Image Map Image Map
Page 4 of 5 FirstFirst 12345 LastLast
Results 31 to 40 of 43

Thread: What should I look for in an Ethernet card for an XT-class computer?

  1. #31
    Join Date
    May 2011
    Location
    Outer Mongolia
    Posts
    1,648

    Default

    Quote Originally Posted by Chuck(G) View Post
    Yes, IRQ2 is used as the master for the second 8259. Best to stick with one of the other IRQs on an XT.
    A Tandy 1000 makes that really tough, alas, the way it ties everything but 2, 3, and 4 from getting to the bus connector.

    I've mucked around with things some more, and so far it seems to boil down to this:

    1: At least the NE2000 driver from the Crynwr collection doesn't like the 8019AS. Running it with the arguments set "correctly" for the card gets it to load, but it says the mac address is AA:AA:AA:AA:AA:AA and nothing actually seems to work.

    2: For laughs I tried programming the card as IRQ 9 instead of 2, results are identical with the PNPPD driver.

    3: If I program it for IRQ3 it seems to basically work. That lands it on top of a serial port, but I guess that's life.

    I'm curious if anyone else has had success using IRQ2 with this card in an 8-bit machine. It kind of reads as a bug in the packet driver instead of a hardware issue.
    Last edited by Eudimorphodon; November 13th, 2019 at 09:32 PM.
    My Retro-computing YouTube Channel (updates... eventually?): Paleozoic PCs

  2. #32

    Default

    Just use IRQ 3 or 4. Unless there is something that actively sets up the interrupt for the UARTs that normally use those interrupts, it will work fine.

    (For example, if you are sharing IRQ 3 with a serial port don't bother loading a serial mouse driver and the packet driver at the same time. But if one is inactive, the other will work.)

  3. #33
    Join Date
    May 2011
    Location
    Outer Mongolia
    Posts
    1,648

    Default

    I wish the design of the machine would let me overload IRQ 7 instead, because I happen to have a serial printer. But you're right, it's DOS on an XT, it's not as if you're likely to really be doing everything at once.

    Still wondering, though, if this is something that could be patched in the driver. I wonder if the source is floating around.
    My Retro-computing YouTube Channel (updates... eventually?): Paleozoic PCs

  4. #34
    Join Date
    Jan 2017
    Location
    Galicia, Spain
    Posts
    162

    Default

    Quote Originally Posted by Eudimorphodon View Post
    A Tandy 1000 makes that really tough, alas, the way it ties everything but 2, 3, and 4 from getting to the bus connector.

    I've mucked around with things some more, and so far it seems to boil down to this:

    1: At least the NE2000 driver from the Crynwr collection doesn't like the 8019AS. Running it with the arguments set "correctly" for the card gets it to load, but it says the mac address is AA:AA:AA:AA:AA:AA and nothing actually seems to work.

    2: For laughs I tried programming the card as IRQ 9 instead of 2, results are identical with the PNPPD driver.

    3: If I program it for IRQ3 it seems to basically work. That lands it on top of a serial port, but I guess that's life.

    I'm curious if anyone else has had success using IRQ2 with this card in an 8-bit machine. It kind of reads as a bug in the packet driver instead of a hardware issue.
    I use IRQ 3. I was intrigued so I reconfigure the card to use IRQ 2 using RSET8019 without problems but the packet driver seems to assume that IRQ 2 == 9, which is true for an AT, but since IRQs on a XT cannot be higher that 7 it exits with an error message. An obvious bug of the packet driver, IMHO. Maybe somebody can patch it to solve the problem.

    Somebody on this forum modified the Crynwr NE2000 driver to work on 8-bit slots:
    http://www.vcfed.org/forum/showthrea...or-8-bit-slots

    I tested it and works fine.

  5. #35
    Join Date
    May 2011
    Location
    Outer Mongolia
    Posts
    1,648

    Default

    Quote Originally Posted by dieymir View Post
    I use IRQ 3. I was intrigued so I reconfigure the card to use IRQ 2 using RSET8019 without problems but the packet driver seems to assume that IRQ 2 == 9, which is true for an AT, but since IRQs on a XT cannot be higher that 7 it exits with an error message. An obvious bug of the packet driver, IMHO. Maybe somebody can patch it to solve the problem.
    I'm glad it's not just me, then. My working knowledge of 8088 machine language is pathetically lacking, but I do wonder how hard this would be to track down.

    I'll try the modded driver and see if that works for me on IRQ2. It'd be nice to not have a conflict even if it doesn't materially matter.
    My Retro-computing YouTube Channel (updates... eventually?): Paleozoic PCs

  6. #36
    Join Date
    Jan 2007
    Location
    Pacific Northwest, USA
    Posts
    32,710
    Blog Entries
    18

    Default

    The issue is that for the high IRQ numbers, the code expects to be vectored to a different location. Some 80286 BIOSes plug the (now unused) IRQ2/INT 0x0a) ISR slot with a redirection from IRQ9/INT 0x71, so some stuff works, but by no means all.

    Clearly, if a interrupt service routine is expected to handle IRQ9, it's going to put the interrupt vector in the wrong place if a system not supporting the second 8259 is used. It's going to plug 0x1c4 instead of 0x28.

  7. #37
    Join Date
    May 2011
    Location
    Outer Mongolia
    Posts
    1,648

    Default

    I'd still like to know why RSET8019 hangs hard on both my Tandy 1000s; mainly curious if it'd hang on the same machines if I had the memory card with a DMA controller (even though I wouldn't think that should matter) instead of my SRAM-based cards.

    Another slight annoyance is to run the PG8019 program I have to pull the XT-CF card out of my system, because it apparently gets tricked by it sitting at 300h into thinking that there's more than one RTL card in the system. Can't say making this all work isn't an adventure!
    My Retro-computing YouTube Channel (updates... eventually?): Paleozoic PCs

  8. #38
    Join Date
    May 2011
    Location
    Outer Mongolia
    Posts
    1,648

    Default

    Quote Originally Posted by Chuck(G) View Post
    Clearly, if a interrupt service routine is expected to handle IRQ9, it's going to put the interrupt vector in the wrong place if a system not supporting the second 8259 is used. It's going to plug 0x1c4 instead of 0x28.
    A thing I did discover: one company that sold these cards had multiple versions of the support software on their website. The documentation, such as it is, for PG8019.EXE is in the form of comments in the config file (it's hardcoded to a specific filename), and I did see that one version of the file listed valid interrupts choices from 2 to 15, but another version started at 3. A jaded person might suspect that they discovered the bug with sussing out the difference between IRQ2 and IRQ9 in their driver and decided the fix was to pretend that you can't program the card for IRQ2 instead of bothering to fix the driver code. (Of course, if RSET8019 still lets you set "2" then they failed to avoid fail.)
    My Retro-computing YouTube Channel (updates... eventually?): Paleozoic PCs

  9. #39
    Join Date
    Jan 2007
    Location
    Pacific Northwest, USA
    Posts
    32,710
    Blog Entries
    18

    Default

    IRQ 0, 1 and 8 are immutable (0 = keyboard; 1 timer tick, 8 = RTC interrupt). The one that starts at 3 probably goes to 15. On a PC AT and above, IRQ 2 doesn't technically exist.

  10. #40
    Join Date
    May 2011
    Location
    Outer Mongolia
    Posts
    1,648

    Default

    Quote Originally Posted by Chuck(G) View Post
    On a PC AT and above, IRQ 2 doesn't technically exist.
    Yeah, and I suppose I can't really blame them for not bothering to check and see if the card is actually in an AT or not. It does detect if it's in an 8-bit slot, prints that out in the "Found Card:" initialization message, though, so in *theory* it seems like they could have done a little detective work to see if there was actually a cascaded 8259 present... But, totally, this is a weird edge case, especially for what seems to be a late 1990's vintage chipset.
    My Retro-computing YouTube Channel (updates... eventually?): Paleozoic PCs

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •