[Bug math/34465] New: math: x86_64 tanh ifunc selection wrong

https://sourceware.org/bugzilla/show_bug.cgi?id=3D34465

            Bug ID: 34465
           Summary: math: x86_64 tanh ifunc selection wrong
           Product: glibc
           Version: 2.44
            Status: NEW
          Severity: normal
          Priority: P2
         Component: math
          Assignee: unassigned at sourceware dot org
          Reporter: adhemerval.zanella at linaro dot org
  Target Milestone: ---

Introduced by ebb12ffa3e8 ("x86: Fix tanh ifunc selection"), tanhf32x/tanhf=
64
bypass the ifunc and bind directly to __tanh_fma. This trigger regressions =
as:

$ qemu-x86_64 -cpu Nehalem math/test-float64-tanh
testing _Float64 (without inline functions)
qemu: uncaught target signal 4 (Illegal instruction) - core dumped
Illegal instruction        qemu-x86_64 -cpu Nehalem math/test-float64-tanh

Initial discussiona at:
https://sourceware.org/pipermail/libc-help/2026-July/007090.html

--=20
You are receiving this mail because:
You are on the CC list for the bug.=