Andrew Eisenberg
4ffd8c62ac
Merge pull request #7579 from github/aeisenberg/changenote-upgrades-removal
...
Changenotes: Add changenotes for upgrades refactoring
2022-01-13 09:09:06 -08:00
Anders Schack-Mulligen
c44cf29992
Merge pull request #7587 from owen-mc/add-default-taint-sanitizer-guard
...
Dataflow: Add default taint sanitizer guard
2022-01-13 14:44:55 +01:00
Mathias Vorreiter Pedersen
6148af4621
C++: Fix join order in 'getConversionType4'.
2022-01-13 13:28:36 +00:00
Anders Schack-Mulligen
f7cf327e71
Dataflow: Sync
2022-01-13 13:28:43 +01:00
Paolo Tranquilli
e6763c858d
C++: add bindingset to private Printf predicate
...
That predicate turned out to create a lot of tuples, of which only a
minimal part was then used in the query.
2022-01-13 11:59:48 +00:00
Paolo Tranquilli
64d15d6226
C++: fix inc.qhelp files and change notes
2022-01-13 11:59:48 +00:00
Paolo Tranquilli
7b4300e4cf
C++: Apply suggestions in documentation
...
Co-authored-by: Sarah Edwards <skedwards88@github.com >
2022-01-13 11:59:48 +00:00
Paolo Tranquilli
9d49ad9f20
C++: use includes in OverrunWrite qhelp files
...
Also added the relevant CERT C _and_ C++ standard references where they
were missing, and did some minor stylistic tweaks to
`OverrunWriteFloat.qhelp`.
2022-01-13 11:59:48 +00:00
Paolo Tranquilli
c117a1e21f
C++: demote VeryLikelyOverrunWrite cast results
...
There were some false positives where something like
int x;
// ...
sprintf(buff, "%ld", (long)x);
was considered as if the parameter had a non-trivial range analysis only
because the range of `int` is smaller than the range for `long`, without
any non-trivial range analysis actually done on `x`.
These will now be reported by `OverrunWrite` instead.
2022-01-13 11:59:48 +00:00
Paolo Tranquilli
630982cc31
C++: auto format Printf.qll
2022-01-13 11:59:48 +00:00
Paolo Tranquilli
9f811b2439
C++: remove unused variables and fix tests
2022-01-13 11:59:47 +00:00
Paolo Tranquilli
013216d5e6
C++: exclude widening from VeryLikelyOverrunWrite
...
This also restrict what we consider "non-trivial" range analysis, as we
now require both ends to be non-trivially bounded for signed integers.
This avoids false positives stemming from a non trivial upper bound but
no meaningful lower bound, for example.
2022-01-13 11:59:47 +00:00
Paolo Tranquilli
aac029841a
C++: doc fixes to VeryLikelyOverrunWrite
2022-01-13 11:59:47 +00:00
Paolo Tranquilli
c8741f6475
C++: update 2021-12-14-overruning-write-split.md
...
Co-authored-by: Mathias Vorreiter Pedersen <mathiasvp@github.com >
2022-01-13 11:59:47 +00:00
Paolo Tranquilli
1e4861a944
C++: shorten VeryLikelyOverrunWrite @name
2022-01-13 11:59:47 +00:00
Paolo Tranquilli
106400238a
C++: tweak overrunning write qhelp files
2022-01-13 11:59:47 +00:00
Paolo Tranquilli
8ac34f3db5
C++: NoSpecifiedEstimateReason→Unspecified...
2022-01-13 11:59:47 +00:00
Paolo Tranquilli
4a85b9b0cc
C++: add VeryLikelyOverrunWrite.ql to cwe-120
2022-01-13 11:59:47 +00:00
Paolo Tranquilli
97f1a5bac0
C++: add VeryLikelyOverrunWrite.qhelp
2022-01-13 11:59:47 +00:00
Paolo Tranquilli
10b62154a1
C++: add cpp/very-likely-overruning-write help
...
Also update the help of `cpp/overruning-write`, as the case shown there
will actually not be flagged by that query any more.
2022-01-13 11:59:47 +00:00
Paolo Tranquilli
b979f02e5d
C++: fix OverrunWrite for backward compatibility
...
Rather than testing for `TypeBoundsAnalysis`, we test that the reason is
not `ValueFlowAnalysis` (which is reported by the new
`cpp/very-likely-overruning-write` query), so that if a client has
overridden `BufferWrite::getMaxData` the `NoSpecifiedEstimateReason` is
taken into account.
2022-01-13 11:59:47 +00:00
Paolo Tranquilli
db6214fdff
C++: add change note for new overrun write query
2022-01-13 11:59:47 +00:00
Paolo Tranquilli
a0059202db
C++: split cpp/overrunning-write into two
...
This splits the `cpp/overruning-write` into two separate queries based
off on the reason for the estimation. If the overrun is detected based
on non-trivial range analysis, the results are now marked by the new
`cpp/very-likely-overruning-write` high precision query. If it is based
on less precise, usually type based bounds, then it will still be marked
by `cpp/overruning-write` which remains at medium precision.
2022-01-13 11:59:47 +00:00
Andrew Eisenberg
e435a3e9c3
Changenotes: Add changenotes for upgrades refactoring
2022-01-12 11:36:31 -08:00
Robert Marsh
5031d6c4a3
Merge pull request #7566 from MathiasVP/smaller-join-in-reachesRefParameter
...
C++: Smaller join in `reachesRefParameter`
2022-01-12 10:04:35 -05:00
Owen Mansel-Chan
8e8278764b
Add predicate defaultTaintSanitizerGuard for each language
...
This was done manually, as these files are not synced by sync-files.py.
2022-01-12 14:44:56 +00:00
Owen Mansel-Chan
c112980b81
Sync TaintTrackingImpl.qll
...
Done automatically using sync-files.py
2022-01-12 14:44:55 +00:00
github-actions[bot]
8a2d92badc
Post-release preparation for codeql-cli-2.7.5
2022-01-12 13:28:43 +00:00
Andrew Eisenberg
07228672df
Merge branch 'main' into aeisenberg/remove-upgrades
2022-01-11 11:25:27 -08:00
Mathias Vorreiter Pedersen
c45127fdd6
Merge pull request #7541 from github/rdmarsh2/dataflow-ipa-params
...
C++: Use an IPA type rather than negative indexes for argument/parameter matching in data flow
2022-01-11 16:52:13 +00:00
Mathias Vorreiter Pedersen
b3a7090068
C++: Fix join in reachesRefParameter by joining with 'getEnd' instead
...
of 'getANode'.
Before:
Tuple counts for FlowVar::FlowVar::reachesRefParameter_dispred#ff/2@956ac39i after 229ms:
24806 ~1% {2} r1 = JOIN FlowVar::FlowVar_internal::parameterIsNonConstReference#f WITH Parameter::Parameter::getFunction_dispred#ff ON FIRST 1 OUTPUT Lhs.0 'p', Rhs.1
56985 ~3% {3} r2 = JOIN r1 WITH num#FlowVar::FlowVar_internal::TBlockVar#fff_12#join_rhs ON FIRST 1 OUTPUT Rhs.1 'this', Lhs.0 'p', Lhs.1
2384489 ~4% {4} r3 = JOIN r2 WITH FlowVar::FlowVar_internal::getAReachedBlockVarSBB#ff ON FIRST 1 OUTPUT Rhs.1, Lhs.2, Lhs.1 'p', Lhs.0 'this'
49457 ~0% {2} r4 = JOIN r3 WITH SubBasicBlocks::SubBasicBlock::getANode_dispred#fb ON FIRST 2 OUTPUT Lhs.3 'this', Lhs.2 'p'
return r4
After:
Tuple counts for FlowVar::FlowVar::reachesRefParameter_dispred#ff/2@46f8bfn7 after 32ms:
24806 ~1% {2} r1 = JOIN FlowVar::FlowVar_internal::parameterIsNonConstReference#f WITH Parameter::Parameter::getFunction_dispred#ff ON FIRST 1 OUTPUT Lhs.0 'p', Rhs.1
56985 ~1% {3} r2 = JOIN r1 WITH num#FlowVar::FlowVar_internal::TBlockVar#fff_12#join_rhs ON FIRST 1 OUTPUT Lhs.1, Lhs.0 'p', Rhs.1 'this'
56985 ~1% {3} r3 = JOIN r2 WITH SubBasicBlocks::SubBasicBlock::getEnd_dispred#fb_10#join_rhs ON FIRST 1 OUTPUT Lhs.2 'this', Rhs.1, Lhs.1 'p'
49457 ~0% {2} r4 = JOIN r3 WITH FlowVar::FlowVar_internal::getAReachedBlockVarSBB#ff ON FIRST 2 OUTPUT Lhs.0 'this', Lhs.2 'p'
return r4
2022-01-11 13:48:20 +00:00
Robert Marsh
fe355a0bc9
C++: update test comments
2022-01-10 12:38:08 -05:00
Mathias Vorreiter Pedersen
2a02ce137a
C++: Fix join orders in 'exprIsSubLeftOrLess'.
...
Before:
Tuple counts for UnsignedDifferenceExpressionComparedZero::exprIsSubLeftOrLess#ff/2@i3#a5071w3a after 24s:
304220 ~2% {2} r1 = JOIN UnsignedDifferenceExpressionComparedZero::exprIsSubLeftOrLess#ff#prev_delta WITH Expr::BinaryOperation#class#f#join_rhs ON FIRST 1 OUTPUT Lhs.1, Rhs.0 'sub'
190061335 ~24% {2} r2 = JOIN r1 WITH DataFlowUtil::localFlowStep#ff ON FIRST 1 OUTPUT Lhs.1 'sub', Rhs.1 'n'
3956 ~0% {2} r3 = JOIN r1 WITH DataFlowUtil::localFlowStep#ff_10#join_rhs ON FIRST 1 OUTPUT Lhs.1 'sub', Rhs.1 'n'
407983 ~1% {2} r4 = JOIN Expr::BinaryOperation#class#f#join_rhs WITH UnsignedDifferenceExpressionComparedZero::exprIsSubLeftOrLess#ff#prev ON FIRST 1 OUTPUT Rhs.1 'n', Lhs.0 'sub'
380823 ~0% {2} r5 = JOIN r4 WITH DataFlowUtil::TExprNode#ff_10#join_rhs ON FIRST 1 OUTPUT Lhs.1 'sub', Rhs.1
0 ~0% {2} r6 = JOIN r5 WITH UnsignedDifferenceExpressionComparedZero::isGuarded#fff#prev_delta ON FIRST 2 OUTPUT Rhs.2, Lhs.0 'sub'
0 ~0% {2} r7 = JOIN r6 WITH DataFlowUtil::TExprNode#ff ON FIRST 1 OUTPUT Lhs.1 'sub', Rhs.1 'n'
3956 ~0% {2} r8 = r3 UNION r7
190065291 ~24% {2} r9 = r2 UNION r8
...
After:
Tuple counts for UnsignedDifferenceExpressionComparedZero::interestingSubExpr#f/1@654e29g3 after 228ms:
370 ~2% {2} r1 = ComparisonOperation::RelationalOperation::getGreaterOperand_dispred#fb AND NOT Exclusions::isFromMacroDefinition#b(Lhs.1 'sub')
370 ~0% {2} r2 = SCAN r1 OUTPUT In.1 'sub', In.0
370 ~3% {3} r3 = JOIN r2 WITH Expr::Expr::getFullyConverted_dispred#ff ON FIRST 1 OUTPUT Rhs.1, Lhs.1, Lhs.0 'sub'
210 ~1% {2} r4 = JOIN r3 WITH SimpleRangeAnalysis::SimpleRangeAnalysisCached::exprMightOverflowNegatively#f ON FIRST 1 OUTPUT Lhs.2 'sub', Lhs.1
210 ~0% {3} r5 = JOIN r4 WITH Expr::Expr::getFullyConverted_dispred#ff ON FIRST 1 OUTPUT Lhs.1, Lhs.0 'sub', Rhs.1
210 ~1% {3} r6 = JOIN r5 WITH ComparisonOperation::RelationalOperation::getLesserOperand_dispred#ff ON FIRST 1 OUTPUT Rhs.1, Lhs.1 'sub', Lhs.2
59 ~2% {4} r7 = JOIN r6 WITH Expr::Expr::getValue_dispred#ff ON FIRST 1 OUTPUT Lhs.1 'sub', Lhs.2, Rhs.1, toInt(Rhs.1)
17 ~0% {4} r8 = SELECT r7 ON In.3 = 0
17 ~0% {2} r9 = SCAN r8 OUTPUT In.1, In.0 'sub'
8 ~0% {2} r10 = JOIN r9 WITH Expr::Expr::getUnspecifiedType_dispred#bb ON FIRST 1 OUTPUT Rhs.1, Lhs.1 'sub'
8 ~0% {1} r11 = JOIN r10 WITH Type::IntegralType::isUnsigned_dispred#f ON FIRST 1 OUTPUT Lhs.1 'sub'
return r11
Tuple counts for UnsignedDifferenceExpressionComparedZero::exprIsSubLeftOrLess#ff/2@i2#61800weu after 1ms:
8 ~0% {2} r1 = JOIN UnsignedDifferenceExpressionComparedZero::exprIsSubLeftOrLess#ff#prev_delta WITH UnsignedDifferenceExpressionComparedZero::interestingSubExpr#f ON FIRST 1 OUTPUT Lhs.1, Lhs.0 'sub'
0 ~0% {2} r2 = JOIN r1 WITH DataFlowUtil::localFlowStep#ff ON FIRST 1 OUTPUT Lhs.1 'sub', Rhs.1 'n'
1 ~0% {2} r3 = JOIN r1 WITH DataFlowUtil::localFlowStep#ff_10#join_rhs ON FIRST 1 OUTPUT Lhs.1 'sub', Rhs.1 'n'
0 ~0% {3} r4 = JOIN UnsignedDifferenceExpressionComparedZero::isGuarded#fff#prev_delta WITH UnsignedDifferenceExpressionComparedZero::interestingSubExpr#f ON FIRST 1 OUTPUT Lhs.1, Lhs.0 'sub', Lhs.2
0 ~0% {3} r5 = JOIN r4 WITH DataFlowUtil::TExprNode#ff ON FIRST 1 OUTPUT Lhs.1 'sub', Rhs.1 'n', Lhs.2
0 ~0% {2} r6 = JOIN r5 WITH UnsignedDifferenceExpressionComparedZero::exprIsSubLeftOrLess#ff#prev ON FIRST 2 OUTPUT Lhs.2, Lhs.0 'sub'
0 ~0% {2} r7 = JOIN r6 WITH DataFlowUtil::TExprNode#ff ON FIRST 1 OUTPUT Lhs.1 'sub', Rhs.1 'n'
1 ~0% {2} r8 = r3 UNION r7
1 ~0% {2} r9 = r2 UNION r8
...
2022-01-10 17:28:14 +00:00
Mathias Vorreiter Pedersen
f2d6bcd767
C++: Fix join order in 'isGuarded'.
...
Before:
Tuple counts for UnsignedDifferenceExpressionComparedZero::isGuarded#bff/3@ec24001m after 1.7s:
97431 ~0% {2} r1 = JOIN UnsignedDifferenceExpressionComparedZero::isGuarded#bff#join_rhs WITH project#BasicBlocks::Cached::basic_block_member ON FIRST 1 OUTPUT Rhs.1, Lhs.0 'sub'
11809769 ~1% {2} r2 = JOIN r1 WITH Guards::GuardCondition::controls_dispred#fff_10#join_rhs ON FIRST 1 OUTPUT Lhs.1 'sub', Rhs.1
11809769 ~0% {4} r3 = JOIN r2 WITH project#BasicBlocks::Cached::basic_block_member ON FIRST 1 OUTPUT Lhs.1, Rhs.1, false, Lhs.0 'sub'
629277 ~4% {7} r4 = JOIN r3 WITH Guards::GuardCondition::ensuresLt_dispred#ffffff_045123#join_rhs ON FIRST 3 OUTPUT Lhs.3 'sub', Lhs.0, Lhs.1, false, Rhs.3 'left', Rhs.4 'right', Rhs.5
628120 ~4% {7} r5 = SELECT r4 ON In.6 >= 0
628120 ~1% {3} r6 = SCAN r5 OUTPUT In.0 'sub', In.4 'left', In.5 'right'
return r6
After:
Tuple counts for UnsignedDifferenceExpressionComparedZero::isGuarded#fff/3@i2#a5071x3a after 392ms:
103763 ~0% {2} r1 = SCAN UnsignedDifferenceExpressionComparedZero::exprIsSubLeftOrLess#ff#prev_delta OUTPUT In.0 'sub', 26
103763 ~0% {1} r2 = JOIN r1 WITH exprs ON FIRST 2 OUTPUT Lhs.0 'sub'
97431 ~0% {3} r3 = JOIN r2 WITH project#BasicBlocks::Cached::basic_block_member ON FIRST 1 OUTPUT Rhs.1, false, Lhs.0 'sub'
629277 ~0% {7} r4 = JOIN r3 WITH Guards::GuardCondition::ensuresLt_dispred#ffffff_450123#join_rhs ON FIRST 2 OUTPUT Lhs.2 'sub', Lhs.0, false, Rhs.2, Rhs.3 'left', Rhs.4 'right', Rhs.5
628120 ~0% {7} r5 = SELECT r4 ON In.6 >= 0
628120 ~1% {6} r6 = SCAN r5 OUTPUT In.0 'sub', In.1, In.3, In.4 'left', In.5 'right', In.6
628120 ~1% {6} r7 = r6 AND NOT UnsignedDifferenceExpressionComparedZero::isGuarded#fff#prev(Lhs.0 'sub', Lhs.3 'left', Lhs.4 'right')
628120 ~0% {5} r8 = SCAN r7 OUTPUT In.2, In.1, In.0 'sub', In.3 'left', In.4 'right'
628120 ~1% {3} r9 = JOIN r8 WITH Guards::GuardCondition::controls_dispred#fff ON FIRST 2 OUTPUT Lhs.2 'sub', Lhs.3 'left', Lhs.4 'right'
return r9
2022-01-10 17:03:40 +00:00
Tom Hvitved
d2ebbe0819
Merge pull request #7469 from hvitved/csharp/promote-adhoc-consistency-checks
...
C#: Promote existing ad-hoc consistency checks to consistency queries
2022-01-10 11:10:25 +01:00
Mathias Vorreiter Pedersen
a5ccd6a23b
Merge pull request #7521 from rdmarsh2/rdmarsh2/cpp/use-guards-in-overflow
2022-01-09 14:09:04 +00:00
Robert Marsh
67fb48fcc1
C++: use range analysis for hex format lengths
...
The "new" result on line 189 is a tighter bound than was previously
established, not a newly introduced location.
2022-01-07 16:16:22 -05:00
Robert Marsh
fa9242befe
C++: Add tests for bounded hex format values
2022-01-07 16:08:53 -05:00
Robert Marsh
673399719e
C++: autoformat DataFlowPrivate
2022-01-07 15:23:24 -05:00
Robert Marsh
78b8d113bb
C++: PR comments on DataFlow Position
2022-01-07 14:21:56 -05:00
Robert Marsh
4322a39807
C++: fix typo in Overflow.qll abs handling
2022-01-07 14:09:47 -05:00
Robert Marsh
a126154dfb
C++: use -1 for this in dataflow Position
2022-01-07 11:39:26 -05:00
Robert Marsh
1890a14026
C++: IPA for pointer arg instead of negative index
...
This takes advantage of the new ArgumentPosition and ParameterPosition
types in the shared DataFlow library interface to represent indirections
with an IPA type rather than the negative-index system in use previously
2022-01-07 11:39:26 -05:00
Robert Marsh
4f23cce63b
C++: Accept more test output
2022-01-07 11:27:45 -05:00
Mathias Vorreiter Pedersen
4ee653378e
Merge pull request #7517 from MathiasVP/avoid-self-joins-in-toctou-query
...
C++: Remove bad self joins in `cpp/toctou-race-condition`.
2022-01-07 13:08:30 +00:00
Robert Marsh
c6da1f2be0
C++: re-add comment
2022-01-06 12:43:22 -05:00
Robert Marsh
355fc0ae63
C++: Use Guards library in Overflow.qll
...
Replaces the ad-hoc guard handling with the Guards library. Fixes an
observed false positive pattern, and (hopefully) means some pragmas are
no longer necessary for performance.
2022-01-06 12:15:37 -05:00
Robert Marsh
617bdbc5ba
C++: test for guard-by-return in Overflow.qll
2022-01-06 12:15:37 -05:00
Mathias Vorreiter Pedersen
173cefd7e4
C++: Respond to PR reviews.
2022-01-06 15:39:40 +00:00
Mathias Vorreiter Pedersen
671954025d
C++: Fix qldoc.
2022-01-06 11:02:15 +00:00