Ticket #4848: solaris-shared-libs-support.dpatch

Hello,

following three patches add support for shared libs on i386/solaris2
platform. One patch fixes CPP misdetection of Solaris in NCG. I've
used this oportunity to rename OSSolaris to OSSolaris2 to be
consistent with other OSes there. Feel free to stay with OSSolaris if
you prefer, but then the patch will require small
modification. Another provides shared libs support which was really
easy to do thanks to Solaris being ELF system and thanks to GHC ELF
support in place. The last fixes linker errors which we get when we
try to link code compiled with -fno-PIC into shared object on
Solaris. i.e. we cannot use this RTS speed hack on Solaris.

Any chance those patches will be reviewed/incorporated?

Thanks,
Karel

3 patches for repository http://darcs.haskell.org/ghc:

Fri Dec 17 09:45:10 CET 2010  Karel Gardas <karel.gardas@centrum.cz>
  * fix CPP detection of Solaris in NCG

Fri Dec 17 09:46:17 CET 2010  Karel Gardas <karel.gardas@centrum.cz>
  * provide shared libraries support on i386-unknown-solaris2 platform

Fri Dec 17 09:51:33 CET 2010  Karel Gardas <karel.gardas@centrum.cz>
  * do not compile part of shared lib RTS with -fno-PIC on Solaris

New patches:

[fix CPP detection of Solaris in NCG
Karel Gardas <karel.gardas@centrum.cz>**20101217084510
 Ignore-this: 9d1ce59d469294eab1f0cbc697e48d69
] {
hunk ./compiler/nativeGen/Platform.hs 49
        = OSUnknown
        | OSLinux
        | OSDarwin
-       | OSSolaris
+       | OSSolaris2
        | OSMinGW32
        | OSFreeBSD
        | OSOpenBSD
hunk ./compiler/nativeGen/Platform.hs 95
 defaultTargetOS        = OSLinux
 #elif darwin_TARGET_OS
 defaultTargetOS        = OSDarwin
-#elif solaris_TARGET_OS
-defaultTargetOS        = OSSolaris
+#elif solaris2_TARGET_OS
+defaultTargetOS        = OSSolaris2
 #elif mingw32_TARGET_OS
 defaultTargetOS        = OSMinGW32
 #elif freebsd_TARGET_OS
}
[provide shared libraries support on i386-unknown-solaris2 platform
Karel Gardas <karel.gardas@centrum.cz>**20101217084617
 Ignore-this: b6079c6a39a71200a1ee863573e40828
] {
hunk ./compiler/HsVersions.h 20
 #include "ghc_boot_platform.h"
 
 /* This macro indicates that the target OS supports ELF-like shared libraries */
-#if linux_TARGET_OS || freebsd_TARGET_OS || openbsd_TARGET_OS
+#if linux_TARGET_OS || freebsd_TARGET_OS || openbsd_TARGET_OS || solaris2_TARGET_OS
 #define elf_OBJ_FORMAT 1
 #endif
 
hunk ./compiler/nativeGen/Platform.hs 61
 osElfTarget OSLinux   = True
 osElfTarget OSFreeBSD = True
 osElfTarget OSOpenBSD = True
+osElfTarget OSSolaris2 = True
 osElfTarget _         = False
 
 -- | This is the target platform as far as the #ifdefs are concerned.
hunk ./mk/config.mk.in 112
        i386-unknown-freebsd x86_64-unknown-freebsd \
        i386-unknown-openbsd x86_64-unknown-openbsd \
        i386-unknown-mingw32 \
+       i386-unknown-solaris2 \
        i386-apple-darwin powerpc-apple-darwin),YES,NO)
 
 # Build a compiler that will build *unregisterised* libraries and
}
[do not compile part of shared lib RTS with -fno-PIC on Solaris
Karel Gardas <karel.gardas@centrum.cz>**20101217085133
 Ignore-this: 8c8dbb45cac0578a58a3557f1e03c66
] {
hunk ./rts/ghc.mk 312
 # by the default small memory can't be resolved at runtime).  So we
 # only do this on i386.
 #
-# This apparently doesn't work on OS X (Darwin) where we get errors of
-# the form
+# This apparently doesn't work on OS X (Darwin) nor on Solaris.
+# On Darwin we get errors of the form
 #
 #  ld: absolute addressing (perhaps -mdynamic-no-pic) used in _stg_ap_0_fast from rts/dist/build/Apply.dyn_o not allowed in slidable image
 #
hunk ./rts/ghc.mk 321
 #
 #  ld: warning codegen in _stg_ap_pppv_fast (offset 0x0000005E) prevents image from loading in dyld shared cache
 #
+# On Solaris we get errors like:
+#
+# Text relocation remains                         referenced
+#     against symbol                  offset      in file
+# .rodata (section)                   0x11        rts/dist/build/Apply.dyn_o
+#   ...
+# ld: fatal: relocations remain against allocatable but non-writable sections
+# collect2: ld returned 1 exit status
+
+ifeq "$(TargetArch_CPP)" "i386"
+i386_SPEED_HACK := "YES"
+ifeq "$(TargetOS_CPP)" "darwin"
+i386_SPEED_HACK := "NO"
+endif
+ifeq "$(TargetOS_CPP)" "solaris2"
+i386_SPEED_HACK := "NO"
+endif
+endif
+
 ifeq "$(TargetArch_CPP)" "i386"
hunk ./rts/ghc.mk 341
-ifneq "$(TargetOS_CPP)" "darwin"
+ifeq "$(i386_SPEED_HACK)" "YES"
 rts/sm/Evac_HC_OPTS           += -fno-PIC
 rts/sm/Evac_thr_HC_OPTS       += -fno-PIC
 rts/sm/Scav_HC_OPTS           += -fno-PIC
}

Context:

[refactor and tidy up the section on RTS options
Simon Marlow <marlowsd@gmail.com>**20101216123151
 Ignore-this: 9cdafd687351d8a3ff879b64347f85d3
] 
[Related to #4826: Some minor tweaks to the wording of the User Guide, section 4.16
Orphi <MathematicalOrchid@hotmail.com>**20101209170440
 Ignore-this: c3d942d58594be7d4c2eb4dc3a22f19
] 
[FIX #4826 partial: Add -rtsopts and -with-rtsopts to User Guide section 4.11.6
Orphi <MathematicalOrchid@hotmail.com>**20101209165152
 Ignore-this: 2fc1c0abbb783695773ab0f9c013bbaa
] 
[FIX #4826 partially: Change -f to -? in User Guide section F4.16
Orphi <MathematicalOrchid@hotmail.com>**20101209144148
 Ignore-this: 73410b350e80c8943ae722dec8dea44b
] 
[fix #3910
Simon Marlow <marlowsd@gmail.com>**20101216114452
 Ignore-this: 410e95e188344a523520e192a3fb58ea
] 
[remove an optimisation that wasn't
Simon Marlow <marlowsd@gmail.com>**20101215152656
 Ignore-this: e8413f58e8292c6e7463087d885b3a7d
] 
[fix a warning
Simon Marlow <marlowsd@gmail.com>**20101216105723
 Ignore-this: ed6024378021a698ce638267ed3e21ab
] 
[use EXTERN_INLINE instead of STATIC_INLINE to avoid some gcc warnings
Simon Marlow <marlowsd@gmail.com>**20101216105709
 Ignore-this: d4e1586cf318883a8e611b55df7fbf10
] 
[remove dead code
Simon Marlow <marlowsd@gmail.com>**20101216104944
 Ignore-this: 97a04a3e37c1b28abc222a28bab3d17d
] 
[fix retainer profiling: add missing case for TSO
Simon Marlow <marlowsd@gmail.com>**20101216103900
 Ignore-this: 11bda81ac159f638d719c1f6177702fb
] 
[add a missing STACK case
Simon Marlow <marlowsd@gmail.com>**20101216102100
 Ignore-this: ac1c036b5cbf4209b1d10b6ab1c83f27
] 
[Remove code that is dead now that we need >= 6.12 to build
Ian Lynagh <igloo@earth.li>**20101215201006] 
[fix for large stack allocations
Simon Marlow <marlowsd@gmail.com>**20101215152419
 Ignore-this: d9aca17d68bd99214c126989a2318e79
] 
[Implement stack chunks and separate TSO/STACK objects
Simon Marlow <marlowsd@gmail.com>**20101215120843
 Ignore-this: 73fa9460314d4a4e54456af12bef7960
 
 This patch makes two changes to the way stacks are managed:
 
 1. The stack is now stored in a separate object from the TSO.
 
 This means that it is easier to replace the stack object for a thread
 when the stack overflows or underflows; we don't have to leave behind
 the old TSO as an indirection any more.  Consequently, we can remove
 ThreadRelocated and deRefTSO(), which were a pain.
 
 This is obviously the right thing, but the last time I tried to do it
 it made performance worse.  This time I seem to have cracked it.
 
 2. Stacks are now represented as a chain of chunks, rather than
    a single monolithic object.
 
 The big advantage here is that individual chunks are marked clean or
 dirty according to whether they contain pointers to the young
 generation, and the GC can avoid traversing clean stack chunks during
 a young-generation collection.  This means that programs with deep
 stacks will see a big saving in GC overhead when using the default GC
 settings.
 
 A secondary advantage is that there is much less copying involved as
 the stack grows.  Programs that quickly grow a deep stack will see big
 improvements.
 
 In some ways the implementation is simpler, as nothing special needs
 to be done to reclaim stack as the stack shrinks (the GC just recovers
 the dead stack chunks).  On the other hand, we have to manage stack
 underflow between chunks, so there's a new stack frame
 (UNDERFLOW_FRAME), and we now have separate TSO and STACK objects.
 The total amount of code is probably about the same as before.
 
 There are new RTS flags:
 
    -ki<size> Sets the initial thread stack size (default 1k)  Egs: -ki4k -ki2m
    -kc<size> Sets the stack chunk size (default 32k)
    -kb<size> Sets the stack chunk buffer size (default 1k)
 
 -ki was previously called just -k, and the old name is still accepted
 for backwards compatibility.  These new options are documented.
] 
[comments on SRC_HC_OPTS (#4829)
Simon Marlow <marlowsd@gmail.com>**20101214101340
 Ignore-this: e2bdec00f07b68e82837e77a4faf6514
] 
[fix another sanity error, and refactor/tidy up
Simon Marlow <marlowsd@gmail.com>**20101209163919
 Ignore-this: d5ce953ac78e90fc0e22cd9848d26e2e
] 
[Fix a bug in functorLikeTraverse, which was giving wrong answer for tuples
simonpj@microsoft.com**20101215123725
 Ignore-this: 560220e92429b5b1a6197a62f94a4ff2
 
 This bug led to Trac #4816, which is hereby fixed
] 
[Improve printing for -ddump-deriv
simonpj@microsoft.com**20101215121955
 Ignore-this: 3181c948c4c2471bd99b32c5ee487a1e
] 
[Tighten up what it means to be an "enumeration data constructor"
simonpj@microsoft.com**20101215121927
 Ignore-this: 459b3f9f7994a13094ed87b0768b33a8
 
 See Note [Enumeration types] in TyCon, and comments in Trac #4528
] 
[Allow enumerations to have phantom arguments.
simonpj@microsoft.com**20101215121817
 Ignore-this: 32ef8cb869e6e38c2e43b3ae87b1b9a8
 
 The bytecode generator was being too eager.
 Fixes Trac #4528, or rather, a near variant.
] 
[Instance declaration overlap allowed if *either* has -XOverlappingInstances
simonpj@microsoft.com**20101214180500
 Ignore-this: f1b1492541a7e0464ebc6adb45510a2e
 
 This satisfies Trac #3877.  Documentation is changed too.
 I'm not sure if this should go in 7.0.2.
] 
[Fix Trac #4841: behave right with TypeSynonymInstances and NoFlexibleInstances
simonpj@microsoft.com**20101214174755
 Ignore-this: dccd707fdca84904b7885170a296ecb6
 
 When we have TypeSynonymInstances without FlexibleInstances we should still
 insist on a H98-style instance head, after looking through the synonym.
 
 This patch also make FlexibleInstances imply TypeSynonymInstances.  Anything
 else is a bit awkward, and not very useful.
 
] 
[Fix Trac #3731: more superclass subtlety (sigh)
simonpj@microsoft.com**20101214180344
 Ignore-this: f4168e59f3164303ba7be022ba19c37b
 
 I will add more comments, but I want to commit this tonight,
 so the overnight builds get it.
] 
[Less verbose debug print
simonpj@microsoft.com**20101214180248
 Ignore-this: e405e8545763e913155abe43daf7e36c
] 
[Wibble to InstEnv.instanceHead
simonpj@microsoft.com**20101214082939
 Ignore-this: 851db517f8638a0aeb7ad461298f7e9f
 
 Fixes an accidental glitch in T1835
] 
[Remove dead code now that we require the bootstrapping compiler be >= 6.12
Ian Lynagh <igloo@earth.li>**20101214011011] 
[GHC 6.12 is now needed to build the HEAD
Ian Lynagh <igloo@earth.li>**20101214010923] 
[Add libstdc++-4.5.0-1-mingw32-dll-6.tar.lzma to mingw tarballs
Ian Lynagh <igloo@earth.li>**20101213223153] 
[Fix recursive superclasses (again).  Fixes Trac #4809.
simonpj@microsoft.com**20101213171511
 Ignore-this: b91651397918fd8f0183812f9a070073
 
 This patch finally deals with the super-delicate question of
 superclases in possibly-recursive dictionaries.  The key idea
 is the DFun Superclass Invariant (see TcInstDcls):
 
      In the body of a DFun, every superclass argument to the
      returned dictionary is
        either   * one of the arguments of the DFun,
        or       * constant, bound at top level
 
 To establish the invariant, we add new "silent" superclass
 argument(s) to each dfun, so that the dfun does not do superclass
 selection internally.  There's a bit of hoo-ha to make sure that
 we don't print those silent arguments in error messages; a knock
 on effect was a change in interface-file format.
 
 A second change is that instead of the complex and fragile
 "self dictionary binding" in TcInstDcls and TcClassDcl,
 using the same mechanism for existential pattern bindings.
 See Note [Subtle interaction of recursion and overlap] in TcInstDcls
 and Note [Binding when looking up instances] in InstEnv.
 
 Main notes are here:
 
   * Note [Silent Superclass Arguments] in TcInstDcls,
     including the DFun Superclass Invariant
 
 Main code changes are:
 
   * The code for MkId.mkDictFunId and mkDictFunTy
 
   * DFunUnfoldings get a little more complicated;
     their arguments are a new type DFunArg (in CoreSyn)
 
   * No "self" argument in tcInstanceMethod
   * No special tcSimplifySuperClasss
   * No "dependents" argument to EvDFunApp
 
 IMPORTANT
    It turns out that it's quite tricky to generate the right
    DFunUnfolding for a specialised dfun, when you use SPECIALISE
    INSTANCE.  For now I've just commented it out (in DsBinds) but
    that'll lose some optimisation, and I need to get back to
    this.
] 
[Doing the smart canonicalization only if we are not simplifying a Rule LHS.
dimitris@microsoft.com**20101210132221
 Also, same thing now applies for adding superclasses.
 
] 
[Moved canonicalisation inside solveInteract
dimitris@microsoft.com**20101209141215
 
 Moreover canonicalisation now is "clever", i.e. it never canonicalizes a class 
 constraint if it can already discharge it from some other inert or previously
 encountered constraints. See Note [Avoiding the superclass explosion]
 
] 
[GHCi linker: Assume non-Haskell libraries are dynamic libs
Ian Lynagh <igloo@earth.li>**20101213124930
 Ignore-this: aa153a8f6e309c7b3dae7e46bb7a9583
 This works around a segfault we get when trying to load libiconv.a on
 some platforms.
] 
[Add --version support to ghc-cabal
Ian Lynagh <igloo@earth.li>**20101212213600
 Ignore-this: ef696dcb1b96a23765f9f18e75a56f5
] 
[Don't link the GHC RTS into our C-only programs
Ian Lynagh <igloo@earth.li>**20101210185402
 Ignore-this: 56f620f7eb16a03e7497a161bc48458e
] 
[Build a copy of ghc-cabal with the in-tree compiler, for the bindist
Ian Lynagh <igloo@earth.li>**20101210181123] 
[Add a test that all programs in the bindist were built with the right GHC
Ian Lynagh <igloo@earth.li>**20101210161218
 They should use the GHC from the build tree, not the bootstrapping compiler.
] 
[Fix Trac #4534: renamer bug
simonpj@microsoft.com**20101210084530
 Ignore-this: 8163bfa3a56344cfe89ad17c62e9655d
   
 The renamer wasn't attaching the right used-variables to a
 TransformStmt constructor.
 
 The real modification is in RnExpr; the rest is just
 pretty-printing and white space.
] 
[White space only
simonpj@microsoft.com**20101210084255
 Ignore-this: 3fcf8a4fc8c15052c379a135951d53ea
] 
[Comments only
simonpj@microsoft.com**20101210084116
 Ignore-this: 55bb1de129b1c9513751885eaa84b884
] 
[Make the case-to-let transformation a little less eager
simonpj@microsoft.com**20101208172251
 Ignore-this: 55eaa1b5753af31aeb32ec792cb6b662
 
 See Note [Case elimination: lifted case].
 Thanks to Roman for identifying this case.
] 
[warning fix: don't redefine BLOCKS_PER_MBLOCK
Simon Marlow <marlowsd@gmail.com>**20101210094002
 Ignore-this: cadba57f1c38f5e2af1de37d0a79c7ee
] 
[Only reset the event log if logging is turned on (addendum to #4512)
Simon Marlow <marlowsd@gmail.com>**20101210093951
 Ignore-this: c9f85f0de2b11a37337672fba59aecc6
] 
[allocate enough room for the longer filename (addendum to #4512)
Simon Marlow <marlowsd@gmail.com>**20101210093906
 Ignore-this: 270dc0219d98f1e0f9e006102ade7087
] 
[Fix Windows build: move rtsTimerSignal to the POSIX-only section
Simon Marlow <marlowsd@gmail.com>**20101210090045
 Ignore-this: aa1844b70b9f1a44447787c4bbe10d44
] 
[Default the value of -dppr-cols when the static flags aren't initialised yet
Ben Lippmeier <benl@ouroborus.net>**20101210060154
 Ignore-this: 4cea29085ef904f379a8829714c9e180
 If GHC's command line options are bad then the options parser uses the
 pretty printer before the -dppr-cols flag has been read.
] 
[Defensify naked read in LLVM mangler
Ben Lippmeier <benl@ouroborus.net>**20101210045922
 Ignore-this: 1373f597863851bd03e7a7254558ed04
] 
[Formatting only
Ben Lippmeier <benl@ouroborus.net>**20101210042600
 Ignore-this: 20bbcd95c70b59094d0bb8a63e459103
] 
[Always ppr case alts on separate lines
Ben Lippmeier <benl@ouroborus.net>**20101208070508
 Ignore-this: 7e2edd57a61427621aeb254aef84f0f7
] 
[Add -dppr-colsN to set width of dumps
Ben Lippmeier <benl@ouroborus.net>**20101208070245
 Ignore-this: edc64fee6c373b895bb80b83b549ce1a
] 
[Add -dppr-case-as-let to print "strict lets" as actual lets
Ben Lippmeier <benl@ouroborus.net>**20101208065548
 Ignore-this: eb1d122dbd73b5263cae3a9f8259a838
] 
[Suppress more info with -dsuppress-idinfo
Ben Lippmeier <benl@ouroborus.net>**20101208063037
 Ignore-this: 5e8213d7b8d2905e245917aa3e83efc5
] 
[Implement -dsuppress-type-signatures
Ben Lippmeier <benl@ouroborus.net>**20101208062814
 Ignore-this: 34dbefe5f8d7fe58ecb26d6a748d1c71
] 
[Add more suppression flags
Ben Lippmeier <benl@ouroborus.net>**20101208020723
 Ignore-this: b010ba9789a2fde6b815f33494fcc23c
  -dsuppress-all
  -dsuppress-type-applications
  -dsuppress-idinfo
] 
[fix ticket number (#4505)
Simon Marlow <marlowsd@gmail.com>**20101209120404
 Ignore-this: 5769c5ce2a8d69d62d977a9ae138ec23
] 
[fix warnings
Simon Marlow <marlowsd@gmail.com>**20101209115844
 Ignore-this: ffff37feb2abbfc5bd12940c7007c208
] 
[Catch too-large allocations and emit an error message (#4505)
Simon Marlow <marlowsd@gmail.com>**20101209114005
 Ignore-this: c9013ab63dd0bd62ea045060528550c6
 
 This is a temporary measure until we fix the bug properly (which is
 somewhat tricky, and we think might be easier in the new code
 generator).
 
 For now we get:
 
 ghc-stage2: sorry! (unimplemented feature or known bug)
   (GHC version 7.1 for i386-unknown-linux):
         Trying to allocate more than 1040384 bytes.
 
 See: http://hackage.haskell.org/trac/ghc/ticket/4550
 Suggestion: read data from a file instead of having large static data
 structures in the code.
] 
[Export the value of the signal used by scheduler (#4504)
Dmitry Astapov <dastapov@gmail.com>**20101208183755
 Ignore-this: 427bf8c2469283fc7a6f759440d07d87
] 
[Tweak the "sorry" message a bit
Simon Marlow <marlowsd@gmail.com>**20101208163212
 Ignore-this: aa1ce5bc3c27111548204b740572efbe
 
 -              "sorry! (this is work in progress)\n"
 +              "sorry! (unimplemented feature or known bug)\n"
] 
[:unset settings support
Boris Lykah <lykahb@gmail.com>**20101123190132
 Ignore-this: 5e97c99238f5d2394592858c34c004d
 Added support for settings [args, prog, prompt, editor and stop].
 Now :unset supports the same set of options as :set.
] 
[Fix Windows memory freeing: add a check for fb == NULL; fixes trac #4506
Ian Lynagh <igloo@earth.li>**20101208152349
 Also added a few comments, and a load of code got indented 1 level deeper.
] 
[Fixes for #4512: EventLog.c - provides ability to terminate event logging, Schedule.c - uses them in forkProcess.
Dmitry Astapov <dastapov@gmail.com>**20101203133950
 Ignore-this: 2da7f215d6c22708a18291a416ba8881
] 
[Make CPPFLAGS variables, as well as CFLAGS and LDFLAGS
Ian Lynagh <igloo@earth.li>**20101207010033
 Ignore-this: 2fc1ca1422aae1988d0fe1d29a8485d9
 This fixes the "does unsetenv return void" test in the unix package on
 OS X, if I tell it to make 10.4-compatible binaries. The test uses
 CPPFLAGS but not CFLAGS, so it thought it returned int (as it was
 in 10.5-mode), but the C compiler (using CFLAGS, so in 10.4 mode)
 thought it returned void.
 
 I also added CONF_LD_OPTS_STAGE$3 to the list of things in LDFLAGS,
 which looks like an accidental ommission.
] 
[Add a configure message
Ian Lynagh <igloo@earth.li>**20101206215201] 
[Link even programs containing no Haskell modules with GHC
Ian Lynagh <igloo@earth.li>**20101206203329
 I don't remember why we made it use gcc instead, but going back to
 using ghc doesn't seem to break anything, and should fix the build
 on OS X 10.6.
] 
[Correct the stage that the includes/ tools are built in
Ian Lynagh <igloo@earth.li>**20101206203125] 
[Tweak the cleaning of inplace/; fixes trac #4320
Ian Lynagh <igloo@earth.li>**20101205212048] 
[Close .ghci files after reading them; fixes trac #4487
Ian Lynagh <igloo@earth.li>**20101205205301] 
[Fix the behaviour of :history for ticks surrounding top level functions
pepeiborra@gmail.com**20101203202346
 Ignore-this: 8059d4859c52c0c9a235b937cb8cde1d
] 
[Don't warn of duplicate exports in case of module exports.
Michal Terepeta <michal.terepeta@gmail.com>**20101127212116
 Ignore-this: ea225d517826f971c400bbb68d1405b8
 
 But only when the module exports refer to different modules.
 See ticket #4478.
] 
[Fix whitespace/layout in RnNames.
Michal Terepeta <michal.terepeta@gmail.com>**20101030171303
 Ignore-this: 707a7955fc4fc51683cc5a1dfe57f93
] 
[Tell gcc to support back to OS X 10.5
Ian Lynagh <igloo@earth.li>**20101203201558
 Ignore-this: f02d70e5b9cce50137981c6cb2b62a18
] 
[Make RelaxedLayout off by default
Ian Lynagh <igloo@earth.li>**20101202140808
 I suspect this is a vary rarely used extension to the official layout
 rule.
] 
[Fix up TcInstDcls
simonpj@microsoft.com**20101203180758
 Ignore-this: 9311aeb4ee67c799704afec90b5982d0
 
 I really don't know how this module got left out of my last
 patch, namely
   Thu Dec  2 12:35:47 GMT 2010  simonpj@microsoft.com
   * Re-jig simplifySuperClass (again)
 
 I suggest you don't pull either the patch above, or this
 one, unless you really have to.  I'm not fully confident
 that it works properly yet.  Ran out of time. Sigh.
] 
[throwTo: report the why_blocked value in the barf()
Simon Marlow <marlowsd@gmail.com>**20101203094840
 Ignore-this: 3b167c581be1c51dfe3586cc6359e1d0
] 
[handle ThreadMigrating in throwTo() (#4811)
Simon Marlow <marlowsd@gmail.com>**20101203094818
 Ignore-this: 8ef8cb7fd3b50a27f83c29968131d461
 If a throwTo targets a thread that has just been created with
 forkOnIO, then it is possible the exception strikes while the thread
 is still in the process of migrating.  throwTo() didn't handle this
 case, but it's fairly straightforward.
] 
[removeThreadFromQueue: stub out the link field before returning (#4813)
Simon Marlow <marlowsd@gmail.com>**20101202160838
 Ignore-this: 653ae17bc1120d7f4130da94665002a1
] 
[small tidyup
Simon Marlow <marlowsd@gmail.com>**20101126140620
 Ignore-this: 70b1d5ed4c81a7b29dd5980a2d84aae1
] 
[Fix a recomp bug: make classes/datatypes depend directly on DFuns (#4469)
Simon Marlow <marlowsd@gmail.com>**20101202122349
 Ignore-this: 61c765583bb1d97caa88cf9b4f45b87c
 And remove the old mechanism of recording dfun uses separately,
 because it didn't work.
 
 This wiki page describes recompilation avoidance and fingerprinting.
 I'll update it to describe the new method and what went wrong with the
 old method:
 
 http://hackage.haskell.org/trac/ghc/wiki/Commentary/Compiler/RecompilationAvoidance
] 
[make a panic message more informative and suggest -dcore-lint (see #4534)
Simon Marlow <marlowsd@gmail.com>**20101201151706
 Ignore-this: 2a10761925d6f9f52675948baa30f7a
] 
[Re-jig simplifySuperClass (again)
simonpj@microsoft.com**20101202123547
 Ignore-this: fe4062b8988258f6748ebd8fbd6515b5
 
 This fixes the current loop in T3731, and will fix other
 reported loops.  The loops show up when we are generating
 evidence for superclasses in an instance declaration.
 
 The trick is to make the "self" dictionary simplifySuperClass
 depend *explicitly* on the superclass we are currently trying
 to build.  See Note [Dependencies in self dictionaries] in TcSimplify.
 
 That in turn means that EvDFunApp needs a dependency-list, used
 when chasing dependencies in isGoodRecEv.
] 
[A little refactoring (remove redundant argument passed to isGoodRecEv)
simonpj@microsoft.com**20101202123110
 Ignore-this: e517c5c12109a230f08dafb4d1e386df
] 
[Make rebindable if-then-else a little more permissive
simonpj@microsoft.com**20101202122540
 Ignore-this: ddb552cfe307607b42d1e4baf4e3bf21
 
 See Note [Rebindable syntax for if].  Fixes Trac #4798.
 Thanks to Nils Schweinsberg <mail@n-sch.de>
] 
[Improve error message (Trac #4799)
simonpj@microsoft.com**20101202102706
 Ignore-this: d9896e4d182936de1f256c820b96a8cf
] 
[Fix a nasty bug in RULE matching: Trac #4814
simonpj@microsoft.com**20101202102618
 Ignore-this: ba058ad46a02bd2faf3a14de93fd19c6
 
 See Note [Matching lets], which explains it all in detail.
 It took me a day to think of a nice way to fix the bug,
 but I think the result is quite respectable. Subtle, though.
] 
[Rename -XPArr to -XParallelArrays
Ben Lippmeier <benl@ouroborus.net>**20101130075415
 Ignore-this: 21b37680a7f25800d1200b59ad0b6b39
] 
[FIX #1845 (unconditional relative branch out of range)
pho@cielonegro.org**20101130143014
 Ignore-this: df234bd8ad937104c455656fe3c33732
 
 Don't use mmap on powerpc-apple-darwin as mmap doesn't support
 reallocating but we need to allocate jump islands just after each
 object images. Otherwise relative branches to jump islands can fail
 due to 24-bits displacement overflow.
] 
[rts/Linker.c (loadArchive):
pho@cielonegro.org**20101130142700
 Ignore-this: bc84f9369ce5c2d289440701b7a3a2ab
 
 This routine should be aware of Mach-O misalignment of malloc'ed memory regions.
] 
[rts/Linker.c (machoGetMisalignment):
pho@cielonegro.org**20101130123355
 Ignore-this: 75425600049efd587e9873578e26392f
 
 Use fseek(3) instead of rewind(3) to move the file position indicator back to the initial position. Otherwise we can't use this function in loadArchive().
] 
[rts/Linker.c (ocFlushInstructionCache):
pho@cielonegro.org**20101130121425
 Ignore-this: 1e2c207e4b1d17387617ec5d645204b7
 
 I found this function causes a segfault when ocAllocateSymbolExtras() has allocated a separate memory region for jump islands.
] 
[Remove NewQualifiedOperators
Ian Lynagh <igloo@earth.li>**20101201181117
 The extension was rejected by Haskell', and deprecated in 7.0.
] 
[fix ref to utils/ext-core, which moved to Hackage (extcore package)
Simon Marlow <marlowsd@gmail.com>**20101201092147
 Ignore-this: 272a7daaa335ef60bcc645db70b4d68b
] 
[fix floating-point/FFI section: fenv is C99, not POSIX
Simon Marlow <marlowsd@gmail.com>**20101201092119
 Ignore-this: ce8b3edd428e4f77691dd739b5b4ae73
] 
[Fixed some 'unused vars' warnings
keller@cse.unsw.edu.au**20101130013425
 Ignore-this: 35790d443faa23b87e4ba442e62376a3
] 
[vectScalarLam handles int, float, and double now
keller@cse.unsw.edu.au**20101129231043
 Ignore-this: 6d67bdc8dd8577184040e791e6f3d0
] 
[Handling of lets, letrec and case when checking if a lambda expr needs to be vectorised
keller@cse.unsw.edu.au**20101115051225
 Ignore-this: 1db6ed63d7b3f6d093e019322b407ff7
] 
[Document the behaviour of fenv.h functions with GHC (#4391)
Simon Marlow <marlowsd@gmail.com>**20101126125336
 Ignore-this: bc4eab49428d567505a28add6fed90f1
] 
[Remove the no-ghci-lib warning in ghc-pkg
Ian Lynagh <igloo@earth.li>**20101127235805
 GHCi libs are no longer necessary, as we can use the .a or .so versions
 instead.
] 
[Add GNU-variant support to the .a parser, and other improvements/tidyups
Ian Lynagh <igloo@earth.li>**20101127223945] 
[Re-indent only
Ian Lynagh <igloo@earth.li>**20101127191646] 
[Improve linker debugging for archive files
Ian Lynagh <igloo@earth.li>**20101127190907] 
[Always enable the archive-loading code
Ian Lynagh <igloo@earth.li>**20101127173000
 If the GHCi .o lib doesn't exist, load the .a instead
] 
[Inherit the ForceSpecConstr flag in non-recursive nested bindings
Roman Leshchinskiy <rl@cse.unsw.edu.au>**20101127125025
 Ignore-this: 401391eae25cefcb4afaba2e357decc1
 
 This makes sure that join points are fully specialised in loops which are
 marked as ForceSpecConstr.
] 
[Document -ddump-rule-firings and -ddump-rule-rewrites
Roman Leshchinskiy <rl@cse.unsw.edu.au>**20101127123528
 Ignore-this: beade2efe0cd767c0ce9d4f45a3380ba
] 
[New flag -dddump-rule-rewrites
Roman Leshchinskiy <rl@cse.unsw.edu.au>**20101127122022
 Ignore-this: c0ef5b8a199fbd1ef020258d2cde85a3
 
 Now, -ddump-rule-firings only shows the names of the rules that fired (it would
 show "before" and "after" with -dverbose-core2core previously) and
 -ddump-rule-rewrites always shows the "before" and "after" bits, even without
 -dverbose-core2core.
] 
[Acutally, wild-card variables *can* have occurrences
simonpj@microsoft.com**20101126162409
 Ignore-this: 544bffed75eeccef03a1097f98524eea
 
 This patch removes the Lint test, and comments why
] 
[Tidy up the handling of wild-card binders, and make Lint check it
simonpj@microsoft.com**20101126133210
 Ignore-this: 9e0be9f7867d53046ee5b0e478a0f433
 
 See Note [WildCard binders] in SimplEnv.  Spotted by Roman.
] 
[Substitution should just substitute, not optimise
simonpj@microsoft.com**20101125172356
 Ignore-this: 657628d9b6796ceb5f915c43d56e4a06
 
 This was causing Trac #4524, by optimising
      (e |> co)  to   e
 on the LHS of a rule.  Result, the template variable
 'co' wasn't bound any more.
 
 Now that substition doesn't optimise, it seems sensible to call
 simpleOptExpr rather than substExpr when substituting in the
 RHS of rules.  Not a big deal either way.
] 
[Make SpecConstr "look through" identity coercions
simonpj@microsoft.com**20101125172138
 Ignore-this: c1cc585ed890a7702c33987e971e0af6
] 
[Comment only
simonpj@microsoft.com**20101125172011
 Ignore-this: 3c7be8791badd00dcca9610ebb8981d1
] 
[White space only
simonpj@microsoft.com**20101101080748
 Ignore-this: f7133fc6b22ae263c6672543a8534a6f
] 
[Keep a maximum of 6 spare worker threads per Capability (#4262)
Simon Marlow <marlowsd@gmail.com>**20101125135729
 Ignore-this: a020786569656bf2f3a1717b65d463bd
] 
[Unicide OtherNumber category should be allowed in identifiers (#4373)
Simon Marlow <marlowsd@gmail.com>**20101115095444
 Ignore-this: e331b6ddb17550163ee91bd283348800
] 
[vectoriser: fix warning
Ben Lippmeier <benl@ouroborus.net>**20101126044036
 Ignore-this: e1a66bb405bf2f3f56b42c3b13fd4bf3
] 
[vectoriser: fix warning
Ben Lippmeier <benl@ouroborus.net>**20101126042950
 Ignore-this: df8dd25bcfb3946c2974b13953a2f2c7
] 
[vectoriser: take class directly from the instance tycon
Ben Lippmeier <benl@ouroborus.net>**20101126042900
 Ignore-this: 626a416717a5a059f39e53f4ec95fc66
] 
[vectoriser: comments only
Ben Lippmeier <benl@ouroborus.net>**20101125073201
 Ignore-this: 8846ea8895307083bd1ebbc5d7fb1c5
] 
[vectoriser: follow changes in mkClass
Ben Lippmeier <benl@ouroborus.net>**20101125062349
 Ignore-this: d5018cc022686d4272e126ca9a12283a
] 
[vectoriser: tracing wibbles
Ben Lippmeier <benl@ouroborus.net>**20101125062332
 Ignore-this: c2024d8f03bc03bee2851f4f1c139fd5
] 
[mkDFunUnfolding wants the type of the dfun to be a PredTy
benl@ouroborus.net**20100914062939
 Ignore-this: 7aa6e6b140746184cf00355b50c83b66
] 
[vectoriser: fix conflicts
Ben Lippmeier <benl@ouroborus.net>**20101125060904
 Ignore-this: cc3decab1affada8629ca3818b76b3bf
] 
[Comments and formatting only
benl@ouroborus.net**20100914062903
 Ignore-this: b0fc25f0952cafd56cc25353936327d4
] 
[Comments and formatting to type environment vectoriser
benl@ouroborus.net**20100909080405
 Ignore-this: ab8549d53f845c9d82ed9a525fda3906
] 
[Don't mix implicit and explicit layout
Ian Lynagh <igloo@earth.li>**20101124231514] 
[Whitespace only
Ian Lynagh <igloo@earth.li>**20101124230655] 
[Separate NondecreasingIndentation out into its own extension
Ian Lynagh <igloo@earth.li>**20101124220507] 
[Add another GHC layout rule relaxation to RelaxedLayout
Ian Lynagh <igloo@earth.li>**20101124205957] 
[Remove an unused build system variable: GhcDir
Ian Lynagh <igloo@earth.li>**20101124140455] 
[Remove unused build system variable: GhcHasEditline
Ian Lynagh <igloo@earth.li>**20101124140415] 
[Remove unused variables from the build system: HBC, NHC, MKDEPENDHS
Ian Lynagh <igloo@earth.li>**20101124140052] 
[Remove references to Haskell 98
Ian Lynagh <igloo@earth.li>**20101123233536
 They are no longer right, as we have Haskell' generating new Haskell
 standards.
] 
[Tweak a configure test
Ian Lynagh <igloo@earth.li>**20101123170621] 
[Add a configure test for the visibility hidden attribute
Ian Lynagh <igloo@earth.li>**20101123170541] 
[sanity: fix places where we weren't filling fresh memory with 0xaa
Simon Marlow <marlowsd@gmail.com>**20101029092843
 Ignore-this: 2cb18f7f5afcaf33371aeffce67e218f
] 
[Just some alpha renaming
Ian Lynagh <igloo@earth.li>**20101121144455
 Ignore-this: d5e807c5470840efc199e29f7d50804c
] 
[Fix bug #3165 (:history throws irrefutable pattern failed)
pepeiborra@gmail.com**20101115223623
 Ignore-this: 73edf56e502b4d0385bc044133b27946
 
 I ran across this bug and took the time to fix it, closing
 a long time due TODO in InteractiveEval.hs
 
 Instead of looking around to find the enclosing declaration
 of a tick, this patch makes use of the information already collected during the
 coverage desugaring phase
] 
[For bindists, build ghc-pwd with stage 1
Ian Lynagh <igloo@earth.li>**20101121183520
 Ignore-this: a3b5c8b78c81ec1b6d5fbf23da346ff5
 rather then the bootstrapping compiler. This fixes problems where the
 bootstrapping compiler dynamically links against libraries not on the
 target machine.
] 
[Makefile tweak
Ian Lynagh <igloo@earth.li>**20101121183342
 Ignore-this: cd55a2819c1a5fd36da1bc7a75d2ded1
] 
[Fix a makefile include ordering sanity check
Ian Lynagh <igloo@earth.li>**20101121174916
 Ignore-this: d0bdd41c4b618944d04ecb4f54fdd0f1
] 
[Add an extension for GHC's layout-rule relaxations
Ian Lynagh <igloo@earth.li>**20101120215340
 Still TODO: Add the other relaxation (#1060) and update the alternative
 layout rule to use the extension.
] 
[Tweak the bindist configure.ac.in
Ian Lynagh <igloo@earth.li>**20101120173735] 
[configure.ac tweaks
Ian Lynagh <igloo@earth.li>**20101120170245] 
[When testing the bindist, tell it where gcc is
Ian Lynagh <igloo@earth.li>**20101120155920
 The location isn't baked into the bindist, as it may differ from
 machine to machine.
] 
[Comments only
simonpj@microsoft.com**20101119100153
 Ignore-this: 7abd5d965ea805770449d6f8dadbb921
] 
[ForceSpecConstr now forces specialisation even for arguments which aren't scrutinised
Roman Leshchinskiy <rl@cse.unsw.edu.au>**20101118212839
 Ignore-this: db45721d29a694e53746f8b76513efa4
] 
[Move the superclass generation to the canonicaliser
simonpj@microsoft.com**20101118120533
 Ignore-this: 5e0e525402a240b709f2b8104c1682b2
 
 Doing superclass generation in the canonicaliser (rather than
 TcInteract) uses less code, and is generally more efficient.
 
 See Note [Adding superclasses] in TcCanonical.
 
 Fixes Trac #4497.
] 
[Fix the generation of in-scope variables for IfaceLint check
simonpj@microsoft.com**20101118090057
 Ignore-this: bbcdba61ddf89d07fe69ca99c2017e3f
] 
[Comments only
simonpj@microsoft.com**20101118090034
 Ignore-this: fa2936d35a0f7be4e4535ea9e2b7bf7b
] 
[Omit bogus test for -XDeriveFunctor
simonpj@microsoft.com**20101118090028
 Ignore-this: a534243011809ebbb788b910961601c5
 
 It was duplicated in the case of 'deriving( Functor )'
 and wrong for 'deriving( Foldable )'
] 
[Improve error message on advice from a user
simonpj@microsoft.com**20101118085306
 Ignore-this: bd4f3858ff24e602e985288f27d536f3
 
 See Trac #4499
] 
[TAG 2010-11-18
Ian Lynagh <igloo@earth.li>**20101118011554
 Ignore-this: ccadbe7fadd1148d2ee3caa8c8821ec5
] 
Patch bundle hash:
49f9e9a49dea515dca76dc9a7bbbe5ac1d6f6ee7