test #7

NHDaly · 2023-05-15T18:17:27Z

No description provided.

(cherry picked from commit b5bbb9f)

The event might have triggered on another thread before we observed it here, or it might have gotten finalized before it got triggered. Either outcome can result in a lost event. (I observed the later situation occurring locally during the Dates test once). (cherry picked from commit 8cc00ff)

…JuliaLang#44972) * [REPL] remove erroneous space in test Introduced by JuliaLang#33805 (74f2de1). * Revert "Temporarily move the `REPL` test suite to node 1, to buy us time until we fix the underlying bugs (JuliaLang#44961)" This reverts commit 322fd70. (cherry picked from commit fbec395)

(cherry picked from commit c589e0d)

…ng#44866. (JuliaLang#45026) (cherry picked from commit d29d2d6)

If we load libjulia-codegen directly, e.g. from opt or llc, we don't call julia_init but init_llvm. Co-authored-by: Keno Fischer <[email protected]> (cherry picked from commit f562ad0)

The change in JuliaLang#44793 changed the signature of the donotdelete intrinsic, but didn't change the corresponding attributes, leaving a `nonnull` attribute on a `void` return, which fails the verifier.

Follows up JuliaLang#45017. We can relax the condition in `add_call_backedges!` for backedge optimization by ignoring the `nonoverlayed` property when the `AbstractInterpreter` doesn't use overlayed method table at all, since the property will never be tainted anyway even if we add a new method later.

Discovered via JET. This also causes `make_aliases` to always return a `Vector{SimpleVector}` as its first argument. Co-authored-by: Shuhei Kadowaki <[email protected]> Co-authored-by: Jameson Nash <[email protected]> Co-authored-by: Kristoffer Carlsson <[email protected]> (cherry picked from commit cc60657)

This prior conditional was a fixed constant branch, so this seems more like the intent. (cherry picked from commit ac51add)

* Add cover page * use tikz logo * Add "Release Notes" to devdocs Co-authored-by: Viral B. Shah <[email protected]> (cherry picked from commit 8a3b4ab)

(cherry picked from commit d8346cc)

(cherry picked from commit 45abec4)

(cherry picked from commit 5e74df5)

(cherry picked from commit b9d8280)

(cherry picked from commit b4eb88a)

…uliaLang#45176) (cherry picked from commit 862018b)

…uliaLang#45149) For quite some time I've been observing Revise randomly not pick up changes to my packages. Usually I just write that off to Revise getting into a bad state and restarting Julia fixes it. Today, I got very annoyed by this again and decided to finally track it down. What turns out to happen here is the packages in question are those which: A. Are being precompiled on load B. Use Requires C. Have an `@require`'d-dependency already loaded D. The change to be revised is made in a file not loaded via Requires.jl. In this case the `__init__` of the package triggers Requires, which in turn calls back to Revise, which tries to start watching the package. However, on the `compilecache` path (but not on the path where we just find a pre-existing cache file), we used to not set the .cachepath property of `pkgorigins` until after the `__init__` callbacks run, causing Revise not to be able to see those files. Infuriatingly, restarting julia fixes this because it just loads the .ji file that was just compiled. Fix this by unifying the point at which the .cachepath is set, always setting it just prior to the module initialization callback. (cherry picked from commit 58ab4c7)

If multiple processes attempt to clean up cache files at the same time, a race condition can result, e.g. https://buildkite.com/clima/climaatmos-ci/builds/812#6a961e99-e2a1-488b-a116-2a45dee26d38/102-104 (cherry picked from commit d4acead)

(cherry picked from commit b6b0874)

…Lang#45199) fix JuliaLang#45171 (cherry picked from commit 5c557b2)

… keywords (JuliaLang#45198) (cherry picked from commit c62ea26)

(cherry picked from commit 6be86a3)

The optimization pass often uses values for phi values (and thus by extension, also for pi, phic and upsilon values) that are invalid. We make sure that these have a null pointer, so that we can detect that case at runtime (at the cost of slightly worse code generation for them), but it means we need to be very careful to check for that. This is identical to JuliaLang#39747, which added the equivalent code to the other side of the conditional there, but missed some additional relevant, but rare, cases that are observed to be possible. The `emit_isa_and_defined` is derived from the LLVM name for this operation: `isa_and_nonnull<T>`. Secondly, we also optimize `emit_unionmove` to change a bad IR case to a better IR form. Fix JuliaLang#44501 (cherry picked from commit 72b80e2)

There's no performance change, if the `indices`s of `Rdest` and `Rsrc` are all `NTuple{N,<:AbstractUnitRange}`. (cherry picked from commit a91be39)

(cherry picked from commit eed2dba)

Fixes JuliaLang#44499 (cherry picked from commit f18324c)

(cherry picked from commit 7f84d46)

Pass the types to the allocator functions. ------- Before this PR, we were missing the types for allocations in two cases: 1. allocations from codegen 2. allocations in `gc_managed_realloc_` The second one is easy: those are always used for buffers, right? For the first one: we extend the allocation functions called from codegen, to take the type as a parameter, and set the tag there. I kept the old interfaces around, since I think that they cannot be removed due to supporting legacy code? ------ An example of the generated code: ```julia %ptls_field6 = getelementptr inbounds {}**, {}*** %4, i64 2 %13 = bitcast {}*** %ptls_field6 to i8** %ptls_load78 = load i8*, i8** %13, align 8 %box = call noalias nonnull dereferenceable(32) {}* @ijl_gc_pool_alloc_typed(i8* %ptls_load78, i32 1184, i32 32, i64 4366152144) #7 ``` Fixes JuliaLang#43688. Fixes JuliaLang#45268. Co-authored-by: Valentin Churavy <[email protected]>

@noinline

E.g. this allows `finalizer` inlining in the following case: ```julia mutable struct ForeignBuffer{T} const ptr::Ptr{T} end const foreign_buffer_finalized = Ref(false) function foreign_alloc(::Type{T}, length) where T ptr = Libc.malloc(sizeof(T) * length) ptr = Base.unsafe_convert(Ptr{T}, ptr) obj = ForeignBuffer{T}(ptr) return finalizer(obj) do obj Base.@assume_effects :notaskstate :nothrow foreign_buffer_finalized[] = true Libc.free(obj.ptr) end end function f_EA_finalizer(N::Int) workspace = foreign_alloc(Float64, N) GC.@preserve workspace begin (;ptr) = workspace Base.@assume_effects :nothrow @noinline println(devnull, "ptr = ", ptr) end end ``` ```julia julia> @code_typed f_EA_finalizer(42) CodeInfo( 1 ── %1 = Base.mul_int(8, N)::Int64 │ %2 = Core.lshr_int(%1, 63)::Int64 │ %3 = Core.trunc_int(Core.UInt8, %2)::UInt8 │ %4 = Core.eq_int(%3, 0x01)::Bool └─── goto #3 if not %4 2 ── invoke Core.throw_inexacterror(:convert::Symbol, UInt64::Type, %1::Int64)::Union{} └─── unreachable 3 ── goto #4 4 ── %9 = Core.bitcast(Core.UInt64, %1)::UInt64 └─── goto #5 5 ── goto #6 6 ── goto #7 7 ── goto #8 8 ── %14 = $(Expr(:foreigncall, :(:malloc), Ptr{Nothing}, svec(UInt64), 0, :(:ccall), :(%9), :(%9)))::Ptr{Nothing} └─── goto #9 9 ── %16 = Base.bitcast(Ptr{Float64}, %14)::Ptr{Float64} │ %17 = %new(ForeignBuffer{Float64}, %16)::ForeignBuffer{Float64} └─── goto #10 10 ─ %19 = $(Expr(:gc_preserve_begin, :(%17))) │ %20 = Base.getfield(%17, :ptr)::Ptr{Float64} │ invoke Main.println(Main.devnull::Base.DevNull, "ptr = "::String, %20::Ptr{Float64})::Nothing │ $(Expr(:gc_preserve_end, :(%19))) │ %23 = Main.foreign_buffer_finalized::Base.RefValue{Bool} │ Base.setfield!(%23, :x, true)::Bool │ %25 = Base.getfield(%17, :ptr)::Ptr{Float64} │ %26 = Base.bitcast(Ptr{Nothing}, %25)::Ptr{Nothing} │ $(Expr(:foreigncall, :(:free), Nothing, svec(Ptr{Nothing}), 0, :(:ccall), :(%26), :(%25)))::Nothing └─── return nothing ) => Nothing ``` However, this is still a WIP. Before merging, I want to improve EA's precision a bit and at least fix the test case that is currently marked as `broken`. I also need to check its impact on compiler performance. Additionally, I believe this feature is not yet practical. In particular, there is still significant room for improvement in the following areas: - EA's interprocedural capabilities: currently EA is performed ad-hoc for limited frames because of latency reasons, which significantly reduces its precision in the presence of interprocedural calls. - Relaxing the `:nothrow` check for finalizer inlining: the current algorithm requires `:nothrow`-ness on all paths from the allocation of the mutable struct to its last use, which is not practical for real-world cases. Even when `:nothrow` cannot be guaranteed, auxiliary optimizations such as inserting a `finalize` call after the last use might still be possible (JuliaLang#55990).

Will Kimmerer and others added 30 commits April 19, 2022 14:00

distributed docs (JuliaLang#44940)

a897847

(cherry picked from commit b5bbb9f)

Fix embedding with MSVC (JuliaLang#44976)

fe0faad

(cherry picked from commit c589e0d)

Update Example header in admonition to render PDF better, see JuliaLa…

26a25e2

…ng#44866. (JuliaLang#45026) (cherry picked from commit d29d2d6)

Initialize jl_page_size when not calling julia_init. (JuliaLang#44699)

cc5743c

If we load libjulia-codegen directly, e.g. from opt or llc, we don't call julia_init but init_llvm. Co-authored-by: Keno Fischer <[email protected]> (cherry picked from commit f562ad0)

Change type of donotdelete intrinsics

f13039e

Fix attributes for donotdelete intrinsic

5073c4e

The change in JuliaLang#44793 changed the signature of the donotdelete intrinsic, but didn't change the corresponding attributes, leaving a `nonnull` attribute on a `void` return, which fails the verifier.

gc-ext: only sweep unmarked objects (JuliaLang#45035)

b1e1c52

This prior conditional was a fixed constant branch, so this seems more like the intent. (cherry picked from commit ac51add)

[doc] Add a Cover page to PDF doc (JuliaLang#45034)

68ddba3

* Add cover page * use tikz logo * Add "Release Notes" to devdocs Co-authored-by: Viral B. Shah <[email protected]> (cherry picked from commit 8a3b4ab)

correct man-custom-indices ref in devdocs (JuliaLang#45071)

eadbeec

(cherry picked from commit d8346cc)

Correct typo in devdocs gc-sa (JuliaLang#45072)

c2ce2a1

(cherry picked from commit 45abec4)

[deps] Fix compillation of LLVM with system zlib (JuliaLang#45119)

515a5ce

(cherry picked from commit 5e74df5)

Create a copy while evaluating eigvals(::Diagonal) (JuliaLang#45048)

f5e292a

(cherry picked from commit b9d8280)

Fix bug in pinv (JuliaLang#45009)

6d4b8d0

(cherry picked from commit b4eb88a)

[deps] Remove Python workaround in LLVM build, no longer necessary (J…

f89f049

…uliaLang#45176) (cherry picked from commit 862018b)

Consider additional default ssh keys for LibGit2. (JuliaLang#44767)

fb45a1a

(cherry picked from commit b6b0874)

syntax: keyword sorter location should just be definition line (Julia…

a380056

…Lang#45199) fix JuliaLang#45171 (cherry picked from commit 5c557b2)

fix JuliaLang#45162, function arg not specialized if only called with…

51f8ac2

… keywords (JuliaLang#45198) (cherry picked from commit c62ea26)

Fix a concurrency bug in iterate(::Dict) (JuliaLang#44534)

b3dde45

(cherry picked from commit 6be86a3)

Bump Documenter to get the new pdf improvements for the manual.

293b513

Use CartesianIndices(Rsrc) as the shared iterator. (JuliaLang#45289)

834c8e6

There's no performance change, if the `indices`s of `Rdest` and `Rsrc` are all `NTuple{N,<:AbstractUnitRange}`. (cherry picked from commit a91be39)

Typo fix. (JuliaLang#45333)

3dfa74d

(cherry picked from commit eed2dba)

Resurrect libunwind patches (JuliaLang#45189)

43ba8ae

Fixes JuliaLang#44499 (cherry picked from commit f18324c)

[deps] Use newer config.sub when building nghttp2 (JuliaLang#45346)

1f9140f

(cherry picked from commit 7f84d46)

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

test #7

test #7

NHDaly commented May 15, 2023

test #7

test #7

Conversation

NHDaly commented May 15, 2023