hohn/codeql
1
0
Fork 0
mirror of https://github.com/github/codeql.git synced 2026-04-25 00:35:20 +02:00
Code Issues Packages Projects Releases Wiki Activity

C++: Fix perf of IR value numbering

Browse Source 
On some snapshots, notably ffmpeg, the IR `ValueNumbering` recursion
would generate billions of tuples and eventually run out of space.

It turns out it was fairly common for an `Instruction` to get more than
one `ValueNumber` in the base cases for `VariableAddressInstruction` and
`InitializeParameterInstruction`, and it could also happen in an
instruction with more than one operand of the same `OperandTag`. When a
binary operation was applied to an instruction with `m` value numbers
and another instruction with `n` value numbers, the result would get
`m * n` value numbers. This led to doubly-exponential growth in the
number of value numbers in rare cases.

The underlying reason why a `VariableAddressInstruction` could get
multiple value numbers is that it was keyed on the associated
`IRVariable`, and the `IRVariable` is defined in part by the type of its
underlying `Variable` (or other AST element). If the extractor defines a
variable to have multiple types because of linker ambiguity, this leads
to the creation of multiple `IRVariable`s. That should ideally be solved
in `TIRVariable.qll`, but for now I've put a workaround in
`ValueNumberingInternal.qll` instead.

To remove the problem with instructions having multiple operands, the
construction in `Operand.qll` will now filter out any such operand. It
wasn't enough to apply that filter to the `raw` stage, so I've applied
it to all three stages.
This commit is contained in:
Jonas Jensen
2020-02-12 14:04:03 +01:00
parent de45b1a08e
commit 033a4c30ea
8 changed files with 91 additions and 43 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
csharp/ql/src/semmle/code/csharp/ir/implementation/raw/Operand.qll
View File

@@ -11,13 +11,19 @@ cached
private newtype TOperand =
TRegisterOperand(Instruction useInstr, RegisterOperandTag tag, Instruction defInstr) {
defInstr = Construction::getRegisterOperandDefinition(useInstr, tag) and
not Construction::isInCycle(useInstr)
not Construction::isInCycle(useInstr) and
strictcount(Construction::getRegisterOperandDefinition(useInstr, tag)) = 1
} or
TNonPhiMemoryOperand(
Instruction useInstr, MemoryOperandTag tag, Instruction defInstr, Overlap overlap
) {
defInstr = Construction::getMemoryOperandDefinition(useInstr, tag, overlap) and
not Construction::isInCycle(useInstr)
not Construction::isInCycle(useInstr) and
(
strictcount(Construction::getMemoryOperandDefinition(useInstr, tag, _)) = 1
or
tag instanceof UnmodeledUseOperandTag
)
} or
TPhiOperand(
PhiInstruction useInstr, Instruction defInstr, IRBlock predecessorBlock, Overlap overlap

10
csharp/ql/src/semmle/code/csharp/ir/implementation/unaliased_ssa/Operand.qll
View File

@@ -11,13 +11,19 @@ cached
private newtype TOperand =
TRegisterOperand(Instruction useInstr, RegisterOperandTag tag, Instruction defInstr) {
defInstr = Construction::getRegisterOperandDefinition(useInstr, tag) and
not Construction::isInCycle(useInstr)
not Construction::isInCycle(useInstr) and
strictcount(Construction::getRegisterOperandDefinition(useInstr, tag)) = 1
} or
TNonPhiMemoryOperand(
Instruction useInstr, MemoryOperandTag tag, Instruction defInstr, Overlap overlap
) {
defInstr = Construction::getMemoryOperandDefinition(useInstr, tag, overlap) and
not Construction::isInCycle(useInstr)
not Construction::isInCycle(useInstr) and
(
strictcount(Construction::getMemoryOperandDefinition(useInstr, tag, _)) = 1
or
tag instanceof UnmodeledUseOperandTag
)
} or
TPhiOperand(
PhiInstruction useInstr, Instruction defInstr, IRBlock predecessorBlock, Overlap overlap