Messages in this thread | | | Date | Fri, 6 Apr 2012 19:21:40 +0200 | From | Alessandro Rubini <> | Subject | Re: [PATCH 0/1] Renaming AMBA to PRIMECELL |
| |
Thank your Rob for replying. I acknowledge the reasoning has not been discussed.
> While I agree this is a more correct name, I think this is pointless > churn and as you highlight above incomplete.
I don't know. I can't evaluate the "pointless" bit. I leave this decision to people more involved with the fpga/asic/cpu environment.
> What's the reasoning for this other than rmk suggested it?
My aim is using primecell for x86: my target chip is a pci-e bridge that exposes a number of amba devices to the PC; some of those are PrimeCell devices for which the existing drivers work fine after you set up stuff correctly. While touching this stuff I"ve been asked to make the rename, and for me "rmk suggested it" is strong enough a reason.
I suspect this misnaming may be disturbing in some technical environments; it's known that amba is being used in many environments without primecell logic blocks, and it's now spreading fast to non-arm SoC's. It may happen that a new user base may be confused by a misnaming (e.g., my ata is under amba, but it's not a primecell block, so it's not including <linux/amba.h>).
It's not unlike how IDE became ATA over time. If there's agreement about its usefulness I can deal with the move, before submitting the bridge. Otherwise I'm also fine with using the old naming in my patches.
/alessandro, feeling flame alert, thus maybe too verbose
| |