Skip to contentSkip to conversation

TPM: STMicroelectronics st33 driver SPI

PATCH In review Matthias Leblanc 2 participants 5 messages May 15, 2013 – May 17, 2013

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 TPM: STMicroelectronics st33 driver SPI

From: Mathias Leblanc <[email protected]>

* STMicroelectronics version 1.2.0, Copyright (C) 2010
* STMicroelectronics comes with ABSOLUTELY NO WARRANTY.
* This is free software, and you are welcome to redistribute it
* under certain conditions.

This is the driver for TPM chip from ST Microelectronics.

If you have a TPM security chip from STMicroelectronics working with an SPI, in menuconfig or .config choose the tpm driver on device --> tpm and activate the protocol of your choice before compiling the kernel.
The driver will be accessible from within Linux.

Tested on linux x86/x64 and beagleboard REV B & XM REV C

---
Remove unused variables
Remove tpm_st33_spi_exit function that cause compilation error Initalise data and temp variable
Change some variable type
Correction of warnings and errors from smatch tools

 drivers/char/tpm/Kconfig            |   10 +
 drivers/char/tpm/Makefile           |    1 +
 drivers/char/tpm/tpm_spi_stm_st33.c |  948 +++++++++++++++++++++++++++++++++++
 drivers/char/tpm/tpm_spi_stm_st33.h |   61 +++
 4 files changed, 1020 insertions(+), 0 deletions(-)
 create mode 100644 drivers/char/tpm/tpm_spi_stm_st33.c
 create mode 100644 drivers/char/tpm/tpm_spi_stm_st33.h
drivers/char/tpm/Kconfig +10 -0
index dbfd564..2fc1254 100644
@@ -91,4 +91,14 @@ config TCG_ST33_I2C
         To compile this driver as a module, choose M here; the module will be
         called tpm_stm_st33_i2c.
 
+config TCG_ST33_SPI
+       tristate "STMicroelectronics ST33 SPI"
+       depends on SPI
+       depends on GPIOLIB
+       ---help---
+       If you have a TPM security chip from STMicroelectronics working with
+       an SPI bus say Yes and it will be accessible from within Linux.
+       To compile this driver as a module, choose M here; the module will be
+       called tpm_spi_stm_st33.
+
 endif # TCG_TPM
drivers/char/tpm/Makefile +1 -0
index a3736c9..a1db3dd 100644
@@ -18,3 +18,4 @@ obj-$(CONFIG_TCG_ATMEL) += tpm_atmel.o
 obj-$(CONFIG_TCG_INFINEON) += tpm_infineon.o
 obj-$(CONFIG_TCG_IBMVTPM) += tpm_ibmvtpm.o
 obj-$(CONFIG_TCG_ST33_I2C) += tpm_i2c_stm_st33.o
+obj-$(CONFIG_TCG_ST33_SPI) += tpm_spi_stm_st33.o
drivers/char/tpm/tpm_spi_stm_st33.c new file +948 -0

948 lines not on this page · raw message

+MODULE_LICENSE("GPL");

drivers/char/tpm/tpm_spi_stm_st33.h new file +61 -0

61 lines not on this page · raw message

Signature · 1 line
1.7.9.5
maintainer
#1 lore raw

Hi Matthias,

Am Mittwoch, 15. Mai 2013, 15:53:19 schrieb Matthias Leblanc:

drivers/char/tpm/tpm_spi_stm_st33.c line 87

Quoted 7 lines
drivers/char/tpm/tpm_spi_stm_st33.c new file +7 -0
+static inline int spi_read_write(struct spi_device *spi,
+	 struct spi_transfer xfer) {
+	struct spi_message msg;
+	spi_message_init(&msg);
+	spi_message_add_tail(&xfer, &msg);
+	return spi_sync(spi, &msg);
+}

This looks extremely like the predefined spi_sync_transfer in include/linux/spi.h

/**
* spi_sync_transfer - synchronous SPI data transfer
* @spi: device with which data will be exchanged
* @xfers: An array of spi_transfers
* @num_xfers: Number of items in the xfer array
* Context: can sleep
*
* Does a synchronous SPI data transfer of the given spi_transfer array. *
* For more specific semantics see spi_sync().
*
* It returns zero on success, else a negative error code. */
static inline int
spi_sync_transfer(struct spi_device *spi, struct spi_transfer *xfers,
unsigned int num_xfers)

{
	struct spi_message msg;

	spi_message_init_with_transfers(&msg, xfers, num_xfers);

	return spi_sync(spi, &msg);
}

where spi_message_init_with_transfers is defined as

static inline void spi_message_init_with_transfers(struct spi_message *m, struct spi_transfer *xfers, unsigned int num_xfers)

{
	unsigned int i;

	spi_message_init(m);
	for (i = 0; i < num_xfers; ++i)
		spi_message_add_tail(&xfers[i], m);
}

If you combine these functions you end up (more or less) with your implementation

So please use spi_sync_transfer
OR as an alternative use spi_write / spi_read as an alternative as mentioned in a previous email

And here is a rather subjective remark:

drivers/char/tpm/tpm_spi_stm_st33.c line 201

Quoted 4 lines
drivers/char/tpm/tpm_spi_stm_st33.c new file +4 -0
+             for (; nbr_dummy_bytes < total_length &&
+                      ((u8 *)xfer.rx_buf)[nbr_dummy_bytes] == 0;
+                      nbr_dummy_bytes++)
+                     ;

Not sure if it would be easier to read using a while and putting the increment into the loop body

drivers/char/tpm/tpm_spi_stm_st33.c line 146

Quoted 3 lines
drivers/char/tpm/tpm_spi_stm_st33.c new file +3 -0
+              while (nbr_dummy_bytes < total_length && 
+                      ((u8 *)xfer.rx_buf)[nbr_dummy_bytes] == 0)
+                       nbr_dummy_bytes++;

I usually don't like empty loop bodies, as they tend to be overlooked.

drivers/char/tpm/tpm_spi_stm_st33.c

Quoted 6 lines
drivers/char/tpm/tpm_spi_stm_st33.c new file +6 -0
+module_spi_driver(tpm_st33_spi_driver);
+
+MODULE_AUTHOR("Christophe Ricard ([email protected])");
+MODULE_DESCRIPTION("STM TPM SPI ST33 Driver");
+MODULE_VERSION("1.2.0");
+MODULE_LICENSE("GPL");

Something is strange here - if I apply your patch, the module license gets lost - so I get a compile warning:

WARNING: modpost: missing MODULE_LICENSE() in /data/data-old/linux-2.6/drivers/char/tpm/tpm_spi_stm_st33.o

Do you edit the patch by hand before sending?

Steps to reproduce:
drivers/char/tpm $ git reset --hard tpmdd/tpmdd-04-26-13
HEAD is now at d224213 tpm_tis: missing platform_driver_unregister() on error in init_tis() drivers/char/tpm $ git am /tmp/\[tpmdd-devel\]\ \[PATCH\ 1_1\]\ TPM_STMicroelectronics\ st33\ driver\ SPI.mbox
Applying: TPM: STMicroelectronics st33 driver SPI

drivers/char/tpm $ tail tpm_spi_stm_st33.c },

        .probe = tpm_st33_spi_probe,
        .remove = tpm_st33_spi_remove,
};

module_spi_driver(tpm_st33_spi_driver);

MODULE_AUTHOR("Christophe Ricard ([email protected])"); MODULE_DESCRIPTION("STM TPM SPI ST33 Driver"); MODULE_VERSION("1.2.0");

And the usual sparse warnings ;)
drivers/char/tpm $ make M=pwd -C /data/data-old/linux-2.6/ modules C=1 CHECK=sparse
make: Entering directory `/data/data-old/linux-2.6'
CHECK /data/data-old/linux-2.6/drivers/char/tpm/tpm_spi_stm_st33.c
/data/data-old/linux-2.6/drivers/char/tpm/tpm_spi_stm_st33.c:104:35: warning: cast removes address space of expression /data/data-old/linux-2.6/drivers/char/tpm/tpm_spi_stm_st33.c:170:35: warning: cast removes address space of expression /data/data-old/linux-2.6/drivers/char/tpm/tpm_spi_stm_st33.c:242:19: warning: cast removes address space of expression /data/data-old/linux-2.6/drivers/char/tpm/tpm_spi_stm_st33.c:235:15: warning: symbol 'wait_for_serirq_timeout' was not declared. Should it be static? /data/data-old/linux-2.6/drivers/char/tpm/tpm_spi_stm_st33.c:482:19: warning: cast removes address space of expression

Regards,
Peter

#2 lore raw

Hi Peter,
Thanks for your feedback.

You are right, this function looks like the spi_sync_transfer; So, I will use it in the next version of the driver.

Regarding the while loop, I don't see how can I check the number of dummy byte differently?

For the module error, Yes I have modified the patch file :s, apparently I haven't correctly modified the number of lines

Finaly, These sparse warning are finaly fix ;).

Best regards,

Mathias Leblanc
________________________________________
From: Peter Hüwe [[email protected]]
Sent: 16 May 2013 00:29
To: [email protected]
Cc: Mathias LEBLANC; Kent Yoder; Rajiv Andrade; Marcel Selhorst; Sirrix AG; Jean-Luc BLANC; [email protected]
Subject: Re: [tpmdd-devel] [PATCH 1/1] TPM: STMicroelectronics st33 driver SPI

Hi Matthias,

Am Mittwoch, 15. Mai 2013, 15:53:19 schrieb Matthias Leblanc:

drivers/char/tpm/tpm_spi_stm_st33.c line 87

Quoted 7 lines
drivers/char/tpm/tpm_spi_stm_st33.c new file +7 -0
+static inline int spi_read_write(struct spi_device *spi,
+      struct spi_transfer xfer) {
+     struct spi_message msg;
+     spi_message_init(&msg);
+     spi_message_add_tail(&xfer, &msg);
+     return spi_sync(spi, &msg);
+}

This looks extremely like the predefined spi_sync_transfer in include/linux/spi.h

/**
* spi_sync_transfer - synchronous SPI data transfer
* @spi: device with which data will be exchanged
* @xfers: An array of spi_transfers
* @num_xfers: Number of items in the xfer array
* Context: can sleep
*
* Does a synchronous SPI data transfer of the given spi_transfer array. *
* For more specific semantics see spi_sync().
*
* It returns zero on success, else a negative error code. */
static inline int
spi_sync_transfer(struct spi_device *spi, struct spi_transfer *xfers,
unsigned int num_xfers)

{
        struct spi_message msg;

        spi_message_init_with_transfers(&msg, xfers, num_xfers);

        return spi_sync(spi, &msg);
}

where spi_message_init_with_transfers is defined as

static inline void spi_message_init_with_transfers(struct spi_message *m, struct spi_transfer *xfers, unsigned int num_xfers)

{
        unsigned int i;

        spi_message_init(m);
        for (i = 0; i < num_xfers; ++i)
                spi_message_add_tail(&xfers[i], m);
}

If you combine these functions you end up (more or less) with your implementation

So please use spi_sync_transfer
OR as an alternative use spi_write / spi_read as an alternative as mentioned in a previous email

And here is a rather subjective remark:

drivers/char/tpm/tpm_spi_stm_st33.c line 201

Quoted 4 lines
drivers/char/tpm/tpm_spi_stm_st33.c new file +4 -0
+             for (; nbr_dummy_bytes < total_length &&
+                      ((u8 *)xfer.rx_buf)[nbr_dummy_bytes] == 0;
+                      nbr_dummy_bytes++)
+                     ;

Not sure if it would be easier to read using a while and putting the increment into the loop body

drivers/char/tpm/tpm_spi_stm_st33.c line 146

Quoted 3 lines
drivers/char/tpm/tpm_spi_stm_st33.c new file +3 -0
+              while (nbr_dummy_bytes < total_length &&
+                      ((u8 *)xfer.rx_buf)[nbr_dummy_bytes] == 0)
+                       nbr_dummy_bytes++;

I usually don't like empty loop bodies, as they tend to be overlooked.

drivers/char/tpm/tpm_spi_stm_st33.c

Quoted 6 lines
drivers/char/tpm/tpm_spi_stm_st33.c new file +6 -0
+module_spi_driver(tpm_st33_spi_driver);
+
+MODULE_AUTHOR("Christophe Ricard ([email protected])");
+MODULE_DESCRIPTION("STM TPM SPI ST33 Driver");
+MODULE_VERSION("1.2.0");
+MODULE_LICENSE("GPL");

Something is strange here - if I apply your patch, the module license gets lost - so I get a compile warning:

WARNING: modpost: missing MODULE_LICENSE() in /data/data-old/linux-2.6/drivers/char/tpm/tpm_spi_stm_st33.o

Do you edit the patch by hand before sending?

Steps to reproduce:
drivers/char/tpm $ git reset --hard tpmdd/tpmdd-04-26-13
HEAD is now at d224213 tpm_tis: missing platform_driver_unregister() on error in init_tis() drivers/char/tpm $ git am /tmp/\[tpmdd-devel\]\ \[PATCH\ 1_1\]\ TPM_STMicroelectronics\ st33\ driver\ SPI.mbox
Applying: TPM: STMicroelectronics st33 driver SPI

drivers/char/tpm $ tail tpm_spi_stm_st33.c },

        .probe = tpm_st33_spi_probe,
        .remove = tpm_st33_spi_remove,
};

module_spi_driver(tpm_st33_spi_driver);

MODULE_AUTHOR("Christophe Ricard ([email protected])"); MODULE_DESCRIPTION("STM TPM SPI ST33 Driver"); MODULE_VERSION("1.2.0");

And the usual sparse warnings ;)
drivers/char/tpm $ make M=pwd -C /data/data-old/linux-2.6/ modules C=1 CHECK=sparse
make: Entering directory `/data/data-old/linux-2.6'
CHECK /data/data-old/linux-2.6/drivers/char/tpm/tpm_spi_stm_st33.c
/data/data-old/linux-2.6/drivers/char/tpm/tpm_spi_stm_st33.c:104:35: warning: cast removes address space of expression /data/data-old/linux-2.6/drivers/char/tpm/tpm_spi_stm_st33.c:170:35: warning: cast removes address space of expression /data/data-old/linux-2.6/drivers/char/tpm/tpm_spi_stm_st33.c:242:19: warning: cast removes address space of expression /data/data-old/linux-2.6/drivers/char/tpm/tpm_spi_stm_st33.c:235:15: warning: symbol 'wait_for_serirq_timeout' was not declared. Should it be static? /data/data-old/linux-2.6/drivers/char/tpm/tpm_spi_stm_st33.c:482:19: warning: cast removes address space of expression

Regards,
Peter

maintainer
#3 lore raw

Hi Matthias:

Quoted 2 lines

Regarding the while loop, I don't see how can I check the number of dummy byte differently?

Checking the number of dummy bytes is fine -
in line #144 you check it with a while loop which looks fine in line #198 you check it with a for loop with an empty body, which in my opinion looks less nice than the while in #144.

So simply replace the second instance in #198 and I'm happy with that part ;)

Maybe also have a look at memchr_inv

/**
* memchr_inv - Find an unmatching character in an area of memory.
* @start: The memory area
* @c: Find a character other than c
* @bytes: The size of the area.
*
* returns the address of the first character other than @c, or %NULL
* if the whole buffer contains just @c.
*/
void *memchr_inv(const void *start, int c, size_t bytes)

so #198 would look something like:
nbr_dummy_bytes = memchr_inv (xfer.rx_buf+nbr_dummy_bytes,

					0,
					 total_length-nbr_dummy_bytes) - xfer.rx_buf;

But I'm not sure whether that's more readable, maybe give it a try.
I'm also fine with the while loop.

Peter

#4 lore raw

Hi Peter,

Thanks for this new function.
I had already changed the for loop but i would to be sure that it's the best way and that you expected.

Don't hesitate if you have any other recommendation or any comments for this driver.

Regards,

Mathias Leblanc
________________________________________
From: Peter Hüwe [[email protected]]
Sent: 16 May 2013 21:03
To: [email protected]
Cc: Mathias LEBLANC; Jean-Luc BLANC; [email protected]; Rajiv Andrade; Sirrix AG
Subject: Re: [tpmdd-devel] [PATCH 1/1] TPM: STMicroelectronics st33 driver SPI

Hi Matthias:

Quoted 2 lines

Regarding the while loop, I don't see how can I check the number of dummy byte differently?

Checking the number of dummy bytes is fine -
in line #144 you check it with a while loop which looks fine in line #198 you check it with a for loop with an empty body, which in my opinion looks less nice than the while in #144.

So simply replace the second instance in #198 and I'm happy with that part ;)

Maybe also have a look at memchr_inv

/**
* memchr_inv - Find an unmatching character in an area of memory.
* @start: The memory area
* @c: Find a character other than c
* @bytes: The size of the area.
*
* returns the address of the first character other than @c, or %NULL
* if the whole buffer contains just @c.
*/
void *memchr_inv(const void *start, int c, size_t bytes)

so #198 would look something like:
nbr_dummy_bytes = memchr_inv (xfer.rx_buf+nbr_dummy_bytes,

                                        0,
                                         total_length-nbr_dummy_bytes) - xfer.rx_buf;

But I'm not sure whether that's more readable, maybe give it a try.
I'm also fine with the while loop.

Peter