Sparse 0.6.0 released

From Sparse Wiki
Revision as of 09:30, 26 December 2018 by Luc Van Oostenryck (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

Sparse v0.6.0 is now out.

The source code can be found at its usual repository:

 git:// v0.6.0

The tarballs are found at:

Many thanks to people who have contributed to the 888 non-merge patches of this release:

 Ramsay Jones, Randy Dunlap, Uwe Kleine-König, Joey Pabalinas,
 John Levon, Ben Dooks, Jann Horn, Logan Gunthorpe, Vincenzo
 Frascino and Tycho Andersen.

Special thanks to Ramsay Jones who has reviewed numerous of my patches, tested many of my series and found many of my typos.

Best wishes for 2019 -- Luc Van Oostenryck

The changes since the pre-release (v0.6.0-rc1) are:

  • add -Wbitwise-pointer to warn on casts to/from bitwise pointers
  • beautify how types are displayed:
  • no more <noident>
  • no more redundant type specifiers
  • remove double space
  • some cleanup of the build system:
  • only need includedir from llvm-config
  • check if sparse-llvm needs libc++
  • remove -finline-functions from CFLAGS
  • some random cleanup:
  • remove unneeded declarations in "compat.h"
  • remove unused arg in add_branch()
  • allocate BBs after the guards
  • remove redundant check of _Bool bitsize
  • remove unused regno()
  • remove self-assignment of base_type
  • some documentation:
  • update comment for struct-expression::cast_expression
  • fix list formatting in inline doc
  • document that identifiers are now OK for address spaces
  • add TODO list.

The main changes since the previous release (v0.5.2) are:

  • by default, disable warnings about unknown attributes
  • no more warnings about sizeof(void) unless -Wpointer-arith is given
  • add support for __has_attribute() & __has_builtin()
  • many fixes for type evaluation/checking
  • the build should be more friendly for distros
  • a huge number of testcases have been added to the testsuite
  • the handling of cast instructions has been completely revamped
  • the SSA conversion has been is now corrected and has been rewritten with a variant of the classical algorithm
  • documentation is now handled with Sphinx and inline doc is extracted from the code.
  • online documentation can be found at

A more complete list of changes is:

  • add predefined macros for __INTMAX_TYPE__, __INT_MAX__, ...
  • prepare to identify & display the address spaces by name
  • fix linearization of non-constant switch-cases
  • manpage: update maintainer info
  • manpage: add AUTHORS section
  • fixes for -dD
  • add support for -dM
  • remove more complex phi-nodes
  • fix linearization/SSA when missing a return
  • fix linearization/SSA of (nested) logical expressions
  • fix linearization of unreachable switch + label
  • add support for __has_attribute()
  • consolidate instruction's properties into an opcode table
  • fix: do not optimize away accesses to volatile bitfields
  • support mode(__pointer__) and mode(__byte__)
  • do 'classical' SSA conversion (via the iterated dominance frontier).
  • fix buggy recursion in kill_dead_stores()
  • kill dead stores again after memops simplification is done.
  • simplify TRUNC((x & M') | y, N)
  • simplify AND(SHIFT(a | b, S), M)
  • simplify TRUNC(SHIFT(a | b, S), N)
  • add simplification of TRUNC(TRUNC((x))
  • add simplification of SHL(LSR(x), S), S)
  • generate integer-wide OP_ADD & OP_SUB in linearize_inc_dec()
  • simplify mask instructions & bitfield accesses
  • fix a few bugs related to the linearization of logical expressions
  • simplify those logical expressions.
  • add optimized version of some list operations
  • some simplifications of OP_SETNE & OP_SETEQ with 0 and 1
  • several simplifications involving casts and/or bitfields
  • give a correct & sensible warning on negative or over-sized shifts.
  • add conservative simplification of such shifts.
  • do not optimize the meaningless shift:
    • any shift with a negative count
    • OP_ASRs with an over-sized shift count.
  • try to give a correct negative/too-big error message during simplification.
  • simplify chains of shifts.
  • simplify ZEXT + ASR into ZEXT + LSR
  • cse: try the next pair instead of keeping the first instruction
  • cse: compare casts only by kind a size, not by C type.
  • optimize away OP_UTPTR & OP_PTRTU which are nops.
  • cleanup of list walking macros:
    • make untagged pointers the normal case
    • use structurally equivalent struct for all pointer lists to avoid needless casting to and fro struct ptrlist
    • simplify PREPARE/NEXT/RESET logic by using common PTR_NEXT()
  • add validation of the IR.
  • improve expansion of builtin dynamic macros (__FILE__, ...)
  • add support for __INCLUDE_LEVEL__ & __BASE_FILE__
  • mprove generation of predefined macros
  • add support for builtins doing overflow checking.
  • add support for the __has_builtin() macro.
  • improve Sphinx doc for IR instructions
    • have those instructions in the index
    • have a nicer presentation of the generated doc thanks to not having to use level-4 headings anymore
  • fixes & improvements to the testsuite; mainly:
    • allow to run the testsuite on all the tests of a subdir
    • teach 'format' to directly append to the testcase
    • validate the 'check-...' tags


Ben Dooks (1):

     tokenize: check show_string() for NULL pointer

Jann Horn (1):

     fix accesses through incorrect union members

Joey Pabalinas (2):

     doc: copy-edit text related to applying sizeof to a _Bool
     sparse: add -Wpointer-arith flag to toggle sizeof(void) warnings

John Levon (2):

     Ignore #ident directives
     Conditionalize 'warning: non-ANSI function ...'

Logan Gunthorpe (1):

     add __builtin functions for isinf_sign, isfinite and isnan

Luc Van Oostenryck (857):

     use long for all mem stats
     diet: use smaller LIST_NODE_NR (29 -> 13)
     diet: remove unused struct scope::token
     diet: remove unused struct symbol::arg_count
     option: add helper to parse/match command line options
     option: rename 'struct warning' to 'struct flag'
     option: let handle_simple_switch() handle an array of flags
     option: extract OPTION_NUMERIC() from handle_switch_fmemcpy_max_count()
     option: add support for options with 'zero is infinity'
     option: add support for '-<some-option>=unlimited'
     option: use OPTION_NUMERIC() for handle_switch_fmemcpy_max_count()
     option: constify match_option()
     option: handle switches by table
     dump-ir: add defines for the compilation passes
     testsuite: 'echo -n' may not be interpreted as '-n'
     testsuite: allow to test a few cases at once
     testsuite: move verbose() & error()
     testsuite: better message for pattern nbr checking
     testsuite: better message for pattern absence/presence
     use shorter name for constexpr tests
     testsuite: new eq/min/max syntax for pattern checking
     testsuite: obsolete old pattern checking syntax
     testsuite: convert to the new patern syntax
     use a specific struct for asm operands
     fix: missing evaluate with '-include' : add testcase
     fix: missing evaluate with '-include'
     fix test case kill-phi-ttsb
     add test case for incomplete type
     add test case for bad return type
     diet: remove unused struct symbol::value
     cclass: char is wide enough
     cclass: cleanup
     remove prototype extern int is_ptr_type()
     remove prototype for unexistant examine_simple_symbol_type()
     graph: do not scan removed instructions
     build: fix effectiveness of generated dependencies
     build: remove unused support for pkgconfig
     cgcc: teach cgcc about freebsd & netbsd
     testsuite: clearer result summary
     testsuite: check error messages first
     testsuite: saner handling of 'must_fail'
     testsuite: allow to parse several options
     testsuite: add support for -q|--quiet
     testsuite: add support for -a|--abort
     testsuite: get options from env too
     testsuite: allow --format & --single
     testsuite: remove useless selftest
     testsuite: remove useless test-be.c
     testsuite: extract disable()
     testsuite: simplify documentation
     testsuite: allow arch-specific tests
     testsuite: save screen real estate
     testsuite: add a blank line before format
     testsuite: 'quiet' must be initialized earlier
     testsuite: move up arg_file()
     testsuite: make do_format() more self-contained
     testsuite: format: saner defaults handling
     testsuite: format: strip .c from default name
     testsuite: add support for 'format -f'
     testsuite: add support for 'format -l'
     remove never-used MOD_TYPEDEF
     MOD_ACCESSED is not a type modifier ...
     reorganize the definition of the modifiers
     remove redundancy in MOD_STORAGE
     associate MOD_RESTRICT with restrict-qualified variables
     add support for C11's _Atomic as type qualifier
     build: use '-objs' instead of '_EXTRA_DEPS'
     build: use '-ldlibs' instead of '_EXTRA_OBJS'
     build: allow target-specific CFLAGS, CPPFLAGS, LDFLAGS & LDLIBS
     build: allow CFLAGS & friends from command line
     build: avoid rule-specific CFLAGS
     build: use $LIBS directly in the dependency list
     build: no need to use wildcards for generated dependencies
     build: reuse rule for ALL_OBJS
     build: CHECKER_FLAGS=-Wno-vla for all targets
     build: move tests near their use
     build: add note about overwritable vars
     build: remove references to unexisting pre-process.h
     build: move clean & clean-check together
     build: make clean targets quieter
     build: remove rule for shared lib, it's unused
     build: normalize rules
     build: remove the dist rule since unused
     build: use one line per item
     build: allow the name '' to be configurable via the environment
     build: use standard rules for install
     build: remove unused QUIET_INST_SH
     build: let quiet commands use less indentation
     build: simplify quiet commands
     build: simplify clean pattern
     build: add *.o to clean-check pattern
     build: avoid foreach
     build: reorg & add comment
     build: use a single space before assignments
     build: add rule to run a single test
     build: let -fno-strict-aliasing be a mandatory flag
     volatile loads are side-effects too
     fix 'simplification' of float-to-int casts
     fix description setval & symaddr
     flush stdout when warning
     add test case for bogus volatile simplification
     fix: volatile stores must not be simplified
     dump-ir: add testcase for option parsing corner case
     dump-ir: allow to specify the passes to execute via cli's options
     dump-ir: activate/deactive pass 'mem2reg'
     dump-ir: allow to skip the optimization pass(es)
     dump-ir: saner use of fdump_linearize
     dump-ir: rename -fdump-linearize to -fdump-ir
     dump-ir: make it more flexible
     dump-ir: activate -fdump-ir=mem2reg
     add test case for using multiple input files
     add test case for VLA sizeof
     add test case for memory to register problem
     add test case for conditionally undefined var
     add test case for incomplete type
     add test case bitfields in K&R decl
     add test case storage specifier in struct member
     add test case using sizeof on incomplete type
     add test case for bad layout of bool in bitfields
     add test case for missed overflow detection
     add test cases for canonicalization of add/sub chains
     add test case for compound literals
     add testcase for __builtin_unreachable()
     add test cases for canonicalization of mul chains
     add test case for pre-processor extra tokens warning
     add testcase for return & inline
     add test cases for simplification of equivalent to 'x == 0' or 'x != 0'
     add test case for superfluous cast with volatiles
     add testcase for mem2reg/SSA conversion
     add test cases for canonicalization of boolean expressions
     add test case for missing conversion to select
     show OP_PHI without VOID
     don't output value of anonymous symbol's pointer
     add table to "negate" some opcode
     use opcode table for compare_opcode()
     canonicalize binops before simplification
     canonicalize compare instructions
     add is_signed_type()
     fix usage of inlined calls
     inlined calls should not block BB packing
     give a type to all function arguments
     give a type to OP_PHISOURCEs
     give a type to OP_SELs, always
     give a type to OP_SWITCHs
     add doc about sparse's instructions/IR
     add support for wider type in switch-case
     llvm: remove unneeded arg 'module'
     llvm: remove unneeded 'generation'
     llvm: remove unneeded function::type
     llvm: reduce scope of 'bb_nr'
     llvm: use pseudo_list_size() instead of open coding it
     llvm: give arguments a name
     llvm: give a name to call's return values
     llvm: avoid useless temp variable
     llvm: extract get_sym_value() from pseudo_to_value()
     llvm: fix test of floating-point type
     llvm: fix translation of PSEUDO_VALs into a ValueRefs
     llvm: fix output_op_store() which modify its operand
     llvm: fix output_op_[ptr]cast()
     llvm: add test cases for symbol's address
     llvm: add test cases for pointers passed as argument
     llvm: add test cases for arrays passed as argument
     llvm: add test cases for degenerated pointers
     llvm: add support for OP_NEG
     llvm: add support for OP_SETVAL with floats
     llvm: add support for OP_SETVAL with labels
     llvm: ignore OP_INLINED_CALL
     llvm: fix pointer/float mixup in comparisons
     llvm: fix type in comparison with an address constant
     llvm: give correct type to binops
     llvm: adjust OP_RET's type
     llvm: variadic functions are not being marked as such
     llvm: fix type of switch constants
     llvm: make pseudo_name() more flexible
     llvm: give a name to all values
     llvm: add support for OP_SWITCH with a range
     llvm: fix OP_SWITCH has no target
     llvm: make value_to_pvalue() more flexible
     llvm: make value_to_ivalue() more flexible
     llvm: add test case pointer compare with cast
     llvm: let pseudo_to_value() directly use the type
     llvm: add small script to test LLVM generated bytecode
     llvm: add testcase for calling variadic functions
     llvm: fix variadic calls with constants
     llvm: take care of degenerated rvalues
     llvm: fix mutating function pointer
     llvm: fix mutated OP_RET
     llvm: fix mutated OP_SEL
     llvm: fix mutated OP_SWITCH
     llvm: fix mutated OP_PHISOURCE
     llvm: fix mutated OP_[PTR]CAST
     llvm: add support for restricted types
     llvm: fix get value from initialized symbol
     llvm: fix get value from non-anonymous symbol
     llvm: fix type of bitfields
     llvm: add support for OP_FPCAST
     llvm: add support for cast from floats
     llvm: cleanup of output_[ptr]cast()
     llvm: fix creation of sparsec's tmp files
     llvm: give names easier to debug
     llvm: gracefully catch impossible type/value
     llvm: warn instead of assert on global inits
     llvm: add support for float initializer
     llvm: only compare void pointers
     fix linearize_inc_dec() with floats
     add test case for boolean negation on float
     fix support of floating-point compare
     add support of floating-point specific arithmetic ops
     testsuite: fix: remove unneeded './' before commands
     fix: build sparse-llvm on i686s too.
     add more testcases for using addresses in conditionals
     add testcases linearization of degenerated arrays/functions
     fix: add missing degenerate() for logical not
     testsuite: make the '%.t' rule depends on PROGRAMS too
     testsuite: fix a few more incorrect check-commands
     testsuite: convert to the new pattern syntax
     testsuite: remove old ugly pattern syntax
     testsuite: move verbose/error() before get_tag_value()
     testsuite: add & use warning()
     testsuite: reset 'quiet' at the start of each testcase
     testsuite: fix invalid 'check-...' tags
     testsuite: validate the 'check-...' tags
     testsuite: early return in getopt loop
     testsuite: move do_test_suite out of the getopt loop
     testsuite: move no-arg out of the getopt loop
     testsuite: change do_usage text
     testsuite: allow to test only a subdir
     testsuite: default to shift in the getopt loop
     testsuite: add support for 'format -a'
     add testcase for 'sparse -D M...'
     fix: accept 'sparse -D M...'
     testsuite: add test case for quoting of command's arguments
     testsuite: process extra options without exec
     testsuite: respect command line's quotes & whitespaces
     testsuite: allow default args from environment for test commands
     add test case for space within command line
     fix: spaces in macro definition on the command line
     add testcases for unexamined base type
     fix: evaluate_dereference() unexamined base type
     add testcases for the linearization of calls
     simplify linearize_call_expression()
     fix linearize (*fun)()
     add testcases for multiple deref of calls
     avoid unneeded alloc on error path
     dereference of a function is a no-op
     add testcase for constant bitfield dereference
     fix expansion of constant bitfield dereference
     add testcase for CSE of floating-point compares
     fix: restore CSE on floating-point compares
     llvm: fix: previous function ref MUST be reused
     llvm: use LLVMModuleRef for get_sym_value()
     llvm: add declares for function prototypes
     testcases: add missing return statements
     warn on empty parenthesized expressions
     fix crash on bad expression in linearize_switch()
     llvm: simplify emit of null pointers
     llvm: default init of arrays & structs
     add more testcases for function designator dereference
     add testcases for type comparison
     fix implicit size of unsized arrays
     let handle_switches() also handle reverse logic
     add support for '-f[no-][un]signed-char'
     fix: dereference null-type
     teach sparse about '-fmax-warnings'
     give a type to builtin functions
     ctags: avoid null deref
     cleanup: make some functions static
     cleanup: remove unused & obsolete symbol_is_typename()
     cleanup: remove unused delete_last_basic_block()
     cleanup: remove declaration of unused merge_phi_sources
     add OP_SETFVAL
     CSE: support CSE of floating-point literal
     doc: fix manpage formatting
     report type & size on non-power-of-2 pointer subtraction
     remove warning "call with no type"
     add testcases for duplicated warning about invalid types
     fix error in bad conditional
     early return if null ctype in evaluate_conditional()
     add helper: valid_type()
     use valid_type to avoid to warn twice on conditionals
     add helpers: valid_expr_type() & valid_subexpr_type()
     do not report bad types twice
     always evaluate both operands
     fix examination of bad typeof
     extract extract eval_insn() from simplify_constant_binop()
     add testcase of dead dominator
     fix dead dominator
     fix missing checks for deleted instructions
     add testcase for bad killing of dominated stores
     add helper for pseudo's user-list's size
     add helper: has_users()
     use has_users() in dead_insn() too
     let kill_instruction() report if changes were made
     add testcases for converted loads
     fix killing of converted loads
     fix usage of deadborn loads
     kill dead loads
     kill dead stores when simplifying symbols
     By default disable the warning flag '-Wunknown-attribute'
     no repetition in unknown attribute warning message
     cleanup: remove unused 'struct pseudo_ptr_list'
     llvm: use list_size() to count the numbers of arguments
     llvm: initialize at declaration time
     show_pseudo(): protect against NULL ->sym
     use show_pseudo() for OP_SYMADDR's symbol
     let struct access_data use a single type
     rename base_type() to bitfield_base_type()
     builtin: add ctype for const {void,char} *
     builtin: make builtins more builtin
     builtin: extract eval_args() from arguments_choose()
     builtin: add typechecking of isnan(), isinf(), ...
     builtin: add testcases for expansion of special FP constants
     builtin: add testcases for expansion of FP classification
     unsigned multiplication is also associative
     no need for signed & unsigned multiplication
     use '%lld' for printing long longs
     build: add -MP for generated dependencies
     build: use -MMD for generated dependencies
     ban use of 'true' or 'false'
     'amd64' is also ok for sparse-llvm
     testsuite: fix typo with 'test-suite format -a'
     rename variable 'optimize' to 'optimize_level'
     move the optimization loop in its own file
     cse: untangle simplification & hashing
     extract cse_eliminate() from cleanup_and_cse()
     expose interface to CSE
     move liveness interface to its own header
     move inner optimization loop into optimize.c
     move the inner optimization loop into the main loop
     remove unneeded cast in calls to free_ptr_list()
     testsuite: add testcase for some random crash
     testsuite: add testcase about CSE problem
     IR: fix typo in IR doc
     IR: remove now unused OP_LNOP & OP_SNOP
     IR: remove never-generated instructions
     IR: let .cond unionize with .src and not .target
     IR: let OP_COMPUTEGOTO use .src instead of .target
     llvm: normalize sparse-llvm-dis' output
     llvm: fix typo for constant addresses
     testsuite: fix problem with double-escaping in patterns
     add a field 'tainted' to struct instruction
     taint: let check_access() warn just once
     fix address_taken()
     fix symbol cleanup
     cleanup deadborn phi-sources
     optim: add some more optimization tests
     optim: add testcase for internal infinite loop
     optim: add timeout for infinite optim loop tests
     optim: kill unreachable BBS after CFG simplification
     optim: no need to kill_unreachable_bbs() after main loop
     optim: fix optimization loop's condition
     optim: pack bb must set REPEAT_CFG
     optim: load simplification should repeat optimization
     optim: fix REPEAT_CFG_CLEANUP
     add an helper to test the value of a pseudo against zero
     optim: simplify null select
     make remove_usage() more generic
     add remove_use()
     show_label: add (and use) show_label()
     extract alloc_phisrc() from alloc_phi()
     small code reorg of add_store()
     alloc: add missing #include "compat.h"
     defer initialization of bb::context
     fix-return: remove special case for single return
     avoid deadborn loads & stores
     doc: is for development
     doc: document the debug flags
     fix missing handling of OP_FNEG
     graph: do not use insn->symbol for memops
     use -Wpointer-arith for tests
     default to LP64 for all and only for 64 bit ABIs
     fix alignment of 64 bit integers on LLP64
     add testcases for verifying ABI's integer size & align
     use an enum for ARCH_LP32 & friends
     add a flag -mx32 ILP32 env on 64 bit archs
     add testcase for enum / int type difference
     add testcase for array size type difference
     add testcase for typedef redefinition
     export check_duplicates()
     fix: warn on typedef redefinition
     do not to ignore old preprocessor testcases
     use also __x86_64 when __x86_64__ is used
     build: use a variable for $(shell uname -m)
     build: use 'filter' to do pattern matching inside the Makefile
     build: disable LLVM on x86-64-x32
     let cgcc use sparse's predefines for i386 & x86-64
     build: use --dirty with 'git describe'
     fix build on Hurd which doesn't define PATH_MAX
     testsuite: add check-cp-if
     testsuite: add check-assert
     teach sparse about _Floatn and _Floatnx
     add test case bug expand union
     alloc: check if size is too big
     fix: don't dump pointer value in error message
     fix missing checks for deleted instructions
     fix comment about PSEUDO_SYM usage
     fix: remove usage when killing symaddr (part 1)
     fix: remove usage when killing symaddr (part 2)
     OP_SYMADDR is simply an unop
     use function-like syntax for __range__
     increment the version number suffix it with -dev
     doc: fix markdown syntax
     doc: fix headings
     doc: add minimal support for sphinx-doc
     doc: add logo
     doc: automatically set the copyright date
     doc: automatically get the version
     doc: set primary domain to C
     doc: allow .md with py3-sphinx
     doc: move sparse.txt to markdown and rename it
     doc: the testsuite doc in reST
     doc: format as a man page
     doc: use reST for manpages
     api: move evaluate interface to its own header file
     doc: add structured doc to ptrlist.c
     autodoc: extract doc from the C files
     autodoc: convert extracted doc to reST
     autodoc: add a sphinx c:autodoc directive for the extracted doc
     autodoc: add doc from ptrlist.c
     autodoc: add markup to argument's references
     autodoc: doc the doc
     autodoc: by default disable syntax highlighting
     autodoc: add a small cheatsheet for reST markup
     autodoc: support muti-line param & return descriptions
     autodoc: document a few more APIs to test multiline
     autodoc: add autodoc tests in the testsuite
     doc: convert to reST
     doc: add sphinx domain for IR instruction indexation
     add helper for new parsing errors: unexpected()
     context: fix parsing of attribute 'context'
     context: __context__(...) expect a constant expression
     context: fix crashes while parsing '__context__;' or '__context__(;'
     context: stricter syntax for __context__ statement
     context: extra warning for __context__() & friends
     label: add testcase for label redefinition
     label: avoid multiple definitions
     vla-sizeof: add test cases
     vla-sizeof: add support for sizeof of VLAs
     fix typing of __builtin_expect()
     fix crash on 'goto <reserved word>'
     give a position to end-of-input
     avoid multiple error message after parsing error
     add test for integer-const-expr-ness
     dyn-macro: add testcase for __LINE__ & friends
     dyn-macro: use a table to expand __DATE__, __FILE__, ...
     dyn-macro: add support for __INCLUDE_LEVEL__
     dyn-macro: add real support for __BASE_FILE__
     utils: add xmemdup() & xstrdup()
     utils: convert alloc + copy to {mem,str}dup_alloc()
     builtin: add testcase for builtin macro expansion
     builtin: extract do_define() from do_handle_define()
     builtin: add builtin types {u,}{int,long,long}_ptr_ctype
     builtin: declare __builtin_[us]{add,sub,mul}{,l,ll}_overflow()
     builtin: rename arguments_choose() to args_triadic()
     builtin: add support for __builtin_{add,sub,mul}_overflow(), ...
     extract replace_with_bool() from replace_with_defined()
     builtin: add support for __has_builtin()
     builtin: add predefine()
     builtin: directly predefine builtin macros
     builtin: consolidate predefined_macros()
     doc: API before IR
     builtin: switch calling order of predefined_macros() & friends
     builtin: merge declare_builtin_function() with declare_builtins()
     teach sparse about -m16
     ptrlist: specialize __add_ptr_list() for tag/notag
     ptrlist: remove now unneeded add_ptr_list_notag()
     ptrlist: add helper PTR_UNTAG()
     ptrlist: rename PTR_ENTRY() to PTR_ENTRY_UNTAG()
     ptrlist: make explicit when tagged pointers are used.
     ptrlist: let FOR_EACH_PTR() ignore tags
     utils: add xasprintf() & xvasprintf()
     add support for -fdiagnostic-prefix[=prefix]
     doc: add doc for the -vcompound flag
     testsuite: fix missing return
     keep the debug flags alphabetically sorted
     testsuite: allow extra/default options to test commands
     ir-validate: add framework for IR validation
     ir-validate: validate pseudo's defining instruction
     ir-validate: add validation of (nbr of) phi operands
     ir-validate: add more validation points
     sparsec: simplify & portable use of mktemp
     add predefines for __INT_WIDTH__ & friends
     ptrlist: remove the now unneeded FOR_EACH_PTR_NOTAG()
     ptrlist: let {first,last}_ptr_list() return the raw pointer
     ptrlist: let sort_list() use the raw pointer
     ptrlist: let all pointer lists have the same parametrized structure
     ptrlist: when possible use the real type of the list
     ptrlist: remove now unneeded CHECK_TYPE()
     ptrlist: make add_ptr_list() more readable
     ptrlist: make free_ptr_list() more readable
     ptrlist: remove some unneeded arg from internal macros.
     ptrlist: remove extra ident level
     ptrlist: simplify loop nesting
     ptrlist: simplify DO_NEXT
     ptrlist: simplify PREPARE/NEXT
     ptrlist: shorter continuated lines
     ptrlist: remove ptr_list_empty()
     ptrlist: make {first,last}_ptr_list() out-of-line functions
     ptrlist: move semi-private prototypes close to their user
     ptrlist: use VRFY_PTR_LIST() for sanity check
     ptrlist: keep declaration of head-list-nr together
     ptrlist: make clear what is API and what is implementation.
     ptrlist: move DO_SPLIT() into DO_INSERT_CURRENT()
     ptrlist: add missing doc for some functions
     add testcase for bad fpcast simplification
     fix bad fpcast simplification
     avoid useless deref in simplify_cond_branch()
     new helper: replace_pseudo()
     remove unused arg in simplify_cond_branch()
     add_uniop() should take a type, not an expression
     rename add_uniop() to add_unop()
     add missing entry for OP_FNEG in kill_insn() & validate_insn()
     ir: define an OP_... range for unops
     ir: case OP_UNOP ... OP_UNOP_END
     cast: reorg testcases related to casts
     cast: add testcase for bad implicit casts to struct/union
     cast: add testcase for cast to bad typeof
     cast: add tests for warnings issued by sparse -v
     cast: rename evaluate_cast()'s vars with slightly more meaningful names
     cast: force_cast are OK on non-scalar values
     cast: prepare finer grained cast instructions
     cast: specialize FPCAST into [USF]CVTF
     cast: handle NO-OP casts
     cast: specialize floats to integer conversion
     cast: specialize casts from unsigned to pointers
     cast: make [u]intptr_ctype alias of [s]size_t_ctype
     cast: make pointer casts always size preserving
     cast: temporary simplify handling cast to/from void*
     cast: specialize cast from pointers
     cast: add support for -Wpointer-to-int-cast
     cast: make casts from pointer always size preserving
     cast: specialize integer casts
     cast: accept null casts
     cast: do not try to linearize illegal casts
     cse: add testcase for missed opportunity
     new helper: def_opcode()
     cast: simplify simplify_cast()
     cast: merge simplification of constant casts with constant unops
     cast: prepare for more cast simplifications
     cast: keep instruction sizes consistent
     cse: move to next comparable instruction
     cast: simplify TRUNC + ZEXT to AND
     add simple testcases for internal infinite loops
     simplify 'x | ~0' to '~0'
     simplify 'x & ~0' to 'x'
     simplify 'x ^ ~0' to '~x'
     bool: add testcase for bool simplification
     bool: fix add missing check in simplify_seteq_setne()
     bool: simplify ZEXT in bool -> int -> bool
     bool: fix missing boolean context for floats
     bool: generate plain OP_{AND,OR} instead of OP_{AND,OR}_BOOL
     bool: remove OP_{AND,OR}_BOOL instructions
     cast: reorganize testcases for cast optimization
     cast: optimize away casts to/from pointers
     cse: let equivalent casts hash & compare identically
     fix killing OP_SWITCH
     fix: remove dead OP_{SETVAL,SETFVAL,SLICE}
     kds: add testcases for kill_dead_stores()
     kds: add explanation to kill_dead_stores()
     kds: rename kill_dead_stores() to kill_dead_stores_bb()
     kds: add interface for kill_dead_stores()
     kds: kill dead stores after memops simplification
     kds: shortcut for kill_dead_stores()
     kds: fix recursion in kill_dead_stores_bb()
     kds: clarify kill_dead_stores_bb()
     testsuite: reorganize tests for compound literals
     testsuite: add a few more tests catching quadratic behaviour
     testsuite: improve mem2reg testcases
     testsuite: remove useless test for loop-linearization
     graph: build the CFG reverse postorder traversal
     graph: add debugging for (reverse) postorder traversal
     dom: calculate the dominance tree
     dom: add some debugging for the dominance tree
     dom: add support for dominance queries
     dom: build the domtree before optimization
     dom: use domtree for bb_dominates()
     sset: add implementation of sparse sets
     idf: compute the iterated dominance frontier
     idf: add test/debug/example
     add new helper: is_integral_type()
     add PSEUDO_UNDEF & undef_pseudo()
     add insert_phi_node()
     ptrmap: core implementation
     ptrmap: add type-safe interface
     ssa: phase 1: phi-nodes placement
     ssa: phase 2: rename load & stores
     ssa: phase 3: rename phi-nodes
     ssa: activate the new SSA conversion
     ssa: remove unused simplify_symbol_usage()
     ssa: phi worklist
     remove unused finish_address_gen()
     remove unused struct access_data::pos
     no need to assign ad->type for EXPR_POS
     remove obsolete comment: /* Dummy pseudo allocator */
     big-shift: add test for shifts with bad count
     big-shift: mark out-of-range OP_{ASR,LSR,SHL} as tainted
     big-shift: do not evaluate negative or over-sized shifts
     big-shift: don't take the modulo at expand time
     big-shift: move the check into check_shift_count()
     big-shift: use the base type for shift-too-big warning
     big-shift: also check shift count of shift-assignment
     big-shift: do not simplify over-sized OP_ASR to zero
     big-shift: reorder the tests in simplify_asr()
     big-shift: reuse simplify_asr() for LSR & SHL
     big-shift: simplify over-sized OP_LSRs
     big-shift: simplify over-sized OP_SHLs
     big-shift: use the type width for too big shift
     big-shift: fix warning message for negative shift count
     big-shift: fix evaluation of shift-assign
     big-shift: do not truncate the count when checking it
     big-shift: add -Wshift-count-{negative,overflow}
     extract nbr_users() from unssa.c
     add testcases for casts & bitfield insertion/extraction
     bitfield: extract linearize_bitfield_extract()
     bitfield: extract linearize_bitfield_insert()
     cast: simplify [SZ]EXT + TRUNC to original size
     cast: simplify [SZ]EXT + TRUNC to a smaller/greater size
     cast: fix shift signedness in cast simplification
     cast: do not compare sizes, test the opcode
     cast: use a switch to handle TRUNC(AND(x,M),N) in simplify_cast()
     cast: preserve the sizes of TRUNC(AND(x,M),N)
     cast: simplify [ZS]EXT(AND(x,M),N)
     cast: simplify AND(ZEXT(x,M),N)
     cast: simplify SEXT(SEXT(x,N),N')
     cast: simplify ZEXT(ZEXT(x,N),N')
     cast: simplify SEXT(ZEXT(x,N),N')
     bits: add helpers for zero & sign-extension
     big-shift: add testcases for simplification of over-sized shifts
     big-shift: add testcases for simplification of negative shifts
     big-shift: move shift count check in a separate function
     big-shift: fix warning message for negative or over-sized shifts
     big-shift: do not optimize negative shifts
     big-shift: do not optimize over-sized ASRs
     use "%Le" to display floats
     add copy_ptr_list()
     testcases: add testcase for missing detection of out-of-bound stores
     testcases: missing evaluation of side effects in typeof(VLA)
     kill dead OP_FADD & friends
     add ptr_list_empty()
     add ptr_list_multiple()
     add lookup_ptr_list_entry()
     shift: simplify LSR(LSR(x,N),N') & friends
     shift: simplify ASR(LSR(x,N),N')
     shift: avoid simplification of ASR(LSR(x,0),N)
     shift: simplify ASR(ZEXT(X, N), C)
     testcase for SET{EQ,NE}([SZ]EXT(x, N),{0,1})'s simplification
     cleanup of simplify_seteq_setne(): remove tmp vars
     simplify SET{EQ,NE}(ZEXT(x, N),{0,1})
     simplify SET{EQ,NE}(SEXT(x, N),{0,1})
     simplify 'x != 0' or 'x == 1' to 'x'
     add testcase for linearize_logical()
     fix size corruption when simplifing 'x != 0' to 'x'
     protect add_convert_to_bool() against bad types / invalid expressions
     conditional branches can't accept arbitrary expressions
     fix linearize_conditional() for logical ops
     expand linearize_conditional() into  linearize_logical()
     simplify linearize_logical()
     simplify SETNE(AND(X,1),0) to AND(X,1)
     simplify SETNE(TRUNC(x,N),{0,1})
     simplify ZEXT(SETCC(x,y), N)
     simplify SEXT(SETCC(x,y), N)
     simplify TRUNC(SETCC(x,y), N)
     simplify AND(SETCC(x,y), M)
     boolean conversion of boolean value is a no-op
     cast: fix warning position in cast_pseudo()
     limit the mask used for bitfield insertion
     expand linearize_position() into linearize_initializer()
     put back the bitfield base type into struct access_data
     fix instruction size & type in linearize_inc_dec()
     fix bad indentation in linearize_inc_dec()
     avoid infinite simplification loops of the second kind
     optim: add a few more testcases for shift & mask
     use multi_users() instead on nbr_users()
     reorg code for shift-shift simplification
     simplify ((x & M') | y ) & M into (y & M) when (M' & M) == 0
     simplify ((x & M) | y) >> S to (y >> S) when (M >> S) == 0
     simplify (x << S) >> S into x & (-1 >> S)
     simplify (x & M) >> S to (x >> S) & (M >> S)
     rename testcase for ((x << S) >> S) simplification
     add testcase for ((x >> S) << S) simplification
     add testcase for TRUNC(TRUNC(x)) simplification
     simpler guard in LSR-SHL simplification
     reorganize shift-shift simplification
     simplify ((x >> S) << S)
     reorganize simplification of ZEXT(TRUNC(x))
     simplify TRUNC(TRUNC(x))
     doc: simplify the creation of the viewlist
     doc: automatically insert the blank line for lists
     doc: convert existing simplify.c doc into ReST autodoc
     doc: reword doc for replace_pseudo()
     doc: add doc for simplification notation
     add testcase for (((x & M') | (y & M)) & M)
     add testcases for bitfield & AND/OR simplification
     unify simplify_lsr_or() & simplify_and_or_mask()
     add simplify_mask_or()
     use better names for simplify_mask_or_and() vars
     document simplify_mask_or() & simplify_mask_or_and()
     switch return order in simplify_mask_or_and()
     allow simplification of OP(((x & y) | (a & M')), K)
     move opcode test inside simplify_mask_or_and()
     simplify OP(((x & M') | y), K) when (M' & M) == M
     simplify OP(((x & M') | y), K) when (M' & M) != M'
     simplify OP((x | C), K) when (C & M) == 0
     simplify OP((x | C), K) when (C & M) == M
     simplify OP((x | C), K) when (C & M) != C
     simplify SHL((x & M') | y, S)
     add testcases for {LSR,SHL}(AND(x, M), S) with shared AND(x, M)
     use an intermediate mask in simplify_shift()
     simplify ((x & M) >> S) when (M >> S) == 0
     simplify ((x & M) >> S) when (M >> S) == (-1 >> S)
     simplify ((x & M) << S) when (M << S) == 0
     simplify ((x & M) << S) when (M << S) == (-1 << S)
     simplify TRUNC((x & M') | y, N)
     doc: extend simplification notation
     prepare simplification of MASK(SHIFT(a | b, S), M)
     simplify AND(SHIFT(a | b, S), M)
     simplify TRUNC(SHIFT(a | b, S), N)
     mode keywords don't need MOD_{CHAR,LONG,...}
     add support for mode __pointer__
     add support for mode __byte__
     add a testcase for enum using a mode
     remove superfluous newline in 'unknown mode attribute' error message
     testsuite: remove useless test for loop-linearization
     symaddr: s/insn->symbol/insn->src/
     add testcase for accesses to volatile bitfields
     split memops from unops
     add a flag for volatile memops
     fix: do not optimize away accesses to volatile bitfields
     opcode: centralize opcode definition
     opcode: add arity info
     opcode: add OPF_TARGET
     add a function to remove deadborn instructions
     fix missing declarations
     has-attr: add testcase for __has_attribute()
     has-attr: move 'mode' next to '__mode__'
     has-attr: add __designated_init__ & transparent_union
     has-attr: add support for __has_attribute()
     ir-validate: add validation branch to dead BB
     ir-validate: ignore dead phis
     ir-validate: validate return value
     add tescase for unreachable label in switch
     fix linearization of unreachable switch (with reachable label).
     move DEF_OPCODE() to header file
     trivial-phi: add testcase for unneeded trivial phi-nodes
     trivial-phi: make clean_up_phi() more sequential
     trivial-phi: extract trivial_phi() from clean_up_phi()
     trivial-phi: early return
     trivial-phi: use a temp var for the real source
     trivial-phi: directly return the unique value
     trivial-phi: remove more complex trivial phi-nodes
     stricter warning for explicit cast to ulong
     add linearization as a pass
     add testcases for missing return in last block
     use a temp var for function's upper-level statement
     topasm: top-level asm is special
     specialize linearize_compound_statement()
     there is always an active BB after linearize_fn_statement()
     the return BB is never terminated
     extract add_return() from linearize_return()
     use UNDEF for missing returns
     use a temp var for the return type/symbol
     return nothing only in void functions
     add testcases for wrong ordering in phi-nodes
     fix ordering of phi-node operand
     add tests for nested logical expr
     fix linearization of nested logical expr
     add testcase for non-constant switch-case
     fix linearization of non-constant switch-cases
     test: make test Waddress-space-strict succeed on 32-bit
     test: use integers of different sizes, even on 32-bit
     test: make 32-bit version of failed test
     ssa: relax what can be promoted
     doc: is_int_type() returns false for SYM_RESTRICTs
     enum: add testcase for UB in oversized shift
     enum: fix UB when rshifting by full width
     enum: add testcase for type of enum members
     enum: add testcase for base & enumerator type
     enum: fix cast_enum_list()
     enum: use the smallest type that fit
     enum: use the values to determine the base type
     enum: only warn (once) when mixing bitwiseness
     enum: warn when mixing different restricted types
     enum: warn on bad enums
     enum: rewrite bound checking
     enum: keep enumerators as int if they fit
     enum: default to unsigned
     enum: more specific error message for empty enum
     __attribute__((fallthrough)) can't simply be ignored
     ptrlist: add ptr_list_nth_entry()
     add testcase for missing function designator expansion
     fix expansion of function designator
     teach sparse about '-o <file>'
     teach sparse about '-x <language>'
     cgcc: add support to ignore argument(s) of options
     cgcc: teach about '-o <file>'
     cgcc: teach about '-x c'
     dump-macro: break the loop at TOKEN_UNTAINT
     dump-macro: simplify processing of whitespace
     fix implicit K&R argument types
     Use -Wimplicit-int when warning about missing K&R argument types
     Accept comma-separated list for function declarations.
     cgcc: use 'i386' for the arch instead of 'i86'
     add testcase for missing deliminator ' or "
     man: add section about reporting bugs
     man: add AUTHORS section
     man: update maintainer info
     don't allow newlines inside string literals
     multi-buffer for idents
     as-name: add and use show_as()
     as-name: use idents for address spaces
     as-name: allow ident as address_space
     as-name: check for multiple address spaces at parsing time
     as-named: warn on bad address space
     add detection of native platform
     Consolidate 'machine detection' into "machine.h"
     test endianness with __BYTE_ORDER__
     testsuite: test predef macros on LP32/LP64/LLP64
     fix '__SIZE_TYPE__' for LLP64
     allow optional "_T" suffix to __SIZEOF_XXX__
     use bits_mask() for predefined_max()
     add builtin_type_suffix()
     teach sparse about asm inline
     remove duplicates from gcc-attr-list.h
     show-parse: strip do_show_type()'s trailing space
     make predefined_type_size() more generic
     give a type to wchar
     use the type for predefined_max()
     add predefined macros for wint_t
     add predefined macros for [u]intptr
     add predefined macros for [u]intmax
     add predefined macros for [u]int{8,16}_t
     add predefined macros for [u]int64_t
     add predefined macros for [u]int32_t
     add predefined macros for char{16,32}_t
     fix the size of long double
     add predefine for __CHAR_UNSIGNED__
     add predefine_min() and use it for __{WCHAR,WINT}_MIN__
     add a flag to warn on casts to/from bitwise pointers
     show-parse: don't display null ident in show_typename()
     show-parse: do not display base type's redundant specifiers
     show-parse: remove string_ctype from typenames
     build: only need includedir from llvm-config
     build: check if sparse-llvm needs libc++
     remove unneeded declarations in "compat.h"
     remove unused arg in add_branch()
     allocate BBs after the guards
     remove redundant check of _Bool bitsize
     remove unused regno()
     remove -finline-functions from CFLAGS
     remove self-assignment of base_type
     doc: fix list formatting
     as-name: document that identifiers are OK for address spaces
     add TODO list.
     Sparse v0.6.0

Ramsay Jones (9):

     Makefile: use locally built sparse in the selfcheck target
     sparsec: use a compatible exception model on cygwin
     sparsei: add the --[no-]jit options
     constant: add -Wconstant-suffix warning
     pre-process: suppress trailing space when dumping macros
     pre-process: print macros containing # and ## correctly
     pre-process: don't put spaces in macro parameter list
     pre-process: print variable argument macros correctly
     pre-process: add the -dM option to dump macro definitions

Randy Dunlap (6):

     sparse: minor manpage corrections
     Documentation: make data-structures.txt easier to read
     Documentation: editing fixes in test-suite
     lib.c: early return from handle_onoff_switch()
     sparse: ignore indirect_branch attribute
     sparse: option to print compound global data symbol info

Thiebaud Weksteen (1):

     Add testcases for bitwise cast on pointer

Tycho Andersen (1):

     expression.h: update comment to include other cast types

Uwe Kleine-König (6):

     build: make PREFIX overwritable from the environment
     build: put comment about to the place where it is included
     build: drop BASIC_CFLAGS and ALL_CFLAGS
     build: drop -g from LDFLAGS
     build: pass CPPFLAGS to compiler
     build: only generate version.h when needed

Vincenzo Frascino (1):

     print address space number for cast-from-AS warnings
Personal tools