Merged imxrt1020 into master
* Mostly arch/arm/src/imxrt and include/imxrt: This commit gets imx1020 family support started. It is pretty similar, but subtly different, to 1050/60 (less PLLs, less GPIO banks, differently numbered ports etc. etc.). I think it's time to at least look at what needs to be done before we can get these into the mainline kernel even though they're not properly mature yet. Core memory functions, PLLs, serial and networking are all working, but there's no stress-testing or similar been done...I'm not aware of anything broken, but there again I haven't looked too hard.
I want to go through and start moving each of the imxrt family-specific files into its own subdirectory to tidy things up a bit, and remove the vast majority of ifdefs from the mainline code...but I think it's appropriate to separate the cleanup steps from the new-chip-suppport steps to keep things a bit more compartmentalized. Next week is a travel week so it's possible I'll get time on the cleanup side of the house.
I can't be 100% certain I didn't inflict casualties on the imxrt1050/60 code in the process of doing this, but nsh for 1050 and netnsh and nsh for the 1060 both compile correctly, so if I did break anything it's going to be properly evil and difficult to find.
* include/nuttx/usb/hid.h: The usbhid_descriptor_s struct defined in hid.h included some optional fields that should not be hard-defined as part of that structure. An arbitrray number of optional entries could be included in the descriptor, but that is not properly represneted. No code on NuttX currently depends on the structure definition with an optional descriptor so it is safe to remove.
* Directory re-arrangement and splitting of each imxrt family variant
* Prevent intermittent SDCard timeouts by increasing timer in line with spec.
* Support data dump on Hardfault
* Fix missing ccm definitions and wrong iomux offset
* Prevent hardfault when can't read from ENET PHY
* More consistent Ethernet pinmux
* Mostly arch/arm/src/imxrt and include/imxrt: This commit gets imx1020 family support started. It is pretty similar, but subtly different, to 1050/60 (less PLLs, less GPIO banks, differently numbered ports etc. etc.). I think it's time to at least look at what needs to be done before we can get these into the mainline kernel even though they're not properly mature yet. Core memory functions, PLLs, serial and networking are all working, but there's no stress-testing or similar been done...I'm not aware of anything broken, but there again I haven't looked too hard.
I want to go through and start moving each of the imxrt family-specific files into its own subdirectory to tidy things up a bit, and remove the vast majority of ifdefs from the mainline code...but I think it's appropriate to separate the cleanup steps from the new-chip-suppport steps to keep things a bit more compartmentalized. Next week is a travel week so it's possible I'll get time on the cleanup side of the house.
I can't be 100% certain I didn't inflict casualties on the imxrt1050/60 code in the process of doing this, but nsh for 1050 and netnsh and nsh for the 1060 both compile correctly, so if I did break anything it's going to be properly evil and difficult to find.
* Directory re-arrangement and splitting of each imxrt family variant
* Prevent intermittent SDCard timeouts by increasing timer in line with spec.
* Support data dump on Hardfault
* Fix missing ccm definitions and wrong iomux offset
* Prevent hardfault when can't read from ENET PHY
* More consistent Ethernet pinmux
* Various small commits starting i2c debug
This commit is contained in:
Gregory Nutt
parent
3e848fb893
commit
cfc254a684