[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#411813: linux-2.6/alpha: math-emu built as a module, should be built-in



Package: linux-2.6
Version: 2.6.18-7
Severity: important

Throughout the 2.6 kernel series, users of alpha systems of ev56-class and
below have been reporting problems with certain fp operations not working
correctly that did work when using 2.4 kernels[1].

The source of this behavior has recently been identified[2] -- the math-emu
driver is configured as a module instead of being built-in, which it never
was before, and without this driver the kernel doesn't provide complete IEEE
semantics.

The upstream description of this config option says:

        This option is required for IEEE compliant floating point arithmetic
        on the Alpha. The only time you would ever not say Y is to say M in
        order to debug the code. Say Y unless you know what you are doing.

Not only that, it's not even *possible* to disable this driver when building
an SMP kernel.  So the behavior is inconsistent between the SMP and UP alpha
kernel images; upstream discourages modularizing this driver at all except
for debugging; there is definitely a wide range of hardware where one would
never want to have this driver disabled as it will break fp math; and users
aren't going to know to look for this module.

Unless someone has a good reason why this should be built as a module, I
plan to update the configs for inclusion in the -11 upload.

Thanks,
-- 
Steve Langasek                   Give me a lever long enough and a Free OS
Debian Developer                   to set it on, and I can move the world.
vorlon@debian.org                                   http://www.debian.org/

[1] http://lists.debian.org/debian-alpha/2006/06/msg00073.html
[2] http://lists.debian.org/debian-alpha/2007/02/msg00016.html



Reply to: