C library: additional floating-point functions and cleanup #8195
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
fd07bb3 to
406a39a
Compare
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## develop #8195 +/- ##
========================================
Coverage 79.66% 79.66%
========================================
Files 1682 1682
Lines 195377 195377
========================================
Hits 155641 155641
Misses 39736 39736 ☔ View full report in Codecov by Sentry. |
kroening
reviewed
Feb 8, 2024
| @@ -854,11 +899,14 @@ float sqrtf(float f) | |||
| // number of exponent and significand bits. Thus they are | |||
| // given implicitly... | |||
|
|
|||
| #pragma CPROVER check push | |||
| #pragma CPROVER check disable "float-overflow" | |||
| float lowerSquare = lower * lower; | |||
| __CPROVER_assume(__CPROVER_isnormalf(lowerSquare)); | |||
There was a problem hiding this comment.
A square root may well be denormal?
There was a problem hiding this comment.
I don't think that's possible for the case that f (which we are taking the square root of) is normal, which this branch assumes: given a denormal number is smaller than 1, any denormal square root s cannot be smaller than f (for f has to be denormal when s is denormal.
kroening
approved these changes
Feb 8, 2024
Follows the same approximation approach as previously taken for log (and logf, logl).
We previously ended up with recursion.
Model floating-point exceptions as failing assertions.
Model fused multiply-add as documented in its man page.
We construct NaN (and Inf) by dividing by zero, which is a standards-compliant way in that Nan (Inf) is the correct result for these cases. Do not flag these operations as division-by-zero, which the user would not expect.
Fail as documented rather than via built-in assertions when overflowing.
We guess upper and lower bounds and check them for infinity afterwards. Those multiplications should not be flagged by auto-generated assertions.
c62afc8 to
085068c
Compare
Visual Studio knows static_assert, but not the standardised _Static_assert.
Visual Studio uses `1.000000e+300 * 1.000000e+300` as value behind the `HUGE_VAL` macro. We should not report an overflow for this approach of producing infinity.
This was referenced Aug 1, 2024
tautschnig
added a commit
to model-checking/kani
that referenced
this pull request
Aug 2, 2024
Implemented in CBMC in diffblue/cbmc#8195.
tautschnig
added a commit
to model-checking/kani
that referenced
this pull request
Aug 2, 2024
Implemented in CBMC in diffblue/cbmc#8195.
tautschnig
added a commit
to model-checking/kani
that referenced
this pull request
Aug 2, 2024
CBMC's sqrt* implementations were fixed in diffblue/cbmc#8195.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Please see individual commit messages.