Marko Koci=C4=87 writes:
> Just one quick question before I try them, do this set of patches
> conflict with relocation patch?
I'm sorry, I don't think I made myself clear. These patches, in and of
themselves, are insufficient to get threading running on windows. They ar=
(at least most of) the compiler changes required, but not the runtime
support. I built and tested everything on x86/linux. If you want to try t=
on win32, just use the :x86-two-arg-passing-regs and :x86-reserve-ebx
features (which won't get you threads, but should test the rest of the
But I can't imagine them conflicting with the relocation patch.
> On 5/9/06, Alastair Bridgewater <nyef_sbcl@...> wrote:
>> Hello all,
>> At http://www.lisphacker.com/temp/sbcl-ebx-threads/ is a set of patche=
>> against 0.9.12 that changes threading on x86 to use EBX as a TLS
>> The changes are mostly conditional-compiled in based on the features
>> :x86-two-arg-passing-regs, :x86-reserve-ebx, and :x86-ebx-threads. We
>> this change or one like it for Win32 threads, as we can't allocate a
>> selector to use for GS (FS is already in use) on that platform.
>> The only known bug with this patch set is that a call-out to C compile=
>> (> space speed) will call call_into_c in x86-assem.S, which hasn't bee=
>> patched not to use EBX.
>> During the course of making these changes I discovered that
>> src/assembly/x86/alloc.lisp is comprised entirely of dead code.
>> Feedback on these patches would be appreciated.
>> --Alastair Bridgewater