| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
Examination of a pointer type doesn't examine the corresponding
base type (this base type may not yet be complete). So, this
examination must be done later, when the base type is needed.
However, in some cases it's possible to call evaluate_dereference()
while the base type is still unexamined.
Fix this by adding the missing examine_symbol_type() on the base type.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
evaluate_dereference() lacks an explicit examination of the
base type. Most of the time, the base type has already been
examined via another path, but in some case, it's not.
The symptom here is the dereferenced value having a null size.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
|
|
GCC's manual or POSIX say about the '-D' option something like:
'−D name[=value]' should be treated as if in directive
'#define name value' (with '1' as default for the value),
including its tokenization.
So an option like '-DM(X, Y)=...' should be processed like a
directive '#define M(X, Y) ...'.
However, the current code treat a space as a separator between
the macro and its definition, just like the '='. As consequence,
the above option is processed like the directive would be
'#define M(X, Y)=...', with 'M(X,' as the macro (name) and
'Y)=...' as its definition.
Fix this by stopping to treat the space character specially,
thus only using '=' as the separator.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
|
|
During testing it's sometimes useful to force some default arguments
for all commands. An example of this is using '-m32' which essentially
allow to run the tessuite on an 64bit machine as-if run a 32-bit one.
Allow this by using the environment variable 'SPARSE_TEST_ARGS' to
hole default arguments for the test commands.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Currently the testsuite use 'eval echo $cmd' to expand
the name of the test file to be given on the command line.
This has the annoying consequence to go a bit too far in
the expansion of variables and to destroy any quotes and
whitespaces escaping that would have done.
Fix this by doing the eval later, when effectively executing
the command.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Commit 399c43889 (testsuite: get options from env too) allowed
the testsuite to takes extra options from the environment but
did it in a crude way involving exec.
Change this by using 'set --' instead of doing an 'exec'.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Till now, sparse was unneedlessly strict in what it accepted in
'-D' options. More specifically, it doesn't accept:
1) separated '-D' and the macro definition, like:
sparse -D MACRO[=definition] ...
2) a space between the '-D' and the macro name, like:
sparse '-D MACRO[=definition] ...
Case 1) is clearly accepted by GCC, clang and should be
accepted for a POSIX's c99. Case 2's status is less clear
but is also accepted by GCC and clang (leaving any validation
to the corresponding internal #define).
Fix this by accepting separated command line argument for '-D'
and the macro (and removing the check that rejected the macro
part if it started with a space).
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Some 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
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
The 'format' command create the information needed for the testcase
from the input file and output this on stdout. The developper must
then add this to the input file.
Let's do this automatically by adding an option '-a' to the 'format'
command to directly append the infos to the input file.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
The getopt loop used to bear by default and only some
options had to explicitly call 'shift' and 'continue'
to process further elements.
Change this to a 'normal' loop, shifting the next arg by default
and breaking of the loop when needed.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
During development, it is very useful to be able to run only
some of the tests, maybe a whole class.
Help this by allowing to run the testsuite on only a subdir
of the 'validation/' directory.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
The text for the testsuite usage used 'none' as if it was an
option/keyword while it only meant the absence of arguments.
Make the text clearer by removing the 'none' and being explicit
about the absence of arguments.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
This is a preparatory step to allow to run only a part
of the testsuite (a subdir).
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
This is a preparatory step to allow to run only a part
of the testsuite (a subdir).
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
This is a preparatory step to allow to run only a part
of the testsuite (a subdir).
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Making a typo in one of the 'check-...' tags can make
a testcase useless and thus incapable of detecting a regression.
Add some validation to these tags in order to detect wrong
tags.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
A few testcases had typos in their 'check-...' tags or
the tag was plainly invalid.
Fix them in accordance to the doc.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
The flag 'quiet' is used to quiets unwanted error messages,
for example for testcases known to fail, but this flag is reset
too late so that the beginning of the next testcases will run
with the value for the previous case.
Fix this by reseting the flag at the begining of each testcase.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Allow this new helper to indicate wich file trigger the
warning and replace the existing call to 'echo "warning: ...'.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
So, we can use them inside get_tag_value().
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
It was too ugly (and a bit longish).
Remove it.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
The old one is too ugly and has to die.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
The exact syntax for commands is:
'check-command: ' <command> <args>...
and the command itself must *not* be prefixed with './'.
Fix the last three that had it wrong.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
The testsuite can be run on a specific testcase directly
via the Makefile via a 'validation/%.t' pattern rule but
this rule had no dependency on the programs being tested
and thus could be run on a previous version.
Fix this by adding the needed dependencies.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Expressions involving the logical-not '!' does not
call degenerate().
Since the result type is always 'int' and thus independent
of the expression being negated, this has no effect on the
type-checking but the linearization is wrong.
For example, code like:
int foo(void)
{
if (!arr) return 1;
return 0;
}
generates:
foo:
load %r6 <- 0[arr]
seteq.32 %r7 <- VOID, $0
ret.32 %r7
The 'load' being obviously wrong.
Fix this by adding the missing degenerate().
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
and unify the existing ones.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
In commit "f1e4ba13d (build: disable sparse-llvm on non-x86)",
we wanted to disable the build of sparse-llvm on non-x86 archs
but the pattern used also excluded machines like 'i686'.
Fix this by using the pattern 'i[3456]86' instead.
Fixes: f1e4ba13d1499407a72349b50052ae818c8d8553
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Some testcases have their command specified as './<command name>'
but the './' part is unneeded as all commands are first prefixed
with '../' before being run.
Furthermore, the presence of these './' inhibit simple
filtering of the disabled commands.
Fix this by stripping the './' where it was used.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Floating-point arithmetic is quite different from the
arithmetic on integers or the one of real numbers.
In particular, most transformations, simplifications that can
be done on integers are invalid when done on floats.
For example:
- associativity doesn't hold
- distributivity doesn't hold
- comparison is tricky & complex
This is because (among others things):
- limited precision, rounding everywhere
- presence of signed zeroes
- presence of infinities
- presence of NaNs (signaling or quiet)
- presence of numbers without inverse
- several kind of exceptions.
Since they don't follow the same rules as their integer
counterpart, better to give them a specific opcode
instead of having to test the type of the operands at
each manipulation.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Comparision of floating-point values can't be done
like for integral values because of the possibility to have
NaNs which can't be ordered with normal values or even between
themselves.
The real difference appears once there is any "reasoning"
done with the result of the comparison. For example, once NaNs
are taken in account: "!(a < b)" and "(a >= b)" are not the same.
In fact the usual comparison operators must be reinterpreted
as implicitely first testing if any of the operand is a Nan
and return 'false' if it is the case. Thus "a < b" becomes
"!isnan(a) && !isnan(b) && (a < b)".
If we need to negate the comparison we get "!(a < b)" which
naturally becomes "isnan(a) || isnan(b) || (a >= b)".
We thus need two sets of operators for comparison of floats:
one for the "ordered" values (only true if neither operand
is a Nan) and one for the "values" (also true if either
operand is a NaN). A negation of the comparison switch from one
of the set to the other.
So, introduce another set of instructions for the comparison
of floats.
Note: the C standard requires that:
*) "x == x" is false if x is a NaN,
*) "x != x" is true if x is a NaN,
and this is coherent with "x != x" <-> "!(x == x)".
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Using the pre or post increment or decrement operator on
floating-point values mix the addition of a floating-point
value with an *integral* constant 1 or -1.
Fix this by checking if we're dealing with fp or not and using
the proper fp constants (1.0 or -1.0) if it is the case.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
LLVM use a stricter type system for its IR than sparse does.
In the present case, sparse allow to compare pointers regardless
of their types.
This create LLVM errors if we try to simply translate, instruction
by instruction, sparse's compare instructions to LLVM ones.
Fix this by casting all pointers to void *, before comparing them.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Currently, sparse-llvm can't emit the needed code for
all initializers and when it's the case it stop abruptly
with an assert(0).
This is kinda useless.
Somehow fix this by issuing a warning instead and try to
continue gracefully.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Until now, in sparse-llvm, an assert(0) was used each time
something which was not yet handled was encountered.
This is doubly annoying because:
- the assert gave no info about the cause
- it also hide possible further issuses.
Fix this by:
- removing the asserts
- use a warnng explaining what's at hand
- try to continue with the following instructions.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
LLVM automatically add an numeric suffix for names
automatically created. So, if intermediate names must
be created for a pseudo whose name was, for example, "%R4",
these new names will be "%R41", "%R42".
This is quite annoying because we can't make the distinction
between these names and the original names, (maybe of some other
pseudos whose names were "%R41" & "%R42).
Change this by adding a "." at the end of each name, as this will
then allow to see what the original name was.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
It seems the intention was to create the tmp files with
the suffixes but the files are create without the suffixes
(while the filename that is used later is with the suffixes).
In all case the non-suffixed file that is created by the
mktemp command is never removed and after a while these
files begin to pile-up. Better to remove them.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
The linearized code, sparse's IR, have no use of C's complex type
system. Those types are checked in previous phases and the pseudos
doesn't have a type directly attached to them as all the needed
typing info info are conveyed by the instructions.
In particular, PSEUDO_VAL (used for integer and address constants)
are completely typeless.
There is a problem with this when calling a variadic function
with a constant argument as in this case there is no type in the
function prototype (for the variadic part, of course) and there is
no defining instructions holding the type of the argument.
Fiw this by using the type of the arguments explicitly given
in the OP_CALL instructions.
Reported-by: Dibyendu Majumdar <mobile@majumdar.org.uk>
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
In sparse-llvm, the type of switch constants are hardcoded
to 'i32'.
Fix this by using the right type as given by the instruction.
Reported-by: Dibyendu Majumdar <mobile@majumdar.org.uk>
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
It appears that when a function is compiled, its variadic flag is not
being set in LLVM so LLVM complains when at call sites variable
arguments are passed.
Reported-by: Dibyendu Majumdar <mobile@majumdar.org.uk>
Originally-by: Dibyendu Majumdar <mobile@majumdar.org.uk>
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Pointer arithmetic and/or simplification can mixup pointer
and integer types.
Fix this by adding casts before all non-floating point binops
and adjust the result type if needed to match the instructio.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Since sparse's constant are typeless comparing a pointer with
an address constant lack correct type information.
Fix this by casting the constant to the same type as the LHS.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
In output_op_compare() everything that is not of interger
type is treated as floats. Pointers disagree.
Fix this by rearranging the code and treat pointers like integers
as required for LLVM's icmp.
Reported-by: Dibyendu Majumdar <mobile@majumdar.org.uk>
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
sparse-llvm has not yet support for OP_NEG and stop on an assert
if one is encountered.
Fix this by invoking the appropriate LLVMBuild[F]Neg().
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
OP_PTRCASTs can't always be directly translated into LLVM bitcasts and
OP_[S]CASTs can't always be translated into LLVM's trunc/sext/zext
because integer to pointer and pointer to integer must be handled too.
Fix this in output_op_ptrcast() & output_op_cast() by issuing
LLVMBuildIntToPtr/PtrToInt when appropriate.
Reported-by: Dibyendu Majumdar <mobile@majumdar.org.uk>
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
In sparse-llvm the field 'priv' of a pseudo is used to store
the corresponding LLVMValueRef. This field is normaly assigned
when processing the instruction that produces the speudo.
In output_op_store(), the field insn->target->priv is overwritten
by the LLVMValueRef returned by LLVMBuildStore().
It's unclear what this return value is:
- this corrupts the pseudo, making it unusable in subsequent
instructions.
- there is no reason to change this field anyway.
Fix this by removing the assignment to insn->target->priv
in output_op_store().
Reported-by: Dibyendu Majumdar <mobile@majumdar.org.uk>
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
In sparse-llvm there is the assumption that a PSEUDO_VAL is always
of integer type. But this is not always the case: constant pointers,
like NULL, are also of the PSEUDO_VAL kind.
Fix this by adding a helper 'val_to_value()' and using the
instruction's type where this pseudo is used as the type of the value.
Note: while this patch improve the situation, like for example for the
test cases added here, it's still not correct because now we're making
the assumption that 'insn->type' is the type we need for the pseudo.
This is often true, but certainly not always.
For example this is not true for:
- OP_STORE/OP_LOAD's insn->src
- OP_SET{EQ,...}'s insn->src[12]
- probably some others ones
- in general, obviously, for any instructions where the target has
a different type than the operands.
Reported-by: Dibyendu Majumdar <mobile@majumdar.org.uk>
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Currently, a name is given to the result of instruction like
binops, compares, ... but not to function calls.
Functionally, it doesn't change anything but those names are
useful by easing reading while debugging, reading the generated
code, ...
Fix this by giving a name to OP_CALL's result too.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Arguments, like all LLVMValues, are given a default name
but these name are simply '%0', '%1', ... and are thus not
very readable.
Fix this by giving them an explicit name 'ARG1', ... to
match the names used in sparse's linearized code.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
It's not needed here since there is no recursive access to BBs.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Currently the different cases of a switch-statement, or more
exactly the 'struct multijmp' that hold the value of these cases
excepted only value of 'int' type. Trying to use a wider value
results in the value being truncated but any integer value should
be valid.
Fix this by unsigned 'long long' to hold these values.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
For consistency and for sparse-LLVM which needs it,
give them a type too.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Currently, when a phi-node is converted into a OP_SEL
this instruction is given a size but not a type but when
created directly it is given a type.
For consistency and for sparse-LLVM which needs it,
give them always a type.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Currently, OP_PHISOURCEs are given a size but not a type.
For consistency and for sparse-LLVM which need it,
give them a type too.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
The linearized code, sparse's IR, have no use of C's complex type
system. Those types are checked in previous phases and the pseudos
doesn't have a type directly attached to them as all the needed
typing info info are conveyed by the instructions.
In particular, PSEUDO_VAL (used for integer and address constants)
are completely typeless.
There is a problem with this when calling a variadic function
with a constant argument as in this case there is no type in the
function prototype (for the variadic part, of course) and there is
no defining instructions holding the type of the argument.
Possible but rejected solutions are:
* give a full type to all pseudos
This is silly as, for example 'int 0' and 'unsigned int 0'
are, really, the same constants.
* give simply a size to all pseudos
This seems to be simple enough but *currently* it negatively
impacts CSE (maybe because of some others deficiencies in
the type system, especially for casts).
* give a type to all function arguments via a new instruction
that would mimic argument passing (OP_ARG or OP_PUSH).
This solution is interesting, especially as it gives a placeholder
for real argument passing at code generation time, but:
0) they can be added, if needed, when some code generation
will be done.
1) it creates a superfluous instruction for every argument
of every function call
2) these instructions are essentially copy instructions in
disguise.
So, since the problem only exist for constants used in variadic
arguments (and currently, it's only a problem for sparse-llvm),
the solution selected is to add to OP_CALLs a list holding the
type of all arguments. More exactly, it reuses the field .fntype
which was used to store the type of the function, and changes
it to a list holding the function type *and* the type of all
effective arguments.
Reported-by: Dibyendu Majumdar <mobile@majumdar.org.uk>
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
OP_INLINED_CALL are there only as a sort of annotation
for debugging purpose.
Their presence should thus not block the packing of
basic blocks.
Fix this by ignoring OP_INLINED_CALL when trying to pack
a basic block.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
OP_INLINED_CALL are there only as a sort of annotation
for debugging purpose. It is thus wrong to associate
pseudo's usage to them (even if the pseudo are the arguments
of the function now inlined).
Fix this by removing the use_pseudo() for each arguments.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Currently only commutative instructions are canonicalized
(the "simpler" operands, often a constant, is forced, if present
to be in the second operand). This improve CSE (more cases are
considered as equivalent) and help to reduce the number of "pattern"
to be handled at simplification.
Do this also for compare instructions since in thsi case we can
swap the order of the operands if at the same time we also swap
the 'direction' on the comparison.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Currently, canonicalization of binops (more specifically
insuring that the operands of binops are in canonical order)
is only done after simplify_binop(). But the goal of
canonicalization is to limit the number of cases/patterns
we need to check/handle during ... simplification.
So canonicalization need to be done before simplification.
Fix this by moving (this part of) canonicalization before
doing simplification.
Note 1: the motivation of this patch is to prepare code for
the canonicalization of compare instructions
Note 2: this patch allow now some simplification of ...
the simplification code (simplify_binop()), this
will be done in a later serie.
Note 3: this patch changes slightly the cost of the CSE/
simplification, positively or negatively, depending
on the ration of simplification/canonicalization
that can be done.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
At the same time, change also the name of the function.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Some optimizations transform an instruction opcode
into another. For example, it may be needed to know
the opcode corresponding to the negation of a comparison.
This patch make this easy and flexible by adding a table
for the relation between opcodes.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
The value of this pointer is of no use unless you're
using a debugger (or just to see if two things are
identical or not) and it's presence produces noise
when comparing the output of two runs for testing.
Change this by issuing it only if 'verbose' is set.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
As the normal result of branch simplification OP_PHI
can have some VOID in their phi_list, sometimes lots of them.
These list can't be simplified, comacted or so because the
address of the pseudos is used for the pseudo tracking.
But it's annoying that these VOID are displayed by
show_instruction(), it make things sometimes hard to read.
Chnage this by ommiting to display them (when not verbose).
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
'testcases-mem2reg' into tip
|
|
This may seems as an missing optimization problem but in truth,
the root cause is a problem with SSA conversion.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
The linearization of 'return' statements must correctly take
in account some implementation details of the inlining.
As such, it deserves its own testcase.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
__builtin_unreachable()'s semantic has consequences on the CFG
and this should be taken in account for:
* checking for undefined variables
* checking when control reaches end of non-void function
* context checking
* ...
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Reported-by: Lance Richardson <lrichard@redhat.com>
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
GCC can be called with several input files in a single invocation
but these files are processed individually.
There is so no reasons to warn about a symbol being already defined
in a previous file.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
as it will be used for dumping the IR not only just after
linearization but after other passes too.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
For experimenting with some optimization and the linearization
process, it's useful to see the raw result, without any kind
of optimization.
This patch test PASS_OPTIM (-foptim-{enable,disable}) to see
if the whole optimization process can be skipped.
Note: ideally, this should be coupled with -O0 but this
could create problems with the 'sparse' tools when used
on file wit an optimization level of 0 (since the sparse
tools depends on the optimization for the context checking
to be accurate and in general diagnostics also need
at least basic optimization).
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
|
|
'fix-volatile-simplification', 'struct-asm-ops', 'restricted-pointers', 'fix-f2i-casts', 'symaddr-description', 'flush-stdout' and 'diet-simple' into tip
|
|
kill_dominated_stores() simplify away redundant stores.
Nice but volatile stores are never redundant and must
never be simplified away.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
This is much needed for tools like test-linearize
which output on stderr & stdout and without this patch
error messages and normal output are out-of-sync.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
'graph-cleanup' and 'cgcc-cleanup' into tip
|
|
Currently, simplify_cast() don't really take floating-points
in account and happily simplify away a float-to-int cast
if both types have the same size.
Fix this by not touching such casts.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Some branch simplifications are only valid if the branch is free
of side-effects. The check is done in flow.c:bb_has_side_effects()
but currently deosn't take in account the fact that a volatile load
is also a side-effect.
Fix this by adding the appropriate check.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Ramsay Jones <ramsay@ramsayjones.plus.com>
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
No functionnal changes here, only shuffling a few lines around,
adding separators and adding a few comments
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
(g)make foreach can be quite handy but it also make
Makefiles much less declarative and thus harder to read.
Avoid them by adding the few needed assignments & dependencies.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
.o files are also created in the validation directory
(currently only tmp.o, for sparse-llvm/sparsec testing).
Aso remove them during make clean-check.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
The current mechanism for the quiet commands, what's
hiding the true command and emitting instead things like:
CC target.o
is, IMO, somehow unneedlessly sophisticated and this doesn't help
to understand what's happening and to adapt things when needed.
Change this by using simple 'echo' commands and make's '@' to
display the short command and hiding the long one.
Warning: There is a small change in behaviour with this:
previously, when displaying the non-quiet commands with
'make V=1' the quiet ones were not emitted. Now, with this
patch, the short/quiet command is emitted in both case.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Now it does the same as done in the kernel: 2 + 8.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
So people who like this file to be a hidden one or to be an
out-of-tree file or simply who don't like the name, can choose
whatever suits them the best.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
This help to minimize merge conflicts.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
The last .tar.gz was for v0.5.0 in 2014.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
This way version.h isn't generated when running $(make clean) but only
when lib.c is about to be compiled.
This simplifies packaging for Debian because the package building programs
abort when there are additional files after $(make clean).
Signed-off-by: Uwe Kleine-König <uwe@kleine-koenig.org>
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
No functional changes here, just moving things around.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Those are not evil but it's for consistency with the remaining.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
This allow distros or devs to override the default CFLAGS,
CPPFLAGS, LDFLAGS, ...
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
This a preparatory step to allow these flags to be overriden
from the command line.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Some of the programs need to link with some external libraries.
For some reasons, these libraries are specified via variables named:
"<target>_EXTRA_OBJS".
Use the '-ldlibs' prefix instead to better reflect the real use.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Some programs are composed of several source files other than the
main one. These files need then to be added at linking time.
These extra file are currently specified using variables named
"<program-name>_EXTRA_DEPS" but the way these variables are used
make that they can only hold .o files (or .a ones) and are not
some kind of generic dependencie.
Make this explicit by using the suffix '-objs' instead.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
This only add the parsing and checks as a type qualifier;
there is no operational semantic associated with it.
Note: this only support _Atomic as *type qualifier*, not
as a *type specifier* (partly because there an
ambiguity on how to parse '_Atomic' when followed
by an open parenthesis (can be valid as qualifier
and as specifier)).
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Note: there is still no semantic associated with 'restrict'
but this is a preparatory step.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
This is slightly shorter (and thus may avoid long lines) and
facilitate the introduction of MOD_RETRICT in a later patch.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
MOD_TOPLEVEL & MOD_INLINE are already include in MOD_STORAGE
so there is no need to repeat them in MOD_IGNORE & elsewhere
where MOD_STORAGE is used.
Change this by removing the redundant MOD_TOPLEVEL & MOD_INLINE.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Now they are grouped a bit more logically.
Note: MOD_ASSIGNED & MOD_ADDRESSABLE are not type modifiers
but properties of the symbol. As such they should be
moved to struct symbol. However, as they should be
correctly propagated to the symbol components if any,
better to leave them as is.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
but is used to track which inline functions are
effectively used. So better remove it from the MOD_...
and implement the same functionality via a flag
in struct symbol.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
as it can, for example, be needed via arg_file() -> error()
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Emit the short command witht he same sacing as done
in the kernel: 2 + 8.
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
|
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
|
