>> @@ -797,7 +734,7 @@ static inline bool is_lec_err(u32 psr)
>>>  static int m_can_handle_bus_errors(struct net_device *dev, u32 irqstatus,
>>>  				   u32 psr)
>>>  {
>>> -	struct m_can_priv *priv = netdev_priv(dev);
>>> +	struct m_can_classdev *priv = netdev_priv(dev);
>>>  	int work_done = 0;
>>>  
>>>  	if (irqstatus & IR_RF0L)
>>> @@ -814,10 +751,9 @@ static int m_can_handle_bus_errors(struct net_device *dev, u32 irqstatus,
>>>  	return work_done;
>>>  }
>>>  
>>> -static int m_can_poll(struct napi_struct *napi, int quota)
>>> +static int m_can_rx_handler(struct net_device *dev, int quota)
>>>  {
>>> -	struct net_device *dev = napi->dev;
>>> -	struct m_can_priv *priv = netdev_priv(dev);
>>> +	struct m_can_classdev *priv = netdev_priv(dev);
>>>  	int work_done = 0;
>>>  	u32 irqstatus, psr;
>>>  
>>> @@ -834,13 +770,33 @@ static int m_can_poll(struct napi_struct *napi, int quota)
>>>  
>>>  	if (irqstatus & IR_RF0N)
>>>  		work_done += m_can_do_rx_poll(dev, (quota - work_done));
>>> +end:
>>> +	return work_done;
>>> +}
>>> +
>>> +static int m_can_rx(struct net_device *dev)
>>
>> m_can_rx_peripheral ?
>>
> 
> Ack
> 
>>> +{
>>> +	struct m_can_classdev *priv = netdev_priv(dev);
>>>  
>>> +	m_can_rx_handler(dev, 1);
>>> +
>>> +	m_can_enable_all_interrupts(priv);
>>> +
>>> +	return 0;
>>> +}
>>> +
>>> +static int m_can_poll(struct napi_struct *napi, int quota)
>>> +{
>>> +	struct net_device *dev = napi->dev;
>>> +	struct m_can_classdev *priv = netdev_priv(dev);
>>> +	int work_done = 0;
>>> +
>>> +	work_done = m_can_rx_handler(dev, quota);
>>>  	if (work_done < quota) {
>>>  		napi_complete_done(napi, work_done);
>>>  		m_can_enable_all_interrupts(priv);
>>>  	}
>>>  
>>> -end:
>>>  	return work_done;
>>>  }
>>>  
>>> @@ -852,7 +808,7 @@ static void m_can_echo_tx_event(struct net_device *dev)
>>>  	int i = 0;
>>>  	unsigned int msg_mark;
>>>  
>>> -	struct m_can_priv *priv = netdev_priv(dev);
>>> +	struct m_can_classdev *priv = netdev_priv(dev);
>>>  	struct net_device_stats *stats = &dev->stats;
>>>  
>>>  	/* read tx event fifo status */
>>> @@ -885,7 +841,7 @@ static void m_can_echo_tx_event(struct net_device *dev)
>>>  static irqreturn_t m_can_isr(int irq, void *dev_id)
>>>  {
>>>  	struct net_device *dev = (struct net_device *)dev_id;
>>> -	struct m_can_priv *priv = netdev_priv(dev);
>>> +	struct m_can_classdev *priv = netdev_priv(dev);
>>>  	struct net_device_stats *stats = &dev->stats;
>>>  	u32 ir;
>>>  
>>> @@ -905,7 +861,10 @@ static irqreturn_t m_can_isr(int irq, void *dev_id)
>>>  	if ((ir & IR_RF0N) || (ir & IR_ERR_ALL_30X)) {
>>>  		priv->irqstatus = ir;
>>>  		m_can_disable_all_interrupts(priv);
>>> -		napi_schedule(&priv->napi);
>>> +		if (!priv->is_peripherial)
>>> +			napi_schedule(&priv->napi);
>>> +		else
>>> +			m_can_rx(dev);
>>>  	}
>>>  
>>>  	if (priv->version == 30) {
>>> @@ -927,6 +886,9 @@ static irqreturn_t m_can_isr(int irq, void *dev_id)
>>>  		}
>>>  	}
>>>  
>>> +	if (priv->clr_dev_interrupts)
>>> +		priv->clr_dev_interrupts(priv);
>>> +
>>>  	return IRQ_HANDLED;
>>>  }
>>>  
>>> @@ -980,7 +942,7 @@ static const struct can_bittiming_const m_can_data_bittiming_const_31X = {
>>>  
>>>  static int m_can_set_bittiming(struct net_device *dev)
>>>  {
>>> -	struct m_can_priv *priv = netdev_priv(dev);
>>> +	struct m_can_classdev *priv = netdev_priv(dev);
>>>  	const struct can_bittiming *bt = &priv->can.bittiming;
>>>  	const struct can_bittiming *dbt = &priv->can.data_bittiming;
>>>  	u16 brp, sjw, tseg1, tseg2;
>>> @@ -1053,7 +1015,7 @@ static int m_can_set_bittiming(struct net_device *dev)
>>>   */
>>>  static void m_can_chip_config(struct net_device *dev)
>>>  {
>>> -	struct m_can_priv *priv = netdev_priv(dev);
>>> +	struct m_can_classdev *priv = netdev_priv(dev);
>>>  	u32 cccr, test;
>>>  
>>>  	m_can_config_endisable(priv, true);
>>> @@ -1165,7 +1127,7 @@ static void m_can_chip_config(struct net_device *dev)
>>>  
>>>  static void m_can_start(struct net_device *dev)
>>>  {
>>> -	struct m_can_priv *priv = netdev_priv(dev);
>>> +	struct m_can_classdev *priv = netdev_priv(dev);
>>>  
>>>  	/* basic m_can configuration */
>>>  	m_can_chip_config(dev);
>>> @@ -1194,20 +1156,17 @@ static int m_can_set_mode(struct net_device *dev, enum can_mode mode)
>>>   * else it returns the release and step coded as:
>>>   * return value = 10 * <release> + 1 * <step>
>>>   */
>>> -static int m_can_check_core_release(void __iomem *m_can_base)
>>> +static int m_can_check_core_release(struct m_can_classdev *priv)
>>>  {
>>>  	u32 crel_reg;
>>>  	u8 rel;
>>>  	u8 step;
>>>  	int res;
>>> -	struct m_can_priv temp_priv = {
>>> -		.base = m_can_base
>>> -	};
>>>  
>>>  	/* Read Core Release Version and split into version number
>>>  	 * Example: Version 3.2.1 => rel = 3; step = 2; substep = 1;
>>>  	 */
>>> -	crel_reg = m_can_read(&temp_priv, M_CAN_CREL);
>>> +	crel_reg = m_can_read(priv, M_CAN_CREL);
>>>  	rel = (u8)((crel_reg & CREL_REL_MASK) >> CREL_REL_SHIFT);
>>>  	step = (u8)((crel_reg & CREL_STEP_MASK) >> CREL_STEP_SHIFT);
>>>  
>>> @@ -1225,18 +1184,22 @@ static int m_can_check_core_release(void __iomem *m_can_base)
>>>  /* Selectable Non ISO support only in version 3.2.x
>>>   * This function checks if the bit is writable.
>>>   */
>>> -static bool m_can_niso_supported(const struct m_can_priv *priv)
>>> +static bool m_can_niso_supported(struct m_can_classdev *priv)
>>>  {
>>> -	u32 cccr_reg, cccr_poll;
>>> -	int niso_timeout;
>>> +	u32 cccr_reg, cccr_poll = 0;
>>> +	int niso_timeout = -ETIMEDOUT;
>>> +	int i;
>>>  
>>>  	m_can_config_endisable(priv, true);
>>>  	cccr_reg = m_can_read(priv, M_CAN_CCCR);
>>>  	cccr_reg |= CCCR_NISO;
>>>  	m_can_write(priv, M_CAN_CCCR, cccr_reg);
>>>  
>>> -	niso_timeout = readl_poll_timeout((priv->base + M_CAN_CCCR), cccr_poll,
>>> -					  (cccr_poll == cccr_reg), 0, 10);
>>> +	for (i = 0; i <= 10; i++) {
>>> +		cccr_poll = m_can_read(priv, M_CAN_CCCR);
>>> +		if (cccr_poll == cccr_reg)
>>> +			niso_timeout = 0;
>>> +	}
>>
>> This change is also unrelated. Should be done in an extra patch.
>>
> 
> Actually it is not.  readl_poll_timeout is not supported in peripherial devices.
> I had to re-write this to poll the periherial devices.  It worked fine for io-mapped devices.

OK.

>>>  	/* Clear NISO */
>>>  	cccr_reg &= ~(CCCR_NISO);
>>> @@ -1248,112 +1211,100 @@ static bool m_can_niso_supported(const struct m_can_priv *priv)
>>>  	return !niso_timeout;
>>>  }
>>>  
>>> -static int m_can_dev_setup(struct platform_device *pdev, struct net_device *dev,
>>> -			   void __iomem *addr)
>>> +static int m_can_dev_setup(struct m_can_classdev *m_can_dev)
>>>  {
>>> -	struct m_can_priv *priv;
>>> +	struct net_device *dev = m_can_dev->net;
>>>  	int m_can_version;
>>>  
>>> -	m_can_version = m_can_check_core_release(addr);
>>> +	m_can_version = m_can_check_core_release(m_can_dev);
>>>  	/* return if unsupported version */
>>>  	if (!m_can_version) {
>>> -		dev_err(&pdev->dev, "Unsupported version number: %2d",
>>> +		dev_err(m_can_dev->dev, "Unsupported version number: %2d",
>>>  			m_can_version);
>>>  		return -EINVAL;
>>>  	}
>>>  
>>> -	priv = netdev_priv(dev);
>>> -	netif_napi_add(dev, &priv->napi, m_can_poll, M_CAN_NAPI_WEIGHT);
>>> +	if (!m_can_dev->is_peripherial)
>>> +		netif_napi_add(dev, &m_can_dev->napi,
>>> +			       m_can_poll, M_CAN_NAPI_WEIGHT);
>>>  
>>>  	/* Shared properties of all M_CAN versions */
>>> -	priv->version = m_can_version;
>>> -	priv->dev = dev;
>>> -	priv->base = addr;
>>> -	priv->can.do_set_mode = m_can_set_mode;
>>> -	priv->can.do_get_berr_counter = m_can_get_berr_counter;
>>> +	m_can_dev->version = m_can_version;
>>> +	m_can_dev->can.do_set_mode = m_can_set_mode;
>>> +	m_can_dev->can.do_get_berr_counter = m_can_get_berr_counter;
>>>  
>>>  	/* Set M_CAN supported operations */
>>> -	priv->can.ctrlmode_supported = CAN_CTRLMODE_LOOPBACK |
>>> +	m_can_dev->can.ctrlmode_supported = CAN_CTRLMODE_LOOPBACK |
>>>  					CAN_CTRLMODE_LISTENONLY |
>>>  					CAN_CTRLMODE_BERR_REPORTING |
>>>  					CAN_CTRLMODE_FD;
>>>  
>>>  	/* Set properties depending on M_CAN version */
>>> -	switch (priv->version) {
>>> +	switch (m_can_dev->version) {
>>>  	case 30:
>>>  		/* CAN_CTRLMODE_FD_NON_ISO is fixed with M_CAN IP v3.0.x */
>>>  		can_set_static_ctrlmode(dev, CAN_CTRLMODE_FD_NON_ISO);
>>> -		priv->can.bittiming_const = &m_can_bittiming_const_30X;
>>> -		priv->can.data_bittiming_const =
>>> +		if (m_can_dev->bit_timing)
>>> +			m_can_dev->can.bittiming_const = m_can_dev->bit_timing;
>>> +		else
>>> +			m_can_dev->can.bittiming_const =
>>> +				&m_can_bittiming_const_30X;
>>> +		if (m_can_dev->data_timing)
>>> +			m_can_dev->can.data_bittiming_const =
>>> +				m_can_dev->data_timing;
>>> +		else
>>> +			m_can_dev->can.data_bittiming_const =
>>>  				&m_can_data_bittiming_const_30X;
>>
>> Should'nt that go to m_can_platform.c?
>>
> 
> No.  This is the original code.  I added the ability for the peripherials to over ride
> the default bit timings provided.
> 
> I had to do this because TCAN needed different bit and data timings.

OK.

>>>  		break;
>>>  	case 31:
>>>  		/* CAN_CTRLMODE_FD_NON_ISO is fixed with M_CAN IP v3.1.x */
>>>  		can_set_static_ctrlmode(dev, CAN_CTRLMODE_FD_NON_ISO);
>>> -		priv->can.bittiming_const = &m_can_bittiming_const_31X;
>>> -		priv->can.data_bittiming_const =
>>> +		if (m_can_dev->bit_timing)
>>> +			m_can_dev->can.bittiming_const = m_can_dev->bit_timing;
>>> +		else
>>> +			m_can_dev->can.bittiming_const =
>>> +				&m_can_bittiming_const_31X;
>>> +		if (m_can_dev->data_timing)
>>> +			m_can_dev->can.data_bittiming_const =
>>> +				m_can_dev->data_timing;
>>> +		else
>>> +			m_can_dev->can.data_bittiming_const =
>>>  				&m_can_data_bittiming_const_31X;
>>>  		break;
>>>  	case 32:
>>> -		priv->can.bittiming_const = &m_can_bittiming_const_31X;
>>> -		priv->can.data_bittiming_const =
>>> +		if (m_can_dev->bit_timing)
>>> +			m_can_dev->can.bittiming_const = m_can_dev->bit_timing;
>>> +		else
>>> +			m_can_dev->can.bittiming_const =
>>> +				&m_can_bittiming_const_31X;
>>> +
>>> +		if (m_can_dev->data_timing)
>>> +			m_can_dev->can.data_bittiming_const =
>>> +				m_can_dev->data_timing;
>>> +		else
>>> +			m_can_dev->can.data_bittiming_const =
>>>  				&m_can_data_bittiming_const_31X;
>>> -		priv->can.ctrlmode_supported |= (m_can_niso_supported(priv)
>>> +
>>> +		m_can_dev->can.ctrlmode_supported |=
>>> +						(m_can_niso_supported(m_can_dev)
>>>  						? CAN_CTRLMODE_FD_NON_ISO
>>>  						: 0);
>>
>> 		if (m_can_niso_supported(m_can_dev)
>> 			m_can_dev->can.ctrlmode_supported |=
>> 		 		CAN_CTRLMODE_FD_NON_ISO;
>>
>>>  		break;
>>>  	default:
>>> -		dev_err(&pdev->dev, "Unsupported version number: %2d",
>>> -			priv->version);
>>> +		dev_err(m_can_dev->dev, "Unsupported version number: %2d",
>>> +			m_can_dev->version);
>>>  		return -EINVAL;
>>>  	}
>>>  
>>> -	return 0;
>>> -}
>>> -
>>> -static int m_can_open(struct net_device *dev)
>>
>> This function has been moved around making it difficult to understand
>> the diffs.
>>
> 
> Yes it had to be broken up as some of the calls needed to be done in the 
> peripherial and some are common and can be done here.
> 
>>> -{
>>> -	struct m_can_priv *priv = netdev_priv(dev);
>>> -	int err;
>>> -
>>> -	err = m_can_clk_start(priv);
>>> -	if (err)
>>> -		return err;
>>> -
>>> -	/* open the can device */
>>> -	err = open_candev(dev);
>>> -	if (err) {
>>> -		netdev_err(dev, "failed to open can device\n");
>>> -		goto exit_disable_clks;
>>> -	}
>>> -
>>> -	/* register interrupt handler */
>>> -	err = request_irq(dev->irq, m_can_isr, IRQF_SHARED, dev->name,
>>> -			  dev);
>>> -	if (err < 0) {
>>> -		netdev_err(dev, "failed to request interrupt\n");
>>> -		goto exit_irq_fail;
>>> -	}
>>> -
>>> -	/* start the m_can controller */
>>> -	m_can_start(dev);
>>> -
>>> -	can_led_event(dev, CAN_LED_EVENT_OPEN);
>>> -	napi_enable(&priv->napi);
>>> -	netif_start_queue(dev);
>>> +	if (m_can_dev->device_init)
>>> +		m_can_dev->device_init(m_can_dev);
>>>  
>>>  	return 0;
>>> -
>>> -exit_irq_fail:
>>> -	close_candev(dev);
>>> -exit_disable_clks:
>>> -	m_can_clk_stop(priv);
>>> -	return err;
>>>  }
>>>  
>>>  static void m_can_stop(struct net_device *dev)
>>>  {
>>> -	struct m_can_priv *priv = netdev_priv(dev);
>>> +	struct m_can_classdev *priv = netdev_priv(dev);
>>>  
>>>  	/* disable all interrupts */
>>>  	m_can_disable_all_interrupts(priv);
>>> @@ -1364,13 +1315,16 @@ static void m_can_stop(struct net_device *dev)
>>>  
>>>  static int m_can_close(struct net_device *dev)
>>>  {
>>> -	struct m_can_priv *priv = netdev_priv(dev);
>>> +	struct m_can_classdev *priv = netdev_priv(dev);
>>>  
>>>  	netif_stop_queue(dev);
>>> -	napi_disable(&priv->napi);
>>> +	if (!priv->is_peripherial)
>>> +		napi_disable(&priv->napi);
>>>  	m_can_stop(dev);
>>>  	m_can_clk_stop(priv);
>>>  	free_irq(dev->irq, dev);
>>> +	destroy_workqueue(priv->wq);
>>
>> if (priv->is_peripherial) ?
> 
> This workqueue is created for both peripherial and io-mapped

Even if the io-mapped does not use it? Ah, you mean it does use it...
which is a bad solution, I think. This adds extra latency!

> 
>>
>>> +	priv->wq = NULL;
>>>  	close_candev(dev);
>>>  	can_led_event(dev, CAN_LED_EVENT_STOP);
>>>  
>>> @@ -1379,7 +1333,7 @@ static int m_can_close(struct net_device *dev)
>>>  
>>>  static int m_can_next_echo_skb_occupied(struct net_device *dev, int putidx)
>>>  {
>>> -	struct m_can_priv *priv = netdev_priv(dev);
>>> +	struct m_can_classdev *priv = netdev_priv(dev);
>>>  	/*get wrap around for loopback skb index */
>>>  	unsigned int wrap = priv->can.echo_skb_max;
>>>  	int next_idx;
>>> @@ -1391,18 +1345,17 @@ static int m_can_next_echo_skb_occupied(struct net_device *dev, int putidx)
>>>  	return !!priv->can.echo_skb[next_idx];
>>>  }
>>>  
>>> -static netdev_tx_t m_can_start_xmit(struct sk_buff *skb,
>>> -				    struct net_device *dev)
>> Ditto
>>
> 
> Same as above
> 
>>
>>> +static void m_can_tx_work_handler(struct work_struct *ws)
>>
>> I think you need a common function to do the tx.
>>
> 
> This is the common function that was in the original code.
> 
>>
>>>  {
>>> -	struct m_can_priv *priv = netdev_priv(dev);
>>> -	struct canfd_frame *cf = (struct canfd_frame *)skb->data;
>>> +	struct m_can_classdev *priv = container_of(ws, struct m_can_classdev,
>>> +						tx_work);
>>> +	struct canfd_frame *cf = (struct canfd_frame *)priv->skb->data;
>>> +	struct net_device *dev = priv->net;
>>> +	struct sk_buff *skb = priv->skb;
>>>  	u32 id, cccr, fdflags;
>>>  	int i;
>>>  	int putidx;
>>>  
>>> -	if (can_dropped_invalid_skb(dev, skb))
>>> -		return NETDEV_TX_OK;
>>> -
>>>  	/* Generate ID field for TX buffer Element */
>>>  	/* Common to all supported M_CAN versions */
>>>  	if (cf->can_id & CAN_EFF_FLAG) {
>>> @@ -1431,7 +1384,8 @@ static netdev_tx_t m_can_start_xmit(struct sk_buff *skb,
>>>  		can_put_echo_skb(skb, dev, 0);
>>>  
>>>  		if (priv->can.ctrlmode & CAN_CTRLMODE_FD) {
>>> -			cccr = m_can_read(priv, M_CAN_CCCR);
>>> +			/*cccr = m_can_read(priv, M_CAN_CCCR);*/
>>> +			cccr = 0;
>>
>> Unrelated change.
>>
> 
> Ah yes debug code.  Can be removed.
> 
>>>  			cccr &= ~(CCCR_CMR_MASK << CCCR_CMR_SHIFT);
>>>  			if (can_is_canfd_skb(skb)) {
>>>  				if (cf->flags & CANFD_BRS)
>>> @@ -1457,7 +1411,7 @@ static netdev_tx_t m_can_start_xmit(struct sk_buff *skb,
>>>  			netif_stop_queue(dev);
>>>  			netdev_warn(dev,
>>>  				    "TX queue active although FIFO is full.");
>>> -			return NETDEV_TX_BUSY;
>>> +			return;
>>>  		}
>>>  
>>>  		/* get put index for frame */
>>> @@ -1498,14 +1452,87 @@ static netdev_tx_t m_can_start_xmit(struct sk_buff *skb,
>>>  		m_can_write(priv, M_CAN_TXBAR, (1 << putidx));
>>>  
>>>  		/* stop network queue if fifo full */
>>> -			if (m_can_tx_fifo_full(priv) ||
>>> -			    m_can_next_echo_skb_occupied(dev, putidx))
>>> -				netif_stop_queue(dev);
>>> +		if (m_can_tx_fifo_full(priv) ||
>>> +		    m_can_next_echo_skb_occupied(dev, putidx))
>>> +			netif_stop_queue(dev);
>>>  	}
>>> +}
>>> +
>>> +static netdev_tx_t m_can_start_xmit(struct sk_buff *skb,
>>> +				    struct net_device *dev)
>>> +{
>>> +	struct m_can_classdev *priv = netdev_priv(dev);
>>> +
>>> +	if (can_dropped_invalid_skb(dev, skb))
>>> +		return NETDEV_TX_BUSY;
>>> +
>>> +	netif_stop_queue(dev);
>>
>> 	if (!priv->is_peripheral) {
>> 		m_can_tx(priv, skb);
>> 	} else {
>> 		netif_stop_queue(dev);
>> 		priv->skb = skb;
>> 		queue_work(priv->wq, &priv->tx_work);
>> 	}
>>
>> I think stopping the queue here is still experimental.
>>
>>>  	return NETDEV_TX_OK;
>>>  }
>>>  
>>> +static int m_can_open(struct net_device *dev)
>>> +{
>>> +	struct m_can_classdev *priv = netdev_priv(dev);
>>> +	int err;
>>> +
>>> +	err = m_can_clk_start(priv);
>>> +	if (err)
>>> +		return err;
>>> +
>>> +	/* open the can device */
>>> +	err = open_candev(dev);
>>> +	if (err) {
>>> +		netdev_err(dev, "failed to open can device\n");
>>> +		goto exit_disable_clks;
>>> +	}
>>> +
>>
>> 	if (priv->is_peripheral) {
>>
> 
> Not here this is needed.  It is original code
> 
>>> +	priv->wq = alloc_workqueue("mcan_wq", WQ_FREEZABLE | WQ_MEM_RECLAIM,
>>> +				   0);
>>> +	if (!priv->wq) {
>>> +		err = -ENOMEM;
>>> +		goto out_wq_fail;
>>> +	}
>>> +
>>> +	INIT_WORK(&priv->tx_work, m_can_tx_work_handler);
>>
>> 	}
>>
>>> +	/* register interrupt handler */
>>> +	if (priv->is_peripherial)
>>> +		err = request_threaded_irq(dev->irq, NULL, m_can_isr,
>>> +					   IRQF_ONESHOT | IRQF_TRIGGER_FALLING,
>>> +					   dev->name, dev);
>>> +	else
>>> +		err = request_irq(dev->irq, m_can_isr, IRQF_SHARED, dev->name,
>>> +				  dev);
>>> +
>>> +	if (err < 0) {
>>> +		netdev_err(dev, "failed to request interrupt\n");
>>> +		goto exit_irq_fail;
>>> +	}
>>> +
>>> +	/* start the m_can controller */
>>> +	m_can_start(dev);
>>> +
>>> +	can_led_event(dev, CAN_LED_EVENT_OPEN);
>>> +
>>> +	if (!priv->is_peripherial)
>>> +		napi_enable(&priv->napi);
>>> +
>>> +	netif_start_queue(dev);
>>> +
>>> +	return 0;
>>> +
>>> +exit_irq_fail:
>>> +	destroy_workqueue(priv->wq);
>>> +out_wq_fail:
>>> +	close_candev(dev);
>>> +exit_disable_clks:
>>> +	m_can_clk_stop(priv);
>>> +	return err;
>>> +}
>>> +
>>>  static const struct net_device_ops m_can_netdev_ops = {
>>>  	.ndo_open = m_can_open,
>>>  	.ndo_stop = m_can_close,
>>> @@ -1521,21 +1548,7 @@ static int register_m_can_dev(struct net_device *dev)
>>>  	return register_candev(dev);
>>>  }
>>>  
>>> -static void m_can_init_ram(struct m_can_priv *priv)
>>> -{
>>> -	int end, i, start;
>>> -
>>> -	/* initialize the entire Message RAM in use to avoid possible
>>> -	 * ECC/parity checksum errors when reading an uninitialized buffer
>>> -	 */
>>> -	start = priv->mcfg[MRAM_SIDF].off;
>>> -	end = priv->mcfg[MRAM_TXB].off +
>>> -		priv->mcfg[MRAM_TXB].num * TXB_ELEMENT_SIZE;
>>> -	for (i = start; i < end; i += 4)
>>> -		writel(0x0, priv->mram_base + i);
>>> -}
>>> -
>>> -static void m_can_of_parse_mram(struct m_can_priv *priv,
>>> +static void m_can_of_parse_mram(struct m_can_classdev *priv,
>>>  				const u32 *mram_config_vals)
>>>  {
>>>  	priv->mcfg[MRAM_SIDF].off = mram_config_vals[0];
>>> @@ -1562,9 +1575,8 @@ static void m_can_of_parse_mram(struct m_can_priv *priv,
>>>  	priv->mcfg[MRAM_TXB].num = mram_config_vals[7] &
>>>  			(TXBC_NDTB_MASK >> TXBC_NDTB_SHIFT);
>>>  
>>> -	dev_dbg(priv->device,
>>> -		"mram_base %p sidf 0x%x %d xidf 0x%x %d rxf0 0x%x %d rxf1 0x%x %d rxb 0x%x %d txe 0x%x %d txb 0x%x %d\n",
>>> -		priv->mram_base,
>>> +	dev_dbg(priv->dev,
>>> +		"sidf 0x%x %d xidf 0x%x %d rxf0 0x%x %d rxf1 0x%x %d rxb 0x%x %d txe 0x%x %d txb 0x%x %d\n",
>>>  		priv->mcfg[MRAM_SIDF].off, priv->mcfg[MRAM_SIDF].num,
>>>  		priv->mcfg[MRAM_XIDF].off, priv->mcfg[MRAM_XIDF].num,
>>>  		priv->mcfg[MRAM_RXF0].off, priv->mcfg[MRAM_RXF0].num,
>>> @@ -1572,63 +1584,55 @@ static void m_can_of_parse_mram(struct m_can_priv *priv,
>>>  		priv->mcfg[MRAM_RXB].off, priv->mcfg[MRAM_RXB].num,
>>>  		priv->mcfg[MRAM_TXE].off, priv->mcfg[MRAM_TXE].num,
>>>  		priv->mcfg[MRAM_TXB].off, priv->mcfg[MRAM_TXB].num);
>>> -
>>> -	m_can_init_ram(priv);
>>>  }
>>>  
>>> -static int m_can_plat_probe(struct platform_device *pdev)
>>> +void m_can_init_ram(struct m_can_classdev *priv)
>>>  {
>>> -	struct net_device *dev;
>>> -	struct m_can_priv *priv;
>>> -	struct resource *res;
>>> -	void __iomem *addr;
>>> -	void __iomem *mram_addr;
>>> -	struct clk *hclk, *cclk;
>>> -	int irq, ret;
>>> -	struct device_node *np;
>>> -	u32 mram_config_vals[MRAM_CFG_LEN];
>>> -	u32 tx_fifo_size;
>>> -
>>> -	np = pdev->dev.of_node;
>>> +	int end, i, start;
>>>  
>>> -	hclk = devm_clk_get(&pdev->dev, "hclk");
>>> -	cclk = devm_clk_get(&pdev->dev, "cclk");
>>> +	/* initialize the entire Message RAM in use to avoid possible
>>> +	 * ECC/parity checksum errors when reading an uninitialized buffer
>>> +	 */
>>> +	start = priv->mcfg[MRAM_SIDF].off;
>>> +	end = priv->mcfg[MRAM_TXB].off +
>>> +		priv->mcfg[MRAM_TXB].num * TXB_ELEMENT_SIZE;
>>>  
>>> -	if (IS_ERR(hclk) || IS_ERR(cclk)) {
>>> -		dev_err(&pdev->dev, "no clock found\n");
>>> -		ret = -ENODEV;
>>> -		goto failed_ret;
>>> -	}
>>> +	for (i = start; i < end; i += 4)
>>> +		m_can_fifo_write_no_off(priv, i, 0x0);
>>> +}
>>> +EXPORT_SYMBOL_GPL(m_can_init_ram);
>>>  
>>> -	res = platform_get_resource_byname(pdev, IORESOURCE_MEM, "m_can");
>>> -	addr = devm_ioremap_resource(&pdev->dev, res);
>>> -	irq = platform_get_irq_byname(pdev, "int0");
>>> +int m_can_core_get_clocks(struct m_can_classdev *m_can_dev)
>>> +{
>>> +	int ret = 0;
>>>  
>>> -	if (IS_ERR(addr) || irq < 0) {
>>> -		ret = -EINVAL;
>>> -		goto failed_ret;
>>> -	}
>>> +	m_can_dev->hclk = devm_clk_get(m_can_dev->dev, "hclk");
>>> +	m_can_dev->cclk = devm_clk_get(m_can_dev->dev, "cclk");
>>>  
>>> -	/* message ram could be shared */
>>> -	res = platform_get_resource_byname(pdev, IORESOURCE_MEM, "message_ram");
>>> -	if (!res) {
>>> +	if (IS_ERR(m_can_dev->cclk)) {
>>> +		dev_err(m_can_dev->dev, "no clock found\n");
>>>  		ret = -ENODEV;
>>> -		goto failed_ret;
>>>  	}
>>>  
>>> -	mram_addr = devm_ioremap(&pdev->dev, res->start, resource_size(res));
>>> -	if (!mram_addr) {
>>> -		ret = -ENOMEM;
>>> -		goto failed_ret;
>>> -	}
>>> +	return ret;
>>> +}
>>> +EXPORT_SYMBOL_GPL(m_can_core_get_clocks);
>>>  
>>> -	/* get message ram configuration */
>>> -	ret = of_property_read_u32_array(np, "bosch,mram-cfg",
>>> -					 mram_config_vals,
>>> -					 sizeof(mram_config_vals) / 4);
>>> +struct m_can_classdev *m_can_core_allocate_dev(struct device *dev)
>>> +{
>>> +	struct m_can_classdev *class_dev = NULL;
>>> +	u32 mram_config_vals[MRAM_CFG_LEN];
>>> +	struct net_device *net_dev;
>>> +	u32 tx_fifo_size;
>>> +	int ret;
>>> +
>>> +	ret = fwnode_property_read_u32_array(dev_fwnode(dev),
>>> +					     "bosch,mram-cfg",
>>> +					     mram_config_vals,
>>> +					     sizeof(mram_config_vals) / 4);
>>>  	if (ret) {
>>> -		dev_err(&pdev->dev, "Could not get Message RAM configuration.");
>>> -		goto failed_ret;
>>> +		dev_err(dev, "Could not get Message RAM configuration.");
>>> +		goto out;
>>>  	}
>>>  
>>>  	/* Get TX FIFO size
>>> @@ -1637,69 +1641,77 @@ static int m_can_plat_probe(struct platform_device *pdev)
>>>  	tx_fifo_size = mram_config_vals[7];
>>>  
>>>  	/* allocate the m_can device */
>>> -	dev = alloc_candev(sizeof(*priv), tx_fifo_size);
>>> -	if (!dev) {
>>> -		ret = -ENOMEM;
>>> -		goto failed_ret;
>>> +	net_dev = alloc_candev(sizeof(*class_dev), tx_fifo_size);
>>> +	if (!net_dev) {
>>> +		dev_err(dev, "Failed to allocate CAN device");
>>
>> No error message in case of ENOMEM. It makes it worse. Also, it's an
>> unrelated change.
>>
> 
> Most likely a copy and paste from original code.
> 
>>> +		goto out;
>>>  	}
>>>  
>>> -	priv = netdev_priv(dev);
>>> -	dev->irq = irq;
>>> -	priv->device = &pdev->dev;
>>> -	priv->hclk = hclk;
>>> -	priv->cclk = cclk;
>>> -	priv->can.clock.freq = clk_get_rate(cclk);
>>> -	priv->mram_base = mram_addr;
>>> +	class_dev = netdev_priv(net_dev);
>>> +	if (!class_dev) {
>>> +		dev_err(dev, "Failed to init netdev private");
>>> +		goto out;
>>> +	}
>>
>> WARN_ON_ONECE() ?
>>
> That would be an unrelated change this is a copy/paste from original code.

OK.

>>>  
>>> -	platform_set_drvdata(pdev, dev);
>>> -	SET_NETDEV_DEV(dev, &pdev->dev);
>>> +	class_dev->net = net_dev;
>>> +	class_dev->dev = dev;
>>> +	SET_NETDEV_DEV(net_dev, dev);
>>>  
>>> -	/* Enable clocks. Necessary to read Core Release in order to determine
>>> -	 * M_CAN version
>>> -	 */
>>> -	pm_runtime_enable(&pdev->dev);
>>> -	ret = m_can_clk_start(priv);
>>> -	if (ret)
>>> -		goto pm_runtime_fail;
>>> +	m_can_of_parse_mram(class_dev, mram_config_vals);
>>> +out:
>>> +	return class_dev;
>>> +}
>>> +EXPORT_SYMBOL_GPL(m_can_core_allocate_dev);
>>> +
>>> +int m_can_core_register(struct m_can_classdev *m_can_dev)
>>> +{
>>> +	int ret;
>>> +
>>> +	if (m_can_dev->pm_clock_support) {
>>> +		pm_runtime_enable(m_can_dev->dev);
>>> +		ret = m_can_clk_start(m_can_dev);
>>> +		if (ret)
>>> +			goto pm_runtime_fail;
>>> +	}
>>>  
>>> -	ret = m_can_dev_setup(pdev, dev, addr);
>>> +	ret = m_can_dev_setup(m_can_dev);
>>>  	if (ret)
>>>  		goto clk_disable;
>>>  
>>> -	ret = register_m_can_dev(dev);
>>> +	ret = register_m_can_dev(m_can_dev->net);
>>>  	if (ret) {
>>> -		dev_err(&pdev->dev, "registering %s failed (err=%d)\n",
>>> -			KBUILD_MODNAME, ret);
>>> +		dev_err(m_can_dev->dev, "registering %s failed (err=%d)\n",
>>> +			m_can_dev->net->name, ret);
>>>  		goto clk_disable;
>>>  	}
>>>  
>>> -	m_can_of_parse_mram(priv, mram_config_vals);
>>> -
>>> -	devm_can_led_init(dev);
>>> +	devm_can_led_init(m_can_dev->net);
>>>  
>>> -	of_can_transceiver(dev);
>>> +	of_can_transceiver(m_can_dev->net);
>>>  
>>> -	dev_info(&pdev->dev, "%s device registered (irq=%d, version=%d)\n",
>>> -		 KBUILD_MODNAME, dev->irq, priv->version);
>>> +	dev_info(m_can_dev->dev, "%s device registered (irq=%d, version=%d)\n",
>>> +		 KBUILD_MODNAME, m_can_dev->net->irq, m_can_dev->version);
>>>  
>>>  	/* Probe finished
>>>  	 * Stop clocks. They will be reactivated once the M_CAN device is opened
>>>  	 */
>>>  clk_disable:
>>> -	m_can_clk_stop(priv);
>>> +	m_can_clk_stop(m_can_dev);
>>>  pm_runtime_fail:
>>>  	if (ret) {
>>> -		pm_runtime_disable(&pdev->dev);
>>> -		free_candev(dev);
>>> +		if (m_can_dev->pm_clock_support)
>>> +			pm_runtime_disable(m_can_dev->dev);
>>> +		free_candev(m_can_dev->net);
>>>  	}
>>> -failed_ret:
>>> +
>>>  	return ret;
>>>  }
>>> +EXPORT_SYMBOL_GPL(m_can_core_register);
>>>  
>>> -static __maybe_unused int m_can_suspend(struct device *dev)
>>> +int m_can_core_suspend(struct device *dev)
>>>  {
>>>  	struct net_device *ndev = dev_get_drvdata(dev);
>>> -	struct m_can_priv *priv = netdev_priv(ndev);
>>> +	struct m_can_classdev *priv = netdev_priv(ndev);
>>>  
>>>  	if (netif_running(ndev)) {
>>>  		netif_stop_queue(ndev);
>>> @@ -1714,11 +1726,12 @@ static __maybe_unused int m_can_suspend(struct device *dev)
>>>  
>>>  	return 0;
>>>  }
>>> +EXPORT_SYMBOL_GPL(m_can_core_suspend);
>>>  
>>> -static __maybe_unused int m_can_resume(struct device *dev)
>>> +int m_can_core_resume(struct device *dev)
>>>  {
>>>  	struct net_device *ndev = dev_get_drvdata(dev);
>>> -	struct m_can_priv *priv = netdev_priv(ndev);
>>> +	struct m_can_classdev *priv = netdev_priv(ndev);
>>>  
>>>  	pinctrl_pm_select_default_state(dev);
>>>  
>>> @@ -1739,78 +1752,17 @@ static __maybe_unused int m_can_resume(struct device *dev)
>>>  
>>>  	return 0;
>>>  }
>>> +EXPORT_SYMBOL_GPL(m_can_core_resume);
>>>  
>>> -static void unregister_m_can_dev(struct net_device *dev)
>>> -{
>>> -	unregister_candev(dev);
>>> -}
>>> -
>>> -static int m_can_plat_remove(struct platform_device *pdev)
>>> -{
>>> -	struct net_device *dev = platform_get_drvdata(pdev);
>>> -
>>> -	unregister_m_can_dev(dev);
>>> -
>>> -	pm_runtime_disable(&pdev->dev);
>>> -
>>> -	platform_set_drvdata(pdev, NULL);
>>> -
>>> -	free_candev(dev);
>>> -
>>> -	return 0;
>>> -}
>>> -
>>> -static int __maybe_unused m_can_runtime_suspend(struct device *dev)
>>> +void m_can_core_unregister(struct m_can_classdev *m_can_dev)
>>>  {
>>> -	struct net_device *ndev = dev_get_drvdata(dev);
>>> -	struct m_can_priv *priv = netdev_priv(ndev);
>>> +	unregister_candev(m_can_dev->net);
>>>  
>>> -	clk_disable_unprepare(priv->cclk);
>>> -	clk_disable_unprepare(priv->hclk);
>>> +	m_can_clk_stop(m_can_dev);
>>>  
>>> -	return 0;
>>> +	free_candev(m_can_dev->net);
>>>  }
>>> -
>>> -static int __maybe_unused m_can_runtime_resume(struct device *dev)
>>> -{
>>> -	struct net_device *ndev = dev_get_drvdata(dev);
>>> -	struct m_can_priv *priv = netdev_priv(ndev);
>>> -	int err;
>>> -
>>> -	err = clk_prepare_enable(priv->hclk);
>>> -	if (err)
>>> -		return err;
>>> -
>>> -	err = clk_prepare_enable(priv->cclk);
>>> -	if (err)
>>> -		clk_disable_unprepare(priv->hclk);
>>> -
>>> -	return err;
>>> -}
>>> -
>>> -static const struct dev_pm_ops m_can_pmops = {
>>> -	SET_RUNTIME_PM_OPS(m_can_runtime_suspend,
>>> -			   m_can_runtime_resume, NULL)
>>> -	SET_SYSTEM_SLEEP_PM_OPS(m_can_suspend, m_can_resume)
>>> -};
>>> -
>>> -static const struct of_device_id m_can_of_table[] = {
>>> -	{ .compatible = "bosch,m_can", .data = NULL },
>>> -	{ /* sentinel */ },
>>> -};
>>> -MODULE_DEVICE_TABLE(of, m_can_of_table);
>>> -
>>> -static struct platform_driver m_can_plat_driver = {
>>> -	.driver = {
>>> -		.name = KBUILD_MODNAME,
>>> -		.of_match_table = m_can_of_table,
>>> -		.pm     = &m_can_pmops,
>>> -	},
>>> -	.probe = m_can_plat_probe,
>>> -	.remove = m_can_plat_remove,
>>> -};
>>> -
>>> -module_platform_driver(m_can_plat_driver);
>>> +EXPORT_SYMBOL_GPL(m_can_core_unregister);
>>>  
>>>  MODULE_AUTHOR("Dong Aisheng <b29396@...escale.com>");
>>>  MODULE_LICENSE("GPL v2");
>>> diff --git a/drivers/net/can/m_can/m_can_platform.h b/drivers/net/can/m_can/m_can_platform.h
>>> index 97e90dd79613..c3dd301756ba 100644
>>> --- a/drivers/net/can/m_can/m_can_platform.h
>>> +++ b/drivers/net/can/m_can/m_can_platform.h
>>> @@ -156,7 +156,7 @@ int m_can_core_register(struct m_can_classdev *m_can_dev);
>>>  void m_can_core_unregister(struct m_can_classdev *m_can_dev);
>>>  int m_can_core_get_clocks(struct m_can_classdev *m_can_dev);
>>>  void m_can_init_ram(struct m_can_classdev *priv);
>>> -void m_can_config_endisable(const struct m_can_classdev *priv, bool enable);
>>> +void m_can_config_endisable(struct m_can_classdev *priv, bool enable);
>>>  
>>>  int m_can_core_suspend(struct device *dev);
>>>  int m_can_core_resume(struct device *dev);
>>
>> If you fix the issues with "is_peripheral" and the TX function, it
>> should already work on standard M_CAN devices as before... at least in
>> theory!
>>
> 
> It does work on the io-mapped devices.  There may be some issues but we did run the test
> and it at least did not bug check.
> 
> Still testing the io-mapped code though

Not too bad!

Thanks,

Wolfgang.

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ