9fans - fans of the OS Plan 9 from Bell Labs
 help / color / mirror / Atom feed
From: "David Leimbach" <leimy2k@gmail.com>
To: "Fans of the OS Plan 9 from Bell Labs" <9fans@cse.psu.edu>
Subject: Re: [9fans] Vmware playground but in qemu.
Date: Tue, 18 Apr 2006 00:37:35 -0700	[thread overview]
Message-ID: <3e1162e60604180037g7ea6b035vb835524be0822129@mail.gmail.com> (raw)
In-Reply-To: <4443ED69.5040904@gmail.com>

When I use the functions below in my ether2000.c I get

ne2000reset: undefined: ioalloc in ne2000reset
ne2000reset: undefined: iofree in ne2000reset...

Dave

On 4/17/06, Lluís Batlle i Rossell <viriketo@gmail.com> wrote:
> Hi,
> I'm using qemu for the vmware playground described in the wiki
> (http://cm.bell-labs.com/wiki/plan9/vmware_playground_for_plan9/index.html).
> Although I found some flaws in the wiki page (aux/timesync _-n_ <host>,
> for instance, or ip/dhcpd ip/tftpd in the same line), I think I got the
> 9pccpuf kernel running fine - I explain here some of the problems I had.
>
> I had to change the ether2000.c file for the pxeload loader in order to
> get a working pxe boot (qemu has a pnp or pci ne2000). I attach the new
> ether2000.c file for /sys/src/boot/pc/ether2000.c. I simply took the 9
> kernel's ether2000.c, corrected the locations of the include files,
> removed the 'static' of ne2000reset(), and removed the function
> ether2000link() (if I recall everything). I really got surprised when
> the new 9pxeload worked.
>
> In order to get the "9pccpu" kernel in a second host working, I had to
> use a qemu floppy image together with the zero-filled small hd. I
> manually created an msdos filesystem there, with a file "plan9.nvr"
> (thanks 20h@#9fans) with silly data on it (using mkfs.msdos in Linux).
> If I didn't have that floppy, and only had a hard disk WITHOUT a plan9
> partition/nvram part, the 9pccpu kernel hanged after asking for the
> root. After getting the first boot done, I partitioned the hard disk,
> prepared the nvram part, and took off the "floppy" on the next boot.
> Although the author says the HD vmware file could be 1KiB long, I needed
> at least 512KiB long, because that's the size of a cylinder according to
> disk/fdisk (and less than that is not allowed as I experienced).
>
> Among my many unsuccesful tries to get everything done as described in
> the wiki, it was dark for me the "Set up the nvram", because I thought I
> should write the lines suggested into the nvram partition. And I also
> expected that to work, so the rebooting's "error in nvram key" made me mad.
>
> There are other details that I didn't understood at first from the wiki
> page, but it would be too a matter of opinion. As an example: what's
> "First boot"? Should I reboot? As 'bootes'?
>
> So, here you have comments from a 'newbie'. I hope they help in some sense.
>
> Thanks!
>
>
> #include "u.h"
> #include "lib.h"
> #include "mem.h"
> #include "dat.h"
> #include "fns.h"
> #include "io.h"
> #include "error.h"
>
> #include "etherif.h"
> #include "ether8390.h"
>
> /*
>  * Driver written for the 'Notebook Computer Ethernet LAN Adapter',
>  * a plug-in to the bus-slot on the rear of the Gateway NOMAD 425DXL
>  * laptop. The manual says NE2000 compatible.
>  * The interface appears to be pretty well described in the National
>  * Semiconductor Local Area Network Databook (1992) as one of the
>  * AT evaluation cards.
>  *
>  * The NE2000 is really just a DP8390[12] plus a data port
>  * and a reset port.
>  */
> enum {
>         Data            = 0x10,         /* offset from I/O base of data port */
>         Reset           = 0x1F,         /* offset from I/O base of reset port */
> };
>
> typedef struct Ctlr Ctlr;
> typedef struct Ctlr {
>         Pcidev* pcidev;
>         Ctlr*   next;
>         int     active;
> } Ctlr;
>
> static Ctlr* ctlrhead;
> static Ctlr* ctlrtail;
>
> static struct {
>         char*   name;
>         int     id;
> } ne2000pci[] = {
>         { "Realtek 8029",       (0x8029<<16)|0x10EC, },
>         { "Winbond 89C940",     (0x0940<<16)|0x1050, },
>         { nil },
> };
>
> static Ctlr*
> ne2000match(Ether* edev, int id)
> {
>         int port;
>         Pcidev *p;
>         Ctlr *ctlr;
>
>         /*
>          * Any adapter matches if no edev->port is supplied,
>          * otherwise the ports must match.
>          */
>         for(ctlr = ctlrhead; ctlr != nil; ctlr = ctlr->next){
>                 if(ctlr->active)
>                         continue;
>                 p = ctlr->pcidev;
>                 if(((p->did<<16)|p->vid) != id)
>                         continue;
>                 port = p->mem[0].bar & ~0x01;
>                 if(edev->port != 0 && edev->port != port)
>                         continue;
>
>                 /*
>                  * It suffices to fill these in,
>                  * the rest is gleaned from the card.
>                  */
>                 edev->port = port;
>                 edev->irq = p->intl;
>
>                 ctlr->active = 1;
>
>                 return ctlr;
>         }
>
>         return nil;
> }
>
> static void
> ne2000pnp(Ether* edev)
> {
>         int i, id;
>         Pcidev *p;
>         Ctlr *ctlr;
>
>         /*
>          * Make a list of all ethernet controllers
>          * if not already done.
>          */
>         if(ctlrhead == nil){
>                 p = nil;
>                 while(p = pcimatch(p, 0, 0)){
>                         if(p->ccrb != 0x02 || p->ccru != 0)
>                                 continue;
>                         ctlr = malloc(sizeof(Ctlr));
>                         ctlr->pcidev = p;
>
>                         if(ctlrhead != nil)
>                                 ctlrtail->next = ctlr;
>                         else
>                                 ctlrhead = ctlr;
>                         ctlrtail = ctlr;
>                 }
>         }
>
>         /*
>          * Is it a card with an unrecognised vid+did?
>          * Normally a search is made through all the found controllers
>          * for one which matches any of the known vid+did pairs.
>          * If a vid+did pair is specified a search is made for that
>          * specific controller only.
>          */
>         id = 0;
>         for(i = 0; i < edev->nopt; i++){
>                 if(cistrncmp(edev->opt[i], "id=", 3) == 0)
>                         id = strtol(&edev->opt[i][3], nil, 0);
>         }
>
>         if(id != 0)
>                 ne2000match(edev, id);
>         else for(i = 0; ne2000pci[i].name; i++){
>                 if(ne2000match(edev, ne2000pci[i].id) != nil)
>                         break;
>         }
> }
>
> int
> ne2000reset(Ether* edev)
> {
>         static int first;
>         ushort buf[16];
>         ulong port;
>         Dp8390 *dp8390;
>         int i;
>         uchar ea[Eaddrlen];
>
>         if(edev->port == 0)
>                 ne2000pnp(edev);
>
>         /*
>          * Set up the software configuration.
>          * Use defaults for irq, mem and size
>          * if not specified.
>          * Must have a port, no more default.
>          */
>         if(edev->port == 0)
>                 return -1;
>         if(edev->irq == 0)
>                 edev->irq = 2;
>         if(edev->mem == 0)
>                 edev->mem = 0x4000;
>         if(edev->size == 0)
>                 edev->size = 16*1024;
>         port = edev->port;
>
>         if(ioalloc(edev->port, 0x20, 0, "ne2000") < 0)
>                 return -1;
>
>         edev->ctlr = malloc(sizeof(Dp8390));
>         dp8390 = edev->ctlr;
>         dp8390->width = 2;
>         dp8390->ram = 0;
>
>         dp8390->port = port;
>         dp8390->data = port+Data;
>
>         dp8390->tstart = HOWMANY(edev->mem, Dp8390BufSz);
>         dp8390->pstart = dp8390->tstart + HOWMANY(sizeof(Etherpkt), Dp8390BufSz);
>         dp8390->pstop = dp8390->tstart + HOWMANY(edev->size, Dp8390BufSz);
>
>         dp8390->dummyrr = 1;
>         for(i = 0; i < edev->nopt; i++){
>                 if(strcmp(edev->opt[i], "nodummyrr"))
>                         continue;
>                 dp8390->dummyrr = 0;
>                 break;
>         }
>
>         /*
>          * Reset the board. This is done by doing a read
>          * followed by a write to the Reset address.
>          */
>         buf[0] = inb(port+Reset);
>         delay(2);
>         outb(port+Reset, buf[0]);
>         delay(2);
>
>         /*
>          * Init the (possible) chip, then use the (possible)
>          * chip to read the (possible) PROM for ethernet address
>          * and a marker byte.
>          * Could just look at the DP8390 command register after
>          * initialisation has been tried, but that wouldn't be
>          * enough, there are other ethernet boards which could
>          * match.
>          */
>         dp8390reset(edev);
>         memset(buf, 0, sizeof(buf));
>         dp8390read(dp8390, buf, 0, sizeof(buf));
>         if((buf[0x0E] & 0xFF) != 0x57 || (buf[0x0F] & 0xFF) != 0x57){
>                 iofree(edev->port);
>                 free(edev->ctlr);
>                 return -1;
>         }
>
>         /*
>          * Stupid machine. Shorts were asked for,
>          * shorts were delivered, although the PROM is a byte array.
>          * Set the ethernet address.
>          */
>         memset(ea, 0, Eaddrlen);
>         if(memcmp(ea, edev->ea, Eaddrlen) == 0){
>                 for(i = 0; i < sizeof(edev->ea); i++)
>                         edev->ea[i] = buf[i];
>         }
>         dp8390setea(edev);
>
>         return 0;
> }
>
>
>
>


  parent reply	other threads:[~2006-04-18  7:37 UTC|newest]

Thread overview: 10+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2006-04-17 19:32 Lluís Batlle i Rossell
2006-04-17 20:05 ` Gabriel Diaz
2006-04-17 20:51   ` Lluís Batlle i Rossell
2006-04-18  7:37 ` David Leimbach [this message]
2006-04-18  7:42   ` geoff
2006-04-18 13:55     ` David Leimbach
2006-04-18 14:26       ` jmk
2006-04-18 15:49         ` David Leimbach
2006-04-18 15:50           ` jmk
2006-04-18 15:51             ` David Leimbach

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=3e1162e60604180037g7ea6b035vb835524be0822129@mail.gmail.com \
    --to=leimy2k@gmail.com \
    --cc=9fans@cse.psu.edu \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).