Re: [9fans] Vmware playground but in qemu.

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

Dave

On 4/17/06, Lluís Batlle i Rossell <viri...@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;
> }
>
>
>
>

; grep -n 'ioalloc|iofree' *.h
fns.h:57: #define ioalloc(addr, len, align, name) (addr)
fns.h:58: #define iofree(addr)

(This is in /sys/src/boot/pc, right?)

It's apparently not a macro on my system though, and the definition
appears to be in /sys/src/9/pc/devarch.c

Parallels, so far in beta, doesn't let me middle click anything with
my current trackball, nor does it have a '|' key. It's really hard to
do any amount of useful searching or editing as a result... especially
without networking support, where I could just drawterm.

Dave

I just opened up /sys/src/boot/pc/fns.h and I see no "ioalloc"
prototype or macro.

ialloc is there... but no ioalloc.

Do I need to get a new CD image of Plan 9? Has this stuff been
updated recently?

yes

argh! Very well then :)