Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Bad native codegen (2017-12-25 Nightly) #47015

Closed
leonardo-m opened this issue Dec 26, 2017 · 1 comment · Fixed by #47828
Closed

Bad native codegen (2017-12-25 Nightly) #47015

leonardo-m opened this issue Dec 26, 2017 · 1 comment · Fixed by #47828
Labels
A-LLVM Area: Code generation parts specific to LLVM. Both correctness bugs and optimization-related issues. I-slow Issue: Problems and improvements with respect to performance of generated code.

Comments

@leonardo-m
Copy link

Given this small program:

#[inline(never)]
fn e97() -> u64 {
    // Input.
    const BASE: f64 = 28_433.0;
    const EXP: u64 = 7_830_457;
    const BIG: f64 = 1e10;

    let mut b = EXP.next_power_of_two();

    let mut n = 1.0;
    while b > 0 {
        if EXP & b != 0 {
            n = (n * n * 2.0) % BIG;
        } else {
            n = (n * n) % BIG;
        }
        b >>= 1;
    }

    ((n * BASE + 1.0) % BIG) as u64
}

fn main() {
    assert_eq!(e97(), 8_739_992_577);
}

I am using:

rustc 1.24.0-nightly (4a7c072fa 2017-12-25)
commit-hash: 4a7c072fa61b42f96d8b75c37fc1edfd71172695
host: x86_64-pc-windows-gnu

If I compile that program with this it works correctly:
rustc -C opt-level=1 test.rs

If I compile it with a native CPU target it asserts:
rustc -C opt-level=1 -C target-cpu=native test.rs

The llvm-ir of the e97 function produced with a default CPU target:

; test::e97
; Function Attrs: noinline norecurse nounwind readnone uwtable
define internal fastcc i64 @_ZN4test3e9717h2a230e2896608cc2E() unnamed_addr #3 {
start:
  br label %bb4

bb3:                                              ; preds = %bb4
  %0 = fmul double %n.1.1, 2.843300e+04
  %1 = fadd double %0, 1.000000e+00
  %2 = frem double %1, 1.000000e+10
  %3 = fptoui double %2 to i64
  ret i64 %3

bb4:                                              ; preds = %bb4, %start
  %b.08 = phi i64 [ 8388608, %start ], [ %12, %bb4 ]
  %n.07 = phi double [ 1.000000e+00, %start ], [ %n.1.1, %bb4 ]
  %4 = and i64 %b.08, 7830457
  %5 = icmp eq i64 %4, 0
  %6 = fmul double %n.07, %n.07
  %7 = fmul double %6, 2.000000e+00
  %n.1.in = select i1 %5, double %6, double %7
  %n.1 = frem double %n.1.in, 1.000000e+10
  %8 = and i64 %b.08, 15660914
  %9 = icmp eq i64 %8, 0
  %10 = fmul double %n.1, %n.1
  %11 = fmul double %10, 2.000000e+00
  %n.1.in.1 = select i1 %9, double %10, double %11
  %n.1.1 = frem double %n.1.in.1, 1.000000e+10
  %12 = lshr i64 %b.08, 2
  %13 = icmp eq i64 %12, 0
  br i1 %13, label %bb3, label %bb4
}

And the llvm-ir using the native CPU target:

; test::e97
; Function Attrs: noinline norecurse nounwind readnone uwtable
define internal fastcc i64 @_ZN4test3e9717h2a230e2896608cc2E() unnamed_addr #3 {
start:
  br label %bb4

bb3:                                              ; preds = %bb4
  %0 = fmul double %n.1.3, 2.843300e+04
  %1 = fadd double %0, 1.000000e+00
  %2 = frem double %1, 1.000000e+10
  %3 = fptoui double %2 to i64
  ret i64 %3

bb4:                                              ; preds = %bb4, %start
  %b.08 = phi i64 [ 8388608, %start ], [ %20, %bb4 ]
  %n.07 = phi double [ 1.000000e+00, %start ], [ %n.1.3, %bb4 ]
  %4 = and i64 %b.08, 7830457
  %5 = icmp eq i64 %4, 0
  %6 = fmul double %n.07, %n.07
  %7 = fmul double %6, 2.000000e+00
  %n.1.in = select i1 %5, double %6, double %7
  %n.1 = frem double %n.1.in, 1.000000e+10
  %8 = and i64 %b.08, 15660914
  %9 = icmp eq i64 %8, 0
  %10 = fmul double %n.1, %n.1
  %11 = fmul double %10, 2.000000e+00
  %n.1.in.1 = select i1 %9, double %10, double %11
  %n.1.1 = frem double %n.1.in.1, 1.000000e+10
  %12 = and i64 %b.08, 31321828
  %13 = icmp eq i64 %12, 0
  %14 = fmul double %n.1.1, %n.1.1
  %15 = fmul double %14, 2.000000e+00
  %n.1.in.2 = select i1 %13, double %14, double %15
  %n.1.2 = frem double %n.1.in.2, 1.000000e+10
  %16 = and i64 %b.08, 62643656
  %17 = icmp eq i64 %16, 0
  %18 = fmul double %n.1.2, %n.1.2
  %19 = fmul double %18, 2.000000e+00
  %n.1.in.3 = select i1 %17, double %18, double %19
  %n.1.3 = frem double %n.1.in.3, 1.000000e+10
  %20 = lshr i64 %b.08, 4
  %21 = icmp eq i64 %20, 0
  br i1 %21, label %bb3, label %bb4
}

Their diff:

8c8
<   %0 = fmul double %n.1.1, 2.843300e+04
---
>   %0 = fmul double %n.1.3, 2.843300e+04
15,16c15,16
<   %b.08 = phi i64 [ 8388608, %start ], [ %12, %bb4 ]
<   %n.07 = phi double [ 1.000000e+00, %start ], [ %n.1.1, %bb4 ]
---
>   %b.08 = phi i64 [ 8388608, %start ], [ %20, %bb4 ]
>   %n.07 = phi double [ 1.000000e+00, %start ], [ %n.1.3, %bb4 ]
29,31c29,43
<   %12 = lshr i64 %b.08, 2
<   %13 = icmp eq i64 %12, 0
<   br i1 %13, label %bb3, label %bb4
---
>   %12 = and i64 %b.08, 31321828
>   %13 = icmp eq i64 %12, 0
>   %14 = fmul double %n.1.1, %n.1.1
>   %15 = fmul double %14, 2.000000e+00
>   %n.1.in.2 = select i1 %13, double %14, double %15
>   %n.1.2 = frem double %n.1.in.2, 1.000000e+10
>   %16 = and i64 %b.08, 62643656
>   %17 = icmp eq i64 %16, 0
>   %18 = fmul double %n.1.2, %n.1.2
>   %19 = fmul double %18, 2.000000e+00
>   %n.1.in.3 = select i1 %17, double %18, double %19
>   %n.1.3 = frem double %n.1.in.3, 1.000000e+10
>   %20 = lshr i64 %b.08, 4
>   %21 = icmp eq i64 %20, 0
>   br i1 %21, label %bb3, label %bb4
@alexcrichton
Copy link
Member

This looks like a bug in LLVM unfortunately. For example:

input.ll

$ opt input.ll -S -O2 -mcpu=haswell -o output.ll

output.ll

$ opt output.ll -s -O2 
; ModuleID = '<stdin>'
source_filename = "a.ll"
target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
target triple = "x86_64-unknown-linux-gnu"

; Function Attrs: norecurse nounwind readnone
define i64 @foo() local_unnamed_addr #0 {
start:
  ret i64 9246255105
}

attributes #0 = { norecurse nounwind readnone "target-cpu"="haswell" }

The major change in the nightly here is that ThinLTO was enabled. ThinLTO can run optimizations twice (why opt is run twice here) and it seems as if some bug is generating this answer.

Using LLVM trunk the intermediate result is the same and the second optimization pass generates ret i64 8739992577 instead of the previous number (which is what we'd expect).

As a result this is probably just a bland "already fixed in LLVM" bug, making this blocked on #43370

@alexcrichton alexcrichton added the A-LLVM Area: Code generation parts specific to LLVM. Both correctness bugs and optimization-related issues. label Dec 26, 2017
@bstrie bstrie added the I-slow Issue: Problems and improvements with respect to performance of generated code. label Jan 7, 2018
alexcrichton added a commit to alexcrichton/rust that referenced this issue Jan 28, 2018
The following submodules have been updated for a new version of LLVM:

- `src/llvm`
- `src/libcompiler_builtins` - transitively contains compiler-rt
- `src/dlmalloc`

This also updates the docker container for dist-i686-freebsd as the old 16.04
container is no longer capable of building LLVM. The
compiler-rt/compiler-builtins and dlmalloc updates are pretty routine without
much interesting happening, but the LLVM update here is of particular note.
Unlike previous updates I haven't cherry-picked all existing patches we had on
top of our LLVM branch as we have a [huge amount][patches4] and have at this
point forgotten what most of them are for. Instead I started from the current
`release_60` branch in LLVM and only applied patches that were necessary to get
our tests working and building.

The current set of custom rustc-specific patches included in this LLVM update are:

* rust-lang/llvm@1187443 - this is how we actually implement
  `cfg(target_feature)` for now and continues to not be upstreamed. While a
  hazard for SIMD stabilization this commit is otherwise keeping the status
  quo of a small rustc-specific feature.
* rust-lang/llvm@013f2ec - this is a rustc-specific optimization that we haven't
  upstreamed, notably teaching LLVM about our allocation-related routines (which
  aren't malloc/free). Once we stabilize the global allocator routines we will
  likely want to upstream this patch, but for now it seems reasonable to keep it
  on our fork.
* rust-lang/llvm@a65bbfd - I found this necessary to fix compilation of LLVM in
  our 32-bit linux container. I'm not really sure why it's necessary but my
  guess is that it's because of the absolutely ancient glibc that we're using.
  In any case it's only updating pieces we're not actually using in LLVM so I'm
  hoping it'll turn out alright. This doesn't seem like something we'll want to
  upstream.c
* rust-lang/llvm@77ab1f0 - this is what's actually enabling LLVM to build in our
  i686-freebsd container, I'm not really sure what's going on but we for sure
  probably don't want to upstream this and otherwise it seems not too bad for
  now at least.
* rust-lang/llvm@9eb9267 - we currently suffer on MSVC from an [upstream bug]
  which although diagnosed to a particular revision isn't currently fixed
  upstream (and the bug itself doesn't seem too active). This commit is a
  partial revert of the suspected cause of this regression (found via a
  bisection). I'm sort of hoping that this eventually gets fixed upstream with a
  similar fix (which we can replace in our branch), but for now I'm also hoping
  it's a relatively harmless change to have.

After applying these patches (plus one [backport] which should be [backported
upstream][llvm-back]) I believe we should have all tests working on all
platforms in our current test suite. I'm like 99% sure that we'll need some more
backports as issues are reported for LLVM 6 when this propagates through
nightlies, but that's sort of just par for the course nowadays!

In any case though some extra scrutiny of the patches here would definitely be
welcome, along with scrutiny of the "missing patches" like a [change to pass
manager order](rust-lang/llvm@27174447533), [another change to pass manager
order](rust-lang/llvm@c782febb7b9), some [compile fixes for
sparc](rust-lang/llvm@1a83de63c42), and some [fixes for
solaris](rust-lang/llvm@c2bfe0abb).

[patches4]: rust-lang/llvm@5401fdf...rust-llvm-release-4-0-1
[backport]: rust-lang/llvm@5c54c252db
[llvm-back]: https://bugs.llvm.org/show_bug.cgi?id=36114
[upstream bug]: https://bugs.llvm.org/show_bug.cgi?id=36096

---

The update to LLVM 6 is desirable for a number of reasons, notably:

* This'll allow us to keep up with the upstream wasm backend, picking up new
  features as they start landing.
* Upstream LLVM has fixed a number of SIMD-related compilation errors,
  especially around AVX-512 and such.
* There's a few assorted known bugs which are fixed in LLVM 5 and aren't fixed
  in the LLVM 4 branch we're using.
* Overall it's not a great idea to stagnate with our codegen backend!

This update is mostly powered by rust-lang#47730 which is allowing us to update LLVM
*independent* of the version of LLVM that Emscripten is locked to. This means
that when compiling code for Emscripten we'll still be using the old LLVM 4
backend, but when compiling code for any other target we'll be using the new
LLVM 6 target. Once Emscripten updates we may no longer need this distinction,
but we're not sure when that will happen!

Closes rust-lang#43370
Closes rust-lang#43418
Closes rust-lang#47015
Closes rust-lang#47683
Closes rust-lang/stdarch#157
Closes rust-lang-nursery/rust-wasm#3
bors added a commit that referenced this issue Jan 28, 2018
rustc: Upgrade to LLVM 6

The following submodules have been updated for a new version of LLVM:

- `src/llvm`
- `src/libcompiler_builtins` - transitively contains compiler-rt
- `src/dlmalloc`

This also updates the docker container for dist-i686-freebsd as the old 16.04
container is no longer capable of building LLVM. The
compiler-rt/compiler-builtins and dlmalloc updates are pretty routine without
much interesting happening, but the LLVM update here is of particular note.
Unlike previous updates I haven't cherry-picked all existing patches we had on
top of our LLVM branch as we have a [huge amount][patches4] and have at this
point forgotten what most of them are for. Instead I started from the current
`release_60` branch in LLVM and only applied patches that were necessary to get
our tests working and building.

The current set of custom rustc-specific patches included in this LLVM update are:

* rust-lang/llvm@1187443 - this is how we actually implement
  `cfg(target_feature)` for now and continues to not be upstreamed. While a
  hazard for SIMD stabilization this commit is otherwise keeping the status
  quo of a small rustc-specific feature.
* rust-lang/llvm@013f2ec - this is a rustc-specific optimization that we haven't
  upstreamed, notably teaching LLVM about our allocation-related routines (which
  aren't malloc/free). Once we stabilize the global allocator routines we will
  likely want to upstream this patch, but for now it seems reasonable to keep it
  on our fork.
* rust-lang/llvm@a65bbfd - I found this necessary to fix compilation of LLVM in
  our 32-bit linux container. I'm not really sure why it's necessary but my
  guess is that it's because of the absolutely ancient glibc that we're using.
  In any case it's only updating pieces we're not actually using in LLVM so I'm
  hoping it'll turn out alright. This doesn't seem like something we'll want to
  upstream.c
* rust-lang/llvm@77ab1f0 - this is what's actually enabling LLVM to build in our
  i686-freebsd container, I'm not really sure what's going on but we for sure
  probably don't want to upstream this and otherwise it seems not too bad for
  now at least.
* rust-lang/llvm@9eb9267 - we currently suffer on MSVC from an [upstream bug]
  which although diagnosed to a particular revision isn't currently fixed
  upstream (and the bug itself doesn't seem too active). This commit is a
  partial revert of the suspected cause of this regression (found via a
  bisection). I'm sort of hoping that this eventually gets fixed upstream with a
  similar fix (which we can replace in our branch), but for now I'm also hoping
  it's a relatively harmless change to have.

After applying these patches (plus one [backport] which should be [backported
upstream][llvm-back]) I believe we should have all tests working on all
platforms in our current test suite. I'm like 99% sure that we'll need some more
backports as issues are reported for LLVM 6 when this propagates through
nightlies, but that's sort of just par for the course nowadays!

In any case though some extra scrutiny of the patches here would definitely be
welcome, along with scrutiny of the "missing patches" like a [change to pass
manager order](rust-lang/llvm@2717444), [another change to pass manager
order](rust-lang/llvm@c782feb), some [compile fixes for
sparc](rust-lang/llvm@1a83de6), and some [fixes for
solaris](rust-lang/llvm@c2bfe0a).

[patches4]: rust-lang/llvm@5401fdf...rust-llvm-release-4-0-1
[backport]: rust-lang/llvm@5c54c25
[llvm-back]: https://bugs.llvm.org/show_bug.cgi?id=36114
[upstream bug]: https://bugs.llvm.org/show_bug.cgi?id=36096

---

The update to LLVM 6 is desirable for a number of reasons, notably:

* This'll allow us to keep up with the upstream wasm backend, picking up new
  features as they start landing.
* Upstream LLVM has fixed a number of SIMD-related compilation errors,
  especially around AVX-512 and such.
* There's a few assorted known bugs which are fixed in LLVM 5 and aren't fixed
  in the LLVM 4 branch we're using.
* Overall it's not a great idea to stagnate with our codegen backend!

This update is mostly powered by #47730 which is allowing us to update LLVM
*independent* of the version of LLVM that Emscripten is locked to. This means
that when compiling code for Emscripten we'll still be using the old LLVM 4
backend, but when compiling code for any other target we'll be using the new
LLVM 6 target. Once Emscripten updates we may no longer need this distinction,
but we're not sure when that will happen!

Closes #43370
Closes #43418
Closes #47015
Closes #47683
Closes rust-lang/stdarch#157
Closes rust-lang-nursery/rust-wasm#3
alexcrichton added a commit to alexcrichton/rust that referenced this issue Jan 29, 2018
The following submodules have been updated for a new version of LLVM:

- `src/llvm`
- `src/libcompiler_builtins` - transitively contains compiler-rt
- `src/dlmalloc`

This also updates the docker container for dist-i686-freebsd as the old 16.04
container is no longer capable of building LLVM. The
compiler-rt/compiler-builtins and dlmalloc updates are pretty routine without
much interesting happening, but the LLVM update here is of particular note.
Unlike previous updates I haven't cherry-picked all existing patches we had on
top of our LLVM branch as we have a [huge amount][patches4] and have at this
point forgotten what most of them are for. Instead I started from the current
`release_60` branch in LLVM and only applied patches that were necessary to get
our tests working and building.

The current set of custom rustc-specific patches included in this LLVM update are:

* rust-lang/llvm@1187443 - this is how we actually implement
  `cfg(target_feature)` for now and continues to not be upstreamed. While a
  hazard for SIMD stabilization this commit is otherwise keeping the status
  quo of a small rustc-specific feature.
* rust-lang/llvm@013f2ec - this is a rustc-specific optimization that we haven't
  upstreamed, notably teaching LLVM about our allocation-related routines (which
  aren't malloc/free). Once we stabilize the global allocator routines we will
  likely want to upstream this patch, but for now it seems reasonable to keep it
  on our fork.
* rust-lang/llvm@a65bbfd - I found this necessary to fix compilation of LLVM in
  our 32-bit linux container. I'm not really sure why it's necessary but my
  guess is that it's because of the absolutely ancient glibc that we're using.
  In any case it's only updating pieces we're not actually using in LLVM so I'm
  hoping it'll turn out alright. This doesn't seem like something we'll want to
  upstream.c
* rust-lang/llvm@77ab1f0 - this is what's actually enabling LLVM to build in our
  i686-freebsd container, I'm not really sure what's going on but we for sure
  probably don't want to upstream this and otherwise it seems not too bad for
  now at least.
* rust-lang/llvm@9eb9267 - we currently suffer on MSVC from an [upstream bug]
  which although diagnosed to a particular revision isn't currently fixed
  upstream (and the bug itself doesn't seem too active). This commit is a
  partial revert of the suspected cause of this regression (found via a
  bisection). I'm sort of hoping that this eventually gets fixed upstream with a
  similar fix (which we can replace in our branch), but for now I'm also hoping
  it's a relatively harmless change to have.

After applying these patches (plus one [backport] which should be [backported
upstream][llvm-back]) I believe we should have all tests working on all
platforms in our current test suite. I'm like 99% sure that we'll need some more
backports as issues are reported for LLVM 6 when this propagates through
nightlies, but that's sort of just par for the course nowadays!

In any case though some extra scrutiny of the patches here would definitely be
welcome, along with scrutiny of the "missing patches" like a [change to pass
manager order](rust-lang/llvm@27174447533), [another change to pass manager
order](rust-lang/llvm@c782febb7b9), some [compile fixes for
sparc](rust-lang/llvm@1a83de63c42), and some [fixes for
solaris](rust-lang/llvm@c2bfe0abb).

[patches4]: rust-lang/llvm@5401fdf...rust-llvm-release-4-0-1
[backport]: rust-lang/llvm@5c54c252db
[llvm-back]: https://bugs.llvm.org/show_bug.cgi?id=36114
[upstream bug]: https://bugs.llvm.org/show_bug.cgi?id=36096

---

The update to LLVM 6 is desirable for a number of reasons, notably:

* This'll allow us to keep up with the upstream wasm backend, picking up new
  features as they start landing.
* Upstream LLVM has fixed a number of SIMD-related compilation errors,
  especially around AVX-512 and such.
* There's a few assorted known bugs which are fixed in LLVM 5 and aren't fixed
  in the LLVM 4 branch we're using.
* Overall it's not a great idea to stagnate with our codegen backend!

This update is mostly powered by rust-lang#47730 which is allowing us to update LLVM
*independent* of the version of LLVM that Emscripten is locked to. This means
that when compiling code for Emscripten we'll still be using the old LLVM 4
backend, but when compiling code for any other target we'll be using the new
LLVM 6 target. Once Emscripten updates we may no longer need this distinction,
but we're not sure when that will happen!

Closes rust-lang#43370
Closes rust-lang#43418
Closes rust-lang#47015
Closes rust-lang#47683
Closes rust-lang/stdarch#157
Closes rust-lang-nursery/rust-wasm#3
bors added a commit that referenced this issue Jan 30, 2018
rustc: Upgrade to LLVM 6

The following submodules have been updated for a new version of LLVM:

- `src/llvm`
- `src/libcompiler_builtins` - transitively contains compiler-rt
- `src/dlmalloc`

This also updates the docker container for dist-i686-freebsd as the old 16.04
container is no longer capable of building LLVM. The
compiler-rt/compiler-builtins and dlmalloc updates are pretty routine without
much interesting happening, but the LLVM update here is of particular note.
Unlike previous updates I haven't cherry-picked all existing patches we had on
top of our LLVM branch as we have a [huge amount][patches4] and have at this
point forgotten what most of them are for. Instead I started from the current
`release_60` branch in LLVM and only applied patches that were necessary to get
our tests working and building.

The [current set of custom rustc-specific patches](rust-lang/llvm@f128612...rust-llvm-release-6-0-0) included in this LLVM update are:

* rust-lang/llvm@1187443 - this is how we actually implement
  `cfg(target_feature)` for now and continues to not be upstreamed. While a
  hazard for SIMD stabilization this commit is otherwise keeping the status
  quo of a small rustc-specific feature.
* rust-lang/llvm@013f2ec - this is a rustc-specific optimization that we haven't
  upstreamed, notably teaching LLVM about our allocation-related routines (which
  aren't malloc/free). Once we stabilize the global allocator routines we will
  likely want to upstream this patch, but for now it seems reasonable to keep it
  on our fork.
* rust-lang/llvm@a65bbfd - I found this necessary to fix compilation of LLVM in
  our 32-bit linux container. I'm not really sure why it's necessary but my
  guess is that it's because of the absolutely ancient glibc that we're using.
  In any case it's only updating pieces we're not actually using in LLVM so I'm
  hoping it'll turn out alright. This doesn't seem like something we'll want to
  upstream.c
* rust-lang/llvm@77ab1f0 - this is what's actually enabling LLVM to build in our
  i686-freebsd container, I'm not really sure what's going on but we for sure
  probably don't want to upstream this and otherwise it seems not too bad for
  now at least.
* rust-lang/llvm@9eb9267 - we currently suffer on MSVC from an [upstream bug]
  which although diagnosed to a particular revision isn't currently fixed
  upstream (and the bug itself doesn't seem too active). This commit is a
  partial revert of the suspected cause of this regression (found via a
  bisection). I'm sort of hoping that this eventually gets fixed upstream with a
  similar fix (which we can replace in our branch), but for now I'm also hoping
  it's a relatively harmless change to have.

After applying these patches (plus one [backport] which should be [backported
upstream][llvm-back]) I believe we should have all tests working on all
platforms in our current test suite. I'm like 99% sure that we'll need some more
backports as issues are reported for LLVM 6 when this propagates through
nightlies, but that's sort of just par for the course nowadays!

In any case though some extra scrutiny of the patches here would definitely be
welcome, along with scrutiny of the "missing patches" like a [change to pass
manager order](rust-lang/llvm@2717444), [another change to pass manager
order](rust-lang/llvm@c782feb), some [compile fixes for
sparc](rust-lang/llvm@1a83de6), and some [fixes for
solaris](rust-lang/llvm@c2bfe0a).

[patches4]: rust-lang/llvm@5401fdf...rust-llvm-release-4-0-1
[backport]: rust-lang/llvm@5c54c25
[llvm-back]: https://bugs.llvm.org/show_bug.cgi?id=36114
[upstream bug]: https://bugs.llvm.org/show_bug.cgi?id=36096

---

The update to LLVM 6 is desirable for a number of reasons, notably:

* This'll allow us to keep up with the upstream wasm backend, picking up new
  features as they start landing.
* Upstream LLVM has fixed a number of SIMD-related compilation errors,
  especially around AVX-512 and such.
* There's a few assorted known bugs which are fixed in LLVM 5 and aren't fixed
  in the LLVM 4 branch we're using.
* Overall it's not a great idea to stagnate with our codegen backend!

This update is mostly powered by #47730 which is allowing us to update LLVM
*independent* of the version of LLVM that Emscripten is locked to. This means
that when compiling code for Emscripten we'll still be using the old LLVM 4
backend, but when compiling code for any other target we'll be using the new
LLVM 6 target. Once Emscripten updates we may no longer need this distinction,
but we're not sure when that will happen!

Closes #43370
Closes #43418
Closes #47015
Closes #47683
Closes rust-lang/stdarch#157
Closes rust-lang-nursery/rust-wasm#3
alexcrichton added a commit to alexcrichton/rust that referenced this issue Feb 5, 2018
The following submodules have been updated for a new version of LLVM:

- `src/llvm`
- `src/libcompiler_builtins` - transitively contains compiler-rt
- `src/dlmalloc`

This also updates the docker container for dist-i686-freebsd as the old 16.04
container is no longer capable of building LLVM. The
compiler-rt/compiler-builtins and dlmalloc updates are pretty routine without
much interesting happening, but the LLVM update here is of particular note.
Unlike previous updates I haven't cherry-picked all existing patches we had on
top of our LLVM branch as we have a [huge amount][patches4] and have at this
point forgotten what most of them are for. Instead I started from the current
`release_60` branch in LLVM and only applied patches that were necessary to get
our tests working and building.

The current set of custom rustc-specific patches included in this LLVM update are:

* rust-lang/llvm@1187443 - this is how we actually implement
  `cfg(target_feature)` for now and continues to not be upstreamed. While a
  hazard for SIMD stabilization this commit is otherwise keeping the status
  quo of a small rustc-specific feature.
* rust-lang/llvm@013f2ec - this is a rustc-specific optimization that we haven't
  upstreamed, notably teaching LLVM about our allocation-related routines (which
  aren't malloc/free). Once we stabilize the global allocator routines we will
  likely want to upstream this patch, but for now it seems reasonable to keep it
  on our fork.
* rust-lang/llvm@a65bbfd - I found this necessary to fix compilation of LLVM in
  our 32-bit linux container. I'm not really sure why it's necessary but my
  guess is that it's because of the absolutely ancient glibc that we're using.
  In any case it's only updating pieces we're not actually using in LLVM so I'm
  hoping it'll turn out alright. This doesn't seem like something we'll want to
  upstream.c
* rust-lang/llvm@77ab1f0 - this is what's actually enabling LLVM to build in our
  i686-freebsd container, I'm not really sure what's going on but we for sure
  probably don't want to upstream this and otherwise it seems not too bad for
  now at least.
* rust-lang/llvm@9eb9267 - we currently suffer on MSVC from an [upstream bug]
  which although diagnosed to a particular revision isn't currently fixed
  upstream (and the bug itself doesn't seem too active). This commit is a
  partial revert of the suspected cause of this regression (found via a
  bisection). I'm sort of hoping that this eventually gets fixed upstream with a
  similar fix (which we can replace in our branch), but for now I'm also hoping
  it's a relatively harmless change to have.

After applying these patches (plus one [backport] which should be [backported
upstream][llvm-back]) I believe we should have all tests working on all
platforms in our current test suite. I'm like 99% sure that we'll need some more
backports as issues are reported for LLVM 6 when this propagates through
nightlies, but that's sort of just par for the course nowadays!

In any case though some extra scrutiny of the patches here would definitely be
welcome, along with scrutiny of the "missing patches" like a [change to pass
manager order](rust-lang/llvm@27174447533), [another change to pass manager
order](rust-lang/llvm@c782febb7b9), some [compile fixes for
sparc](rust-lang/llvm@1a83de63c42), and some [fixes for
solaris](rust-lang/llvm@c2bfe0abb).

[patches4]: rust-lang/llvm@5401fdf...rust-llvm-release-4-0-1
[backport]: rust-lang/llvm@5c54c252db
[llvm-back]: https://bugs.llvm.org/show_bug.cgi?id=36114
[upstream bug]: https://bugs.llvm.org/show_bug.cgi?id=36096

---

The update to LLVM 6 is desirable for a number of reasons, notably:

* This'll allow us to keep up with the upstream wasm backend, picking up new
  features as they start landing.
* Upstream LLVM has fixed a number of SIMD-related compilation errors,
  especially around AVX-512 and such.
* There's a few assorted known bugs which are fixed in LLVM 5 and aren't fixed
  in the LLVM 4 branch we're using.
* Overall it's not a great idea to stagnate with our codegen backend!

This update is mostly powered by rust-lang#47730 which is allowing us to update LLVM
*independent* of the version of LLVM that Emscripten is locked to. This means
that when compiling code for Emscripten we'll still be using the old LLVM 4
backend, but when compiling code for any other target we'll be using the new
LLVM 6 target. Once Emscripten updates we may no longer need this distinction,
but we're not sure when that will happen!

Closes rust-lang#43370
Closes rust-lang#43418
Closes rust-lang#47015
Closes rust-lang#47683
Closes rust-lang/stdarch#157
Closes rust-lang-nursery/rust-wasm#3
bors added a commit that referenced this issue Feb 5, 2018
rustc: Upgrade to LLVM 6

The following submodules have been updated for a new version of LLVM:

- `src/llvm`
- `src/libcompiler_builtins` - transitively contains compiler-rt
- `src/dlmalloc`

This also updates the docker container for dist-i686-freebsd as the old 16.04
container is no longer capable of building LLVM. The
compiler-rt/compiler-builtins and dlmalloc updates are pretty routine without
much interesting happening, but the LLVM update here is of particular note.
Unlike previous updates I haven't cherry-picked all existing patches we had on
top of our LLVM branch as we have a [huge amount][patches4] and have at this
point forgotten what most of them are for. Instead I started from the current
`release_60` branch in LLVM and only applied patches that were necessary to get
our tests working and building.

The [current set of custom rustc-specific patches](rust-lang/llvm@f128612...rust-llvm-release-6-0-0) included in this LLVM update are:

* rust-lang/llvm@1187443 - this is how we actually implement
  `cfg(target_feature)` for now and continues to not be upstreamed. While a
  hazard for SIMD stabilization this commit is otherwise keeping the status
  quo of a small rustc-specific feature.
* rust-lang/llvm@013f2ec - this is a rustc-specific optimization that we haven't
  upstreamed, notably teaching LLVM about our allocation-related routines (which
  aren't malloc/free). Once we stabilize the global allocator routines we will
  likely want to upstream this patch, but for now it seems reasonable to keep it
  on our fork.
* rust-lang/llvm@a65bbfd - I found this necessary to fix compilation of LLVM in
  our 32-bit linux container. I'm not really sure why it's necessary but my
  guess is that it's because of the absolutely ancient glibc that we're using.
  In any case it's only updating pieces we're not actually using in LLVM so I'm
  hoping it'll turn out alright. This doesn't seem like something we'll want to
  upstream.c
* rust-lang/llvm@77ab1f0 - this is what's actually enabling LLVM to build in our
  i686-freebsd container, I'm not really sure what's going on but we for sure
  probably don't want to upstream this and otherwise it seems not too bad for
  now at least.
* rust-lang/llvm@9eb9267 - we currently suffer on MSVC from an [upstream bug]
  which although diagnosed to a particular revision isn't currently fixed
  upstream (and the bug itself doesn't seem too active). This commit is a
  partial revert of the suspected cause of this regression (found via a
  bisection). I'm sort of hoping that this eventually gets fixed upstream with a
  similar fix (which we can replace in our branch), but for now I'm also hoping
  it's a relatively harmless change to have.

After applying these patches (plus one [backport] which should be [backported
upstream][llvm-back]) I believe we should have all tests working on all
platforms in our current test suite. I'm like 99% sure that we'll need some more
backports as issues are reported for LLVM 6 when this propagates through
nightlies, but that's sort of just par for the course nowadays!

In any case though some extra scrutiny of the patches here would definitely be
welcome, along with scrutiny of the "missing patches" like a [change to pass
manager order](rust-lang/llvm@2717444), [another change to pass manager
order](rust-lang/llvm@c782feb), some [compile fixes for
sparc](rust-lang/llvm@1a83de6), and some [fixes for
solaris](rust-lang/llvm@c2bfe0a).

[patches4]: rust-lang/llvm@5401fdf...rust-llvm-release-4-0-1
[backport]: rust-lang/llvm@5c54c25
[llvm-back]: https://bugs.llvm.org/show_bug.cgi?id=36114
[upstream bug]: https://bugs.llvm.org/show_bug.cgi?id=36096

---

The update to LLVM 6 is desirable for a number of reasons, notably:

* This'll allow us to keep up with the upstream wasm backend, picking up new
  features as they start landing.
* Upstream LLVM has fixed a number of SIMD-related compilation errors,
  especially around AVX-512 and such.
* There's a few assorted known bugs which are fixed in LLVM 5 and aren't fixed
  in the LLVM 4 branch we're using.
* Overall it's not a great idea to stagnate with our codegen backend!

This update is mostly powered by #47730 which is allowing us to update LLVM
*independent* of the version of LLVM that Emscripten is locked to. This means
that when compiling code for Emscripten we'll still be using the old LLVM 4
backend, but when compiling code for any other target we'll be using the new
LLVM 6 target. Once Emscripten updates we may no longer need this distinction,
but we're not sure when that will happen!

Closes #43370
Closes #43418
Closes #47015
Closes #47683
Closes rust-lang/stdarch#157
Closes rust-lang-nursery/rust-wasm#3
alexcrichton added a commit that referenced this issue Feb 7, 2018
The following submodules have been updated for a new version of LLVM:

- `src/llvm`
- `src/libcompiler_builtins` - transitively contains compiler-rt
- `src/dlmalloc`

This also updates the docker container for dist-i686-freebsd as the old 16.04
container is no longer capable of building LLVM. The
compiler-rt/compiler-builtins and dlmalloc updates are pretty routine without
much interesting happening, but the LLVM update here is of particular note.
Unlike previous updates I haven't cherry-picked all existing patches we had on
top of our LLVM branch as we have a [huge amount][patches4] and have at this
point forgotten what most of them are for. Instead I started from the current
`release_60` branch in LLVM and only applied patches that were necessary to get
our tests working and building.

The current set of custom rustc-specific patches included in this LLVM update are:

* rust-lang/llvm@1187443 - this is how we actually implement
  `cfg(target_feature)` for now and continues to not be upstreamed. While a
  hazard for SIMD stabilization this commit is otherwise keeping the status
  quo of a small rustc-specific feature.
* rust-lang/llvm@013f2ec - this is a rustc-specific optimization that we haven't
  upstreamed, notably teaching LLVM about our allocation-related routines (which
  aren't malloc/free). Once we stabilize the global allocator routines we will
  likely want to upstream this patch, but for now it seems reasonable to keep it
  on our fork.
* rust-lang/llvm@a65bbfd - I found this necessary to fix compilation of LLVM in
  our 32-bit linux container. I'm not really sure why it's necessary but my
  guess is that it's because of the absolutely ancient glibc that we're using.
  In any case it's only updating pieces we're not actually using in LLVM so I'm
  hoping it'll turn out alright. This doesn't seem like something we'll want to
  upstream.c
* rust-lang/llvm@77ab1f0 - this is what's actually enabling LLVM to build in our
  i686-freebsd container, I'm not really sure what's going on but we for sure
  probably don't want to upstream this and otherwise it seems not too bad for
  now at least.
* rust-lang/llvm@9eb9267 - we currently suffer on MSVC from an [upstream bug]
  which although diagnosed to a particular revision isn't currently fixed
  upstream (and the bug itself doesn't seem too active). This commit is a
  partial revert of the suspected cause of this regression (found via a
  bisection). I'm sort of hoping that this eventually gets fixed upstream with a
  similar fix (which we can replace in our branch), but for now I'm also hoping
  it's a relatively harmless change to have.

After applying these patches (plus one [backport] which should be [backported
upstream][llvm-back]) I believe we should have all tests working on all
platforms in our current test suite. I'm like 99% sure that we'll need some more
backports as issues are reported for LLVM 6 when this propagates through
nightlies, but that's sort of just par for the course nowadays!

In any case though some extra scrutiny of the patches here would definitely be
welcome, along with scrutiny of the "missing patches" like a [change to pass
manager order](rust-lang/llvm@27174447533), [another change to pass manager
order](rust-lang/llvm@c782febb7b9), some [compile fixes for
sparc](rust-lang/llvm@1a83de63c42), and some [fixes for
solaris](rust-lang/llvm@c2bfe0abb).

[patches4]: rust-lang/llvm@5401fdf...rust-llvm-release-4-0-1
[backport]: rust-lang/llvm@5c54c252db
[llvm-back]: https://bugs.llvm.org/show_bug.cgi?id=36114
[upstream bug]: https://bugs.llvm.org/show_bug.cgi?id=36096

---

The update to LLVM 6 is desirable for a number of reasons, notably:

* This'll allow us to keep up with the upstream wasm backend, picking up new
  features as they start landing.
* Upstream LLVM has fixed a number of SIMD-related compilation errors,
  especially around AVX-512 and such.
* There's a few assorted known bugs which are fixed in LLVM 5 and aren't fixed
  in the LLVM 4 branch we're using.
* Overall it's not a great idea to stagnate with our codegen backend!

This update is mostly powered by #47730 which is allowing us to update LLVM
*independent* of the version of LLVM that Emscripten is locked to. This means
that when compiling code for Emscripten we'll still be using the old LLVM 4
backend, but when compiling code for any other target we'll be using the new
LLVM 6 target. Once Emscripten updates we may no longer need this distinction,
but we're not sure when that will happen!

Closes #43370
Closes #43418
Closes #47015
Closes #47683
Closes rust-lang/stdarch#157
Closes rust-lang-nursery/rust-wasm#3
alexcrichton added a commit to alexcrichton/rust that referenced this issue Feb 9, 2018
The following submodules have been updated for a new version of LLVM:

- `src/llvm`
- `src/libcompiler_builtins` - transitively contains compiler-rt
- `src/dlmalloc`

This also updates the docker container for dist-i686-freebsd as the old 16.04
container is no longer capable of building LLVM. The
compiler-rt/compiler-builtins and dlmalloc updates are pretty routine without
much interesting happening, but the LLVM update here is of particular note.
Unlike previous updates I haven't cherry-picked all existing patches we had on
top of our LLVM branch as we have a [huge amount][patches4] and have at this
point forgotten what most of them are for. Instead I started from the current
`release_60` branch in LLVM and only applied patches that were necessary to get
our tests working and building.

The current set of custom rustc-specific patches included in this LLVM update are:

* rust-lang/llvm@1187443 - this is how we actually implement
  `cfg(target_feature)` for now and continues to not be upstreamed. While a
  hazard for SIMD stabilization this commit is otherwise keeping the status
  quo of a small rustc-specific feature.
* rust-lang/llvm@013f2ec - this is a rustc-specific optimization that we haven't
  upstreamed, notably teaching LLVM about our allocation-related routines (which
  aren't malloc/free). Once we stabilize the global allocator routines we will
  likely want to upstream this patch, but for now it seems reasonable to keep it
  on our fork.
* rust-lang/llvm@a65bbfd - I found this necessary to fix compilation of LLVM in
  our 32-bit linux container. I'm not really sure why it's necessary but my
  guess is that it's because of the absolutely ancient glibc that we're using.
  In any case it's only updating pieces we're not actually using in LLVM so I'm
  hoping it'll turn out alright. This doesn't seem like something we'll want to
  upstream.c
* rust-lang/llvm@77ab1f0 - this is what's actually enabling LLVM to build in our
  i686-freebsd container, I'm not really sure what's going on but we for sure
  probably don't want to upstream this and otherwise it seems not too bad for
  now at least.
* rust-lang/llvm@9eb9267 - we currently suffer on MSVC from an [upstream bug]
  which although diagnosed to a particular revision isn't currently fixed
  upstream (and the bug itself doesn't seem too active). This commit is a
  partial revert of the suspected cause of this regression (found via a
  bisection). I'm sort of hoping that this eventually gets fixed upstream with a
  similar fix (which we can replace in our branch), but for now I'm also hoping
  it's a relatively harmless change to have.

After applying these patches (plus one [backport] which should be [backported
upstream][llvm-back]) I believe we should have all tests working on all
platforms in our current test suite. I'm like 99% sure that we'll need some more
backports as issues are reported for LLVM 6 when this propagates through
nightlies, but that's sort of just par for the course nowadays!

In any case though some extra scrutiny of the patches here would definitely be
welcome, along with scrutiny of the "missing patches" like a [change to pass
manager order](rust-lang/llvm@27174447533), [another change to pass manager
order](rust-lang/llvm@c782febb7b9), some [compile fixes for
sparc](rust-lang/llvm@1a83de63c42), and some [fixes for
solaris](rust-lang/llvm@c2bfe0abb).

[patches4]: rust-lang/llvm@5401fdf...rust-llvm-release-4-0-1
[backport]: rust-lang/llvm@5c54c252db
[llvm-back]: https://bugs.llvm.org/show_bug.cgi?id=36114
[upstream bug]: https://bugs.llvm.org/show_bug.cgi?id=36096

---

The update to LLVM 6 is desirable for a number of reasons, notably:

* This'll allow us to keep up with the upstream wasm backend, picking up new
  features as they start landing.
* Upstream LLVM has fixed a number of SIMD-related compilation errors,
  especially around AVX-512 and such.
* There's a few assorted known bugs which are fixed in LLVM 5 and aren't fixed
  in the LLVM 4 branch we're using.
* Overall it's not a great idea to stagnate with our codegen backend!

This update is mostly powered by rust-lang#47730 which is allowing us to update LLVM
*independent* of the version of LLVM that Emscripten is locked to. This means
that when compiling code for Emscripten we'll still be using the old LLVM 4
backend, but when compiling code for any other target we'll be using the new
LLVM 6 target. Once Emscripten updates we may no longer need this distinction,
but we're not sure when that will happen!

Closes rust-lang#43370
Closes rust-lang#43418
Closes rust-lang#47015
Closes rust-lang#47683
Closes rust-lang/stdarch#157
Closes rust-lang-nursery/rust-wasm#3
bors added a commit that referenced this issue Feb 9, 2018
rustc: Upgrade to LLVM 6

The following submodules have been updated for a new version of LLVM:

- `src/llvm`
- `src/libcompiler_builtins` - transitively contains compiler-rt
- `src/dlmalloc`

This also updates the docker container for dist-i686-freebsd as the old 16.04
container is no longer capable of building LLVM. The
compiler-rt/compiler-builtins and dlmalloc updates are pretty routine without
much interesting happening, but the LLVM update here is of particular note.
Unlike previous updates I haven't cherry-picked all existing patches we had on
top of our LLVM branch as we have a [huge amount][patches4] and have at this
point forgotten what most of them are for. Instead I started from the current
`release_60` branch in LLVM and only applied patches that were necessary to get
our tests working and building.

The [current set of custom rustc-specific patches](rust-lang/llvm@f128612...rust-llvm-release-6-0-0) included in this LLVM update are:

* rust-lang/llvm@1187443 - this is how we actually implement
  `cfg(target_feature)` for now and continues to not be upstreamed. While a
  hazard for SIMD stabilization this commit is otherwise keeping the status
  quo of a small rustc-specific feature.
* rust-lang/llvm@013f2ec - this is a rustc-specific optimization that we haven't
  upstreamed, notably teaching LLVM about our allocation-related routines (which
  aren't malloc/free). Once we stabilize the global allocator routines we will
  likely want to upstream this patch, but for now it seems reasonable to keep it
  on our fork.
* rust-lang/llvm@a65bbfd - I found this necessary to fix compilation of LLVM in
  our 32-bit linux container. I'm not really sure why it's necessary but my
  guess is that it's because of the absolutely ancient glibc that we're using.
  In any case it's only updating pieces we're not actually using in LLVM so I'm
  hoping it'll turn out alright. This doesn't seem like something we'll want to
  upstream.c
* rust-lang/llvm@77ab1f0 - this is what's actually enabling LLVM to build in our
  i686-freebsd container, I'm not really sure what's going on but we for sure
  probably don't want to upstream this and otherwise it seems not too bad for
  now at least.
* rust-lang/llvm@9eb9267 - we currently suffer on MSVC from an [upstream bug]
  which although diagnosed to a particular revision isn't currently fixed
  upstream (and the bug itself doesn't seem too active). This commit is a
  partial revert of the suspected cause of this regression (found via a
  bisection). I'm sort of hoping that this eventually gets fixed upstream with a
  similar fix (which we can replace in our branch), but for now I'm also hoping
  it's a relatively harmless change to have.

After applying these patches (plus one [backport] which should be [backported
upstream][llvm-back]) I believe we should have all tests working on all
platforms in our current test suite. I'm like 99% sure that we'll need some more
backports as issues are reported for LLVM 6 when this propagates through
nightlies, but that's sort of just par for the course nowadays!

In any case though some extra scrutiny of the patches here would definitely be
welcome, along with scrutiny of the "missing patches" like a [change to pass
manager order](rust-lang/llvm@2717444), [another change to pass manager
order](rust-lang/llvm@c782feb), some [compile fixes for
sparc](rust-lang/llvm@1a83de6), and some [fixes for
solaris](rust-lang/llvm@c2bfe0a).

[patches4]: rust-lang/llvm@5401fdf...rust-llvm-release-4-0-1
[backport]: rust-lang/llvm@5c54c25
[llvm-back]: https://bugs.llvm.org/show_bug.cgi?id=36114
[upstream bug]: https://bugs.llvm.org/show_bug.cgi?id=36096

---

The update to LLVM 6 is desirable for a number of reasons, notably:

* This'll allow us to keep up with the upstream wasm backend, picking up new
  features as they start landing.
* Upstream LLVM has fixed a number of SIMD-related compilation errors,
  especially around AVX-512 and such.
* There's a few assorted known bugs which are fixed in LLVM 5 and aren't fixed
  in the LLVM 4 branch we're using.
* Overall it's not a great idea to stagnate with our codegen backend!

This update is mostly powered by #47730 which is allowing us to update LLVM
*independent* of the version of LLVM that Emscripten is locked to. This means
that when compiling code for Emscripten we'll still be using the old LLVM 4
backend, but when compiling code for any other target we'll be using the new
LLVM 6 target. Once Emscripten updates we may no longer need this distinction,
but we're not sure when that will happen!

Closes #43370
Closes #43418
Closes #47015
Closes #47683
Closes rust-lang/stdarch#157
Closes rust-lang-nursery/rust-wasm#3
alexcrichton added a commit to alexcrichton/rust that referenced this issue Feb 10, 2018
The following submodules have been updated for a new version of LLVM:

- `src/llvm`
- `src/libcompiler_builtins` - transitively contains compiler-rt
- `src/dlmalloc`

This also updates the docker container for dist-i686-freebsd as the old 16.04
container is no longer capable of building LLVM. The
compiler-rt/compiler-builtins and dlmalloc updates are pretty routine without
much interesting happening, but the LLVM update here is of particular note.
Unlike previous updates I haven't cherry-picked all existing patches we had on
top of our LLVM branch as we have a [huge amount][patches4] and have at this
point forgotten what most of them are for. Instead I started from the current
`release_60` branch in LLVM and only applied patches that were necessary to get
our tests working and building.

The current set of custom rustc-specific patches included in this LLVM update are:

* rust-lang/llvm@1187443 - this is how we actually implement
  `cfg(target_feature)` for now and continues to not be upstreamed. While a
  hazard for SIMD stabilization this commit is otherwise keeping the status
  quo of a small rustc-specific feature.
* rust-lang/llvm@013f2ec - this is a rustc-specific optimization that we haven't
  upstreamed, notably teaching LLVM about our allocation-related routines (which
  aren't malloc/free). Once we stabilize the global allocator routines we will
  likely want to upstream this patch, but for now it seems reasonable to keep it
  on our fork.
* rust-lang/llvm@a65bbfd - I found this necessary to fix compilation of LLVM in
  our 32-bit linux container. I'm not really sure why it's necessary but my
  guess is that it's because of the absolutely ancient glibc that we're using.
  In any case it's only updating pieces we're not actually using in LLVM so I'm
  hoping it'll turn out alright. This doesn't seem like something we'll want to
  upstream.c
* rust-lang/llvm@77ab1f0 - this is what's actually enabling LLVM to build in our
  i686-freebsd container, I'm not really sure what's going on but we for sure
  probably don't want to upstream this and otherwise it seems not too bad for
  now at least.
* rust-lang/llvm@9eb9267 - we currently suffer on MSVC from an [upstream bug]
  which although diagnosed to a particular revision isn't currently fixed
  upstream (and the bug itself doesn't seem too active). This commit is a
  partial revert of the suspected cause of this regression (found via a
  bisection). I'm sort of hoping that this eventually gets fixed upstream with a
  similar fix (which we can replace in our branch), but for now I'm also hoping
  it's a relatively harmless change to have.

After applying these patches (plus one [backport] which should be [backported
upstream][llvm-back]) I believe we should have all tests working on all
platforms in our current test suite. I'm like 99% sure that we'll need some more
backports as issues are reported for LLVM 6 when this propagates through
nightlies, but that's sort of just par for the course nowadays!

In any case though some extra scrutiny of the patches here would definitely be
welcome, along with scrutiny of the "missing patches" like a [change to pass
manager order](rust-lang/llvm@27174447533), [another change to pass manager
order](rust-lang/llvm@c782febb7b9), some [compile fixes for
sparc](rust-lang/llvm@1a83de63c42), and some [fixes for
solaris](rust-lang/llvm@c2bfe0abb).

[patches4]: rust-lang/llvm@5401fdf...rust-llvm-release-4-0-1
[backport]: rust-lang/llvm@5c54c252db
[llvm-back]: https://bugs.llvm.org/show_bug.cgi?id=36114
[upstream bug]: https://bugs.llvm.org/show_bug.cgi?id=36096

---

The update to LLVM 6 is desirable for a number of reasons, notably:

* This'll allow us to keep up with the upstream wasm backend, picking up new
  features as they start landing.
* Upstream LLVM has fixed a number of SIMD-related compilation errors,
  especially around AVX-512 and such.
* There's a few assorted known bugs which are fixed in LLVM 5 and aren't fixed
  in the LLVM 4 branch we're using.
* Overall it's not a great idea to stagnate with our codegen backend!

This update is mostly powered by rust-lang#47730 which is allowing us to update LLVM
*independent* of the version of LLVM that Emscripten is locked to. This means
that when compiling code for Emscripten we'll still be using the old LLVM 4
backend, but when compiling code for any other target we'll be using the new
LLVM 6 target. Once Emscripten updates we may no longer need this distinction,
but we're not sure when that will happen!

Closes rust-lang#43370
Closes rust-lang#43418
Closes rust-lang#47015
Closes rust-lang#47683
Closes rust-lang/stdarch#157
Closes rust-lang-nursery/rust-wasm#3
bors added a commit that referenced this issue Feb 10, 2018
rustc: Upgrade to LLVM 6

The following submodules have been updated for a new version of LLVM:

- `src/llvm`
- `src/libcompiler_builtins` - transitively contains compiler-rt
- `src/dlmalloc`

This also updates the docker container for dist-i686-freebsd as the old 16.04
container is no longer capable of building LLVM. The
compiler-rt/compiler-builtins and dlmalloc updates are pretty routine without
much interesting happening, but the LLVM update here is of particular note.
Unlike previous updates I haven't cherry-picked all existing patches we had on
top of our LLVM branch as we have a [huge amount][patches4] and have at this
point forgotten what most of them are for. Instead I started from the current
`release_60` branch in LLVM and only applied patches that were necessary to get
our tests working and building.

The [current set of custom rustc-specific patches](rust-lang/llvm@f128612...rust-llvm-release-6-0-0) included in this LLVM update are:

* rust-lang/llvm@1187443 - this is how we actually implement
  `cfg(target_feature)` for now and continues to not be upstreamed. While a
  hazard for SIMD stabilization this commit is otherwise keeping the status
  quo of a small rustc-specific feature.
* rust-lang/llvm@013f2ec - this is a rustc-specific optimization that we haven't
  upstreamed, notably teaching LLVM about our allocation-related routines (which
  aren't malloc/free). Once we stabilize the global allocator routines we will
  likely want to upstream this patch, but for now it seems reasonable to keep it
  on our fork.
* rust-lang/llvm@a65bbfd - I found this necessary to fix compilation of LLVM in
  our 32-bit linux container. I'm not really sure why it's necessary but my
  guess is that it's because of the absolutely ancient glibc that we're using.
  In any case it's only updating pieces we're not actually using in LLVM so I'm
  hoping it'll turn out alright. This doesn't seem like something we'll want to
  upstream.c
* rust-lang/llvm@77ab1f0 - this is what's actually enabling LLVM to build in our
  i686-freebsd container, I'm not really sure what's going on but we for sure
  probably don't want to upstream this and otherwise it seems not too bad for
  now at least.
* rust-lang/llvm@9eb9267 - we currently suffer on MSVC from an [upstream bug]
  which although diagnosed to a particular revision isn't currently fixed
  upstream (and the bug itself doesn't seem too active). This commit is a
  partial revert of the suspected cause of this regression (found via a
  bisection). I'm sort of hoping that this eventually gets fixed upstream with a
  similar fix (which we can replace in our branch), but for now I'm also hoping
  it's a relatively harmless change to have.

After applying these patches (plus one [backport] which should be [backported
upstream][llvm-back]) I believe we should have all tests working on all
platforms in our current test suite. I'm like 99% sure that we'll need some more
backports as issues are reported for LLVM 6 when this propagates through
nightlies, but that's sort of just par for the course nowadays!

In any case though some extra scrutiny of the patches here would definitely be
welcome, along with scrutiny of the "missing patches" like a [change to pass
manager order](rust-lang/llvm@2717444), [another change to pass manager
order](rust-lang/llvm@c782feb), some [compile fixes for
sparc](rust-lang/llvm@1a83de6), and some [fixes for
solaris](rust-lang/llvm@c2bfe0a).

[patches4]: rust-lang/llvm@5401fdf...rust-llvm-release-4-0-1
[backport]: rust-lang/llvm@5c54c25
[llvm-back]: https://bugs.llvm.org/show_bug.cgi?id=36114
[upstream bug]: https://bugs.llvm.org/show_bug.cgi?id=36096

---

The update to LLVM 6 is desirable for a number of reasons, notably:

* This'll allow us to keep up with the upstream wasm backend, picking up new
  features as they start landing.
* Upstream LLVM has fixed a number of SIMD-related compilation errors,
  especially around AVX-512 and such.
* There's a few assorted known bugs which are fixed in LLVM 5 and aren't fixed
  in the LLVM 4 branch we're using.
* Overall it's not a great idea to stagnate with our codegen backend!

This update is mostly powered by #47730 which is allowing us to update LLVM
*independent* of the version of LLVM that Emscripten is locked to. This means
that when compiling code for Emscripten we'll still be using the old LLVM 4
backend, but when compiling code for any other target we'll be using the new
LLVM 6 target. Once Emscripten updates we may no longer need this distinction,
but we're not sure when that will happen!

Closes #43370
Closes #43418
Closes #47015
Closes #47683
Closes rust-lang/stdarch#157
Closes rust-lang-nursery/rust-wasm#3
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
A-LLVM Area: Code generation parts specific to LLVM. Both correctness bugs and optimization-related issues. I-slow Issue: Problems and improvements with respect to performance of generated code.
Projects
None yet
Development

Successfully merging a pull request may close this issue.

3 participants