Re: Arm ldaxr / stxr loop question

In article <YfxXO.384093$EEm7.56154@fx16.iad>,
Scott Lurndal <slp53@pacbell.net> wrote:

Do read B2.3 Definition of the Arm memory model.  It's only 32 pages,
and very clearly defines the memory model.

Your definition of "clearly" differs from mine.

Look at Pick dependencies on page B2-239 and B2-240:
(I'm replacing complicating details with "blah blah" or "A, B, C", to
highlight the issue I want to point out)

---
Pick Basic dependency:
There is A, B, C, or a Pick dependency between E1 and E2
Pick Data dependency:
There is a Pick Basic dependency from E1 to E2 and blah blah.
Pick Address dependency:
There is a Pick Data dependency from E1 to E3 and E2 is blah blah
Pick Control dependency:
This is a Pick Basic dependency from E1 to E3 and E2 is blah blah
Pick Dependency:
There is a Pick Basic, Pick Address, Pick Data, or Pick Control
dependency from E1 to E2
---

This is completely circular, and never defines what "pick" is.

Even better, let's look at the actual words for Pick Basic Dependency:

---
Pick Basic Dependency:
There is a Pick Basic dependency from an effect E1 to an effect
E2 if one of the following applies:
1) One of the following applies:
a) E1 is an Explicit Memory Read effect
b) E1 is a Register Read effect
2) One of the following applies:
a) There is a Pick dependency through registers and memory
from E1 to E2
b) E1 and E2 are the same effect
---

Using the words as they are written, if any of 1a, 1b, 2a, or 2a is
true, a Pick Basic dependency exists between E1 and E2.  To give
background, E1 and E2 are any events (effects) and not necessarily in
program order (E2 could be before E1, and can be on another CPU, the
event numbering system is not defined to indicate program order and when
they want to say E1 is in program order before E2 it seems to always
explicitly say so, and there are LOTS of other places where they create
a third event, E3, which may be between E1 and E2).  I'm using event
interchangeably with effect since I think effect is a terrible term.

So by rule 1a by itself, a Pick Basic Dependency exists
between a Load instruction (an example of an Explicit Memory Read, I'm
assuming, as best as I can tell, an Explicit Memory Read is not really
defined) and every other possible event in that system happening before or
after that load.

So what does this mean?  I literally have no idea what they are trying to
get at here.

If E1 and E2 are the "same effect", does that mean it's the same
instruction/operation, or just the same type of operation (like two loads),
or what?  If there was an "overview" summarizing ordering in English,
then it I could interpret the looseness better.

I want to make it clear that I don't want a formal grammar, I just think
this is a particularly poor way to try to present this information.

What it reads like to me like a bad one of those logic puzzles but with info
missing: The cookie was eaten by someone wearing a red coat.  Susan wears
a hat.  Who ate the cookie?

Kent