Stm32 spi : use LL API to improve performances #4375
Conversation
|
@LMESTM Could you investigate please |
|
@adbridge I believe because of the dependency mentioned above, it's now ready to be merged, then this should be rebased and make it green. |
|
Hey |
|
#3445 is also being fixed as part of this PR |
|
/morph test |
Result: SUCCESSYour command has finished executing! Here's what you wrote!
OutputAll builds and test passed! |
targets/TARGET_STM/stm_spi_api.c
Outdated
| struct spi_s *spiobj = SPI_S(obj); | ||
| SPI_HandleTypeDef *handle = &(spiobj->handle); | ||
|
|
||
| size = (handle->Init.DataSize == SPI_DATASIZE_16BIT) ? 2 : 1; | ||
| #ifdef STM32_SPI_KEEP_HAL_USAGE |
There was a problem hiding this comment.
Where is this macro coming from ? IS it documented anywhere? This changes the implementation .
There was a problem hiding this comment.
It differs from what was before (RX an error handling ? )
There was a problem hiding this comment.
The commit message describes it:
This commit implements a SPI mode which will offer better performance
thanks to usage of Lower Layer API which use fewer registers access,
at the cost of lower robustness (no error management at all).
The SPI driver can be configured to re-use the HAL and error management by
defining the STM32_SPI_KEEP_HAL_USAGE compilation switch
There was a problem hiding this comment.
Does this macro change the implementation? That wouldn't be good. We should choose one implementation and stick to it.
|
CI Jenkins reports a build failure - can you share the error ? |
|
@LMESTM There is the old jenkins config, can you rebase from master to pick it up, that should resolve the error. |
This commit implements a SPI mode which will offer better performance thanks to usage of Lower Layer API which use fewer registers access, at the cost of lower robustness (no error management).
|
@0xc0170 ok thx - rebased |
theotherjimmy
changed the title
|
/morph test |
Result: SUCCESSYour command has finished executing! Here's what you wrote!
OutputAll builds and test passed! |
Description
This PR contains a change in SPI driver to improve performances by accessing to registers directly (using LL layer in STM32 SDK, which provides inline functions to access it.
See #4246
Resolves #3445
Status
READY
Waiting for 2 other PRS:
Related PRs
There were 2 pending dependencies of upgrading STM32 SDK for
F4 (#4299) => MERGED
F2 (#4402) => MERGED
This SDK updates will include the LL API that is used here.
I will rebase when this is done - but this PR can be used for test by anyone interested in it, and review can happen in parallel.
Test results
CI tests passed ok
+-------------------+---------------+---------------+--------+--------------------+-------------+
| target | platform_name | test suite | result | elapsed_time (sec) | copy_method |
+-------------------+---------------+---------------+--------+--------------------+-------------+
| NUCLEO_F091RC-ARM | NUCLEO_F091RC | tests-api-spi | OK | 60.5 | shell |
| NUCLEO_F103RB-ARM | NUCLEO_F103RB | tests-api-spi | OK | 61.56 | shell |
| NUCLEO_F303ZE-ARM | NUCLEO_F303ZE | tests-api-spi | OK | 61.3 | shell |
| NUCLEO_F446RE-ARM | NUCLEO_F446RE | tests-api-spi | OK | 59.17 | shell |
| NUCLEO_F767ZI-ARM | NUCLEO_F767ZI | tests-api-spi | OK | 58.7 | shell |
| NUCLEO_L073RZ-ARM | NUCLEO_L073RZ | tests-api-spi | OK | 64.15 | shell |
| NUCLEO_L152RE-ARM | NUCLEO_L152RE | tests-api-spi | OK | 61.34 | shell |
| NUCLEO_L476RG-ARM | NUCLEO_L476RG | tests-api-spi | OK | 60.87 | shell |
+-------------------+---------------+---------------+--------+--------------------+-------------+
Also MBED OS2 SPI tests verified OK.