Number: 259

Date: 22-Mar-84 11':49':10

Submitter: Sannella.PA

Source: Masinter.pa

Subject: GC under TEdit causes 9305

Assigned To: 

Attn: Sybalsky.pa

Status: Incomplete

In/By: 

Problem Type: Bug

Impact: Serious

Difficulty: Moderate

Frequency: Intermittent

Priority: Hopefully

System: Text

Subsystem: TEdit

Machine: 1108

Disk: 

Lisp Version: 15-Mar-84

Source Files: 

Microcode Version: 

Memory Size: 

File Server: 

Server Software Version: 

Disposition: 

Description: '
To': LispSupport.pa'
cc': Masinter.pa'
Lisp-System-Date': 15-Mar-84 00':13':18'
Machine-Type': Dandelion'
'
I dunno if this is DLion microcode or TEdit or a combination of both. Something that TEDIT has on its stack when a GC gets invoked causes the DLion GC to get a 9305. Probably it is trying to reference count on the stack something that isn''t a valid pointer. I guess this is a DLion microcode problem, but it is pretty serious. Could be fixed by either fixing DLion microcode or TEdit, I would guess.'
'
Intermittent, Serious, Attn': Charnley, Sybalsky, Hopefully, Difficulty': Moderate'
'
-----'
'
Date':  4 Apr 84 19':34 PST'
From': vanMelle.pa'
...'
259': Comment': if this is a Dandelion ucode problem, it is also a Dolphin/Dorado ucode problem, because they fall into Raid when GC tries to mark the stack bit on a bogus pointer.  So it seems clearly in Tedit''s domain.  On the other hand, you might add an AR wish for all three microcodes, "GCREF should not cause invalid address for pointers out of bounds".'


Workaround: 

Test Case: 

Edit-By: Sannella.PA

Edit-Date:  5-Apr-84 12':08':57