Bug 15750 - Various functions raise FE_INEXACT when the results are exact and representable
Summary: Various functions raise FE_INEXACT when the results are exact and representable
Status: RESOLVED INVALID
Alias: None
Product: glibc
Classification: Unclassified
Component: math (show other bugs)
Version: 2.16
: P2 normal
Target Milestone: ---
Assignee: Not yet assigned to anyone
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2013-07-17 13:02 UTC by Jeremy
Modified: 2014-06-13 09:23 UTC (History)
1 user (show)

See Also:
Host:
Target:
Build:
Last reconfirmed:
fweimer: security-

Attachments
Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Jeremy 2013-07-17 13:02:43 UTC 
Examples are log10(100), cbrt(27), exp10(2), log2(2), pow(4.0, 0.5)

These all return the correct exact value, but FE_INEXACT is raised.
Comment 1 Joseph Myers 2013-07-17 13:27:44 UTC 
None of these are functions (such as fma or sqrt) for which Annex F specifies correct "inexact" exceptions, and as such correctly raising or not raising "inexact" for them is outside the goals for correctness of results that I described at <http://sourceware.org/ml/libc-alpha/2013-05/msg00132.html>.

If you find bugs that are bugs according to the goals as described there, please file separate bugs in Bugzilla for each function, and each distinct issue (missing exceptions / spurious exceptions / inaccurate return value / missing errno / incorrect errno), after checking the bug isn't already filed; omnibus bug reports about "various functions" should be avoided, as liable to lead to a long time when a bug is partly fixed, a state Bugzilla is not good at tracking.  (A single bug can cover all of the float, double and long double versions of a function, if they appear to share an issue.)