Modular•16mo ago

`llvm.add` is interpreted as `llvm.addrspacecast`

./std/primitive.mojo:71:64: error: 'llvm.addrspacecast' op requires a single operand
            value: __mlir_op.`llvm.add`[_type: __mlir_type.i32](self.value, rhs.value)

./std/primitive.mojo:71:64: error: 'llvm.addrspacecast' op requires a single operand
            value: __mlir_op.`llvm.add`[_type: __mlir_type.i32](self.value, rhs.value)

This feels like it's not working correctly, considering add is documented as something else.

62 Replies

jmky•16mo ago

Can you share the actual docs on LLVM dialect for this Op? I also find that the translation between Mojo syntax to dialect is a black box to us For eg the index.casts index.castu syntax are different from Mojo's

TeamPuzelOP•16mo ago

this is why they're different

TeamPuzelOP•16mo ago

both llvm and index dialects are off by one at least that's what it looks like :D well, sometimes

ModularBot•16mo ago

Congrats @TeamPuzel, you just advanced to level 13!

TeamPuzelOP•16mo ago

it seems like you have to use the one before what you actually want

TeamPuzelOP•16mo ago

https://mlir.llvm.org/docs/Dialects/LLVM/

'llvm' Dialect - MLIR

Multi-Level IR Compiler Framework

TeamPuzelOP•16mo ago

it's the second one

TeamPuzelOP•16mo ago

whichever one I use mojo seems to take the next

Stole•16mo ago

This is quite funny. On my machine, I can't seem to use anything from the llvm dialect (and haven't tried to before), it can't find the llvm.add and other __mlir_ops, but the index dialect works fine and isn't one off as yours seems to be. For example, I can run the following, and it prints 64:

print(__mlir_op.`index.sizeof`[_type : __mlir_type.index]())

print(__mlir_op.`index.sizeof`[_type : __mlir_type.index]())

Okay, now I get the same one-off error with llvm dialect. Strange. (just in case, my mojo --version is mojo 0.2.1 (64d14e85))

TeamPuzelOP•16mo ago

odd, I have the exact same version Not all of them are one off which makes it even more weird to me Though if I actually try to use llvm.ashr (which I assume must be add for me) it gives me an add error!

Stole•16mo ago

First my LSP says that llvm.add is unregistered, then running it in console says llvm.addrspacecast wants a single operand.

TeamPuzelOP•16mo ago

Yeah, I'm using ssh into a linux laptop with too little ram to handle the lsp so I don't even see those :D if you use ashr you will get add

Stole•16mo ago

On my end it crashes if I replace add with ashr 🤷‍♂️

TeamPuzelOP•16mo ago

unfortunately trying to add i32 in this way crashes llvm

mojo: /__w/modular/modular/third-party/llvm-project/llvm/include/llvm/Support/Casting.h:572: decltype(auto) llvm::cast(From &) [To = mlir::BytecodeOpInterface, From = mlir::Operation]: Assertion `isa<To>(Val) && "cast<Ty>() argument of incompatible type!"' failed.
Please submit a bug report to https://github.com/modularml/mojo/issues and include the crash backtrace along with all the relevant source codes.
Stack dump:
0.      Program arguments: mojo run ./test.mojo
Stack dump without symbol names (ensure you have llvm-symbolizer in your PATH or set the environment var `LLVM_SYMBOLIZER_PATH` to point to it):
0  mojo      0x0000561a0e3f6957
1  mojo      0x0000561a0e3f452e
2  mojo      0x0000561a0e3f702f
3  libc.so.6 0x00007f5596442520
4  libc.so.6 0x00007f5596496a7c pthread_kill + 300
5  libc.so.6 0x00007f5596442476 raise + 22
6  libc.so.6 0x00007f55964287f3 abort + 211
7  libc.so.6 0x00007f559642871b
8  libc.so.6 0x00007f5596439e96
9  mojo      0x0000561a0fcffd35
10 mojo      0x0000561a0fd00ed8
11 mojo      0x0000561a0fcffe8f
12 mojo      0x0000561a0fcff0e0
13 mojo      0x0000561a0fcf2c70
14 mojo      0x0000561a0e849174
15 mojo      0x0000561a0e7b8fcc
16 mojo      0x0000561a0fd53913
17 mojo      0x0000561a0fd53b79
18 mojo      0x0000561a0fd198c4
19 mojo      0x0000561a0e7b7344
20 mojo      0x0000561a0e83e854
21 mojo      0x0000561a0e81caef
22 mojo      0x0000561a0e81d2ab
23 mojo      0x0000561a0e3c14ff
24 mojo      0x0000561a0e3a8a50
25 mojo      0x0000561a0e3c0460
26 mojo      0x0000561a0e3a2926
27 libc.so.6 0x00007f5596429d90
28 libc.so.6 0x00007f5596429e40 __libc_start_main + 128
29 mojo      0x0000561a0e3a212e
Aborted

mojo: /__w/modular/modular/third-party/llvm-project/llvm/include/llvm/Support/Casting.h:572: decltype(auto) llvm::cast(From &) [To = mlir::BytecodeOpInterface, From = mlir::Operation]: Assertion `isa<To>(Val) && "cast<Ty>() argument of incompatible type!"' failed.
Please submit a bug report to https://github.com/modularml/mojo/issues and include the crash backtrace along with all the relevant source codes.
Stack dump:
0.      Program arguments: mojo run ./test.mojo
Stack dump without symbol names (ensure you have llvm-symbolizer in your PATH or set the environment var `LLVM_SYMBOLIZER_PATH` to point to it):
0  mojo      0x0000561a0e3f6957
1  mojo      0x0000561a0e3f452e
2  mojo      0x0000561a0e3f702f
3  libc.so.6 0x00007f5596442520
4  libc.so.6 0x00007f5596496a7c pthread_kill + 300
5  libc.so.6 0x00007f5596442476 raise + 22
6  libc.so.6 0x00007f55964287f3 abort + 211
7  libc.so.6 0x00007f559642871b
8  libc.so.6 0x00007f5596439e96
9  mojo      0x0000561a0fcffd35
10 mojo      0x0000561a0fd00ed8
11 mojo      0x0000561a0fcffe8f
12 mojo      0x0000561a0fcff0e0
13 mojo      0x0000561a0fcf2c70
14 mojo      0x0000561a0e849174
15 mojo      0x0000561a0e7b8fcc
16 mojo      0x0000561a0fd53913
17 mojo      0x0000561a0fd53b79
18 mojo      0x0000561a0fd198c4
19 mojo      0x0000561a0e7b7344
20 mojo      0x0000561a0e83e854
21 mojo      0x0000561a0e81caef
22 mojo      0x0000561a0e81d2ab
23 mojo      0x0000561a0e3c14ff
24 mojo      0x0000561a0e3a8a50
25 mojo      0x0000561a0e3c0460
26 mojo      0x0000561a0e3a2926
27 libc.so.6 0x00007f5596429d90
28 libc.so.6 0x00007f5596429e40 __libc_start_main + 128
29 mojo      0x0000561a0e3a212e
Aborted

Stole•16mo ago

Yep lol

TeamPuzelOP•16mo ago

(Yeah if you pass wrong arguments the error message says add) maybe just the error checking is off and it's actually still trying to use the correct operation

Stole•16mo ago

For some reason, it seems to work fine in the REPL.

TeamPuzelOP•16mo ago

huh Well everything I was planning on writing is now impossible until this is fixed :( rebind doesn't want to work for me with the built in UInt8 alias which makes ffi with C very painful

Stole•16mo ago

I think you should use the pop dialect for now. It looks like Mojo will lower the pop dialect down to llvm at least in some cases, and it has the add operation you want Wait, it doesn't seem to support i32 Darn it

TeamPuzelOP•16mo ago

I think only index and llvm are supported right now and only llvm can handle differently sized numbers (and floats) I'm curious how floats are implemented by the standard library if this doesn't work

Stole•16mo ago

pop, kgen, and lit are the internal dialects Mojo uses

TeamPuzelOP•16mo ago

Do they show up in the lsp?

Stole•16mo ago

The LSP doesn't annotate like the inside of `` for all __mlir_things, you just get a normal error

TeamPuzelOP•16mo ago

I think I've seen those in some error messages before

Stole•16mo ago

From what I've seen, pop is where a lot of this scalar stuff happens, along with more basic types. kgen is where more complex types reside, along with some language specific things. lit is for constructing and dealing with things that act like structs as mlir attrs There actually is one more dialect, hlcf, but it looks like it's for the language itself. Like there are hlcf.if and hlcf.for ops

TeamPuzelOP•16mo ago

hm that's interesting, are they going to be documented at some point?

Stole•16mo ago

No idea, I've just been going through a lot of trial and error

TeamPuzelOP•16mo ago

Should I report the llvm dialect bug on github too?

Stole•16mo ago

I guess so

TeamPuzelOP•16mo ago

I opened an issue now

rd4com•16mo ago

do you get code completion on kgen ? could you share a snippet please @Stole @TeamPuzel i'd like to play too

TeamPuzelOP•16mo ago

I do not get code completion for anything, the lsp is broken for me

JIMC•16mo ago

very interesting discussion here loving it very useful comment i cannot actually find the lit dialect probably not documented but its mojo's internal @TeamPuzel you have any lots of lit dialect etc?

TeamPuzelOP•16mo ago

what do you mean?

Stole•16mo ago

No, there is no code completion while you're typing inside any __mlir_thing

TeamPuzelOP•16mo ago

I had to do this for now, seems to be adding correctly

from std.primitive import UInt8

fn main():
    let x: UInt8 = 255
    let y: UInt8 = 2
    let z: UInt8 = x + y
    
    print(z.to_int())

from std.primitive import UInt8

fn main():
    let x: UInt8 = 255
    let y: UInt8 = 2
    let z: UInt8 = x + y
    
    print(z.to_int())

It technically works but it looks really bad:

@register_passable("trivial")
struct UInt8:
    alias Data = __mlir_type.ui8
    
    var value: Self.Data
    
    fn __init__(value: Int) -> Self:
        return Self {
            value: __mlir_op.`index.castu`[_type: Self.Data](value.__mlir_index__())
        }
    
    fn __add__(self, rhs: Self) -> Self:
        let result = __mlir_op.`index.add`[_type: __mlir_type.index](
            __mlir_op.`index.castu`[_type: __mlir_type.index](self.value),
            __mlir_op.`index.castu`[_type: __mlir_type.index](rhs.value)
        )
        return Self(result)
    
    fn to_int(self) -> Int:
        return Int(__mlir_op.`index.castu`[_type: __mlir_type.index](self.value))

@register_passable("trivial")
struct UInt8:
    alias Data = __mlir_type.ui8
    
    var value: Self.Data
    
    fn __init__(value: Int) -> Self:
        return Self {
            value: __mlir_op.`index.castu`[_type: Self.Data](value.__mlir_index__())
        }
    
    fn __add__(self, rhs: Self) -> Self:
        let result = __mlir_op.`index.add`[_type: __mlir_type.index](
            __mlir_op.`index.castu`[_type: __mlir_type.index](self.value),
            __mlir_op.`index.castu`[_type: __mlir_type.index](rhs.value)
        )
        return Self(result)
    
    fn to_int(self) -> Int:
        return Int(__mlir_op.`index.castu`[_type: __mlir_type.index](self.value))

I'm not sure if this would get optimised, but either way it's going to be very hard to read once it has all of the methods. And then I need to copy and paste this for every type :/ at least the alias will help

Stole•16mo ago

Yeah, just doing the adding through index isn't ideal

rd4com•16mo ago

hello for llvm it is var f = intrinsics.llvm_intrinsic"llvm.cos.f64",Float64

jmky•16mo ago

intrinsics? can you show us the full codes so we can study it deeply? @rd4com did u try if it correctly rejects compile time values that are not UInt8 eg alias my_int = 256? it should reject as 256 > 255

rd4com•16mo ago

yes sorry you need import sys.intrinsics

import sys.intrinsics
var f = intrinsics.llvm_intrinsic["llvm.cos.f64",Float64](1.4)

jmky•16mo ago

so this is doing cosine(1.4)?

rd4com•16mo ago

yes, i just try to figure out where are all the intrinsic we could use

TeamPuzelOP•16mo ago

I didn’t add any checks I just let it wrap around Huh, is that intended? Aren’t those actual llvm intrinsics as opposed to mlir Or am I missing something But if that works I guess that’s cool 😄 I will have to write a constructor to convert between different numbers for every possible combination 😨

jmky•16mo ago

yes u must check and also, there is no known way to raise compile time error

TeamPuzelOP•16mo ago

If there's no way to compile error it will have to wrap :/

rd4com•16mo ago

you can use raise in @parameter if loop no

Stole•16mo ago

https://docs.modular.com/mojo/stdlib/builtin/constrained.html

Modular Docs - constrained

Module

Stole•16mo ago

Is this what you're looking for?

rd4com•16mo ago

🙂

JIMC•16mo ago

https://github.com/modularml/mojo/issues/776

GitHub

[BUG]: Unable to convert a MLIR type to another MLIR type at compil...

Bug description I am attempting to create my own UInt8 following the tutorial on Mojo's documentation on the creation of a new boolean. Below is my implementation: @register_passable('trivi...

JIMC•16mo ago

i created an issue

rd4com•16mo ago

https://github.com/rd4com/mojo-learning/tree/main/FFI%20shared%20libraries%20load%20and%20use

GitHub

mojo-learning/FFI shared libraries load and use at main · rd4com/mo...

Learn some mojo. Contribute to rd4com/mojo-learning development by creating an account on GitHub.

Stole•16mo ago

I'm just wondering, did you try @parameter if as well?

JIMC•16mo ago

@register_passable('trivial')
struct UInt8:
   var value: __mlir_type.ui8

   fn __init__(value: Int) -> Self:
      let value_as_index = value.__mlir_index__()
      let _255 = __mlir_op.`index.castu`[_type : __mlir_type.ui8](__mlir_attr.`255:index`)
      let _255_index = __mlir_op.`index.castu`[_type : __mlir_type.index](_255)
      let greater_than_255 = __mlir_op.`index.cmp`[pred : __mlir_attr.`#index<cmp_predicate ugt>`](value_as_index, _255_index)
      let _0 = __mlir_op.`index.castu`[_type : __mlir_type.index](__mlir_attr.`0`)
      let lesser_than_0 = __mlir_op.`index.cmp`[pred : __mlir_attr.`#index<cmp_predicate ult>`](value_as_index, _0)
      @parameter
      if greater_than_255 or lesser_than_0:
         constrained[False, "compile error"]()
      return Self {
         value: __mlir_op.`index.castu`[_type: __mlir_type.ui8](value_as_index)
      }

   fn into_int(self) -> Int:
      return __mlir_op.`index.castu`[_type : __mlir_type.index](self.value)
      
fn main():
   alias a: UInt8 = 4
   print(a.into_int())

@register_passable('trivial')
struct UInt8:
   var value: __mlir_type.ui8

   fn __init__(value: Int) -> Self:
      let value_as_index = value.__mlir_index__()
      let _255 = __mlir_op.`index.castu`[_type : __mlir_type.ui8](__mlir_attr.`255:index`)
      let _255_index = __mlir_op.`index.castu`[_type : __mlir_type.index](_255)
      let greater_than_255 = __mlir_op.`index.cmp`[pred : __mlir_attr.`#index<cmp_predicate ugt>`](value_as_index, _255_index)
      let _0 = __mlir_op.`index.castu`[_type : __mlir_type.index](__mlir_attr.`0`)
      let lesser_than_0 = __mlir_op.`index.cmp`[pred : __mlir_attr.`#index<cmp_predicate ult>`](value_as_index, _0)
      @parameter
      if greater_than_255 or lesser_than_0:
         constrained[False, "compile error"]()
      return Self {
         value: __mlir_op.`index.castu`[_type: __mlir_type.ui8](value_as_index)
      }

   fn into_int(self) -> Int:
      return __mlir_op.`index.castu`[_type : __mlir_type.index](self.value)
      
fn main():
   alias a: UInt8 = 4
   print(a.into_int())

Error:

/marray# mojo array3.mojo
array3.mojo:13:10: error: cannot use a dynamic value in operator argument
      if greater_than_255 or lesser_than_0:
         ^~~~~~~~~~~~~~~~
array3.mojo:13:10: error: cannot use a dynamic value in '@parameter if' condition
      if greater_than_255 or lesser_than_0:
         ^~~~~~~~~~~~~~~~
array3.mojo:13:30: error: cannot use a dynamic value in boolean condition
      if greater_than_255 or lesser_than_0:
                             ^~~~~~~~~~~~~
mojo: error: failed to parse the provided Mojo

/marray# mojo array3.mojo
array3.mojo:13:10: error: cannot use a dynamic value in operator argument
      if greater_than_255 or lesser_than_0:
         ^~~~~~~~~~~~~~~~
array3.mojo:13:10: error: cannot use a dynamic value in '@parameter if' condition
      if greater_than_255 or lesser_than_0:
         ^~~~~~~~~~~~~~~~
array3.mojo:13:30: error: cannot use a dynamic value in boolean condition
      if greater_than_255 or lesser_than_0:
                             ^~~~~~~~~~~~~
mojo: error: failed to parse the provided Mojo

Stole•16mo ago

Oh, I did forget that you're using a mlir op here, which isn't going to be done at compile time. I'll see if I can find a mlir attr that does the comparing work, but yeah this might not be possible There's a lot else here that isn't compile time as well I mean, @JIMC, do you really need to use MLIR here? Is that a requirement of yours? because things would be much simpler if you just didn't use MLIR here TBH I don't think there's a way to cast between integer types like i1, ui8, etc. at compile time through MLIR At least if there is, it's ultra obscure, and not documented at all if it's part of Mojo's dialects

jmky•16mo ago

it is certainly very possible and a requirement to be able to cast an index to ui8. think about blockchain address that are exactly 20 bytes. many languages like rust, allows for dev to create types that are exactly 20 bytes and checks them at compile time

Stole•16mo ago

You can cast values between ui8 and index fine, but they're mlir ops and that happens at runtime right now, not at compile time

JIMC•16mo ago

no MLIR ops are supposed and can happen at compile time have u read up on MLIR?

Stole•16mo ago

How do you call that op at compile time in Mojo? It seems like every __mlir_op call is interpreted as dynamic

jmky•16mo ago

no, it is not interpreted as dynamic, it is a bug at the mlir level that the foldr optimization is not working for index type you can check out the issue i raised and if you do alias A: U8 = 4; you will see #lit.value{4} being passed into the constructor

TeamPuzelOP•16mo ago

Still broken after the new update 😔

jmky•16mo ago

oh theres a new update?

Gaming

Programming

`llvm.add` is interpreted as `llvm.addrspacecast`