llvm dev - Jun 2017 - Wrong description about getelementptr arguments?

Hi All,

  While reading "Arguments" section of getelementptr [1], the
description
is a little wired to me.

    The first argument is always a type used as the basis for the
calculations. The second argument
    is always a pointer or a vector of pointers, and is the base address to
start from.

    ...

    The first index always indexes the pointer value given as the first
argument, ...

Here I see the paragraph says the first argument is a *type*, then says the
first index always indexes
*the pointer value* given as the first argument. From getelementptr FAQ
[2], I feel the first arguement
(first operand in FAQ) should be *the pointer value*, and the following
sentence seems not so right to
me.

    The first argument is always a type used as the basis for the
calculations.

Am I reading it wrong, or it is wrong. Thanks.

[1] http://llvm.org/docs/LangRef.html#getelementptr-instruction
[2] http://llvm.org/docs/GetElementPtr.html

Regards,
chenwj

-- 
Wei-Ren Chen (陳韋任)
Homepage: https://people.cs.nctu.edu.tw/~chenwj
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20170617/16f3d7fc/attachment.html>

Documentation's a bit stale - it'd probably be enlightening/helpful to
look
at some real world examples, such as:

%arrayidx = getelementptr inbounds i32*, i32** %0, i64 3

So the first argument (i32*) is a type, currently it's the same type as the
type that the second argument points to (or, in the case of a vector gep,
it's the type of the pointee of the elements of the vector argument).

The GEP FAQ docs are a bit out of date, if you'd like to send a patch for
them, that'd be great! :) Otherwise I should probably get to them myself...

On Sat, Jun 17, 2017 at 6:09 AM 陳韋任 via llvm-dev <llvm-dev at
lists.llvm.org>
wrote:
> Hi All,
>
>   While reading "Arguments" section of getelementptr [1], the
description
> is a little wired to me.
>
>     The first argument is always a type used as the basis for the
> calculations. The second argument
>     is always a pointer or a vector of pointers, and is the base address
> to start from.
>
>     ...
>
>     The first index always indexes the pointer value given as the first
> argument, ...
>
> Here I see the paragraph says the first argument is a *type*, then says
> the first index always indexes
> *the pointer value* given as the first argument. From getelementptr FAQ
> [2], I feel the first arguement
> (first operand in FAQ) should be *the pointer value*, and the following
> sentence seems not so right to
> me.
>
>     The first argument is always a type used as the basis for the
> calculations.
>
> Am I reading it wrong, or it is wrong. Thanks.
>
> [1] http://llvm.org/docs/LangRef.html#getelementptr-instruction
> [2] http://llvm.org/docs/GetElementPtr.html
>
> Regards,
> chenwj
>
> --
> Wei-Ren Chen (陳韋任)
> Homepage: https://people.cs.nctu.edu.tw/~chenwj
> _______________________________________________
> LLVM Developers mailing list
> llvm-dev at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20170617/71765989/attachment.html>

> Documentation's a bit stale - it'd probably be enlightening/helpful
to
> look at some real world examples, such as:
>
> %arrayidx = getelementptr inbounds i32*, i32** %0, i64 3
>
> So the first argument (i32*) is a type, currently it's the same type as
> the type that the second argument points to (or, in the case of a vector
> gep, it's the type of the pointee of the elements of the vector
argument).
>
​So you mean we actually have three arguments here, ​i32*, i32** %0 and i64
3? The how about the following statement

    The first index always indexes the pointer value given as the first
argument

it should be

    The first index always indexes the pointer value given as the second
argument

​right?

I am okay sending patch to fix the document, btw. :)

Regards,
chenwj​


-- 
Wei-Ren Chen (陳韋任)
Homepage: https://people.cs.nctu.edu.tw/~chenwj
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20170618/529d629a/attachment.html>