Re: icc bug on test-frexp-nolibm

[Eric Blake]

On 11/05/2010 01:57 PM, Eric Blake wrote:
> That is, with no options, icc defaults to -O2.  With -g it implies -O0
> which in turn implies -mieee-fp.  It is also possible to use -mieee-fp
> in isolation, or alongside an explicit -O2.  The "bug", then, is that
> when optimization is enabled and -mieee-fp is not in effect, the
> compiler changes behavior of casting float to double such that denormal
> floats become 0.0 rather than the proper full-precision double
> representation.  And, since the ABI passes all floats to function calls
> by converting to double, we've already lost - the denormal float has
> already been converted to zero prior to the gnulib function receiving an
> argument.

Correction - that is only in the case of passing a float to a varargs
function like printf; regular functions do have a difference of passing
only the 4-byte float instead of the 8-byte double, but with printf
behaving differently, it was harder to see when the float->double
conversion was truncating denormal floats.

> 
> I'm debating whether the fix is to extend the fpieee module to recognize
> the ICC compiler, and make any <math.h> float-related tests depend on
> fpieee, or to relax the testsuite to somehow expose when a compiler is
> not performing IEEE-compliant conversion of float denormals to double
> and skipping that part of the test.

I'm still trying to figure out a reliable method for doing runtime
testing with ICC code to sniff out whether float denormals might be
chopped by the current compiler options.

-- 
Eric Blake   address@hidden    +1-801-349-2682
Libvirt virtualization library http://libvirt.org

signature.asc
Description: OpenPGP digital signature