[llvm-dev] [GSOC] "Project: Improve inter-procedural analyses and optimisations"

[Johannes Doerfert via llvm-dev]

On 3/18/20 7:47 PM, Fahad Nayyar wrote:
 > Hi Johannes,
 >
 > Thanks for your comments!
 >
 >     > I think the AAReachability TODO is being worked on but #179 
not, as far as I know. Would you be interested in taking this one? If 
so, make sure to split it in multiple smaller patches, starting with one 
for the LangRef.doc and the Attributor.
 >
 > Sure!. I can take this up. I'll put up the patches asap. I have a
 > doubt about #task1 of #179 "Write a short RFC for a the attribute and
 > send it to llvm-dev". What kind of description of the attribute should
 > I put there? I was thinking about the meaning of the attribute and its
 > purpose. Please clarify.

Yes. Feel free to share a draft with me. If you want I can also write
the RFC but it might take longer. My suggestion that you write a short
motivation for the attribute and the intended semantics and share that
on this thread. We go over it and send it out as an RFC after. You
should at the same time write/port the LangRef patch so we can point
people there.


 >     > FIWI, I think we want the attribute to mean 1) below. I note this
 >     because 2) is "the opposite" of dereferenceable.
 >       1) the underlying object is at most this large, where the pointer
 >          points to doesn't matter.
 >       2) the underlying object has at most X more dereferenceable 
bytes from
 >          this point forward.
 >     I think we want 1) and later the opposite of 1) as well.
 >     Does this make sense?
 >
 > Yes. I understood this. Just for clarification the opposite of 1) 
will be "the underlying object is at least this large"?

Yes. Which we for now can approximate with `dereferenceable` as it says,
there are at least that many bytes accessible from the current pointer.

 > Thanks and regards.
 > Fahad Nayyar
 >
 >
 >
 >
 >
 >
 >
 >
 > On Thu, Mar 19, 2020 at 5:11 AM Johannes Doerfert 
<johannesdoerfert at gmail.com> wrote:
 >
 >     On 03/16, Fahad Nayyar wrote:
 >     > I can see that Johanned have put up some issues for GSOC 
aspirants. I think
 >     > that [2] <https://github.com/llvm/llvm-project/issues/179> 
([Attributor]
 >     > Cleanup and upstream `Attribute::MaxObjectSize`) will be a very 
good issue
 >     > for me, It seems doable and I can get familiar with the whole 
process of
 >     > writing a patch for an issue. How should I indicate to the 
community that I
 >     > have started working towards this issue (should I comment on 
the issue page
 >     > on github?)? I can try to work on AAReachability TODO after 
solving this
 >     > issue.
 >
 >     I think the AAReachability TODO is being worked on but #179 not, 
as far
 >     as I know. Would you be interested in taking this one? If so, 
make sure
 >     to split it in multiple smaller patches, starting with one for the
 >     LangRef.doc and the Attributor.
 >
 >     FIWI, I think we want the attribute to mean 1) below. I note this
 >     because 2) is "the opposite" of dereferenceable.
 >       1) the underlying object is at most this large, where the pointer
 >          points to doesn't matter.
 >       2) the underlying object has at most X more dereferenceable 
bytes from
 >          this point forward.
 >     I think we want 1) and later the opposite of 1) as well.
 >
 >     Does this make sense?
 >
 >     Cheers,
 >       Johannes
 >