[coreboot-gerrit] [XL] Change in coreboot[master]: x86: Change MMIO addr in readN(addr)/writeN(addr, val) to pointer

[Nico Huber (Code Review)]

Attention is currently required from: Kevin Paul Herbert.

Nico Huber has posted comments on this change by Kevin Paul Herbert. ( https://review.coreboot.org/c/coreboot/+/7784?usp=email )

Change subject: x86: Change MMIO addr in readN(addr)/writeN(addr, val) to pointer
......................................................................

Patch Set 10:

(1 comment)

File src/device/azalia_device.c:

https://review.coreboot.org/c/coreboot/+/7784/comment/e60449bd_cb3f826b?usp=... :
PS10, Line 62:  codec_detect(u8 *base)
> Please why the "base" changed from `u32` to `u8`?

It didn't. It was changed to a pointer to `u8`. That's a different type.
The reason `u8 *` is used is that there is pointer arithmetic going on, e.g.
```
if (set_bits(base + 0x08, ...
```
If we want this `+ 0x08` to mean 8 *bytes* higher in memory, the pointer needs
to point to something of 1B size. (some people use `void *` instead which works
too if one knows that the compiler implies a size of 1B but it's not standard C).
> shouldn't be `uintptr_t`

Not in the context of this patch. The summary says: "x86: Change [...] to pointer".
So the intention was to make more use of pointers. Back then we didn't have the
readXXp()/writeXXp() versions, so using pointers avoided a lot of casts, I guess.