svn commit: r227538 - head/tools/build

[John Baldwin]

On Tuesday, November 15, 2011 4:52:02 pm Dimitry Andric wrote:
> On 2011-11-15 22:14, David Schultz wrote:
> > On Tue, Nov 15, 2011, Dimitry Andric wrote:
> >> Note all the final executables will use 'real' atomic operations.  That
> >> is, unless you compile with CPUTYPE?=i386, and I wish you the best of
> >> luck in that case, you'll need it. :)
> > 
> > I thought we dropped support for anything less than a 486DX years ago.
> 
> Well, theoretically you could still attempt to build for i386, but it is
> almost sure to have bitrotted beyond any working state.  The CPUTYPE
> stuff in bsd.cpu.mk still seems to support it, as does gcc itself, of
> course.  And as mentioned in the commit message, until 2 years ago, the
> default CPU for gcc was i386; which is the only reason for this commit.
> 
> Indeed, I386_CPU support was removed from GENERIC more than 10 years
> ago, in r71025.  But even there it says:
> 
> "Remove I386_CPU from GENERIC.  Support for the 386 seriously pessimizes
> performance on other x86 processors.  Custom kernels can still be built
> that will run on the 386."
> 
> I wonder if that's still true... maybe it is time to really clean up any
> pre-i486 leftovers. :)

I think I removed all the i386 bits from the kernel a few years ago already:

r137784 | jhb | 2004-11-16 15:42:32 -0500 (Tue, 16 Nov 2004) | 3 lines

Initiate deorbit burn sequence for 80386 support in FreeBSD:  Remove
80386 (I386_CPU) support from the kernel.

-- 
John Baldwin