[Bug math/21171] New: log10, log2 and lgamma

["tuliom at linux dot vnet.ibm.com" <sourceware-bugzilla at sourceware dot org>]

https://sourceware.org/bugzilla/show_bug.cgi?id=21171

            Bug ID: 21171
           Summary: log10, log2 and lgamma
           Product: glibc
           Version: 2.25
            Status: NEW
          Severity: normal
          Priority: P2
         Component: math
          Assignee: unassigned at sourceware dot org
          Reporter: tuliom at linux dot vnet.ibm.com
  Target Milestone: ---

This can be reproduced after applying the following patch:
https://sourceware.org/ml/libc-alpha/2017-02/msg00241.html

Or using the branch tuliom/float128 in a powerpc64le machine.

It affects float, double and long double. I tested on powerpc, powerpc64,
powerpc64le and x86_64.
In the float128 branch it also affects _Float128 on powerpc64le.

The following is a fragment of the affected functions:

Failure: Test: log10_downward (0)
Result:
 is:          inf   inf
 should be:  -inf  -inf
infinity has wrong sign.
Failure: Test: log10_downward (-0)
Result:
 is:          inf   inf
 should be:  -inf  -inf
infinity has wrong sign.
Failure: Test: log2_downward (0)
Result:
 is:          inf   inf
 should be:  -inf  -inf
infinity has wrong sign.
Failure: Test: log2_downward (-0)
Result:
 is:          inf   inf
 should be:  -inf  -inf
infinity has wrong sign.
Failure: Test: lgamma (-max_value)
Result:
 is:         -inf  -inf
 should be:   inf   inf
infinity has wrong sign.
Failure: Test: lgamma_downward (-max_value)
Result:
 is:         -inf  -inf
 should be:   inf   inf
infinity has wrong sign.
Failure: Test: lgamma_towardzero (-max_value)
Result:
 is:         -inf  -inf
 should be:   inf   inf
infinity has wrong sign.
Failure: Test: lgamma_upward (-max_value)
Result:
 is:         -inf  -inf
 should be:   inf   inf

-- 
You are receiving this mail because:
You are on the CC list for the bug.