Number: 1401

Date: 12-Jun-84 16':57':03

Submitter: Sannella.PA

Source: Masinter.pa

Subject: GCSCANSTACK needs to avoid reference counts that would cause MP halts

Assigned To: 

Attn: 

Status: Closed

In/By: 

Problem Type: 

Impact: 

Difficulty: 

Frequency: 

Priority: 

System: Language Support

Subsystem: Stack and Interpreter

Machine: 

Disk: 

Lisp Version: 

Source Files: 

Microcode Version: 

Memory Size: 

File Server: 

Server Software Version: 

Disposition: '
["Sannella" "20-Aug-84 19':02':28" Status':(Fixed->Closed)]

Description: '
Date': 10 Jun 84 17':08 PDT'
From': Masinter.pa'
Subject': GCSCANSTACK needs to avoid reference counts that would cause MP halts'
To': LispSupport, vanMelle, charnley'
cc': Masinter.pa'
'
I think all that''s necessary in GCSCANSTACK is to turn the \GETBASEPTR into a (\VAG2 (LOGAND 77Q (\GETBASE --)) (\GETBASE --]'
'
It would be nicer to fix the microcode not to crash when reference counting a non-pointer, but this is probably easier and means that the microcode has fewer dependencies on a 22-bit address space.'
'
-----'
'
Date': 11 Jun 84 14':47 PDT'
From': vanMelle.pa'
Subject': Re': GCSCANSTACK needs to avoid reference counts that would cause MP halts'
In-reply-to': Masinter.pa''s message of 10 Jun 84 17':08 PDT'
To': Masinter.pa'
cc': LispSupport.pa, vanMelle.pa, charnley.pa'
'
Fine.  I have fixed \GCSCANSTACK to stkref only the bottom 22 bits of random pointers on the stack.  Onward!'
'
	Bill'
'
-----'
'
Date': 11 Jun 84 15':25 PDT'
From': JonL.pa'
Subject': Re': GCSCANSTACK needs to avoid reference counts that would cause MP halts'
In-reply-to': vanMelle.pa''s message of 11 Jun 84 14':47 PDT'
To': vanMelle.pa'
cc': Masinter.pa, LispSupport.pa, charnley.pa'
'
I presume this problem is solely due to the "kludge" of trying unboxed floating point by leaving random bits on the stack?'
'
If so, it would be wise to leave some kind of marker at each place where the 22-bit limitation gets built-in, so that we can change it as-and-when (I seem to remember Fall 84 as the promised delivery time for 24-bit pointers).'
'
-- JonL --'
'


Workaround: 

Test Case: 

Edit-By: Sannella

Edit-Date: 20-Aug-84 19':02':29