bcm63xx gpio issue on 3.19
Hello Alexandre,
Using the latest 3.19 kernel, the bcm63xx GPIO code under arch/mips/bcm63xx/gpio.c is unable to register the gpio chip via gpiochip_add(), as it returns -ENOMEM. The kcalloc call for the gpio_desc array fails, as during prom code, it is too early for the kmalloc to work.
It looks like the issue is caused by your patch: "gpio: remove gpio_descs global array"
Could you please advise on how to fix/workaround that ? (ideally while keeping the possibility to invoke the gpiolib code from the setup/prom code).
Regards,
Signature · 2 lines
Nicolas Schichan Freebox SAS
Hi Nicolas,
(adding the linux-gpio mailing-list and Linus W.)
Quoted 10 lines · Nicolas Schichan
Hello Alexandre,
Using the latest 3.19 kernel, the bcm63xx GPIO code under
arch/mips/bcm63xx/gpio.cis unable to register the gpio chip viagpiochip_add(), as it returns-ENOMEM. The kcalloc call for thegpio_descarray fails, as during prom code, it is too early for the kmalloc to work.It looks like the issue is caused by your patch: "gpio: remove
gpio_descsglobal array"
Indeed. This happens because we removed the global GPIO array and replaced it with a more flexible per-chip array of GPIOs. We were hoping that issues like this one would have been caught in -next, but sadly the problem with bcm63xx went unnoticed until now. :(
Quoted 3 lines
Could you please advise on how to fix/workaround that ? (ideally while keeping the possibility to invoke the gpiolib code from the setup/prom code).
The only allocation performed by gpiochip_add() is the array of gpio_descs. Having this array pre-allocated in your early code (maybe by using a static array variable) and passing it to a gpiochip_add_early() function would do the trick.
However, it is not that simple since gpio_desc is a private structure which details (including its size) are not visible outside of drivers/gpio.
Another solution I could see would be to have a kernel config option that would make gpiolib "pre-allocate" a number of gpio descriptors as a static array for such cases - similar to the global GPIO array, but not as big.
Finally, we can also restore the global GPIO array as a config option for the few architectures that need it.
Of course, I would prefer a solution based on dynamic allocation - is there a kind a primitive memory allocator that we can use at this early stage of boot? I.e. would alloc_pages() maybe work?
How do other subsystems that rely on dynamic allocation for registering their resources handle this? I guess regulator must fall in the same use-case, doesn't it?
Quoted 18 lines · Alexandre Courbot
Quoted 2 lines
Could you please advise on how to fix/workaround that ? (ideally while keeping the possibility to invoke the gpiolib code from the setup/prom code).
The only allocation performed by
gpiochip_add()is the array ofgpio_descs. Having this array pre-allocated in your early code (maybe by using a static array variable) and passing it to agpiochip_add_early()function would do the trick.However, it is not that simple since
gpio_descis a private structure which details (including its size) are not visible outside ofdrivers/gpio.Another solution I could see would be to have a kernel config option that would make gpiolib "pre-allocate" a number of gpio descriptors as a static array for such cases - similar to the global GPIO array, but not as big.
Finally, we can also restore the global GPIO array as a config option for the few architectures that need it.
Quoted 7 lines
Of course, I would prefer a solution based on dynamic allocation - is there a kind a primitive memory allocator that we can use at this early stage of boot? I.e. would
alloc_pages()maybe work?How do other subsystems that rely on dynamic allocation for registering their resources handle this? I guess regulator must fall in the same use-case, doesn't it?
Hi Alexandre,
Moving the bcm63xx_gpio_init() call from board_prom_init() to bcm63xx_register_devices() (an arch_initcall) is enough to get called when kmalloc is working. If code poking GPIOs is invoked earlier by a board code, it will have to move in the board_register_devices() function though there doesn't seem to any problems with the mainline bcm63xx board code in that regard.
I can produce a patch for that if it is an accepted solution. It has the advantage of not requiring changes to the gpiolib code.
Regards,
Signature · 2 lines
Nicolas Schichan Freebox SAS
Quoted 29 lines · Nicolas Schichan
Quoted 18 lines · Alexandre Courbot
Quoted 2 lines
Could you please advise on how to fix/workaround that ? (ideally while keeping the possibility to invoke the gpiolib code from the setup/prom code).
The only allocation performed by
gpiochip_add()is the array ofgpio_descs. Having this array pre-allocated in your early code (maybe by using a static array variable) and passing it to agpiochip_add_early()function would do the trick.However, it is not that simple since
gpio_descis a private structure which details (including its size) are not visible outside ofdrivers/gpio.Another solution I could see would be to have a kernel config option that would make gpiolib "pre-allocate" a number of gpio descriptors as a static array for such cases - similar to the global GPIO array, but not as big.
Finally, we can also restore the global GPIO array as a config option for the few architectures that need it.
Quoted 7 lines
Of course, I would prefer a solution based on dynamic allocation - is there a kind a primitive memory allocator that we can use at this early stage of boot? I.e. would
alloc_pages()maybe work?How do other subsystems that rely on dynamic allocation for registering their resources handle this? I guess regulator must fall in the same use-case, doesn't it?
Hi Alexandre,
Quoted 8 lines
Moving the
bcm63xx_gpio_init()call fromboard_prom_init()tobcm63xx_register_devices()(anarch_initcall) is enough to get called when kmalloc is working. If code poking GPIOs is invoked earlier by a board code, it will have to move in theboard_register_devices()function though there doesn't seem to any problems with the mainline bcm63xx board code in that regard.I can produce a patch for that if it is an accepted solution. It has the advantage of not requiring changes to the gpiolib code.
If that works for you and doesn't break bcm63xx, then yes that would be great. GPIO chips are indeed devices, so moving their registration to bcm63xx_register_devices() seems to make sense.
The GPIO subsystem might require more dynamic allocations in the near future (including maybe when requesting GPIOs), but hopefully all consumers will be safe with that. We'll find a solution for early GPIO support if they aren't.
Quoted 8 lines · Nicolas Schichan
Moving the
bcm63xx_gpio_init()call fromboard_prom_init()tobcm63xx_register_devices()(anarch_initcall) is enough to get called when kmalloc is working. If code poking GPIOs is invoked earlier by a board code, it will have to move in theboard_register_devices()function though there doesn't seem to any problems with the mainline bcm63xx board code in that regard.I can produce a patch for that if it is an accepted solution. It has the advantage of not requiring changes to the gpiolib code.
It would be *awesome* if you can come up with a patch like this. Because it makes our life so much more simple.
(Maybe there is already such a patch in my INBOX somewhere...)
Yours,
Linus Walleij