[Vm-dev] All white screen on Arm 32 with commit 71ffd3fc0eb74519e2ddd8ea4de87503e925fa16

Eliot Miranda eliot.miranda at gmail.com
Sun Feb 6 19:00:07 UTC 2022


Hi Bruce,

    thanks for the prompt feedback.  I think I've fixed it.  Alas I can't
run 32-bit binaries on my mac because the Cocoa GUi no longer compiles
under supported MacOS versions.  I should have checked with the simulator.
Please check commit a29dc4bea18c728d40f8b16dfe879642d0410d91

    CogVM sources as per VMMaker.oscog-eem.3146

    InterpreterPrimitives: fix the 64-bit large integer support on 32-bit
platforms.

On Sun, Feb 6, 2022 at 7:02 AM Bruce O'Neel <bruce.oneel at pckswarms.ch>
wrote:

>
>
> Hi all,
>
> The commit below causes an all white screen on 32 bit Squeak.  Well at
> least Arm 32.
>
> Thanks.
>
> bruce
>
> CogVM source as per Name: VMMaker.oscog-eem.3144
>
> Spur CoInterpreter: harmonize named and numbered primitive Spur primitive
> property flags. Use a common pragma, primitiveMetadata:, to mark named
> primitives and numbered primitives.  Mark several methods with the
> PrimCallOnSmalltalkStack (a.k.a. FastCPrimitiveFlag) flag, especially
> BoxedFloat, SmallFloat, and 64-bit LargeInteger primitives in
> InterpreterPrimitives.  Make sure that vmParameterAt: uses that flag, even
> though the other versions of primitive 254 do not.
>
> Nuke support for the old pre-Alien callbacks.  Only very old versions of
> two plugins use this, and it's easy to upgrade to use Alien callbacks.
> (missing from the Monticello commit message)
>
> Spur Cogit: Instead of setting newMethod & primitiveFunctionPointer for all
> primitives, only set newMethod  & primitiveFunctionPointer on failure (unless
> a primitive uses the PrimCallNeedsNewMethod flag).  Use a CallRT to invoke
> internal primitives if it spans far enough.
>
>
>

-- 
_,,,^..^,,,_
best, Eliot
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.squeakfoundation.org/pipermail/vm-dev/attachments/20220206/b8d9ba07/attachment-0001.html>


More information about the Vm-dev mailing list