29 October 2017, 06:42 | #1 |
Registered User
Join Date: May 2013
Location: Grimstad / Norway
Posts: 839
|
ReSource bug - can you replicate?
I run KS 40.68 (well, my buggy 4000 grinds to a halt if I use MapRom for more than an hour or so) on my CS060 but I always get a bus error when I do this:
-Disassemble memory $f80000,$ffffff -Search, binary, $6303, forwards Both virgin 6.06 and my old 060 patched version. Instant $80000004 and halt/reboot requester or full self-reboot. Anyone who can confirm/deny? (Works just fine with WinUAE ofc...) EDIT: I wonder if I'm running into 68060 hw bugs here. Having patiently done "Run History Task" in BDebug (which is probably single-stepping instructions) I get the same result as in WinUAE and no errors. CPU060 reports my chip as Mask 0-1f43G,2f43G,0g65V Revision 1. I use "CPU060 NOSTOREBYPASS" from a recommendation way back. Are there other options I should be turning off? EDIT2: Bug went away with "CPU060 NoAllocateOperand". Machine feels a bit more sluggish now. EDIT3: Except it did only half-way.If I search forwards and then backwards it seems like ReSource bugs big time by going into non-mapped address areas. And this time WinUAE agrees with me! Last edited by NorthWay; 29 October 2017 at 09:03. |
29 October 2017, 20:50 | #2 |
Registered User
Join Date: Aug 2004
Location:
Posts: 3,333
|
If you're wanting to disassemble Kickstart, specify $F80000,$1000000 otherwise you miss the last byte.
I could get ReSource to hang doing that search (testing in WinUAE), so it looks like there's a bug in it. Maybe when disassembling memory, it doesn't stop searching on reaching the end of the region? The problem doesn't happen if you dump the memory region to a file, and load that file into ReSource. Last edited by mark_k; 29 October 2017 at 21:01. |
15 November 2017, 02:56 | #3 |
Registered User
Join Date: May 2013
Location: Grimstad / Norway
Posts: 839
|
Rev6 cpu arrived. Much breath holding to get old out and new in.
Boot up. Start Resource. Fall over again. Not cool. I _do_ see a difference in that it will do the search and jump to the correct spot before guruing now."NoAllocateOperand" still fixes the guru. |
14 October 2018, 04:52 | #4 |
Registered User
Join Date: May 2013
Location: Grimstad / Norway
Posts: 839
|
Futzing with my Amiga again and I might have gotten something more aligned as I now see two byte writes to $FFFFFE and $FFFFFF that CyberGuard is catching. Exact same procedure to provoke it.
(Hunk 0000, Offset $000155D4 - I see that is in the code for binary search. There must be some logic breakdown bug here.) |
18 November 2022, 01:59 | #5 |
Registered User
Join Date: May 2013
Location: Grimstad / Norway
Posts: 839
|
Oh, and I tested this with my new BFG9060 - still fails in the exact same way.
Hm. Who was it that was the inofficial ReSource updater nowadays again? |
18 November 2022, 06:01 | #6 |
Registered User
Join Date: Jan 2019
Location: Germany
Posts: 3,214
|
First, you should use a decent 68060.library. I suggest the MMULib provided ones, they map all unused pages to a dummy page which should avoid this alert, despite potentially Resource being buggy. You will not able to see such problems on WinUAE, unless you run it with MMU emulation enabled.
|
Currently Active Users Viewing This Thread: 1 (0 members and 1 guests) | |
Thread Tools | |
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
Bug in x64 file requester and bug in Blizzard PPC ROM filesize | headkase | support.WinUAE | 5 | 26 June 2016 14:17 |
ReSource 6.06 | yoki | request.Apps | 6 | 04 November 2009 17:29 |
ReSource | drago | request.Apps | 2 | 08 July 2008 00:30 |
ReSource 6.XX | MrZammler | request.Apps | 7 | 30 October 2007 14:15 |
ReSource v6.06 | A.I | request.Apps | 3 | 08 October 2005 21:25 |
|
|