Improve wording around buffer mutexes

[mkinsner]

No description provided.

[keryell]

Thanks.

[gmlueck]

This PR needs one more reviewer. @TApplencourt, I see that you left comments. If you are happy with this PR, could you approve?

[TApplencourt]

I still dislike the host allocation. host allocation for me should be reserved for memory allocated via sycl::mallloc_host and not the regular malloc memory.

I think we use hostData in other parts of the spec to refer to memory on the host regardless of how it has been allocated.

IMO it's important to have a distinction between the two. Or maybe just to define what we mean by host allocation somewhere.

But my rant can be considered orthogonal with this issue. We can deal with it another time.

[gmlueck]

I still dislike the host allocation. host allocation for me should be reserved for memory allocated via sycl::mallloc_host and not the regular malloc memory.

We should probably use the term "host USM" or "host USM allocation" when we mean memory from sycl::malloc_host. This is a bigger change, though, that is outside of the scope of this PR.

[gmlueck]

We agreed in the WG meeting last week that this PR can be merged once @mkinsner makes the wording change that the WG proposed. He has now done this in 01e0449, so I think this is ready to merge.

I'll wait another day for comments. If there are no more comments, I'll merge it.

[keryell]

Thanks for the even better clarification.

[gmlueck]

Merging as agreed in last WG meeting.