Hello Cedric,
On Thu, 2022-07-07 at 10:32 +0200, Cedric Berger wrote:
> Hello,
>
> On 07.07.22 09:35, Duc Doan wrote:
> > Actually my project is focused on STM32F4, but it is nice that the
> > code
> > could be extensible to H7. However, I can't think of a good way to
> > use
> > that code with H7 right now; do you think that the code can be
> > copied
> > and pasted to H7 BSP? Or is there a better way to share the code
> > among
> > all STM32 BSPs?
>
> I've a lot of experience with STM32 dev, but not with RTEMS, so I'm
> not
> the best person to answer that question.
>
> If you want to expand support for the H7 now, maybe Karel or someone
> else on the list can point you to the right way.
>
> What I can say by reading your "[PATCH 4/4]" file is that I don't see
> what would not work for the H7. If it compiles on H7, it's likely to
> work as is (except for the GPIO port lists, you'd need other ifdefs
> like
> you do for STM32F429X, or maybe just keep
> stm32f4_gpio_init/stm32f4_gpio_deinit separate for F4 and H7).
>
> Maybe it's as simple as renaming the identifiers from stm32f4_gpio =>
> stm32_gpio for clarity for most functions, and declaring these files
> in
> the H7 scripts.
>
I think I will wait for Karel's opinion about that.
> BTW: by rereading your "[PATCH 4/4]" file I noticed that there is
> something else you need to be careful:
>
> I don't know how the lifecycle of init/deinit is intended to work,
> but
> the stm32f4_gpio_deinit() function is dangerous:
>
> The problem is that if you disable the clock of, say, GPIOA, it will
> also make all alternate function IPs that uses these pins stop
> working.
>
> So if you are using the HAL for example for the Ethernet IP, the
> board
> I'm working with will also call the following functions in it's
> initialization routine:
>
> HAL_ETH_MspInit()
> {
> ...
> __HAL_RCC_GPIOG_CLK_ENABLE();
> __HAL_RCC_GPIOC_CLK_ENABLE();
> __HAL_RCC_GPIOA_CLK_ENABLE();
> ...
> }
>
> So your "stm32f4_gpio_deinit" routine will make all IPs that share
> the
> pins with the GPIO port also stop working.
>
> The trivial fix is of course to make "stm32f4_gpio_deinit" do
> nothing.
>
> Cedric
>
You are right, I forgot that disabling the clock affects the whole
port. I made deinit() do nothing.
Best,
Duc Doan
_______________________________________________
devel mailing list
devel@rtems.org
http://lists.rtems.org/mailman/listinfo/devel
