Is READ/WRITE_REGISTER_UXYZ still best practice?

NTDEV
1

Based on my searches of this forum, it appears that the answer is “YES”, just checking, since I’ve only recently returned to bit banging after more than a decade writing software much higher in the stack …

my_struct-\>member = 42 is considerably easier to read than WRITE_REGISTER_ULONG(PULONG(my_struct + offsetof(MY_STRUCT, member)), 42).

Maybe I need to make an addressof() thing for that …

Thanks in advance for any discussion you might offer,

Phil B

Not speaking for AMD
Phil Barila
Senior Software Engineer | AMD
Adaptive and Embedded Computing Group
C +(1) 303-249-9361

2

“WRITE_REGISTER_ULONG(&my_struct->member, 42);” works just fine, reads pretty easily, and compiles to exactly the same code you have above (assuming the platform doesn’t have memory barriers).

1 Like
3

@Tim_Roberts said:
“WRITE_REGISTER_ULONG(&my_struct->member, 42);” works just fine, reads pretty easily, and compiles to exactly the same code you have above (assuming the platform doesn’t have memory barriers).

Yeah, that is quite readable. Wish I had thought of that … :slight_smile: