[LinuxBIOS] Pm49FL002 Flash

[Hugo MERCIER]

Hi,

I'm new to this list and to the project. I'd like to add support for
the Asus A7V600 motherboard (VT8237 and KT600 based). I don't have many
time, I'am not very familiar with chipset low level programming, but ...
I found LinuxBios is an interesting project ;)

Until now, I spend my time gathering informations (manuals and
datasheet - with a request to VIA) and looking to the linuxbios code.

I just wanted to add flash support to the flash_rom util to start
something. My motherboard is using the Pm49FL002 flash memory. It's the
same as the Pm49FL004 with 2Mb instead of 4Mb.

With respect to the Pm49FL002 datasheet, it seems to act as Jedec. So,
I added this to flash_rom.c:
        {"Pm49FL002",   PMC_ID,         PMC_49FL002,    NULL, 256, 16 *
1024,
         probe_jedec,   erase_chip_jedec, NULL,NULL},
and to flash_rom.h:
#define PMC_49FL002       0x6D  /* PMC 49FL002 device code      */

I just wanted to test the probing (No erasing or programming yet).
And ... it fails. flash_rom doesn't find anything : the reported IDs
bytes are the same as those reported without the probing memory cycles
(0x555,0xAAA, ...)

So ... what am I doing wrong ? Do I have to set some bit in the
Northbridge registers ?

[Ronald G. Minnich]

On Wed, 1 Jun 2005, Hugo MERCIER wrote:
> I just wanted to test the probing (No erasing or programming yet).
> And ... it fails. flash_rom doesn't find anything : the reported IDs
> bytes are the same as those reported without the probing memory cycles
> (0x555,0xAAA, ...)

asus loves to put GPIO enables ont their boards to control flash writes. 
My guess is you are getting no write cycles to the flash part.

This is pretty easy to probe for, find the GPIO registers and starting 
changing them one at a time, and see if at some point you have flash 
writes working. Some of them, when changed, may reset the board ... so be 
prepared for that.

ron

[Bari Ari]

Hugo MERCIER wrote:
> Hi,
>
> I'm new to this list and to the project. I'd like to add support for
> the Asus A7V600 motherboard (VT8237 and KT600 based). I don't have many
> time, I'am not very familiar with chipset low level programming, but ...
> I found LinuxBios is an interesting project ;)

I've been working on VIA VT8237 support. Should be up soon. Lots of 
boards using the vt8237 now.

-Bari

[Li-Ta Lo]

On Fri, 2005-06-03 at 08:49 -0600, Ronald G. Minnich wrote:
> On Wed, 1 Jun 2005, Hugo MERCIER wrote:
> > I just wanted to test the probing (No erasing or programming yet).
> > And ... it fails. flash_rom doesn't find anything : the reported IDs
> > bytes are the same as those reported without the probing memory cycles
> > (0x555,0xAAA, ...)
>
> asus loves to put GPIO enables ont their boards to control flash writes. 
> My guess is you are getting no write cycles to the flash part.
>
> This is pretty easy to probe for, find the GPIO registers and starting 
> changing them one at a time, and see if at some point you have flash 
> writes working. Some of them, when changed, may reset the board ... so be 
> prepared for that.

I don't think we have flash_enable for the chipset. So writing command 
to the flash part has no effect.

[Bari Ari]

Li-Ta Lo wrote:
> I don't think we have flash_enable for the chipset. So writing command 
> to the flash part has no effect.

I'm working on flash_enable for VT8237 as well as support for new the 
16Mb LPC Flash.

There are some issues with getting the vt8237 to unlock the block 
locking registers in FWH mode.

-Bari

[Bari Ari]

Hugo MERCIER wrote:
> Hi,
>
> I'm new to this list and to the project. I'd like to add support for
> the Asus A7V600 motherboard (VT8237 and KT600 based). I don't have many
> time, I'am not very familiar with chipset low level programming, but ...
> I found LinuxBios is an interesting project ;)
>
> Until now, I spend my time gathering informations (manuals and
> datasheet - with a request to VIA) and looking to the linuxbios code.
>
> I just wanted to add flash support to the flash_rom util to start
> something. My motherboard is using the Pm49FL002 flash memory. It's the
> same as the Pm49FL004 with 2Mb instead of 4Mb.
>
> With respect to the Pm49FL002 datasheet, it seems to act as Jedec. So,
> I added this to flash_rom.c:
>         {"Pm49FL002",   PMC_ID,         PMC_49FL002,    NULL, 256, 16 *
> 1024,
>          probe_jedec,   erase_chip_jedec, NULL,NULL},
> and to flash_rom.h:
> #define PMC_49FL002       0x6D  /* PMC 49FL002 device code      */
>
> I just wanted to test the probing (No erasing or programming yet).
> And ... it fails. flash_rom doesn't find anything : the reported IDs
> bytes are the same as those reported without the probing memory cycles
> (0x555,0xAAA, ...)
>
> So ... what am I doing wrong ? Do I have to set some bit in the
> Northbridge registers ?

Here's a copy of flash_enable.c with vt8237 support.

-Bari

#include <sys/io.h>
#include <stdio.h>
#include <pci/pci.h>
#include <stdlib.h>

/* Bari- added support for vt8237 */

static int enable_flash_sis630(struct pci_dev *dev, char *name)
{
    char b;

/* get io privilege access PCI configuration space */
    if (iopl(3) != 0) {
    	perror("Can not set io priviliage");
    	exit(1);
    }

/* Enable 0xFFF8000~0xFFFF0000 decoding on SiS 540/630 */
    outl(0x80000840, 0x0cf8);
    b = inb(0x0cfc) | 0x0b;
    outb(b, 0xcfc);
    /* Flash write enable on SiS 540/630 */
    outl(0x80000845, 0x0cf8);
    b = inb(0x0cfd) | 0x40;
    outb(b, 0xcfd);

/* The same thing on SiS 950 SuperIO side */
    outb(0x87, 0x2e);
    outb(0x01, 0x2e);
    outb(0x55, 0x2e);
    outb(0x55, 0x2e);

if (inb(0x2f) != 0x87) {
    	outb(0x87, 0x4e);
    	outb(0x01, 0x4e);
    	outb(0x55, 0x4e);
    	outb(0xaa, 0x4e);
    	if (inb(0x4f) != 0x87) {
    		printf("Can not access SiS 950\n");
    		return -1;
    	}
    	outb(0x24, 0x4e);
    	b = inb(0x4f) | 0xfc;
    	outb(0x24, 0x4e);
    	outb(b, 0x4f);
    	outb(0x02, 0x4e);
    	outb(0x02, 0x4f);
    }

outb(0x24, 0x2e);
    printf("2f is %#x\n", inb(0x2f));
    b = inb(0x2f) | 0xfc;
    outb(0x24, 0x2e);
    outb(b, 0x2f);

outb(0x02, 0x2e);
    outb(0x02, 0x2f);

return 0;
}

static int enable_flash_e7500(struct pci_dev *dev, char *name)
{
    /* register 4e.b gets or'ed with one */
    unsigned char old, new;
    /* if it fails, it fails. There are so many variations of broken mobos
     * that it is hard to argue that we should quit at this point. 
     */

old = pci_read_byte(dev, 0x4e);

new = old | 1;

if (new == old)
    	return 0;

pci_write_byte(dev, 0x4e, new);

if (pci_read_byte(dev, 0x4e) != new) {
    	printf("tried to set 0x%x to 0x%x on %s failed (WARNING ONLY)\n",
    	       0x4e, new, name);
    	return -1;
    }
    return 0;
}

static int enable_flash_ich4(struct pci_dev *dev, char *name)
{
    /* register 4e.b gets or'ed with one */
    unsigned char old, new;
    /* if it fails, it fails. There are so many variations of broken mobos
     * that it is hard to argue that we should quit at this point. 
     */

old = pci_read_byte(dev, 0x4e);

new = old | 1;

if (new == old)
    	return 0;

pci_write_byte(dev, 0x4e, new);

if (pci_read_byte(dev, 0x4e) != new) {
    	printf("tried to set 0x%x to 0x%x on %s failed (WARNING ONLY)\n",
    	       0x4e, new, name);
    	return -1;
    }
    return 0;
}

static int enable_flash_vt8235(struct pci_dev *dev, char *name)
{
    unsigned char old, new, val;
    unsigned int base;
    int ok;

/* get io privilege access PCI configuration space */
    if (iopl(3) != 0) {
    	perror("Can not set io priviliage");
    	exit(1);
    }

old = pci_read_byte(dev, 0x40);

new = old | 0x10;

if (new == old)
    	return 0;

ok = pci_write_byte(dev, 0x40, new);
    if (ok != 0) {
    	printf("tried to set 0x%x to 0x%x on %s failed (WARNING ONLY)\n",
    	       old, new, name);
    }

/* enable GPIO15 which is connected to write protect. */
    base = ((pci_read_byte(dev, 0x88) & 0x80) | pci_read_byte(dev, 0x89) << 8);
    val = inb(base + 0x4d);
    val |= 0x80;
    outb(val, base + 0x4d);

if (ok != 0) {
    	return -1;
    } else {
    	return 0;
    }
}

static int enable_flash_vt8231(struct pci_dev *dev, char *name)
{
    unsigned char val;

val = pci_read_byte(dev, 0x40);
    val |= 0x10;
    pci_write_byte(dev, 0x40, val);

if (pci_read_byte(dev, 0x40) != val) {
    	printf("tried to set 0x%x to 0x%x on %s failed (WARNING ONLY)\n",
    	       0x40, val, name);
    	return -1;
    }
    return 0;
}

static int enable_flash_vt8237(struct pci_dev *dev, char *name)
{
    unsigned char val;

val = pci_read_byte(dev, 0x40);
    val |= 0x10;
    pci_write_byte(dev, 0x40, val);

if (pci_read_byte(dev, 0x40) != val) {
    	printf("tried to set 0x%x to 0x%x on %s failed (WARNING ONLY)\n",
    	       0x40, val, name);
    	return -1;
    }
    return 0;
}

static int enable_flash_cs5530(struct pci_dev *dev, char *name)
{
    unsigned char new;

pci_write_byte(dev, 0x52, 0xee);

new = pci_read_byte(dev, 0x52);

if (new != 0xee) {
    	printf("tried to set register 0x%x to 0x%x on %s failed (WARNING ONLY)\n",
    	       0x52, new, name);
    	return -1;
    }
    return 0;
}

static int enable_flash_sc1100(struct pci_dev *dev, char *name)
{
    unsigned char new;

pci_write_byte(dev, 0x52, 0xee);

new = pci_read_byte(dev, 0x52);

if (new != 0xee) {
    	printf("tried to set register 0x%x to 0x%x on %s failed (WARNING ONLY)\n",
    	       0x52, new, name);
    	return -1;
    }
    return 0;
}

static int enable_flash_sis5595(struct pci_dev *dev, char *name)
{
    unsigned char new, newer;

new = pci_read_byte(dev, 0x45);

/* clear bit 5 */
    new &= (~0x20);
    /* set bit 2 */
    new |= 0x4;

pci_write_byte(dev, 0x45, new);

newer = pci_read_byte(dev, 0x45);
    if (newer != new) {
    	printf("tried to set register 0x%x to 0x%x on %s failed (WARNING ONLY)\n",
    	       0x45, new, name);
    	printf("Stuck at 0x%x\n", newer);
    	return -1;
    }
    return 0;
}

static int enable_flash_amd8111(struct pci_dev *dev, char *name)
{
    /* register 4e.b gets or'ed with one */
    unsigned char old, new;
    /* if it fails, it fails. There are so many variations of broken mobos
     * that it is hard to argue that we should quit at this point. 
     */

/* enable decoding at 0xffb00000 to 0xffffffff */
    old = pci_read_byte(dev, 0x43);
    new = old | 0xC0;
    if (new != old) {
    	pci_write_byte(dev, 0x43, new);
    	if (pci_read_byte(dev, 0x43) != new) {
    		printf("tried to set 0x%x to 0x%x on %s failed (WARNING ONLY)\n",
    		       0x43, new, name);
    	}
    }

old = pci_read_byte(dev, 0x40);
    new = old | 0x01;
    if (new == old)
    	return 0;
    pci_write_byte(dev, 0x40, new);

if (pci_read_byte(dev, 0x40) != new) {
    	printf("tried to set 0x%x to 0x%x on %s failed (WARNING ONLY)\n",
    	       0x40, new, name);
    	return -1;
    }
    return 0;
}

//By yhlu
static int enable_flash_ck804(struct pci_dev *dev, char *name)
{
        /* register 4e.b gets or'ed with one */
        unsigned char old, new;
        /* if it fails, it fails. There are so many variations of broken mobos
         * that it is hard to argue that we should quit at this point. 
         */

//dump_pci_device(dev);

old = pci_read_byte(dev, 0x88);
        new = old | 0xc0;
        if (new != old) {
                pci_write_byte(dev, 0x88, new);
                if (pci_read_byte(dev, 0x88) != new) {
                        printf("tried to set 0x%x to 0x%x on %s failed (WARNING ONLY)\n",
                               0x88, new, name);
                }
        }

old = pci_read_byte(dev, 0x6d);
        new = old | 0x01;
        if (new == old)
                return 0;
        pci_write_byte(dev, 0x6d, new);

if (pci_read_byte(dev, 0x6d) != new) {
                printf("tried to set 0x%x to 0x%x on %s failed (WARNING ONLY)\n",
                       0x6d, new, name);
                return -1;
        }
        return 0;
}

typedef struct penable {
    unsigned short vendor, device;
    char *name;
    int (*doit) (struct pci_dev * dev, char *name);
} FLASH_ENABLE;

static FLASH_ENABLE enables[] = {
    {0x1039, 0x0630, "sis630", enable_flash_sis630},
    {0x8086, 0x2480, "E7500", enable_flash_e7500},
    {0x8086, 0x24c0, "ICH4", enable_flash_ich4},
    {0x1106, 0x8231, "VT8231", enable_flash_vt8231},
    {0x1106, 0x3177, "VT8235", enable_flash_vt8235},
      {0x1106, 0x3227, "VT8237", enable_flash_vt8237},
    {0x1078, 0x0100, "CS5530", enable_flash_cs5530},
    {0x100b, 0x0510, "SC1100", enable_flash_sc1100},
    {0x1039, 0x0008, "SIS5595", enable_flash_sis5595},
    {0x1022, 0x7468, "AMD8111", enable_flash_amd8111},
      {0x10de, 0x0050, "NVIDIA CK804", enable_flash_ck804}, // LPC
      {0x10de, 0x0051, "NVIDIA CK804", enable_flash_ck804}, // Pro
      {0x10de, 0x00d3, "NVIDIA CK804", enable_flash_ck804}, // Slave, should not be here, to fix known bug for A01.
};

int enable_flash_write()
{
    int i;
    struct pci_access *pacc;
    struct pci_dev *dev = 0;
    FLASH_ENABLE *enable = 0;

pacc = pci_alloc();	/* Get the pci_access structure */
    /* Set all options you want -- here we stick with the defaults */
    pci_init(pacc);		/* Initialize the PCI library */
    pci_scan_bus(pacc);	/* We want to get the list of devices */

/* now let's try to find the chipset we have ... */
    for (i = 0; i < sizeof(enables) / sizeof(enables[0]) && (!dev);
         i++) {
    	struct pci_filter f;
    	struct pci_dev *z;
    	/* the first param is unused. */
    	pci_filter_init((struct pci_access *) 0, &f);
    	f.vendor = enables[i].vendor;
    	f.device = enables[i].device;
    	for (z = pacc->devices; z; z = z->next)
    		if (pci_filter_match(&f, z)) {
    			enable = &enables[i];
    			dev = z;
    		}
    }

/* now do the deed. */
    if (enable) {
    	printf("Enabling flash write on %s...", enable->name);
    	if (enable->doit(dev, enable->name) == 0)
    		printf("OK\n");
    }
    return 0;
}