summaryrefslogtreecommitdiff
path: root/compiler/cmm
Commit message (Collapse)AuthorAgeFilesLines
...
* Update Trac ticket URLs to point to GitLabRyan Scott2019-03-153-3/+3
| | | | | This moves all URL references to Trac tickets to their corresponding GitLab counterparts.
* Rip out object splittingBen Gamari2019-03-051-3/+1
| | | | | | | | | | | | | | | The splitter is an evil Perl script that processes assembler code. Its job can be done better by the linker's --gc-sections flag. GHC passes this flag to the linker whenever -split-sections is passed on the command line. This is based on @DemiMarie's D2768. Fixes Trac #11315 Fixes Trac #9832 Fixes Trac #8964 Fixes Trac #8685 Fixes Trac #8629
* Fix warnings and fatal parsing errorsVladislav Zavialov2019-02-172-6/+3
|
* Cmm: Promote stack arguments to word sizePeter Trommler2019-02-161-7/+29
| | | | | | | | | | | | | | | | | Smaller than word size integers must be promoted to word size when passed on the stack. While on little endian systems we can get away with writing a small integer to a word size stack slot and read it as a word ignoring the upper bits, on big endian systems a small integer write ends up in the most significant bits and a word size read that ignores the upper bits delivers a random value. On little endian systems a smaller than word size write to the stack might be more efficient but that decision is system specific and should be done as an optimization in the respective backends. Fixes #16258
* PPC NCG: Promote integers to word size in C callsPeter Trommler2019-01-311-2/+6
| | | | Fixes #16222
* Use ByteString to represent Cmm string literals (#16198)Sylvain Henry2019-01-317-33/+23
| | | | Also used ByteString in some other relevant places
* Prepare source-tree for base-4.13 MFP bumpHerbert Valerio Riedel2019-01-181-0/+4
|
* PPC NCG: Remove Darwin supportPeter Trommler2019-01-011-7/+0
| | | | | | | Support for Mac OS X on PowerPC has been dropped by Apple years ago. We follow suit and remove PowerPC support for Darwin. Fixes #16106.
* Fix unused-import warningsDavid Eichmann2018-11-225-5/+3
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This patch fixes a fairly long-standing bug (dating back to 2015) in RdrName.bestImport, namely commit 9376249b6b78610db055a10d05f6592d6bbbea2f Author: Simon Peyton Jones <simonpj@microsoft.com> Date: Wed Oct 28 17:16:55 2015 +0000 Fix unused-import stuff in a better way In that patch got the sense of the comparison back to front, and thereby failed to implement the unused-import rules described in Note [Choosing the best import declaration] in RdrName This led to Trac #13064 and #15393 Fixing this bug revealed a bunch of unused imports in libraries; the ones in the GHC repo are part of this commit. The two important changes are * Fix the bug in bestImport * Modified the rules by adding (a) in Note [Choosing the best import declaration] in RdrName Reason: the previosu rules made Trac #5211 go bad again. And the new rule (a) makes sense to me. In unravalling this I also ended up doing a few other things * Refactor RnNames.ImportDeclUsage to use a [GlobalRdrElt] for the things that are used, rather than [AvailInfo]. This is simpler and more direct. * Rename greParentName to greParent_maybe, to follow GHC naming conventions * Delete dead code RdrName.greUsedRdrName Bumps a few submodules. Reviewers: hvr, goldfire, bgamari, simonmar, jrtc27 Subscribers: rwbarton, carter Differential Revision: https://phabricator.haskell.org/D5312
* UNREG: PprC: Add support for adjacent floatsJames Clarke2018-11-221-1/+23
| | | | | | | | | | | | | | | | When two 32-bit floats are adjacent for a 64-bit target, there is no padding between them to force alignment, so we must combine their bit representations into a single word. Reviewers: bgamari, simonmar Reviewed By: simonmar Subscribers: rwbarton, carter GHC Trac Issues: #15853 Differential Revision: https://phabricator.haskell.org/D5306
* Remove warnings-silencing flags for code generated by AlexSimon Jakobi2018-11-221-7/+0
| | | | | | | | | | | | | | | | | | Current versions of Alex don't seem to produce as many warnings any more. In order to silence a warning and to avoid overlong lines, I've taken the liberty of refactoring 'tok_num'. Test Plan: ./validate Reviewers: bgamari, simonmar Reviewed By: simonmar Subscribers: erikd, rwbarton, carter Differential Revision: https://phabricator.haskell.org/D5319
* Rename literal constructorsSylvain Henry2018-11-222-3/+3
| | | | | | | | | | | | | | | | | | | | | | | | In a previous patch we replaced some built-in literal constructors (MachInt, MachWord, etc.) with a single LitNumber constructor. In this patch we replace the `Mach` prefix of the remaining constructors with `Lit` for consistency (e.g., LitChar, LitLabel, etc.). Sadly the name `LitString` was already taken for a kind of FastString and it would become misleading to have both `LitStr` (literal constructor renamed after `MachStr`) and `LitString` (FastString variant). Hence this patch renames the FastString variant `PtrString` (which is more accurate) and the literal string constructor now uses the least surprising `LitString` name. Both `Literal` and `LitString/PtrString` have recently seen breaking changes so doing this kind of renaming now shouldn't harm much. Reviewers: hvr, goldfire, bgamari, simonmar, jrtc27, tdammers Subscribers: tdammers, rwbarton, thomie, carter Differential Revision: https://phabricator.haskell.org/D4881
* Introduce Int16# and Word16#Abhiroop Sarkar2018-11-171-0/+4
| | | | | | | | | | | | This builds off of D4475. Bumps binary submodule. Reviewers: carter, AndreasK, hvr, goldfire, bgamari, simonmar Subscribers: rwbarton, thomie Differential Revision: https://phabricator.haskell.org/D5006
* Minor refactoringGabor Greif2018-11-171-13/+13
| | | | PR: https://github.com/ghc/ghc/pull/223/
* NCG: New code layout algorithm.Andreas Klebinger2018-11-175-6/+49
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Summary: This patch implements a new code layout algorithm. It has been tested for x86 and is disabled on other platforms. Performance varies slightly be CPU/Machine but in general seems to be better by around 2%. Nofib shows only small differences of about +/- ~0.5% overall depending on flags/machine performance in other benchmarks improved significantly. Other benchmarks includes at least the benchmarks of: aeson, vector, megaparsec, attoparsec, containers, text and xeno. While the magnitude of gains differed three different CPUs where tested with all getting faster although to differing degrees. I tested: Sandy Bridge(Xeon), Haswell, Skylake * Library benchmark results summarized: * containers: ~1.5% faster * aeson: ~2% faster * megaparsec: ~2-5% faster * xml library benchmarks: 0.2%-1.1% faster * vector-benchmarks: 1-4% faster * text: 5.5% faster On average GHC compile times go down, as GHC compiled with the new layout is faster than the overhead introduced by using the new layout algorithm, Things this patch does: * Move code responsilbe for block layout in it's own module. * Move the NcgImpl Class into the NCGMonad module. * Extract a control flow graph from the input cmm. * Update this cfg to keep it in sync with changes during asm codegen. This has been tested on x64 but should work on x86. Other platforms still use the old codelayout. * Assign weights to the edges in the CFG based on type and limited static analysis which are then used for block layout. * Once we have the final code layout eliminate some redundant jumps. In particular turn a sequences of: jne .foo jmp .bar foo: into je bar foo: .. Test Plan: ci Reviewers: bgamari, jmct, jrtc27, simonmar, simonpj, RyanGlScott Reviewed By: RyanGlScott Subscribers: RyanGlScott, trommler, jmct, carter, thomie, rwbarton GHC Trac Issues: #15124 Differential Revision: https://phabricator.haskell.org/D4726
* Fix a bug in SRT generation (#15892)Simon Marlow2018-11-151-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Summary: The logic in `Note [recursive SRTs]` was correct. However, my implementation of it wasn't: I got the associativity of `Set.difference` wrong, which led to an extremely subtle and difficult to find bug. Fortunately now we have a test case. I was able to cut down the code to something manageable, and I've added it to the test suite. Test Plan: Before (using my stage 1 compiler without the fix): ``` ====> T15892(normal) 1 of 1 [0, 0, 0] cd "T15892.run" && "/home/smarlow/ghc/inplace/bin/ghc-stage1" -o T15892 T15892.hs -dcore-lint -dcmm-lint -no-user-package-db -rtsopts -fno-warn-missed-specialisations -fshow-warning-groups -fdiagnostics-color=never -fno-diagnostics-show-caret -Werror=compat -dno-debug-output -O cd "T15892.run" && ./T15892 +RTS -G1 -A32k -RTS Wrong exit code for T15892(normal)(expected 0 , actual 134 ) Stderr ( T15892 ): T15892: internal error: evacuate: strange closure type 0 (GHC version 8.7.20181113 for x86_64_unknown_linux) Please report this as a GHC bug: http://www.haskell.org/ghc/reportabug Aborted (core dumped) *** unexpected failure for T15892(normal) =====> T15892(g1) 1 of 1 [0, 1, 0] cd "T15892.run" && "/home/smarlow/ghc/inplace/bin/ghc-stage1" -o T15892 T15892.hs -dcore-lint -dcmm-lint -no-user-package-db -rtsopts -fno-warn-missed-specialisations -fshow-warning-groups -fdiagnostics-color=never -fno-diagnostics-show-caret -Werror=compat -dno-debug-output -O cd "T15892.run" && ./T15892 +RTS -G1 -RTS +RTS -G1 -A32k -RTS Wrong exit code for T15892(g1)(expected 0 , actual 134 ) Stderr ( T15892 ): T15892: internal error: evacuate: strange closure type 0 (GHC version 8.7.20181113 for x86_64_unknown_linux) Please report this as a GHC bug: http://www.haskell.org/ghc/reportabug Aborted (core dumped) ``` After (using my stage 2 compiler with the fix): ``` =====> T15892(normal) 1 of 1 [0, 0, 0] cd "T15892.run" && "/home/smarlow/ghc/inplace/test spaces/ghc-stage2" -o T15892 T15892.hs -dcore-lint -dcmm-lint -no-user-package-db -rtsopts -fno-warn-missed-specialisations -fshow-warning-groups -fdiagnostics-color=never -fno-diagnostics-show-caret -Werror=compat -dno-debug-output cd "T15892.run" && ./T15892 +RTS -G1 -A32k -RTS =====> T15892(g1) 1 of 1 [0, 0, 0] cd "T15892.run" && "/home/smarlow/ghc/inplace/test spaces/ghc-stage2" -o T15892 T15892.hs -dcore-lint -dcmm-lint -no-user-package-db -rtsopts -fno-warn-missed-specialisations -fshow-warning-groups -fdiagnostics-color=never -fno-diagnostics-show-caret -Werror=compat -dno-debug-output cd "T15892.run" && ./T15892 +RTS -G1 -RTS +RTS -G1 -A32k -RTS ``` Reviewers: bgamari, osa1, erikd Reviewed By: osa1 Subscribers: rwbarton, carter GHC Trac Issues: #15892 Differential Revision: https://phabricator.haskell.org/D5334
* Add Int8# and Word8#Michal Terepeta2018-11-025-13/+84
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This is the first step of implementing: https://github.com/ghc-proposals/ghc-proposals/pull/74 The main highlights/changes: primops.txt.pp gets two new sections for two new primitive types for signed and unsigned 8-bit integers (Int8# and Word8 respectively) along with basic arithmetic and comparison operations. PrimRep/RuntimeRep get two new constructors for them. All of the primops translate into the existing MachOPs. For CmmCalls the codegen will now zero-extend the values at call site (so that they can be moved to the right register) and then truncate them back their original width. x86 native codegen needed some updates, since it wasn't able to deal with the new widths, but all the changes are quite localized. LLVM backend seems to just work. This is the second attempt at merging this, after the first attempt in D4475 had to be backed out due to regressions on i386. Bumps binary submodule. Signed-off-by: Michal Terepeta <michal.terepeta@gmail.com> Test Plan: ./validate (on both x86-{32,64}) Reviewers: bgamari, hvr, goldfire, simonmar Subscribers: rwbarton, carter Differential Revision: https://phabricator.haskell.org/D5258
* Generate correct relocation for external cost centreZejun Wu2018-10-151-4/+11
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | We used to always generate direct access for cost centre labels. We fixed this by generating indirect data load for cost centre defined in external module. Test Plan: The added test used to fail with error message ``` /bin/ld.gold: error: T15723B.o: requires dynamic R_X86_64_PC32 reloc against 'T15723A_foo1_EXPR_cc' which may overflow at runtime; recompile with -fPIC ``` and now passes. Also check that `R_X86_64_PC32` is generated for CostCentre from the same module and `R_X86_64_GOTPCREL` is generated for CostCentre from external module: ``` $ objdump -rdS T15723B.o 0000000000000028 <T15723B_test_info>: 28: 48 8d 45 f0 lea -0x10(%rbp),%rax 2c: 4c 39 f8 cmp %r15,%rax 2f: 72 70 jb a1 <T15723B_test_info+0x79> 31: 48 83 ec 08 sub $0x8,%rsp 35: 48 8d 35 00 00 00 00 lea 0x0(%rip),%rsi # 3c <T15723B_test_info+0x14> 38: R_X86_64_PC32 T15723B_test1_EXPR_cc-0x4 3c: 49 8b bd 60 03 00 00 mov 0x360(%r13),%rdi 43: 31 c0 xor %eax,%eax 45: e8 00 00 00 00 callq 4a <T15723B_test_info+0x22> 46: R_X86_64_PLT32 pushCostCentre-0x4 4a: 48 83 c4 08 add $0x8,%rsp 4e: 48 ff 40 30 incq 0x30(%rax) 52: 49 89 85 60 03 00 00 mov %rax,0x360(%r13) 59: 48 83 ec 08 sub $0x8,%rsp 5d: 49 8b bd 60 03 00 00 mov 0x360(%r13),%rdi 64: 48 8b 35 00 00 00 00 mov 0x0(%rip),%rsi # 6b <T15723B_test_info+0x43> 67: R_X86_64_GOTPCREL T15723A_foo1_EXPR_cc-0x4 6b: 31 c0 xor %eax,%eax 6d: e8 00 00 00 00 callq 72 <T15723B_test_info+0x4a> 6e: R_X86_64_PLT32 pushCostCentre-0x4 72: 48 83 c4 08 add $0x8,%rsp 76: 48 ff 40 30 incq 0x30(%rax) ``` Reviewers: simonmar, bgamari Reviewed By: simonmar Subscribers: rwbarton, carter GHC Trac Issues: #15723 Differential Revision: https://phabricator.haskell.org/D5214
* Deprecate -fllvm-pass-vectors-in-regsBen Gamari2018-10-151-12/+5
| | | | | | | | | | | | Summary: The behavior previously enabled by this flag is as been the default since 8.6.1. Reviewers: simonmar Subscribers: rwbarton, carter Differential Revision: https://phabricator.haskell.org/D5193
* Revert "Add Int8# and Word8#"Ben Gamari2018-10-095-84/+13
| | | | | | | | | This unfortunately broke i386 support since it introduced references to byte-sized registers that don't exist on that architecture. Reverts binary submodule This reverts commit 5d5307f943d7581d7013ffe20af22233273fba06.
* Add Int8# and Word8#Michal Terepeta2018-10-075-13/+84
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This is the first step of implementing: https://github.com/ghc-proposals/ghc-proposals/pull/74 The main highlights/changes: - `primops.txt.pp` gets two new sections for two new primitive types for signed and unsigned 8-bit integers (`Int8#` and `Word8` respectively) along with basic arithmetic and comparison operations. `PrimRep`/`RuntimeRep` get two new constructors for them. All of the primops translate into the existing `MachOP`s. - For `CmmCall`s the codegen will now zero-extend the values at call site (so that they can be moved to the right register) and then truncate them back their original width. - x86 native codegen needed some updates, since it wasn't able to deal with the new widths, but all the changes are quite localized. LLVM backend seems to just work. Bumps binary submodule. Signed-off-by: Michal Terepeta <michal.terepeta@gmail.com> Test Plan: ./validate with new tests Reviewers: hvr, goldfire, bgamari, simonmar Subscribers: Abhiroop, dfeuer, rwbarton, thomie, carter Differential Revision: https://phabricator.haskell.org/D4475
* UNREG: don't prefix asm prefixes in via-C modeSergei Trofimovich2018-10-061-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | commit 64c54fff2d6534e1229359a8d357ec1dc6c21b73 ("Mark system and internal symbols as private symbols in asm") Added `internalNamePrefix` helper. Unfortunately it generates invalid label in unregisterised mode: ``` $ ./configure --enable-unregisterised /tmp/ghc19372_0/ghc_4.hc:2831:22: error: error: expected identifier or '(' before '.' token static const StgWord .Lcl3_info[]__attribute__((aligned(8)))= { ^ ``` Here asm-style prefix is applied to C symbol. The fix is simple: apply asm-style labels only to assembly code. Signed-off-by: Sergei Trofimovich <slyfox@gentoo.org> Reviewers: simonmar, last_g, bgamari Reviewed By: simonmar Subscribers: rwbarton, carter Differential Revision: https://phabricator.haskell.org/D5207
* Don't shortcut SRTs for static functions (#15544)Simon Marlow2018-09-181-22/+130
| | | | | | | | | | | | | | | | | | | | Shortcutting the SRT for a static function can lead to resurrecting a static object at runtime, which violates assumptions in the GC. See comments for details. Test Plan: - manual testing (in progress) - validate Reviewers: osa1, bgamari, erikd Reviewed By: bgamari Subscribers: rwbarton, carter GHC Trac Issues: #15544 Differential Revision: https://phabricator.haskell.org/D5145
* Mark code related symbols as @function not @objectSergei Azovskov2018-09-141-1/+16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Summary: This diff is a part of the bigger project which goal is to improve common profiling tools support (perf) for GHC binaries. A similar job was already done and reverted in the past: * https://phabricator.haskell.org/rGHCb1f453e16f0ce11a2ab18cc4c350bdcbd36299a6 * https://phabricator.haskell.org/rGHCf1f3c4f50650110ad0f700d6566a44c515b0548f Reasoning: `Perf` and similar tools build in memory symbol table from the .symtab section of the ELF file to display human-readable function names instead of the addresses in the output. `Perf` uses only two types of symbols: `@function` and `@notype` but GHC is not capable to produce any `@function` symbols so the `perf` output is pretty useless (All the haskell symbols that you can see in `perf` now are `@notype` internal symbols extracted by mistake/hack). The changes: * mark code related symbols as @function * small hack to mark InfoTable symbols as code if TABLES_NEXT_TO_CODE is true Limitations: * The perf symbolization support is not complete after this patch but I'm working on the second patch. * Constructor symbols are not supported. To fix that we can issue extra local symbols which mark code sections as code and will be only used for debug. Test Plan: tests any additional ideas? Perf output on stock ghc 8.4.1: ``` 9.78% FibbSlow FibbSlow [.] ckY_info 9.59% FibbSlow FibbSlow [.] cjqd_info 7.17% FibbSlow FibbSlow [.] c3sg_info 6.62% FibbSlow FibbSlow [.] c1X_info 5.32% FibbSlow FibbSlow [.] cjsX_info 4.18% FibbSlow FibbSlow [.] s3rN_info 3.82% FibbSlow FibbSlow [.] c2m_info 3.68% FibbSlow FibbSlow [.] cjlJ_info 3.26% FibbSlow FibbSlow [.] c3sb_info 3.19% FibbSlow FibbSlow [.] cjPQ_info 3.05% FibbSlow FibbSlow [.] cjQd_info 2.97% FibbSlow FibbSlow [.] cjAB_info 2.78% FibbSlow FibbSlow [.] cjzP_info 2.40% FibbSlow FibbSlow [.] cjOS_info 2.38% FibbSlow FibbSlow [.] s3rK_info 2.27% FibbSlow FibbSlow [.] cjq0_info 2.18% FibbSlow FibbSlow [.] cKQ_info 2.13% FibbSlow FibbSlow [.] cjSl_info 1.99% FibbSlow FibbSlow [.] s3rL_info 1.98% FibbSlow FibbSlow [.] c2cC_info 1.80% FibbSlow FibbSlow [.] s3rO_info 1.37% FibbSlow FibbSlow [.] c2f2_info ... ``` Perf output on patched ghc: ``` 7.97% FibbSlow FibbSlow [.] c3rM_info 6.75% FibbSlow FibbSlow [.] 0x000000000032cfa8 6.63% FibbSlow FibbSlow [.] cifA_info 4.98% FibbSlow FibbSlow [.] integerzmgmp_GHCziIntegerziType_eqIntegerzh_info 4.55% FibbSlow FibbSlow [.] chXn_info 4.52% FibbSlow FibbSlow [.] c3rH_info 4.45% FibbSlow FibbSlow [.] chZB_info 4.04% FibbSlow FibbSlow [.] Main_fibbzuslow_info 4.03% FibbSlow FibbSlow [.] stg_ap_0_fast 3.76% FibbSlow FibbSlow [.] chXA_info 3.67% FibbSlow FibbSlow [.] cifu_info 3.25% FibbSlow FibbSlow [.] ci4r_info 2.64% FibbSlow FibbSlow [.] s3rf_info 2.42% FibbSlow FibbSlow [.] s3rg_info 2.39% FibbSlow FibbSlow [.] integerzmgmp_GHCziIntegerziType_eqInteger_info 2.25% FibbSlow FibbSlow [.] integerzmgmp_GHCziIntegerziType_minusInteger_info 2.17% FibbSlow FibbSlow [.] ghczmprim_GHCziClasses_zeze_info 2.09% FibbSlow FibbSlow [.] cicc_info 2.03% FibbSlow FibbSlow [.] 0x0000000000331e15 2.02% FibbSlow FibbSlow [.] s3ri_info 1.91% FibbSlow FibbSlow [.] 0x0000000000331bb8 1.89% FibbSlow FibbSlow [.] ci4N_info ... ``` Reviewers: simonmar, niteria, bgamari, goldfire Reviewed By: simonmar, bgamari Subscribers: lelf, rwbarton, thomie, carter GHC Trac Issues: #15501 Differential Revision: https://phabricator.haskell.org/D4713
* Mark system and internal symbols as private symbols in asmSergei Azovskov2018-09-141-17/+38
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Summary: This marks system and internal symbols as private in asm output so those random generated sysmbols won't appear in .symtab Reasoning: * internal symbols don't help to debug because names are just random * the symbols style breaks perf logic * internal symbols can take ~75% of the .symtab. In the same time .symtab can take about 20% of the binary file size Notice: This diff mostly makes sense on top of the D4713 (or similar) Test Plan: tests Perf from D4713 ``` 7.97% FibbSlow FibbSlow [.] c3rM_info 6.75% FibbSlow FibbSlow [.] 0x000000000032cfa8 6.63% FibbSlow FibbSlow [.] cifA_info 4.98% FibbSlow FibbSlow [.] integerzmgmp_GHCziIntegerziType_eqIntegerzh_info 4.55% FibbSlow FibbSlow [.] chXn_info 4.52% FibbSlow FibbSlow [.] c3rH_info 4.45% FibbSlow FibbSlow [.] chZB_info 4.04% FibbSlow FibbSlow [.] Main_fibbzuslow_info 4.03% FibbSlow FibbSlow [.] stg_ap_0_fast 3.76% FibbSlow FibbSlow [.] chXA_info 3.67% FibbSlow FibbSlow [.] cifu_info 3.25% FibbSlow FibbSlow [.] ci4r_info 2.64% FibbSlow FibbSlow [.] s3rf_info 2.42% FibbSlow FibbSlow [.] s3rg_info 2.39% FibbSlow FibbSlow [.] integerzmgmp_GHCziIntegerziType_eqInteger_info 2.25% FibbSlow FibbSlow [.] integerzmgmp_GHCziIntegerziType_minusInteger_info 2.17% FibbSlow FibbSlow [.] ghczmprim_GHCziClasses_zeze_info 2.09% FibbSlow FibbSlow [.] cicc_info 2.03% FibbSlow FibbSlow [.] 0x0000000000331e15 2.02% FibbSlow FibbSlow [.] s3ri_info 1.91% FibbSlow FibbSlow [.] 0x0000000000331bb8 1.89% FibbSlow FibbSlow [.] ci4N_info ... ``` Perf from this patch: ``` 15.37% FibbSlow FibbSlow [.] Main_fibbzuslow_info 15.33% FibbSlow FibbSlow [.] integerzmgmp_GHCziIntegerziType_minusInteger_info 13.34% FibbSlow FibbSlow [.] integerzmgmp_GHCziIntegerziType_eqIntegerzh_info 9.24% FibbSlow FibbSlow [.] integerzmgmp_GHCziIntegerziType_plusInteger_info 9.08% FibbSlow FibbSlow [.] frame_dummy 8.25% FibbSlow FibbSlow [.] integerzmgmp_GHCziIntegerziType_eqInteger_info 4.29% FibbSlow FibbSlow [.] 0x0000000000321ab0 3.84% FibbSlow FibbSlow [.] stg_ap_0_fast 3.07% FibbSlow FibbSlow [.] ghczmprim_GHCziClasses_zeze_info 2.39% FibbSlow FibbSlow [.] 0x0000000000321ab7 1.90% FibbSlow FibbSlow [.] 0x00000000003266b8 1.88% FibbSlow FibbSlow [.] base_GHCziNum_zm_info 1.83% FibbSlow FibbSlow [.] 0x0000000000326915 1.34% FibbSlow FibbSlow [.] 0x00000000003248cc 1.07% FibbSlow FibbSlow [.] base_GHCziNum_zp_info 0.98% FibbSlow FibbSlow [.] 0x00000000003247c8 0.80% FibbSlow FibbSlow [.] 0x0000000000121498 0.79% FibbSlow FibbSlow [.] stg_gc_noregs 0.75% FibbSlow FibbSlow [.] 0x0000000000321ad6 0.67% FibbSlow FibbSlow [.] 0x0000000000321aca 0.64% FibbSlow FibbSlow [.] 0x0000000000321b4a 0.61% FibbSlow FibbSlow [.] 0x00000000002ff633 ``` Reviewers: simonmar, niteria, bgamari Reviewed By: simonmar Subscribers: lelf, angerman, olsner, rwbarton, thomie, carter Differential Revision: https://phabricator.haskell.org/D4722
* A few typos [ci skip]Gabor Greif2018-08-301-1/+1
|
* Fix precision of asinh/acosh/atanh by making them primopsArtem Pelenitsyn2018-08-212-0/+12
| | | | | | | | | | Reviewers: hvr, bgamari, simonmar, jrtc27 Reviewed By: bgamari Subscribers: alpmestan, rwbarton, thomie, carter Differential Revision: https://phabricator.haskell.org/D5034
* Replace most occurences of foldl with foldl'.klebinger.andreas@gmx.at2018-08-214-4/+0
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This patch adds foldl' to GhcPrelude and changes must occurences of foldl to foldl'. This leads to better performance especially for quick builds where GHC does not perform strictness analysis. It does change strictness behaviour when we use foldl' to turn a argument list into function applications. But this is only a drawback if code looks ONLY at the last argument but not at the first. And as the benchmarks show leads to fewer allocations in practice at O2. Compiler performance for Nofib: O2 Allocations: -1 s.d. ----- -0.0% +1 s.d. ----- -0.0% Average ----- -0.0% O2 Compile Time: -1 s.d. ----- -2.8% +1 s.d. ----- +1.3% Average ----- -0.8% O0 Allocations: -1 s.d. ----- -0.2% +1 s.d. ----- -0.1% Average ----- -0.2% Test Plan: ci Reviewers: goldfire, bgamari, simonmar, tdammers, monoidal Reviewed By: bgamari, monoidal Subscribers: tdammers, rwbarton, thomie, carter Differential Revision: https://phabricator.haskell.org/D4929
* Minor refactoring in CmmUtils.mkLivenessÖmer Sinan Ağacan2018-07-122-12/+9
| | | | | | | | | | | | Test Plan: validate Reviewers: bgamari, simonmar Reviewed By: simonmar Subscribers: rwbarton, thomie, carter Differential Revision: https://phabricator.haskell.org/D4957
* Typofixes in comments and whitespace only [ci skip]Gabor Greif2018-06-261-1/+1
|
* Use __FILE__ for Cmm assertion locations, fix #8619Ömer Sinan Ağacan2018-06-171-2/+0
| | | | | | | | | | | | | | | | | | | | It seems like we currently support string literals in Cmm, so we can use __LINE__ CPP macro in assertion macros. This improves error messages that previously looked like ASSERTION FAILED: file (null), line 1302 (null) part now shows the actual file name. Also inline some single-use string literals in PrimOps.cmm. Reviewers: bgamari, simonmar, erikd Reviewed By: bgamari Subscribers: rwbarton, thomie, carter Differential Revision: https://phabricator.haskell.org/D4862
* UNREG: fix CmmRegOff large offset handling on W64 platformsSergei Trofimovich2018-06-171-8/+4
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Gabor noticed C warning when building unregisterised 64-bit compiler on GHC.Integer.Types (from integer-simple). Minimised example with a warning: ```haskell {-# LANGUAGE MagicHash #-} {-# LANGUAGE NoImplicitPrelude #-} {-# OPTIONS_GHC -Wall #-} module M (bug) where import GHC.Prim (Word#, minusWord#, ltWord#) import GHC.Types (isTrue#) -- assume Word = Word64 bug :: Word# -> Word# bug x = if isTrue# (x `ltWord#` 0x8000000000000000##) then 0## else x `minusWord#` 0x8000000000000000## ``` ``` $ LANG=C x86_64-UNREG-linux-gnu-ghc -O1 -c M.hs -fforce-recomp /tmp/ghc30219_0/ghc_1.hc: In function 'M_bug_entry': /tmp/ghc30219_0/ghc_1.hc:20:14: error: warning: integer constant is so large that it is unsigned ``` It's caused by limited handling of integer literals in CmmRegOff. This change switches to use standard integer literal pretty-printer. C code before the change: ```c FN_(M_bug_entry) { W_ _sAg; _cAr: _sAg = *Sp; switch ((W_)(_sAg < 0x8000000000000000UL)) { case 0x1UL: goto _cAq; default: goto _cAp; } _cAp: R1.w = _sAg+-9223372036854775808; // ... ``` C code after the change: ```c FN_(M_bug_entry) { W_ _sAg; _cAr: _sAg = *Sp; switch ((W_)(_sAg < 0x8000000000000000UL)) { case 0x1UL: goto _cAq; default: goto _cAp; } _cAp: R1.w = _sAg+(-0x8000000000000000UL); ``` URL: https://mail.haskell.org/pipermail/ghc-devs/2018-June/015875.html Reported-by: Gabor Greif Signed-off-by: Sergei Trofimovich <slyfox@gentoo.org> Test Plan: test generated code on unregisterised mips64 and amd64 Reviewers: simonmar, ggreif, bgamari Reviewed By: ggreif, bgamari Subscribers: rwbarton, thomie, carter Differential Revision: https://phabricator.haskell.org/D4856
* UNREG: PprC: add support for of W16 literals (Ticket #15237)Sergei Trofimovich2018-06-151-0/+8
| | | | | | | | | | | | | | | Fix UNREG build failure for 32-bit targets. This change is an equivalent of commit 0238a6c78102d43dae2f56192bd3486e4f9ecf1d ("UNREG: PprC: add support for of W32 literals") The change allows combining two subwords into one word on 32-bit targets. Tested on nios2-unknown-linux-gnu. GHC Trac Issues: #15237 Signed-off-by: Sergei Trofimovich <slyfox@gentoo.org>
* UNREG: PprC: add support for of W32 literalsSergei Trofimovich2018-06-141-0/+8
| | | | | | | | | | | | | | | | | | | | | | | | | Today UNREG build fails to generate sub-word literals: ``` rts_dist_HC rts/dist/build/StgStartup.o ghc-stage1: panic! (the 'impossible' happened) (GHC version 8.5.20180612 for x86_64-unknown-linux): pprStatics: cannot emit a non-word-sized static literal W32 ``` The change allows combining two subwords into one word. Signed-off-by: Sergei Trofimovich <slyfox@gentoo.org> Reviewers: simonmar, bgamari Reviewed By: bgamari Subscribers: rwbarton, thomie, carter GHC Trac Issues: #15237 Differential Revision: https://phabricator.haskell.org/D4837
* Check if both branches of an Cmm if have the same target.klebinger.andreas@gmx.at2018-06-071-4/+9
| | | | | | | | | | | | | | | | | | | | This for some reason or the other and makes it into the final binary. I've added the check to ContFlowOpt as that seems like a logical place for this. In a regular nofib run there were 30 occurences of this pattern. Test Plan: ci Reviewers: bgamari, simonmar, dfeuer, jrtc27, tdammers Reviewed By: bgamari, simonmar Subscribers: tdammers, dfeuer, rwbarton, thomie, carter GHC Trac Issues: #15188 Differential Revision: https://phabricator.haskell.org/D4740
* Rename some mutable closure types for consistencyÖmer Sinan Ağacan2018-06-051-10/+11
| | | | | | | | | | | | | | | | | | | | | | | SMALL_MUT_ARR_PTRS_FROZEN0 -> SMALL_MUT_ARR_PTRS_FROZEN_DIRTY SMALL_MUT_ARR_PTRS_FROZEN -> SMALL_MUT_ARR_PTRS_FROZEN_CLEAN MUT_ARR_PTRS_FROZEN0 -> MUT_ARR_PTRS_FROZEN_DIRTY MUT_ARR_PTRS_FROZEN -> MUT_ARR_PTRS_FROZEN_CLEAN Naming is now consistent with other CLEAR/DIRTY objects (MVAR, MUT_VAR, MUT_ARR_PTRS). (alternatively we could rename MVAR_DIRTY/MVAR_CLEAN etc. to MVAR0/MVAR) Removed a few comments in Scav.c about FROZEN0 being on the mut_list because it's now clear from the closure type. Reviewers: bgamari, simonmar, erikd Reviewed By: simonmar Subscribers: rwbarton, thomie, carter Differential Revision: https://phabricator.haskell.org/D4784
* Also suppress uniques in cmm dumps with `-dsuppress-uniques`.klebinger.andreas@gmx.at2018-06-042-3/+12
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Allows easier structural comparison of Cmm code. Before: ``` cxCH: // global _suEU::P64 = R1; if ((Sp + -16) < SpLim) (likely: False) goto cxCI; else goto cxCJ; ``` After ``` _lbl_: // global __locVar_::P64 = R1; if ((Sp + -16) < SpLim) (likely: False) goto cxBf; else goto cxBg; ``` Test Plan: Looking at dumps, ci Reviewers: bgamari, simonmar Reviewed By: bgamari Subscribers: rwbarton, thomie, carter Differential Revision: https://phabricator.haskell.org/D4786
* C codegen: print details of pprStatics panicsSergei Trofimovich2018-06-021-2/+2
| | | | Signed-off-by: Sergei Trofimovich <slyfox@gentoo.org>
* Optimizations for CmmBlockElim.klebinger.andreas@gmx.at2018-06-022-17/+35
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | * Use toBlockList instead of revPostorder. Block elimination works on a given Cmm graph by: * Getting a list of blocks. * Looking for duplicates in these blocks. * Removing all but one instance of duplicates. There are two (reasonable) ways to get the list of blocks. * The fast way: `toBlockList` This just flattens the underlying map into a list. * The convenient way: `revPostorder` Start at the entry label, scan for reachable blocks and return only these. This has the advantage of removing all dead code. If there is dead code the later is better. Work done on unreachable blocks is clearly wasted work. However by the point we run the common block elimination pass the input graph already had all dead code removed. This is done during control flow optimization in CmmContFlowOpt which is our first Cmm pass. This means common block elimination is free to use toBlockList because revPostorder would return the same blocks. (Although in a different order). * Change the triemap used for grouping by a label list from `(TM.ListMap UniqDFM)` to `ListMap (GenMap LabelMap)`. * Using GenMap offers leaf compression. Which is a trie optimization described by the Note [Compressed TrieMap] in CoreSyn/TrieMap.hs * Using LabelMap removes the overhead associated with UniqDFM. This is deterministic since if we have the same input keys the same LabelMap will be constructed. Test Plan: ci, profiling output Reviewers: bgamari, simonmar Reviewed By: bgamari Subscribers: dfeuer, thomie, carter GHC Trac Issues: #15103 Differential Revision: https://phabricator.haskell.org/D4597
* UNREG: mark SRT as writable in generated C codeSergei Trofimovich2018-06-011-2/+0
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Noticed section mismatch on UNREG build failure: ``` HC [stage 1] libraries/integer-gmp/dist-install/build/GHC/Integer/Type.o error: conflicting types for 'ufu0_srt' static StgWord ufu0_srt[]__attribute__((aligned(8)))= { ^~~~~~~~ note: previous declaration of 'ufu0_srt' was here IRO_(ufu0_srt); ^~~~~~~~ ``` `IRO_` is a 'const' qualifier. The error is a leftover from commit 838b69032566ce6ab3918d70e8d5e098d0bcee02 "Merge FUN_STATIC closure with its SRT" where part of SRT was moved into closure itself and made SRTs writable. This change puts all SRTs into writable section. Signed-off-by: Sergei Trofimovich <slyfox@gentoo.org> Reviewers: simonmar, bgamari Subscribers: rwbarton, thomie, carter Differential Revision: https://phabricator.haskell.org/D4731
* Cleanups [ci skip]Gabor Greif2018-06-011-18/+16
|
* Disable the SRT offset optimisation on MachO platformsBen Gamari2018-05-233-5/+40
| | | | | | | | | | | | | | | | | | | | | | | | | | Unfortunately, this optimisation is infeasible on MachO platforms (e.g. Darwin) due to an object format limitation. Specifically, linking fails with errors of the form: error: unsupported relocation with subtraction expression, symbol '_integerzmgmp_GHCziIntegerziType_quotInteger_closure' can not be undefined in a subtraction expression Apparently MachO does not permit relocations' subtraction expressions to refer to undefined symbols. As far as I can tell this means that it is essentially impossible to express an offset between symbols living in different compilation units. This means that we lively can't use this optimisation on MachO platforms. Test Plan: Validate on Darwin Reviewers: simonmar, erikd Subscribers: rwbarton, thomie, carter, angerman GHC Trac Issues: #15169 Differential Revision: https://phabricator.haskell.org/D4715
* Fix a bug in SRT generationSimon Marlow2018-05-231-48/+51
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Summary: I had good intentions, but they were not being followed. In particular, this comment: ``` --- - we never resolve a reference to a CAF to the contents of its SRT, since --- the point of SRTs is to keep CAFs alive. ``` was not true, because we updated the srtMap after generating the SRT for a CAF. Therefore it was possible for another CAF to refer to an earlier CAF, and the reference to the earlier CAF would be shortcutted to refer to its SRT instead of pointing to the CAF itself. The fix is just to not update the srtMap when generating the SRT for a CAF, but I also refactored the code and comments around this to be a bit better organised. Test Plan: Harbourmaster Reviewers: bgamari, michalt, simonpj, erikd Subscribers: rwbarton, thomie, carter GHC Trac Issues: #15173, #15168 Differential Revision: https://phabricator.haskell.org/D4721
* Typo in commentsGabor Greif2018-05-231-1/+1
|
* Comments and refactoring onlySimon Marlow2018-05-172-12/+30
| | | | Addressing review comments on D4637
* Merge FUN_STATIC closure with its SRTSimon Marlow2018-05-164-95/+151
| | | | | | | | | | | | | | | | | | | | | | | | | | Summary: The idea here is to save a little code size and some work in the GC, by collapsing FUN_STATIC closures and their SRTs. This is (4) in a series; see D4632 for more details. There's a tradeoff here: more complexity in the compiler in exchange for a modest code size reduction (probably around 0.5%). Results: * GHC binary itself (statically linked) is 1% smaller * -0.2% binary sizes in nofib (-0.5% module sizes) Full nofib results comparing D4634 with this: P177 (ignore runtimes, these aren't stable on my laptop) Test Plan: validate, nofib Reviewers: bgamari, niteria, simonpj, erikd Subscribers: thomie, carter Differential Revision: https://phabricator.haskell.org/D4637
* Save a word in the info table on x86_64Simon Marlow2018-05-163-25/+54
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Summary: An info table with an SRT normally looks like this: StgWord64 srt_offset StgClosureInfo layout StgWord32 layout StgWord32 has_srt But we only need 32 bits for srt_offset on x86_64, because the small memory model requires that code segments are at most 2GB. So we can optimise this to StgClosureInfo layout StgWord32 layout StgWord32 srt_offset saving a word. We can tell whether the info table has an SRT or not, because zero is not a valid srt_offset, so zero still indicates that there's no SRT. Test Plan: * validate * For results, see D4632. Reviewers: bgamari, niteria, osa1, erikd Subscribers: thomie, carter Differential Revision: https://phabricator.haskell.org/D4634
* Allow CmmLabelDiffOff with different widthsSimon Marlow2018-05-167-12/+15
| | | | | | | | | | | | | | | | | | Summary: This change makes it possible to generate a static 32-bit relative label offset on x86_64. Currently we can only generate word-sized label offsets. This will be used in D4634 to shrink info tables. See D4632 for more details. Test Plan: See D4632 Reviewers: bgamari, niteria, michalt, erikd, jrtc27, osa1 Subscribers: thomie, carter Differential Revision: https://phabricator.haskell.org/D4633
* An overhaul of the SRT representationSimon Marlow2018-05-169-368/+681
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Summary: - Previously we would hvae a single big table of pointers per module, with a set of bitmaps to reference entries within it. The new representation is identical to a static constructor, which is much simpler for the GC to traverse, and we get to remove the complicated bitmap-traversal code from the GC. - Rewrite all the code to generate SRTs in CmmBuildInfoTables, and document it much better (see Note [SRTs]). This has been something I've wanted to do since we moved to the new code generator, I finally had the opportunity to finish it while on a transatlantic flight recently :) There are a series of 4 diffs: 1. D4632 (this one), which does the bulk of the changes 2. D4633 which adds support for smaller `CmmLabelDiffOff` constants 3. D4634 which takes advantage of D4632 and D4633 to save a word in info tables that have an SRT on x86_64. This is where most of the binary size improvement comes from. 4. D4637 which makes a further optimisation to merge some SRTs with static FUN closures. This adds some complexity and the benefits are fairly modest, so it's not clear yet whether we should do this. Results (after (3), on x86_64) - GHC itself (staticaly linked) is 5.2% smaller - -1.7% binary sizes in nofib, -2.9% module sizes. Full nofib results: P176 - I measured the overhead of traversing all the static objects in a major GC in GHC itself by doing `replicateM_ 1000 performGC` as the first thing in `Main.main`. The new version was 5-10% faster, but the results did vary quite a bit. - I'm not sure if there's a compile-time difference, the results are too unreliable. Test Plan: validate Reviewers: bgamari, michalt, niteria, simonpj, erikd, osa1 Subscribers: thomie, carter Differential Revision: https://phabricator.haskell.org/D4632
* Add 'addWordC#' PrimOpSebastian Graf2018-05-052-0/+2
| | | | | | | | | | | | | | | | | | | This is mostly for congruence with 'subWordC#' and '{add,sub}IntC#'. I found 'plusWord2#' while implementing this, which both lacks documentation and has a slightly different specification than 'addWordC#', which means the generic implementation is unnecessarily complex. While I was at it, I also added lacking meta-information on PrimOps and refactored 'subWordC#'s generic implementation to be branchless. Reviewers: bgamari, simonmar, jrtc27, dfeuer Reviewed By: bgamari, dfeuer Subscribers: dfeuer, thomie, carter Differential Revision: https://phabricator.haskell.org/D4592
View Lorry Controller Status