arch/arm64/imx9: give the ELE a physical address and the cache a virtual one

[royzah]

Why

The ELE addresses memory physically. Cache maintenance takes a virtual address. Each buffer call supplies one and uses it for both, in opposite directions.

	cache maintenance	address given to the enclave
imx9_ele_get_random()	a physical address	correct
imx9_ele_get_key()	correct	a virtual address

up_flush_dcache() and up_invalidate_dcache() document their arguments as virtual. So imx9_ele_get_random() maintains whatever lines its physical values happen to name, and a caller reads back what its own line still holds rather than what the ELE wrote, with nothing reported.

Correct only while the two addresses are equal. True of every config in tree today; not true of a kernel build with address environments.

How

Both take the virtual address, maintain the cache on it, and translate for the message. imx9_ele_get_random() gains the cache line check imx9_ele_get_key() already has, because a line shared with live data can be written back over the transfer.

imx9_ele_get_random() changes prototype. Neither function has an in-tree caller.

Depends on

#20192

Tested

Compiles for imx93-evk:nsh. tools/nxstyle reports only the two diagnostics already on master, neither in changed lines.

[github-actions]

MemBrowse Memory Report

No memory changes detected for:

• arduino-mega2560
• esp32-devkitc
• hifive1-revb
• mirtoo
• qemu-armv8a
• qemu-intel64
• rx65n-rsk2mb
• s698pm-dkit
• stm32-nucleo-f103rb

[royzah]

Run on hardware alongside #20191: i.MX93 Cortex-A55, PX4 kernel build, NuttX 12.11.0.

hexdump /dev/random -c 32
/dev/random at 00000000:
0000: 7c 9a f9 b0 fb 89 be 9d 23 bf 17 28 88 43 67 10
0010: c5 79 10 9e d7 be ca 42 2f 51 00 bd bc ce 77 54

Note this depends on #20192: imx9_ele.c calls up_addrenv_va_to_pa(), so on a base predating that commit a kernel-mode build fails to link with undefined reference to 'up_addrenv_va_to_pa'. A flat build hides it because addrenv compiles out there. Both are merged now, so it only affects backports.

[royzah]

@acassis @xiaoxiang781216 this one is out of draft too, it is the small ELE address fix that #20191 needs. Hardware log is in the comment above.

No rush, just flagging it is ready.