drivers: PMC MSP71xx TWI driver]
Quoted 37 lines · Jean Delvare
Marc,
Quoted 1 line
----- Forwarded message from Marc St-Jean <[email protected]>
-----
Quoted 26 lines
From: Marc St-Jean <[email protected]> Date: Fri, 16 Mar 2007 15:39:51 -0600 To: [email protected]Cc: [email protected]
Subject: [PATCH 8/12] drivers: PMC MSP71xx TWI driver[PATCH 8/12] drivers: PMC MSP71xx TWI driver
Patch to add TWI driver for the PMC-Sierra MSP71xx devices.
Reposting patches as a single set at the request of akpm. Only 9 of 12 will be posted at this time, 3 more to follow when cleanups are complete.
CCing linux-mips.org since minor changes have occured during cleanup of driver patches for other lists.
Thanks,
MarcSigned-off-by Marc St-Jean <[email protected]>---
Re-posting patch with recommended changes:
-No changes.Quoted 6 lines
drivers/i2c/algos/Kconfig | 9 drivers/i2c/algos/Makefile | 1 drivers/i2c/algos/i2c-algo-pmctwi.c | 197 ++++++++++++++++ drivers/i2c/busses/Kconfig | 7 drivers/i2c/busses/Makefile | 1 drivers/i2c/busses/i2c-pmcmsp.c | 419++++++++++++++++++++++++++++++++++++
Quoted 8 lines
Quoted 2 lines
include/linux/i2c-algo-pmctwi.h | 135 +++++++++++ 7 files changed, 768 insertions(+), 1 deletion(-)Why are you making a separate algorithm driver? This should really only be done when the algorithm is very generic. This is the exception, not the rule. These days I tend to move algorithm code back into the only bus driver that uses them (i2c-algo-sibyte done recently, i2c-algo-sgi is next on my list.)
I believe it was done to separate the algorithm for the TWI sub-system, which is used in several devices, from the device family code. If this is no longer acceptable I will merge the algo code with the adapter.
Quoted 8 lines
Quoted 8 lines
drivers/i2c/algos/Kconfig +0
-0@@ -49,5 +49,12 @@ config I2C_ALGO_SGI Supports the SGI interfaces like the ones found on SGI Indy VINO or SGI O2 MACE.
Quoted 8 lines
drivers/i2c/algos/Kconfig config I2C_ALGO_SGI · line 54
Quoted 4 lines
drivers/i2c/algos/Kconfig +3
-1-endmenu +config I2C_ALGO_PMCTWI + tristate "I2C PMC TWI interfaces" + depends on I2C && PMC_MSPAs a matter of fact, each time an algorithm depends on specific hardware, you can be reasonably certain that it should NOT be made a generic algorithm driver.
OK.
Quoted 11 lines
drivers/i2c/algos/Kconfig config I2C_ALGO_SGI · line 56
Quoted 2 lines
drivers/i2c/algos/Kconfig +2
-0+ help + Implements the PMC TWI SoC algorithm for variousimplementations.
Quoted 4 lines
drivers/i2c/algos/Kconfig +3
-0+ Be sure to select the proper bus for your platform below. + +endmenuSignature · 1 line
Jean Delvare
Hi Marc,
Quoted 10 lines · Marc St-Jean
Quoted 5 lines · Jean Delvare
Why are you making a separate algorithm driver? This should really only be done when the algorithm is very generic. This is the exception, not the rule. These days I tend to move algorithm code back into the only bus driver that uses them (i2c-algo-sibyte done recently, i2c-algo-sgi is next on my list.)
I believe it was done to separate the algorithm for the TWI sub-system, which is used in several devices, from the device family code. If this is no longer acceptable I will merge the algo code with the adapter.
You should be able to have a single i2c bus driver handling all the different devices, by using the platform data to provide the device-specific configuration. The i2c-algorithm concept is quite old, now that we have a clean new device driver model it should no longer be necessary in most cases.
Signature · 1 line
Jean Delvare