[Git][ghc/ghc][wip/backports-9.8] 30 commits: Add -finfo-table-map-with-fallback -finfo-table-map-with-stack
Zubin (@wz1000)
gitlab at gitlab.haskell.org
Thu Feb 8 12:09:07 UTC 2024
Zubin pushed to branch wip/backports-9.8 at Glasgow Haskell Compiler / GHC
Commits:
16c42d04 by Finley McIlwaine at 2024-02-08T17:38:50+05:30
Add -finfo-table-map-with-fallback -finfo-table-map-with-stack
The -fno-info-table-map-with-stack flag omits STACK info tables from the info
table map, and the -fno-info-table-map-with-fallback flag omits info tables
with defaulted source locations from the map. In a test on the Agda codebase
the build results were about 7% smaller when both of those types of tables
were omitted.
Adds a test that verifies that passing each combination of these flags
results in the correct output for -dipe-stats, which is disabled for the js
backend since profiling is not implemented.
This commit also refactors a lot of the logic around extracting info tables
from the Cmm results and building the info table map.
This commit also fixes some issues in the users guide rst source to fix
warnings that were noticed while debugging the documentation for these flags.
Fixes #23702
(cherry picked from commit 261c4acbfdaf5babfc57ab0cef211edb66153fb1)
- - - - -
aeb82fab by Finley McIlwaine at 2024-02-08T17:38:50+05:30
Refactor estimation of stack info table provenance
This commit greatly refactors the way we compute estimated provenance for stack
info tables. Previously, this process was done using an entirely separate traversal
of the whole Cmm code stream to build the map from info tables to source locations.
The separate traversal is now fused with the Cmm code generation pipeline in
GHC.Driver.Main.
This results in very significant code generation speed ups when -finfo-table-map is
enabled. In testing, this patch reduces code generation times by almost 30% with
-finfo-table-map and -O0, and 60% with -finfo-table-map and -O1 or -O2 .
Fixes #23103
(cherry picked from commit d99c816f7b5727a3f344960e02a1932187ea093f)
- - - - -
3f8020d8 by Finley McIlwaine at 2024-02-08T17:38:50+05:30
Add a test checking overhead of -finfo-table-map
We want to make sure we don't end up with poor codegen performance resulting from
-finfo-table-map again as in #23103. This test adds a performance test tracking
total allocations while compiling ExactPrint with -finfo-table-map.
(cherry picked from commit d3e0124c1157a4a423d86a1dc1d7e82c6d32ef06)
- - - - -
31a292ba by Matthew Pickering at 2024-02-08T17:38:51+05:30
Add aarch64-deb11 bindist
This adds a debian 11 release job for aarch64.
Fixes #22005
(cherry picked from commit 02c87213e1215520d5496130a3082143f27035ae)
- - - - -
12006e08 by sheaf at 2024-02-08T17:38:51+05:30
Pass quantified tyvars in tcDefaultAssocDecl
This commit passes the correct set of quantified type variables written
by the user in associated type default declarations for validity
checking. This ensures that validity checking of associated type defaults
mirrors that of standalone type family instances.
Fixes #23768 (see testcase T23734 in subsequent commit)
(cherry picked from commit 5f826c180aa9b42ccb0af5763d87a48bee5e4083)
- - - - -
274104a1 by sheaf at 2024-02-08T17:38:51+05:30
Avoid panic in mkGADTVars
This commit avoids panicking in mkGADTVars when we encounter
a type variable as in #23784 that is bound by a user-written
forall but not actually used.
Fixes #23784
(cherry picked from commit aba184241a80e5abc804c4f22db91a16be215c11)
- - - - -
55bfa58e by sheaf at 2024-02-08T17:38:51+05:30
Adjust reporting of unused tyvars in data FamInsts
This commit adjusts the validity checking of data family
instances to improve the reporting of unused type variables.
See Note [Out of scope tvs in data family instances] in GHC.Tc.Validity.
The problem was that, in a situation such as
data family D :: Type
data instance forall (d :: Type). D = MkD
the RHS passed to 'checkFamPatBinders' would be the TyCon app
R:D d
which mentions the type variable 'd' quantified in the user-written
forall. Thus, when computing the set of unused type variables in
the RHS of the data family instance, we would find that 'd' is used,
and report a strange error message that would say that 'd' is not
bound on the LHS.
To fix this, we special-case the data-family instance case,
manually extracting all the type variables that appear in the
arguments of all the data constructores of the data family instance.
Fixes #23778
(cherry picked from commit a525a92a37cc202b6db09a827ba9df9eb8c9fa79)
- - - - -
9a6edec3 by Antoine Leblanc at 2024-02-08T17:38:51+05:30
Be more eager in TyCon boot validity checking
This commit performs boot-file consistency checking for TyCons into
checkValidTyCl. This ensures that we eagerly catch any mismatches,
which prevents the compiler from seeing these inconsistencies and
panicking as a result.
See Note [TyCon boot consistency checking] in GHC.Tc.TyCl.
Fixes #16127
(cherry picked from commit 1420b8cb8a7d6196eec80dc5293864c780379560)
- - - - -
87d78a5a by sheaf at 2024-02-08T17:38:51+05:30
Unused tyvars in FamInst: only report user tyvars
This commit changes how we perform some validity checking for
coercion axioms to mirror how we handle default declarations for
associated type families. This allows us to keep track of whether
type variables in type and data family instances were user-written
or not, in order to only report the user-written ones in
"unused type variable" error messages.
Consider for example:
{-# LANGUAGE PolyKinds #-}
type family F
type instance forall a. F = ()
In this case, we get two quantified type variables,
(k :: Type) and (a :: k); the second being user-written, but the first
is introduced by the typechecker. We should only report 'a' as being
unused, as the user has no idea what 'k' is.
Fixes #23734
(cherry picked from commit 28dd52eec98e50c711cd00df22f6ab9e054c8b75)
- - - - -
b7949e85 by sheaf at 2024-02-08T17:38:51+05:30
Validity: refactor treatment of data families
This commit refactors the reporting of unused type variables in type
and data family instances to be more principled. This avoids ad-hoc
logic in the treatment of data family instances.
(cherry picked from commit 1eed645c8b03b19a14cf58d9be5317cb81cbd30a)
- - - - -
63d6ed11 by Krzysztof Gogolewski at 2024-02-08T17:38:51+05:30
docs: fix ScopedTypeVariables example (#24101)
The previous example didn't compile.
Furthermore, it wasn't demonstrating the point properly.
I have changed it to an example which shows that 'a' in the signature
must be the same 'a' as in the instance head.
(cherry picked from commit 7a90020f167ab016cbfa95decafaa1a54a974bc6)
- - - - -
c06c13ee by PHO at 2024-02-08T17:38:51+05:30
Don't assume the current locale is *.UTF-8, set the encoding explicitly
primops.txt contains Unicode characters:
> LC_ALL=C ./genprimopcode --data-decl < ./primops.txt
> genprimopcode: <stdin>: hGetContents: invalid argument (cannot decode byte sequence starting from 226)
Hadrian must also avoid using readFile' to read primops.txt because it
tries to decode the file with a locale-specific encoding.
(cherry picked from commit 52c0fc691e6501e99a96693ec1fc02e3c93a4fbc)
- - - - -
3fec4c2e by Simon Peyton Jones at 2024-02-08T17:38:51+05:30
Add an extra check in kcCheckDeclHeader_sig
Fix #24083 by checking for a implicitly-scoped type variable that is not
actually bound. See Note [Disconnected type variables] in GHC.Tc.Gen.HsType
For some reason, on aarch64-darwin we saw a 2.8% decrease in compiler
allocations for MultiLayerModulesTH_Make; but 0.0% on other architectures.
Metric Decrease:
MultiLayerModulesTH_Make
(cherry picked from commit 6dbab1808bfbe484b3fb396aab1d105314f918d8)
- - - - -
83f78d73 by Simon Peyton Jones at 2024-02-08T17:38:51+05:30
Second fix to #24083
My earlier fix turns out to be too aggressive for data/type families
See wrinkle (DTV1) in Note [Disconnected type variables]
(cherry picked from commit 2776920e642544477a38d0ed9205d4f0b48a782e)
(cherry picked from commit 110efc9813ba5f68668ab7fc3f95b9febca52880)
- - - - -
9209706f by Matthew Pickering at 2024-02-08T17:38:51+05:30
libraries: Bump filepath to 1.4.200.1 and unix to 2.8.4.0
Updates filepath submodule
Updates unix submodule
Fixes #24240
(cherry picked from commit 36b9a38cc45a26865c4e45f4949e519a5dede76d)
- - - - -
4d158d9a by Matthew Pickering at 2024-02-08T17:38:51+05:30
Submodule linter: Allow references to tags
We modify the submodule linter so that if the bumped commit is a
specific tag then the commit is accepted.
Fixes #24241
(cherry picked from commit 91ff0971df64b04938d011fe1562320c5d90849a)
- - - - -
e5cfbf2a by Matthew Craven at 2024-02-08T17:38:51+05:30
Fix loopification in the presence of void arguments
This also removes Note [Void arguments in self-recursive tail calls],
which was just misleading. It's important to count void args both
in the function's arity and at the call site.
Fixes #24295.
(cherry picked from commit ae9cc1a84c9f470b77d98423400e6dfa95b2449b)
- - - - -
2e009a4a by Andreas Klebinger at 2024-02-08T17:38:51+05:30
Aarch64: Enable -mfma by default.
Fixes #24311
(cherry picked from commit 7e95f738620dc805868d198f980e1bdd53e27a2d)
- - - - -
4b26ce8d by Matthew Pickering at 2024-02-08T17:38:51+05:30
eventlog: Fix off-by-one error in postIPE
We were missing the extra_comma from the calculation of the size of the
payload of postIPE. This was causing assertion failures when the event
would overflow the buffer by one byte, as ensureRoomForVariable event
would report there was enough space for `n` bytes but then we would
write `n + 1` bytes into the buffer.
Fixes #24287
(cherry picked from commit 5776008c7a5581193c3e72e59451ad49abac9d81)
- - - - -
282fb895 by Jade at 2024-02-08T17:38:51+05:30
Enhance Documentation of functions exported by Data.Function
This patch aims to improve the documentation of functions exported
in Data.Function
Tracking: #17929
Fixes: #10065
(cherry picked from commit 1fa1c00c95325761a5aa914af53d71ba5e7072b3)
- - - - -
6222f9f9 by Jade at 2024-02-08T17:38:51+05:30
Improve documentation of hGetLine.
- Add explanation for whether a newline is returned
- Add examples
Fixes #14804
(cherry picked from commit ab47a43d64f6b7d4fc181645171c31ba2db1eebe)
- - - - -
db636e44 by sheaf at 2024-02-08T17:38:51+05:30
Fix FMA instruction on LLVM
We were emitting the wrong instructions for fused multiply-add
operations on LLVM:
- the instruction name is "llvm.fma.f32" or "llvm.fma.f64", not "fmadd"
- LLVM does not support other instructions such as "fmsub"; instead
we implement these by flipping signs of some arguments
- the instruction is an LLVM intrinsic, which requires handling it
like a normal function call instead of a machine instruction
Fixes #24223
(cherry picked from commit a40f4ab21bcc088e63892cd5e85edbec20d3fc69)
- - - - -
b5ea7063 by Hécate Moonlight at 2024-02-08T17:38:51+05:30
Clarification for newtype constructors when using `coerce`
(cherry picked from commit 699da01bbbf3e42c7d38b9cfe443dd0e8e256342)
- - - - -
fbb7e049 by Andreas Klebinger at 2024-02-08T17:38:51+05:30
Fix fma warning when using llvm on aarch64.
On aarch64 fma is always on so the +fma flag doesn't exist for that
target. Hence no need to try and pass +fma to llvm.
Fixes #24379
(cherry picked from commit 9294a08643b89509a0e0957cb73c186a39d4f3db)
- - - - -
cd1acedd by Patrick at 2024-02-08T17:38:51+05:30
Fix bug wrong span of nested_doc_comment #24378
close #24378
1. Update the start position of span in `nested_doc_comment` correctly.
and hence the spans of identifiers of haddoc can be computed correctly.
2. add test `HaddockSpanIssueT24378`.
(cherry picked from commit 8eeadfad3a0035f8c5b339782676ff23572e0e5e)
- - - - -
77b1c0f6 by sheaf at 2024-02-08T17:38:51+05:30
No shadowing warnings for NoFieldSelector fields
This commit ensures we don't emit shadowing warnings when a user
shadows a field defined with NoFieldSelectors.
Fixes #24381
(cherry picked from commit ced2e7312b692e3f5402e4db6cfec390653a6a06)
- - - - -
9b1b1a54 by Teo Camarasu at 2024-02-08T17:38:51+05:30
doc: Add -Dn flag to user guide
Resolves #24394
(cherry picked from commit 94ce031ddc84ee702c12a11793028ef21e65fa00)
- - - - -
8298a9f9 by Rodrigo Mesquita at 2024-02-08T17:38:51+05:30
Work around autotools setting C11 standard in CC/CXX
In autoconf >=2.70, C11 is set by default for $CC and $CXX via the
-std=...11 flag. In this patch, we split the "-std" flag out of the $CC
and $CXX variables, which we traditionally assume to be just the
executable name/path, and move it to $CFLAGS/$CXXFLAGS instead.
Fixes #24324
(cherry picked from commit cdddeb0f1280b40cc194028bbaef36e127175c4c)
- - - - -
a29e056c by Zubin Duggal at 2024-02-08T17:38:51+05:30
driver: Really don't lose track of nodes when we fail to resolve cycles
This fixes a bug in 8db8d2fd1c881032b1b360c032b6d9d072c11723, where we could lose
track of acyclic components at the start of an unresolved cycle. We now ensure we
never loose track of any of these components.
As T24275 demonstrates, a "cyclic" SCC might not really be a true SCC:
When viewed without boot files, we have a single SCC
```
[REC main:T24275B [main:T24275B {-# SOURCE #-},
main:T24275A {-# SOURCE #-}]
main:T24275A [main:T24275A {-# SOURCE #-}]]
```
But with boot files this turns into
```
[NONREC main:T24275B {-# SOURCE #-} [],
REC main:T24275B [main:T24275B {-# SOURCE #-},
main:T24275A {-# SOURCE #-}]
main:T24275A {-# SOURCE #-} [main:T24275B],
NONREC main:T24275A [main:T24275A {-# SOURCE #-}]]
```
Note that this is truly not an SCC, as no nodes are reachable from T24275B.hs-boot.
However, we treat this entire group as a single "SCC" because it seems so when we
analyse the graph without taking boot files into account.
Indeed, we must return a single ResolvedCycle element in the BuildPlan for this
as described in Note [Upsweep].
However, since after resolving this is not a true SCC anymore, `findCycle` fails
to find a cycle and we have a sub-optimal error message as a result.
To handle this, I extended `findCycle` to not assume its input is an SCC, and to
try harder to find cycles in its input.
Fixes #24275
(cherry picked from commit 532993c8160d960f848e7abd401774b6879e3ee8)
- - - - -
626a925d by Matthew Pickering at 2024-02-08T17:38:51+05:30
distrib/configure: Fix typo in CONF_GCC_LINKER_OPTS_STAGE2 variable
Instead we were setting CONF_GCC_LINK_OPTS_STAGE2 which meant that we
were missing passing `--target` when invoking the linker.
Fixes #24414
(cherry picked from commit d309f4e7e37f3795c6d6b150c407d5a9b332854e)
- - - - -
30 changed files:
- .gitlab/gen_ci.hs
- .gitlab/jobs.yaml
- compiler/GHC/Builtin/primops.txt.pp
- compiler/GHC/Cmm.hs
- compiler/GHC/CmmToLlvm/CodeGen.hs
- compiler/GHC/Core/Class.hs
- compiler/GHC/Core/Coercion/Axiom.hs
- compiler/GHC/Data/Graph/Directed.hs
- compiler/GHC/Driver/Config/StgToCmm.hs
- compiler/GHC/Driver/DynFlags.hs
- compiler/GHC/Driver/Flags.hs
- compiler/GHC/Driver/GenerateCgIPEStub.hs
- compiler/GHC/Driver/Main.hs
- compiler/GHC/Driver/Make.hs
- compiler/GHC/Driver/Pipeline/Execute.hs
- compiler/GHC/Driver/Session.hs
- compiler/GHC/IfaceToCore.hs
- compiler/GHC/Llvm/Ppr.hs
- compiler/GHC/Llvm/Syntax.hs
- compiler/GHC/Parser/Lexer.x
- compiler/GHC/Rename/Module.hs
- compiler/GHC/Rename/Utils.hs
- compiler/GHC/Runtime/Heap/Layout.hs
- compiler/GHC/Stg/Debug.hs
- compiler/GHC/StgToCmm/Bind.hs
- compiler/GHC/StgToCmm/Closure.hs
- compiler/GHC/StgToCmm/Config.hs
- compiler/GHC/StgToCmm/Expr.hs
- compiler/GHC/StgToCmm/Heap.hs
- compiler/GHC/StgToCmm/Monad.hs
The diff was not included because it is too large.
View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/993e2abbe47c2c78842f9c706e8ba0f367c2e0e7...626a925d6707a8f16ee798a70b8dad60afe33217
--
View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/993e2abbe47c2c78842f9c706e8ba0f367c2e0e7...626a925d6707a8f16ee798a70b8dad60afe33217
You're receiving this email because of your account on gitlab.haskell.org.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.haskell.org/pipermail/ghc-commits/attachments/20240208/65680d82/attachment-0001.html>
More information about the ghc-commits
mailing list