Re: E15 Carolina Frame Buffer Driver for 2.4

Re: E15 Carolina Frame Buffer Driver for 2.4

[Robert E Brose II]

One caution:

When I was doing some testing of the driver this afternoon in X 3.3.6
repeatedly hitting the forced reload button on netscape while on the
netscape site killed my machine (hard freeze, no keyboard activity).

I'm not sure it's related to the driver because I'm running the
bleeding edge kernel as well but beware... I should be able to do
more testing on it on Thursday. As a precaution, I converted the
43p filesystem to EXT3, nice that EXT3 is in the 2.4 kernel.

A big difference between the ibm_e15 fb driver running on 2.4 as opposed
to 2.2 is that this code that tries to get the frame buffer base address
succeeds in 2.4 but failed (and was hacked to a "default address") in
2.2.

Could anyone more familiar with this machine take a look at the
frame buffer memory initialization below and see if anything looks
funny??


    retval = pci_read_config_dword(pcidev, PCI_BASE_ADDRESS_0, &fb_info.fb_phys_
membase);
.
.
.
/* THE FOLLOWING CODE GETS USED IN 2.2.X ONLY*/

   if (fb_info.fb_phys_membase == 0) {
        printk(KERN_WARNING "%s: No PCI linear framebuffer base address "
            " assigned to 864.\n", ibm_e15_name);
        printk(KERN_WARNING "%s: This is normal. Using default of 0x%08x.\n",
            ibm_e15_name, DEFAULT_VIDBASE);
        fb_info.fb_phys_membase = DEFAULT_VIDBASE;
        retval = pci_write_config_dword(pcidev, PCI_BASE_ADDRESS_0,
                fb_info.fb_phys_membase);
/* END OF 2.2.X ONLY CODE */
.
.
.
    }
    printk(KERN_INFO "%s: detected S3 864, linear frame buffer at 0x%08x\n",
        ibm_e15_name, fb_info.fb_phys_membase);

    fb_info.fb_phys_membase += isa_mem_base;

    device_present = 1;
    request_region(0x3c0, 32, "ibm_e15");
    fb_info.fb_virt_membase = (u32)ioremap(fb_info.fb_phys_membase, VIDSIZE);
    printk(KERN_INFO "Video ram virt base = %p\n", (char*)fb_info.fb_virt_membas
e);
    return;


Here are the values I get when the system comes up...

PCI: Probing PCI hardware
Fixup res 0 (200) of dev 00:0e.0: 2800000 -> c2800000
Fixup res 6 (7200) of dev 00:0e.0: c0000 -> c00c0000
IBM ID: 000000f7
Setting PCI interrupts for a "IBM PS830/PS850/7248 (Carolina)"
Relocating PCI address 20000000 -> 1000000
PCI:00:0b.0: Resource 0: 00001000-00001007 (f=101)
PCI:00:0e.0: Resource 0: c2800000-c2ffffff (f=200)
PCI:00:0c.0: Resource 0: 01000200-0100021f (f=101)
PCI:00:10.0: Resource 0: 01000000-010000ff (f=101)
.
.
.
IBM_E15: initializing
IBM_E15: detected S3 864, linear frame buffer at 0x02800000
Video ram virt base = f2800000
IBM_E15: Text console mode: 8 bit color depth
Console: switching to colour frame buffer device 80x30
fb0: IBM_E15 frame buffer device

Regards,
Bob
--
Robert E. Brose II    N0QBJ
http://www.jriver.com/~bob/
bob@kunk.jriver.com

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: E15 Carolina Frame Buffer Driver for 2.4

[David Monro]

Robert E Brose II wrote:

> One caution:
>
[..]
>
> A big difference between the ibm_e15 fb driver running on 2.4 as opposed
> to 2.2 is that this code that tries to get the frame buffer base address
> succeeds in 2.4 but failed (and was hacked to a "default address") in
> 2.2.


There's a really good reason why 2.4 kernels don't need the hack which
maps it to the default; 2.4 kernels will assign PCI BARs which weren't
set up by the firmware. This code should be removed completely in the
2.4 version.


Cheers,


	David


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: E15 Carolina Frame Buffer Driver for 2.4

[Robert E Brose II]

User David Monro says:
> Robert E Brose II wrote:
>
> > One caution:
> >
> [..]
> >
> > A big difference between the ibm_e15 fb driver running on 2.4 as opposed
> > to 2.2 is that this code that tries to get the frame buffer base address
> > succeeds in 2.4 but failed (and was hacked to a "default address") in
> > 2.2.
>
>
> There's a really good reason why 2.4 kernels don't need the hack which
> maps it to the default; 2.4 kernels will assign PCI BARs which weren't
> set up by the firmware. This code should be removed completely in the
> 2.4 version.

Makes sense to me. It APPEARS the crashing problems I was having
were with the 2.4.15-pre4 kernel. I'm running a kernel.org 2.4.14 kernel
with only the ext3 patch and the addition of the e15 driver and
it works fine so far...

I was thinking of hacking the mmap stuff into it to give the illusion
of acceleration. That's probably as far as I can take it. I'm sure it
would be nice to merge it with the S3 code, whether in the 2.4 kernel
trio  code or elsewhere however that's likely more that I can budget
time for now... If you change the driver more perhaps you can put it
back up on your carolina page?

Regards,
Bob
--
Robert E. Brose II    N0QBJ
http://www.jriver.com/~bob/
bob@kunk.jriver.com

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

E15 Carolina Frame Buffer Driver for 2.4

[Robert E Brose II]

I have gotten the 2.2 framebuffer driver for the E15 (internal)
display adapter in the IBM 7248-133 43p (carolina Prep MB) to
work (on my machine anyway) in 2.4. This provides a way to
run X with the 2.4 kernels on this hardware.

Here is a patch against the benh rsync tree from last night...

http://www.qbjnet.com/~bob/e15fb-2.4.15-pre3-benh-patch

It would be interesting to hear if it works on other variations
of the 7248.

Bob
--
Robert E. Brose II    N0QBJ
http://www.jriver.com/~bob/
bob@kunk.jriver.com

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: E15 Carolina Frame Buffer Driver for 2.4

[Geert Uytterhoeven]

On Tue, 13 Nov 2001, Robert E Brose II wrote:
> I have gotten the 2.2 framebuffer driver for the E15 (internal)
> display adapter in the IBM 7248-133 43p (carolina Prep MB) to
> work (on my machine anyway) in 2.4. This provides a way to
> run X with the 2.4 kernels on this hardware.
>
> Here is a patch against the benh rsync tree from last night...
>
> http://www.qbjnet.com/~bob/e15fb-2.4.15-pre3-benh-patch
>
> It would be interesting to hear if it works on other variations
> of the 7248.

Can't this be merged with the various frame buffer devices for S3 that are
available on the net?

Gr{oetje,eeting}s,

						Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							    -- Linus Torvalds


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: E15 Carolina Frame Buffer Driver for 2.4

[Robert E Brose II]

User Geert Uytterhoeven says:
>
> Can't this be merged with the various frame buffer devices for S3 that are
> available on the net?
>
> Gr{oetje,eeting}s,
>
> 						Geert

Can you give me more info on these? I see only the S3Trio entry in the
2.4 kernel.

Bob
--
Robert E. Brose II    N0QBJ
http://www.jriver.com/~bob/
bob@kunk.jriver.com

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: E15 Carolina Frame Buffer Driver for 2.4

[Geert Uytterhoeven]

On Wed, 14 Nov 2001, Robert E Brose II wrote:
> User Geert Uytterhoeven says:
> > Can't this be merged with the various frame buffer devices for S3 that are
> > available on the net?
>
> Can you give me more info on these? I see only the S3Trio entry in the
> 2.4 kernel.

Yes, there's only one S3 driver in the kernel.

But several (3?) half-finished drivers are floating around the net.
Unfortunately I never got any of them to work on the cards I have.

Anyone who still remembers the URLs?

Gr{oetje,eeting}s,

						Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							    -- Linus Torvalds


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: E15 Carolina Frame Buffer Driver for 2.4

[David Monro]

Robert E Brose II wrote:

> I have gotten the 2.2 framebuffer driver for the E15 (internal)
> display adapter in the IBM 7248-133 43p (carolina Prep MB) to
> work (on my machine anyway) in 2.4. This provides a way to
> run X with the 2.4 kernels on this hardware.
>


Thanks Robert - I was planning to do this work in the next few days, but
it seems you've saved me the work :-)


[..]
> It would be interesting to hear if it works on other variations
> of the 7248.
>

I'll be trying it on my 6070 in the next day or so. Should work fine given its the same motherboard.


Cheers,

	David


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/