mimxrt: Rework flash configuration. #7767

alphaFred · 2021-09-10T05:51:22Z

Moves definition of BOARD_FLASH_SIZE and other header files related to
flash configuration into the Makefile.
Adds board specific clock_config.h.
Adds board.h, pin_mux.h, and peripherals.h as they are
required by NXP MCU SDK in order to use our own clock_config.h.
Renames board specific FlexSPI configuration files.
Updates flash frequency of MIMXRT1020_EVK
Creates separated flash_config files for QSPI NOR and
QSPI Hyper flash.
Unifies VFS start address to be @ 1M for 1010 and 1020 boards.
Unifies 1050EVK boards
Adds support to both NOR and HyperFlash on boards with
both capabilities.
Adds automatic FlexRAM initialization to start-up code based on
linker script and NXP HAL.

With this change the flash configuration is restructured.
This simplifies the configuration process and
provides a better overview of each board's settings. With the integration
of clock_config.h, board.h, pin_mux.h, and peripherals.h we gain better
control of the settings and clock configurations. Furthermore the
implementation of an explicit FlexRAM setup improves the system
performance and allows for performance tuning.

Signed-off-by: Philipp Ebensberger

robert-hh · 2021-09-10T06:14:28Z

Just a few additional notes on that PR:
This PR looks huge, but is an attempt to clean up the various board file trees to a common structure. Over the time, things got a little bit fuzzy. And so we decided to make this PR. It covers the fact, that the boards can be configured with different flash setting. It also includes a change to make better use of the built-in RAM. That causes most of the added lines of code, because it moves the startup_xxx.s files from the nxp_lib to the port file system. That seems necessary since the RAM re-config has to happen before any RAM is used.

dpgeorge · 2021-09-13T13:30:42Z

ports/mimxrt/boards/make-flexram-config.py

@@ -0,0 +1,194 @@
+#!/usr/bin/env python
+"""Evaluate FlexRAM configuration and generate startup code."""


what is the license/copyright of this file?

It was written by @alphaFred, so I assume it's MIT, and Philipp will add the appropriate License.

Yes - I will add the license

please make the style of the license/copyright match existing files, eg tools/pyboard.py and tools/mpy_ld.py (use # at the start of the line)

dpgeorge · 2021-09-13T13:34:54Z

Regarding the startup_MIMXRTxxx.S files: in the HAL versions of these files the Reset_Handler function is defined as a weak symbol. So that means you can reimplement just this function in your own code, which is exactly what you need. This is how stm32 works, it uses the HAL's startup code to get the IRQ table, but reimplements Reset_Handler in ports/stm32/resethandler.s. I'd suggest mimxrt also do it this way so that there is less code, and maybe the same resethandler.s (or .S if it needs to use the C preprocessor) can be used for all MCU variants.

Also, why exactly does the clock_config.h file need to be copied out of the HAL into each board directory? As far as I can tell it's copied without change, so why can't it reuse the HAL version?

alphaFred · 2021-09-13T13:58:12Z

Hi @dpgeorge. Thanks for your review.

Regarding your first point - I will give it a try and check if it works for us.

Regarding clock_config.h. This is part of a set of four files (clock_config.h, board.h,peripherals.h,pin_mux.h) which are provided as templates. So far this worked fine for us but with this cleanup I noticed that the board.h defines BOARD_FLASH_SIZE with out #ifdefguard. This is a problem when we want to configure for different flash sizes (e.g. QPSI vs. HyperFlash). To get around that I needed to create my own set of these header files. Since they are all located in one folder I could not easily exclude clock_config.h.

On a side note:
I already have another PR ready which builds on top of this one where we will refactor the clock configuration. in that PR you will see differences in clock_config.h.

robert-hh · 2021-09-13T14:19:12Z

maybe the same resethandler.s (or .S if it needs to use the C preprocessor) can be used for all MCU variants.

That is indeed the case. I just tested that (and compared the files) and it works for all MCU's I have here.

alphaFred · 2021-09-13T18:35:06Z

@dpgeorge I pushed an update with fixes. Tbh I was not sure about the format of the license statement for the Python script since I could not find a single other Python file in the project which had one. So let me know if it is ok for you 😄

codecov-commenter · 2021-09-13T19:13:22Z

Codecov Report

Merging #7767 (f8bdfad) into master (0a51073) will increase coverage by 0.00%.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master    #7767   +/-   ##
=======================================
  Coverage   98.24%   98.25%           
=======================================
  Files         154      154           
  Lines       20040    20042    +2     
=======================================
+ Hits        19689    19692    +3     
+ Misses        351      350    -1

Impacted Files	Coverage Δ
py/objlist.c	`99.23% <0.00%> (-0.39%)`	⬇️
py/runtime.c	`99.39% <0.00%> (ø)`
py/emitnative.c	`99.26% <0.00%> (+<0.01%)`	⬆️
py/obj.c	`97.22% <0.00%> (+0.39%)`	⬆️
py/bc.c	`89.69% <0.00%> (+1.03%)`	⬆️

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 0a51073...f8bdfad. Read the comment docs.

dpgeorge · 2021-09-14T05:45:34Z

ports/mimxrt/boards/make-flexram-config.py

@@ -0,0 +1,194 @@
+#!/usr/bin/env python
+"""Evaluate FlexRAM configuration and generate startup code."""


please make the style of the license/copyright match existing files, eg tools/pyboard.py and tools/mpy_ld.py (use # at the start of the line)

ports/mimxrt/hal/board.h

ports/mimxrt/hal/peripherals.h

ports/mimxrt/hal/pin_mux.h

ports/mimxrt/hal/startup_MIMXRT10xx.S

dpgeorge · 2021-09-14T05:52:45Z

ports/mimxrt/hal/startup_MIMXRT10xx.S

+    ldr    r2, =__data_start__
+    ldr    r3, =__data_end__
+
+#ifdef __PERFORMANCE_IMPLEMENTATION


The difference in code size between __PERFORMANCE_IMPLEMENTATION enabled and disabled is very small, and these chips have a huge amount of flash, so I think it's simpler/cleaner to just unconditionally use the performance version of the code (ie remove the other code).

Hi, I tried it out now but it does not work. Even when just compiling with __PERFORMANCE_IMPLEMENTATION defined, MicroPython crashes.
If you are ok with it, let us merge it as it is and we will look into it later and fix it.

alphaFred · 2021-09-14T07:02:40Z

@dpgeorge I pushed an update with the required fixes (:smile: )

alphaFred · 2021-09-14T09:19:42Z

@dpgeorge @robert-hh
I pushed an update with an update to code formatting (added "ports/mimxrt/**/*.[ch]", to tools/codeformat.pyas in #7785) and ran the formatting on all the files in mimxrt port. Should be indentical with #7785 now.

Btw. whenever I execute codeformat.py it also creates changes in other files:

	modified:   ../../extmod/machine_i2c.c
	modified:   ../../extmod/vfs_lfsx.c
	modified:   ../stm32/mpconfigport.h
	modified:   ../unix/mpconfigport.h
	modified:   ../../py/compile.

dpgeorge · 2021-09-14T11:04:16Z

Btw. whenever I execute codeformat.py it also creates changes in other files:

It's not showing on the CI here, so could be your local uncrustify version.

alphaFred · 2021-09-14T11:05:51Z

Ok, then never mind :-)

ports/mimxrt/Makefile

ports/mimxrt/boards/MIMXRT1050_EVK/mpconfigboard.h

ports/mimxrt/boards/MIMXRT1062.ld

ports/mimxrt/boards/MIMXRT1064.ld

ports/mimxrt/hal/flexspi_hyper_flash.c

ports/mimxrt/boards/MIMXRT1021.ld

ports/mimxrt/boards/MIMXRT1011.ld

dpgeorge · 2021-09-14T11:40:35Z

@alphaFred at this stage I'd still need to rebase it and reword the commit message (to something like `mimxrt: Rework flash configuration." to get the tense correct). But feel free to do that yourself if you like.

alphaFred · 2021-09-14T11:41:57Z

Ok. I hope I get the tense right the next time at first try 😂.

- Moves definition of BOARD_FLASH_SIZE and other header files related to flash configuration into the Makefile. - Adds board specific clock_config.h. - Adds board.h, pin_mux.h, and peripherals.h as they are required by NXP MCU SDK in order to use our own clock_config.h. - Renames board specific FlexSPI configuration files. - Updates flash frequency of MIMXRT1020_EVK - Creates separated flash_config files for QSPI NOR and QSPI Hyper flash. - Unifies VFS start address to be @ 1M for 1010 and 1020 boards. - Unifies 1050EVK boards - Adds support to both NOR and HyperFlash on boards with both capabilities. - Adds automatic FlexRAM initialization to start-up code based on linker script and NXP HAL. - Applies code formatting to all files in mimxrt port. With this change the flash configuration is restructured and organized. This simplifies the configuration process and provides a better overview of each board's settings. With the integration of clock_config.h, board.h, pin_mux.h, and peripherals.h we gain better control of the settings and clock configurations. Furthermore the implementation of an explicit FlexRAM setup improves the system performance and allows for performance tuning. Signed-off-by: Philipp Ebensberger

alphaFred · 2021-09-14T11:57:03Z

@dpgeorge
Ok, I fixed your latest comments, rebased and fixed the title of the commit. Should be good to go. Many thanks for your quick review.

dpgeorge · 2021-09-14T12:18:04Z

Merged!

Make set_stack_limit respect fixed stack

alphaFred force-pushed the mimxrt/refactoring_flash_config branch 2 times, most recently from cc1a1f3 to a88b94d Compare September 10, 2021 05:57

dpgeorge added the port-mimxrt label Sep 10, 2021

alphaFred force-pushed the mimxrt/refactoring_flash_config branch from a88b94d to 13e9a27 Compare September 10, 2021 06:55

alphaFred marked this pull request as ready for review September 10, 2021 07:52

dpgeorge reviewed Sep 13, 2021

View reviewed changes

robert-hh mentioned this pull request Sep 13, 2021

extmod: Move modnetwork, modusocket to extmod. #7698

Closed

alphaFred force-pushed the mimxrt/refactoring_flash_config branch from 13e9a27 to b59a909 Compare September 13, 2021 18:33

alphaFred force-pushed the mimxrt/refactoring_flash_config branch from b59a909 to 27ecec4 Compare September 13, 2021 19:02

dpgeorge reviewed Sep 14, 2021

View reviewed changes

dpgeorge mentioned this pull request Sep 14, 2021

mimxrt: reformat boards and hal source files #7785

Closed

alphaFred force-pushed the mimxrt/refactoring_flash_config branch from 27ecec4 to 45f22f8 Compare September 14, 2021 07:01

alphaFred force-pushed the mimxrt/refactoring_flash_config branch 2 times, most recently from c1c76b7 to f8bdfad Compare September 14, 2021 09:17

dpgeorge reviewed Sep 14, 2021

View reviewed changes

alphaFred force-pushed the mimxrt/refactoring_flash_config branch from f8bdfad to 0d7366c Compare September 14, 2021 11:55

alphaFred changed the title ~~mimxrt: Rework of flash configuration.~~ mimxrt: Rework flash configuration. Sep 14, 2021

dpgeorge merged commit 0d7366c into micropython:master Sep 14, 2021

alphaFred deleted the mimxrt/refactoring_flash_config branch September 14, 2021 14:33

tannewt added a commit to tannewt/circuitpython that referenced this pull request Mar 22, 2023

Merge pull request micropython#7767 from tannewt/imx_fixed_stack

53b1661

Make set_stack_limit respect fixed stack

		@@ -0,0 +1,194 @@
		#!/usr/bin/env python
		"""Evaluate FlexRAM configuration and generate startup code."""

Uh oh!

mimxrt: Rework flash configuration. #7767

mimxrt: Rework flash configuration. #7767

Uh oh!

Conversation

alphaFred commented Sep 10, 2021

Uh oh!

robert-hh commented Sep 10, 2021

Uh oh!

dpgeorge Sep 13, 2021

Choose a reason for hiding this comment

Uh oh!

robert-hh Sep 13, 2021

Choose a reason for hiding this comment

Uh oh!

alphaFred Sep 13, 2021

Choose a reason for hiding this comment

Uh oh!

dpgeorge Sep 14, 2021

Choose a reason for hiding this comment

Uh oh!

dpgeorge commented Sep 13, 2021

Uh oh!

alphaFred commented Sep 13, 2021

Uh oh!

robert-hh commented Sep 13, 2021

Uh oh!

alphaFred commented Sep 13, 2021

Uh oh!

codecov-commenter commented Sep 13, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

dpgeorge Sep 14, 2021

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

dpgeorge Sep 14, 2021

Choose a reason for hiding this comment

Uh oh!

alphaFred Sep 14, 2021

Choose a reason for hiding this comment

Uh oh!

alphaFred commented Sep 14, 2021

Uh oh!

alphaFred commented Sep 14, 2021

Uh oh!

dpgeorge commented Sep 14, 2021

Uh oh!

alphaFred commented Sep 14, 2021

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

dpgeorge commented Sep 14, 2021

Uh oh!

alphaFred commented Sep 14, 2021

Uh oh!

alphaFred commented Sep 14, 2021

Uh oh!

dpgeorge commented Sep 14, 2021

Uh oh!

Uh oh!

codecov-commenter commented Sep 13, 2021 •

edited

Loading