Previous Thread
Next Thread
Print Thread
Page 17 of 18 1 2 15 16 17 18
Re: Siemens PC-D [Re: rfka01] #117515 07/07/20 03:01 AM
Joined: Jun 2020
Posts: 13
A
anyfoo Offline
Member
Offline
Member
A
Joined: Jun 2020
Posts: 13
And coming back to 84xx vs 86xx: I have no good explanation. Only the last few pages of segment 3, which are part of the massive linear kernel code mapping starting in segment 1, are mapped using 86xx. I have not seen any other pages being mapped through 86xx, and the page boundary between the last 84xx and the first 86xx page is just in the middle of what very much still looks to be linear kernel code (in fact, the EXOS 206 driver seems to be in there at the end).

Really weird. So this is either some kind of hint bit, or equivalent to 84xx. I almost wonder whether this is a flipped bit from the original media? I haven't checked if there's anything guarding against that (and then I haven't checked why only those last pages end up using the different port).

Re: Siemens PC-D [Re: anyfoo] #117599 07/27/20 11:58 PM
Joined: Jan 2012
Posts: 844
C
crazyc Offline
Senior Member
Offline
Senior Member
C
Joined: Jan 2012
Posts: 844
Quote
Bit 5 is either the opposite "read allow" bit, or maybe a more general "present" bit. The difference would be that a "present" bit would fault any access, read or write, as long as this bit is not set, whereas "read allow" would still allow writes as long as "write allow" is set. The distinction seems moot for SINIX, as I haven't observed it ever trying to write to a page that did not also have this bit set. I mentally treat it as "present".


This bit has to be just read allow because there are pages with this bit unset that are written to at start.

Re: Siemens PC-D [Re: rfka01] #117600 07/28/20 12:41 PM
Joined: May 2009
Posts: 1,872
J
Just Desserts Offline
Very Senior Member
Offline
Very Senior Member
J
Joined: May 2009
Posts: 1,872
What if the bit gets set upon write, to flag that the page has been written to and is therefore present? Kind of like a write-through sort of thing.

Re: Siemens PC-D [Re: rfka01] #117604 07/30/20 02:14 AM
Joined: Jan 2012
Posts: 844
C
crazyc Offline
Senior Member
Offline
Senior Member
C
Joined: Jan 2012
Posts: 844
Doubtful. Most pages have that bit set, just a handful don't.

Re: Siemens PC-D [Re: rfka01] #117605 07/30/20 03:16 AM
Joined: Mar 2001
Posts: 16,576
R
R. Belmont Offline
Very Senior Member
Offline
Very Senior Member
R
Joined: Mar 2001
Posts: 16,576
The 68k MMUs have a "page has been accessed" bit that the hardware sets (NeXTStep gets quite angry if you don't support it), but if the software's setting it that's probably not what it is.

Re: Siemens PC-D [Re: rfka01] #117606 07/30/20 05:32 PM
Joined: Jan 2012
Posts: 844
C
crazyc Offline
Senior Member
Offline
Senior Member
C
Joined: Jan 2012
Posts: 844
The 386 has a dirty bit in the page table too. Windows 3.1 enhanced mode starts to discard pages with changed bytes if its not implemented which crashes the os pretty quickly.

Re: Siemens PC-D [Re: rfka01] #117620 08/03/20 03:40 AM
Joined: Jun 2020
Posts: 13
A
anyfoo Offline
Member
Offline
Member
A
Joined: Jun 2020
Posts: 13
Are they written before the MMU is enabled (bit 0x20 in 0x8000)?

Re: Siemens PC-D [Re: rfka01] #117621 08/03/20 03:46 AM
Joined: Jun 2020
Posts: 13
A
anyfoo Offline
Member
Offline
Member
A
Joined: Jun 2020
Posts: 13
R. Belmont: Very sure it's either "present" or "read allow" (more towards the latter if crazyc is right), because it's entirely consistent with how the system works. If this was an "accessed" bit, there would still have to be another present/read allow bit somewhere for the MMU to really make sense.

Just Desserts/crazyc: Possible, but I didn't actually spot any page getting (intentionally) written when the bit wasn't set, at least once the MMU was enabled. Very possible that I missed something. If so, "read allow" seems to be the simplest interpretation (I assume that even then the pages were not intentionally read before the bit was set.)

Last edited by anyfoo; 08/03/20 03:47 AM.
Re: Siemens PC-D [Re: rfka01] #117908 09/26/20 08:00 PM
Joined: Jun 2020
Posts: 13
A
anyfoo Offline
Member
Offline
Member
A
Joined: Jun 2020
Posts: 13
A bit of a long shot, but maybe something:

On some pictures of a PC-X that likely has the MMU on the motherboard itself I noticed three MBM2149 ICs somewhat in the middle of the board. According to the data sheet, those are 4096bit static RAMs. If we divide by 1024 entries, then with 3 of these chips, that would resolve to 12 bit per entry, and could well be the memory for the page table entries. The placement of the 74 series ICs between the (dumb) memory controller and those static RAMs would also be convenient for the MMU logic.

There is also a lone jumper over 3 pins (so three different ways to set it, if you include removing the jumper) among those logic chips in between, right above a multiplexer. It's a lot of speculation, but maybe that's what actually selects the width of the task identifier.

Last edited by anyfoo; 09/26/20 08:02 PM.
Re: Siemens PC-D [Re: rfka01] #117909 09/26/20 08:36 PM
Joined: Jan 2012
Posts: 844
C
crazyc Offline
Senior Member
Offline
Senior Member
C
Joined: Jan 2012
Posts: 844
If it is a jumper there must be some documentation somewhere telling people installing sinix how to set them.

Page 17 of 18 1 2 15 16 17 18

Who's Online Now
3 registered members (AaronGiles, mahlemiut, 1 invisible), 47 guests, and 3 spiders.
Key: Admin, Global Mod, Mod
ShoutChat Box
Comment Guidelines: Do post respectful and insightful comments. Don't flame, hate, spam.
Forum Statistics
Forums9
Topics8,814
Posts115,990
Members4,915
Most Online890
Jan 17th, 2020
Powered by UBB.threads™ PHP Forum Software 7.7.3