RuleDivOpt Class Reference

Convert INT_MULT and shift forms into INT_DIV or INT_SDIV. More...

Inheritance diagram for RuleDivOpt:

Collaboration diagram for RuleDivOpt:

Public Member Functions
	RuleDivOpt (const string &g)
	Constructor.
virtual Rule *	clone (const ActionGroupList &grouplist) const
	Clone the Rule. More...
virtual void	getOpList (vector< uint4 > &oplist) const
	List of op codes this rule operates on. More...
virtual int4	applyOp (PcodeOp *op, Funcdata &data)
	Attempt to apply this Rule. More...
Public Member Functions inherited from Rule
	Rule (const string &g, uint4 fl, const string &nm)
	Construct given group, properties name. More...
virtual	~Rule (void)
	Destructor.
const string &	getName (void) const
	Return the name of this Rule.
const string &	getGroup (void) const
	Return the group this Rule belongs to.
uint4	getNumTests (void)
	Get number of attempted applications.
uint4	getNumApply (void)
	Get number of successful applications.
void	setBreak (uint4 tp)
	Set a breakpoint on this Rule.
void	clearBreak (uint4 tp)
	Clear a breakpoint on this Rule.
void	turnOnWarnings (void)
	Enable warnings for this Rule.
void	turnOffWarnings (void)
	Disable warnings for this Rule.
bool	isDisabled (void) const
	Return true if this Rule is disabled.
void	setDisable (void)
	Disable this Rule (within its pool)
void	clearDisable (void)
	Enable this Rule (within its pool)
bool	checkActionBreak (void)
	Check if an action breakpoint is turned on. More...
uint4	getBreakPoint (void) const
	Return breakpoint toggles.
virtual void	reset (Funcdata &data)
	Reset this Rule. More...
virtual void	resetStats (void)
	Reset Rule statistics. More...
virtual void	printStatistics (ostream &s) const
	Print statistics for this Rule. More...

Static Public Member Functions
static Varnode *	findForm (PcodeOp *op, int4 &n, uintb &y, int4 &xsize, OpCode &extopc)
	Check for INT_(S)RIGHT and/or SUBPIECE followed by INT_MULT. More...

Additional Inherited Members
Public Types inherited from Rule
enum	typeflags { type_disable = 1, rule_debug = 2, warnings_on = 4, warnings_given = 8 }
	Properties associated with a Rule. More...

Detailed Description

Convert INT_MULT and shift forms into INT_DIV or INT_SDIV.

The unsigned and signed variants are:

• sub( (zext(V)*c)>>n, 0) => V / (2^n/(c-1))
• sub( (sext(V)*c)s>>n, 0) => V s/ (2^n/(c-1))

Member Function Documentation

applyOp()

int4 RuleDivOpt::applyOp ( PcodeOp *  op, Funcdata &  data  )

virtual

Attempt to apply this Rule.

This method contains the main logic for applying the Rule. It must use a given PcodeOp as the point at which the Rule applies. If it does apply, changes are made directly to the function and 1 (non-zero) is returned, otherwise 0 is returned.

Parameters

op	is the given PcodeOp where the Rule may apply
data	is the function to which to apply

Reimplemented from Rule.

clone()

virtual Rule* RuleDivOpt::clone ( const ActionGroupList &  grouplist ) const

inlinevirtual

Clone the Rule.

If this Rule is a member of one of the groups in the grouplist, this returns a clone of the Rule, otherwise NULL is returned.

Parameters

grouplist

is the list of groups being cloned

Returns

the cloned Rule or NULL

Implements Rule.

findForm()

Varnode * RuleDivOpt::findForm ( PcodeOp *  op, int4 &  n, uintb &  y, int4 &  xsize, OpCode &  extopc  )

static

Check for INT_(S)RIGHT and/or SUBPIECE followed by INT_MULT.

Look for the forms:

• sub(ext(X) * #y,#c) or
• sub(ext(X) * #y,#c) >> n or
• (ext(X) * #y) >> n

Looks for truncation/multiplication consistent with an optimized division. The truncation can come as either a SUBPIECE operation and/or right shifts. The numerand and the amount it has been extended is discovered. The extension can be, but doesn't have to be, an explicit INT_ZEXT or INT_SEXT. If the form doesn't match NULL is returned. If the Varnode holding the extended numerand matches the final operand size, it is returned, otherwise the unextended numerand is returned. The total truncation, the multiplicative constant, the numerand size, and the extension type are all passed back.

Parameters

op	is the root of the expression
n	is the reference that will hold the total number of bits of truncation
y	will hold the multiplicative constant
xsize	will hold the number of (non-zero) bits in the numerand
extopc	holds whether the extension is INT_ZEXT or INT_SEXT

Returns

the extended numerand if possible, or the unextended numerand, or NULL

getOpList()

void RuleDivOpt::getOpList ( vector< uint4 > &  oplist ) const

virtual

List of op codes this rule operates on.

Populate the given array with all possible OpCodes this Rule might apply to. By default, this method returns all possible OpCodes

Parameters

oplist

is the array to populate

Reimplemented from Rule.

---

The documentation for this class was generated from the following files:

• ruleaction.hh
• ruleaction.cc