busmaster/latency for video

busmaster/latency for video

[Ryan Nielsen]

PCI_COMMAND_MASTER is not enabled and latency = 0 for the ultimate rez video card, and pci_write_config_*
do not change them. latency = 32 and mastering is enabled for all the other cards in the machine,

should they be enabled ?

a machine check happens when calling sys_idle after using offb with the ultimate rez.

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

Re: busmaster/latency for video

[Michel Lanners]

On   6 Oct, this message from Ryan Nielsen echoed through cyberspace:
> 
> PCI_COMMAND_MASTER is not enabled and latency = 0 for the ultimate rez video card, and pci_write_config_*
> do not change them. latency = 32 and mastering is enabled for all the other cards in the machine,
> 
> should they be enabled ?

That probably means the card doesn't support them.

Normal graphic cards have no need for mastership. Only those with video
inputs need it, in order to transfer video data into the framebuffer
without CPU help.

Michel

-------------------------------------------------------------------------
Michel Lanners                 |  " Read Philosophy.  Study Art.
23, Rue Paul Henkes            |    Ask Questions.  Make Mistakes.
L-1710 Luxembourg              |
email   mlan@cpu.lu            |
http://www.cpu.lu/~mlan        |                     Learn Always. "


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

Re: busmaster/latency for video

[Geert Uytterhoeven]

On Thu, 7 Oct 1999, Michel Lanners wrote:
> On   6 Oct, this message from Ryan Nielsen echoed through cyberspace:
> > PCI_COMMAND_MASTER is not enabled and latency = 0 for the ultimate rez video card, and pci_write_config_*
> > do not change them. latency = 32 and mastering is enabled for all the other cards in the machine,
> > 
> > should they be enabled ?
> 
> That probably means the card doesn't support them.
> 
> Normal graphic cards have no need for mastership. Only those with video
> inputs need it, in order to transfer video data into the framebuffer
> without CPU help.

Many recent graphics chips to support bus mastering, especially the AGP ones.

Greetings,

						Geert

--
Geert Uytterhoeven ----------------- Sony Suprastructure Center Europe (SUPC-E)
Geert.Uytterhoeven@sonycom.com ------------------- Sint Stevens Woluwestraat 55
Phone +32-2-7248632 Fax +32-2-7262686 ---------------- B-1130 Brussels, Belgium


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

Re: busmaster/latency for video

[Michel Lanners]

On   7 Oct, this message from Geert Uytterhoeven echoed through cyberspace:
> On Thu, 7 Oct 1999, Michel Lanners wrote:
>> On   6 Oct, this message from Ryan Nielsen echoed through cyberspace:
>> > PCI_COMMAND_MASTER is not enabled and latency = 0 for the ultimate rez video card, and pci_write_config_*
>> > do not change them. latency = 32 and mastering is enabled for all the other cards in the machine,
>> > 
>> > should they be enabled ?
>> 
>> That probably means the card doesn't support them.
>> 
>> Normal graphic cards have no need for mastership. Only those with video
>> inputs need it, in order to transfer video data into the framebuffer
>> without CPU help.
> 
> Many recent graphics chips to support bus mastering, especially the AGP ones.

Ah, but in that case Apple's OF should enable bus mastering. If I read
well between the lines of Apple's docs, OF should set a few default
parameters on all PCI devices, among which bus mastering. I'm not sure
about latency, though...

Michel

-------------------------------------------------------------------------
Michel Lanners                 |  " Read Philosophy.  Study Art.
23, Rue Paul Henkes            |    Ask Questions.  Make Mistakes.
L-1710 Luxembourg              |
email   mlan@cpu.lu            |
http://www.cpu.lu/~mlan        |                     Learn Always. "


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

Re: busmaster/latency for video

[Geert Uytterhoeven]

On Thu, 7 Oct 1999, Michel Lanners wrote:
> On   7 Oct, this message from Geert Uytterhoeven echoed through cyberspace:
> > On Thu, 7 Oct 1999, Michel Lanners wrote:
> >> On   6 Oct, this message from Ryan Nielsen echoed through cyberspace:
> >> > PCI_COMMAND_MASTER is not enabled and latency = 0 for the ultimate rez video card, and pci_write_config_*
> >> > do not change them. latency = 32 and mastering is enabled for all the other cards in the machine,
> >> > 
> >> > should they be enabled ?
> >> 
> >> That probably means the card doesn't support them.
> >> 
> >> Normal graphic cards have no need for mastership. Only those with video
> >> inputs need it, in order to transfer video data into the framebuffer
> >> without CPU help.
> > 
> > Many recent graphics chips to support bus mastering, especially the AGP ones.
> 
> Ah, but in that case Apple's OF should enable bus mastering. If I read
> well between the lines of Apple's docs, OF should set a few default
> parameters on all PCI devices, among which bus mastering. I'm not sure
> about latency, though...

I know my (non-Apple) OF warns me when I try to put a bus mastering-capable PCI
board in the right-most PCI slot (which doesn't support bus mastering).

Greetings,

						Geert

--
Geert Uytterhoeven ----------------- Sony Suprastructure Center Europe (SUPC-E)
Geert.Uytterhoeven@sonycom.com ------------------- Sint Stevens Woluwestraat 55
Phone +32-2-7248632 Fax +32-2-7262686 ---------------- B-1130 Brussels, Belgium


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

Re: busmaster/latency for video

[Benjamin Herrenschmidt]

On Thu, Oct 7, 1999, Michel Lanners <mlan@cpu.lu> wrote:

>Ah, but in that case Apple's OF should enable bus mastering. If I read
>well between the lines of Apple's docs, OF should set a few default
>parameters on all PCI devices, among which bus mastering. I'm not sure
>about latency, though...

OF tends to ignore devices that don't have an OF expansion ROM. (For
example, it won't enable memory or IO accesses in the command register).
I beleive it won't enable bus mastering neither but I may be wrong.

-- 
           Perso. e-mail: <mailto:bh40@calva.net>
           Work   e-mail: <mailto:benh@mipsys.com>
BenH.      Web   : <http://calvaweb.calvacom.fr/bh40/>


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

Re: busmaster/latency for video

[Michel Lanners]

On   8 Oct, this message from Benjamin Herrenschmidt echoed through cyberspace:
> On Thu, Oct 7, 1999, Michel Lanners <mlan@cpu.lu> wrote:
> 
>>Ah, but in that case Apple's OF should enable bus mastering. If I read
>>well between the lines of Apple's docs, OF should set a few default
>>parameters on all PCI devices, among which bus mastering. I'm not sure
>>about latency, though...
> 
> OF tends to ignore devices that don't have an OF expansion ROM. (For
> example, it won't enable memory or IO accesses in the command register).
> I beleive it won't enable bus mastering neither but I may be wrong.

OF itself doesn't enable any memory or IO region; it's the
device-specific init code (which might be on-board for PCI cards, or in
OF for motherboard devices).

There are, however, a few things OF does for _every_ PCI device, among
which enabling bus mastering and assigning memory and/or IO resources.

Michel

-------------------------------------------------------------------------
Michel Lanners                 |  " Read Philosophy.  Study Art.
23, Rue Paul Henkes            |    Ask Questions.  Make Mistakes.
L-1710 Luxembourg              |
email   mlan@cpu.lu            |
http://www.cpu.lu/~mlan        |                     Learn Always. "


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

Re: busmaster/latency for video

[Geert Uytterhoeven]

On Fri, 8 Oct 1999, Michel Lanners wrote:
> OF itself doesn't enable any memory or IO region; it's the
> device-specific init code (which might be on-board for PCI cards, or in
> OF for motherboard devices).
> 
> There are, however, a few things OF does for _every_ PCI device, among
> which enabling bus mastering and assigning memory and/or IO resources.
                                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
assigning some memory and/or IO resources? Or is this only the case in the
Longtrail-OF-which-is-better-than-Apple-OF-except-maybe-for-assigning-resources
case?

Greetings,

						Geert

--
Geert Uytterhoeven ----------------- Sony Suprastructure Center Europe (SUPC-E)
Geert.Uytterhoeven@sonycom.com ------------------- Sint Stevens Woluwestraat 55
Phone +32-2-7248632 Fax +32-2-7262686 ---------------- B-1130 Brussels, Belgium


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

Re: busmaster/latency for video

[Michel Lanners]

On  11 Oct, this message from Geert Uytterhoeven echoed through cyberspace:
> On Fri, 8 Oct 1999, Michel Lanners wrote:
>> OF itself doesn't enable any memory or IO region; it's the
>> device-specific init code (which might be on-board for PCI cards, or in
>> OF for motherboard devices).
>> 
>> There are, however, a few things OF does for _every_ PCI device, among
>> which enabling bus mastering and assigning memory and/or IO resources.
>                                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> assigning some memory and/or IO resources? Or is this only the case in the
> Longtrail-OF-which-is-better-than-Apple-OF-except-maybe-for-assigning-resources
> case?

Hm, my 7600's OF version 1.05 assigns memory and IO regions for two
boards that don't have any OF code: a Matrox Millenium I and a Promise
Ultra/66 IDE controller. Never had any problem in that area..

Michel

-------------------------------------------------------------------------
Michel Lanners                 |  " Read Philosophy.  Study Art.
23, Rue Paul Henkes            |    Ask Questions.  Make Mistakes.
L-1710 Luxembourg              |
email   mlan@cpu.lu            |
http://www.cpu.lu/~mlan        |                     Learn Always. "


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