<html><head></head><body style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space; "><div>ARMv6 binaries updated accordingly.</div><br><div><div>On Dec 6, 2014, at 19:17 , Eliot Miranda wrote:</div><br class="Apple-interchange-newline"><blockquote type="cite"><div dir="ltr">Oops, for <a href="http://www.mirandabanda.org/files/Cog/VM/VM.r3163/">http://www.mirandabanda.org/files/Cog/VM/VM.r3163/</a> read <a href="http://www.mirandabanda.org/files/Cog/VM/VM.r3164">http://www.mirandabanda.org/files/Cog/VM/VM.r3164</a><div class="gmail_extra"><br><div class="gmail_quote">On Sat, Dec 6, 2014 at 1:53 PM, Eliot Miranda <span dir="ltr"><<a href="mailto:eliot.miranda@gmail.com" target="_blank">eliot.miranda@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex"><div dir="ltr">... at <a href="http://www.mirandabanda.org/files/Cog/VM/VM.r3163/" target="_blank">http://www.mirandabanda.org/files/Cog/VM/VM.r3163/</a>.<div><br></div><div><div>CogVM binaries as per VMMaker.oscog-eem.972/r3163<br></div><div><br></div><div>Fix a devilishly tricky bug with 32-bit Spur and PICs in the Cogit. In 32-bit</div><div>Spur only zero is a valid cache tag, that for Characters, due to SmallIntegers</div><div>having both 2r11 and 2r01 as valid tags. The cache check prolog collapses</div><div>these down onto 1 by anding with 1 and hence collapses Character's 2r10 to 0.</div><div><br></div><div>The cache tag is also tested in the PIC abort routine to distinguish between</div><div>an MNU abort and an interpret abort. A PIC MNU entry jumps to code that first</div><div>zeros the cache tag (held in ClassReg) before calling the abort routine, which</div><div>tests ClassReg and selects the relevant abort processing.</div><div><br></div><div>Hence in 32-bit Spur if a send site is linked for Character and later extended</div><div>to a PIC with an interpret case, that interpret case will instead invoke MNU</div><div>processing. This is rare; first there needs to be a send to a CHaracter, then</div><div>a send to something else which binds to something that must be interpreted.</div><div>Ouch.</div><div><br></div><div>The fix is simple; if a send site that was linked with a zero cache tag (which</div><div>can only happen in 32-bit Spur) link to an open PIC instead of to a closed PIC,</div><div>or in the case of an MNU do not link.</div><div><br></div><div>Add a debug lookup routine that only looks up (and follows forwarding pointers).</div><div><br></div><div>Add a -breakmnu vm argument that calls warning on an mnu of a specific selector.</div><div><br></div><div>Fix SmartSyntaxPlugin code generation for Spur of</div><div><br></div><div>- asOop: old code assumed 4 byte oops & headers</div><div>- Unsigned old code assumed 32-bit oops</div><div><br></div><div><br></div><div>Affects AsyncFilePlugin, BochsIA32Plugin & FT2Plgin.</div><div><br></div><div>Avoid duplication of effort in special selector send forwarding fault handling.</div><div><br></div><div>Improve stack trace printing for Newspeak and Glue by providing something that</div><div>assumes the penultimate literal is the selector or an AdditionalMethodState-like</div><div>thing containing it.</div><div><br></div><div>Reimplement primitiveCopyObject to work for both pointer and bits objects.</div><div>This to support a good Object>>clone for Spur.</div><div><br></div><div>Fix Spur's isWordsOrBytes[NonImm]:. Old code answered true for CompiledMethods.</div><span class=""><font color="#888888">-- <br><div>best,<div>Eliot</div></div>
</font></span></div></div>
</blockquote></div><br><br clear="all"><div><br></div>-- <br><div class="gmail_signature">best,<div>Eliot</div></div>
</div></div>
<br></blockquote></div><br></body></html>