PCI IRQ -MPC8540

PCI IRQ -MPC8540

[agnel juni]

[-- Attachment #1: Type: text/plain, Size: 2506 bytes --]

Hello all:

I would like to understand if there is a way to program the PCI interrupt pin register of MPC8540. I did look for this in the datasheet, but couldn't find any positive answer.

Currently, I am working in a set-up where MPC8540 is an agent, plugged into the PC host. 
I couldn't get an IRQ for the card when plugged in some PCI slots, whereas in one of the slots I do see a non-zero value.

It would be great if you could help me understand the issue.

I am sorry if the question is inapporpriate in the group. I am desperate to solve the issue.

Thanks for any help.
J.Joseph

----- Original Message ----
From: Kumar Gala <galak@kernel.crashing.org>
To: Wang Matthew-R59995 <Qi.W@freescale.com>
Cc: linuxppc-embedded@ozlabs.org
Sent: Wednesday, 18 October, 2006 6:54:24 PM
Subject: Re: Linuxppc-embedded Digest, Vol 26, Issue 36


On Oct 18, 2006, at 8:10 PM, Wang Matthew-R59995 wrote:

> Hi Kumar,
>
> Actually I do many trials about it. Vxwerks Bootrom is smaller than
> U-Boot. The key difference between Bootrom and U-boot is that some
> source code of Bootrom is invisible to the users.
>
> Actually the rfi instruction which I point out is the first rfi
> instruction of Linux PowerPC bringup.
>
> Before that, it's TLB entry invalidation and temp TLB entry mapping.
>
> I check MMU setting carefully before coming Linux Kernel.
>
> I just want to know if other guys met similar scenario like me. I  
> don't
> need the precise answer, just overall suggestion about it because I
> understand that not everyone has the same bootloader of mine, that
> bootloader is actually a customized bootloader.

I understand that, thus I was asking what exact problem you were  
seeing to try and help.

> Anyway thank you.
>
> R9 point to LR register, mask the high 20 bit of r9 and send to r7,  
> and
> then add 24, which means stride 6 instructions for rfi instruction
> execution.
>
> Of course, rfi can switch the TLB entry, both the previous TLB  
> entry and
> the temp TLB entry point to the same physical address.
>
> I've checked it.

I know what the code does, I wrote it :)

- kumar



_______________________________________________
Linuxppc-embedded mailing list
Linuxppc-embedded@ozlabs.org
https://ozlabs.org/mailman/listinfo/linuxppc-embedded







		
__________________________________________________________
Yahoo! India Answers: Share what you know. Learn something new
http://in.answers.yahoo.com/

[-- Attachment #2: Type: text/html, Size: 3262 bytes --]

Re: PCI IRQ -MPC8540

[Steve Iribarne (GMail)]

Is this the pin used for the external messaging?


On 1/16/07, agnel juni <junijoseph@yahoo.co.in> wrote:
>
> Hello all:
>
> I would like to understand if there is a way to program the PCI interrupt
> pin register of MPC8540. I did look for this in the datasheet, but couldn't
> find any positive answer.
>
> Currently, I am working in a set-up where MPC8540 is an agent, plugged into
> the PC host.
> I couldn't get an IRQ for the card when plugged in some PCI slots, whereas
> in one of the slots I do see a non-zero value.
>
> It would be great if you could help me understand the issue.
>
> I am sorry if the question is inapporpriate in the group. I am desperate to
> solve the issue.
>
> Thanks for any help.
> J.Joseph
>
> ----- Original Message ----
> From: Kumar Gala <galak@kernel.crashing.org>
> To: Wang Matthew-R59995 <Qi.W@freescale.com>
> Cc: linuxppc-embedded@ozlabs.org
> Sent: Wednesday, 18 October, 2006 6:54:24 PM
> Subject: Re: Linuxppc-embedded Digest, Vol 26, Issue 36
>
>
> On Oct 18, 2006, at 8:10 PM, Wang Matthew-R59995 wrote:
>
> > Hi Kumar,
> >
> > Actually I do many trials about it. Vxwerks Bootrom is smaller than
> > U-Boot. The key difference between Bootrom and U-boot is that some
> > source code of Bootrom is invisible to the users.
> >
> > Actually the rfi instruction which I point out is the first rfi
> > instruction of Linux PowerPC bringup.
> >
> > Before that, it's TLB entry invalidation and temp TLB entry mapping.
> >
> > I check MMU setting carefully before coming Linux Kernel.
> >
> > I just want to know if other guys met similar scenario like me. I
> > don't
> > need the precise answer, just overall suggestion about it because I
> > understand that not everyone has the same bootloader of mine, that
> > bootloader is actually a customized bootloader.
>
> I understand that, thus I was asking what exact problem you were
> seeing to try and help.
>
> > Anyway thank you.
> >
> > R9 point to LR register, mask the high 20 bit of r9 and send to r7,
> > and
> > then add 24, which means stride 6 instructions for rfi instruction
> > execution.
> >
> > Of course, rfi can switch the TLB entry, both the previous TLB
> > entry and
> > the temp TLB entry point to the same physical address.
> >
> > I've checked it.
>
> I know what the code does, I wrote it :)
>
> - kumar
>
>
>
> _______________________________________________
> Linuxppc-embedded mailing list
> Linuxppc-embedded@ozlabs.org
> https://ozlabs.org/mailman/listinfo/linuxppc-embedded
>
>
>  ________________________________
>  Here's a new way to find what you're looking for - Yahoo! Answers
> _______________________________________________
> Linuxppc-embedded mailing list
> Linuxppc-embedded@ozlabs.org
> https://ozlabs.org/mailman/listinfo/linuxppc-embedded
>
>


-- 
/*
 * Steve Iribarne
 * Software Engineer
 * (aka Grunt)
 */

Re: PCI IRQ -MPC8540

[Kumar Gala]

On Jan 16, 2007, at 5:20 PM, agnel juni wrote:

> Hello all:
>
> I would like to understand if there is a way to program the PCI =20
> interrupt pin register of MPC8540. I did look for this in the =20
> datasheet, but couldn't find any positive answer.
>
> Currently, I am working in a set-up where MPC8540 is an agent, =20
> plugged into the PC host.
> I couldn't get an IRQ for the card when plugged in some PCI slots, =20
> whereas in one of the slots I do see a non-zero value.
>
> It would be great if you could help me understand the issue.
>
> I am sorry if the question is inapporpriate in the group. I am =20
> desperate to solve the issue.
>
> Thanks for any help.

You need to find out how the PCI edge interrupts are wired on you =20
board than we can talk about how to generate an interrupt based on =20
that.  I'd guess IRQ_OUT is wired to INTA, but it would be good to =20
verify that.

- k

> J.Joseph
>
> ----- Original Message ----
> From: Kumar Gala <galak@kernel.crashing.org>
> To: Wang Matthew-R59995 <Qi.W@freescale.com>
> Cc: linuxppc-embedded@ozlabs.org
> Sent: Wednesday, 18 October, 2006 6:54:24 PM
> Subject: Re: Linuxppc-embedded Digest, Vol 26, Issue 36
>
>
> On Oct 18, 2006, at 8:10 PM, Wang Matthew-R59995 wrote:
>
> > Hi Kumar,
> >
> > Actually I do many trials about it. Vxwerks Bootrom is smaller than
> > U-Boot. The key difference between Bootrom and U-boot is that some
> > source code of Bootrom is invisible to the users.
> >
> > Actually the rfi instruction which I point out is the first rfi
> > instruction of Linux PowerPC bringup.
> >
> > Before that, it's TLB entry invalidation and temp TLB entry mapping.
> >
> > I check MMU setting carefully before coming Linux Kernel.
> >
> > I just want to know if other guys met similar scenario like me. I
> > don't
> > need the precise answer, just overall suggestion about it because I
> > understand that not everyone has the same bootloader of mine, that
> > bootloader is actually a customized bootloader.
>
> I understand that, thus I was asking what exact problem you were
> seeing to try and help.
>
> > Anyway thank you.
> >
> > R9 point to LR register, mask the high 20 bit of r9 and send to r7,
> > and
> > then add 24, which means stride 6 instructions for rfi instruction
> > execution.
> >
> > Of course, rfi can switch the TLB entry, both the previous TLB
> > entry and
> > the temp TLB entry point to the same physical address.
> >
> > I've checked it.
>
> I know what the code does, I wrote it :)
>
> - kumar
>
>
>
> _______________________________________________
> Linuxppc-embedded mailing list
> Linuxppc-embedded@ozlabs.org
> https://ozlabs.org/mailman/listinfo/linuxppc-embedded
>
>
> Here=92s a new way to find what you're looking for - Yahoo! Answers

Re: PCI IRQ -MPC8540

[Kumar Gala]

What kinda of interrupt do you want on IRQ_OUT/INTA?

- k

On Jan 16, 2007, at 7:54 PM, agnel juni wrote:

> Yes IRQ_OUT is wired to INTA.
>
> Could you please advise on how to generate interrupt for this.
>
> Thanks
> J-
>
> ----- Original Message ----
> From: Kumar Gala <galak@kernel.crashing.org>
> To: agnel juni <junijoseph@yahoo.co.in>
> Cc: linuxppc-embedded@ozlabs.org
> Sent: Tuesday, 16 January, 2007 5:29:01 PM
> Subject: Re: PCI IRQ -MPC8540
>
>
> On Jan 16, 2007, at 5:20 PM, agnel juni wrote:
>
> > Hello all:
> >
> > I would like to understand if there is a way to program the PCI
> > interrupt pin register of MPC8540. I did look for this in the
> > datasheet, but couldn't find any positive answer.
> >
> > Currently, I am working in a set-up where MPC8540 is an agent,
> > plugged into the PC host.
> > I couldn't get an IRQ for the card when plugged in some PCI slots,
> > whereas in one of the slots I do see a non-zero value.
> >
> > It would be great if you could help me understand the issue.
> >
> > I am sorry if the question is inapporpriate in the group. I am
> > desperate to solve the issue.
> >
> > Thanks for any help.
>
> You need to find out how the PCI edge interrupts are wired on you
> board than we can talk about how to generate an interrupt based on
> that.  I'd guess IRQ_OUT is wired to INTA, but it would be good to
> verify that.
>
> - k
>
> > J.Joseph
> >
> > ----- Original Message ----
> > From: Kumar Gala <galak@kernel.crashing.org>
> > To: Wang Matthew-R59995 <Qi.W@freescale.com>
> > Cc: linuxppc-embedded@ozlabs.org
> > Sent: Wednesday, 18 October, 2006 6:54:24 PM
> > Subject: Re: Linuxppc-embedded Digest, Vol 26, Issue 36
> >
> >
> > On Oct 18, 2006, at 8:10 PM, Wang Matthew-R59995 wrote:
> >
> > > Hi Kumar,
> > >
> > > Actually I do many trials about it. Vxwerks Bootrom is smaller =20
> than
> > > U-Boot. The key difference between Bootrom and U-boot is that some
> > > source code of Bootrom is invisible to the users.
> > >
> > > Actually the rfi instruction which I point out is the first rfi
> > > instruction of Linux PowerPC bringup.
> > >
> > > Before that, it's TLB entry invalidation and temp TLB entry =20
> mapping.
> > >
> > > I check MMU setting carefully before coming Linux Kernel.
> > >
> > > I just want to know if other guys met similar scenario like me. I
> > > don't
> > > need the precise answer, just overall suggestion about it =20
> because I
> > > understand that not everyone has the same bootloader of mine, that
> > > bootloader is actually a customized bootloader.
> >
> > I understand that, thus I was asking what exact problem you were
> > seeing to try and help.
> >
> > > Anyway thank you.
> > >
> > > R9 point to LR register, mask the high 20 bit of r9 and send to =20=

> r7,
> > > and
> > > then add 24, which means stride 6 instructions for rfi instruction
> > > execution.
> > >
> > > Of course, rfi can switch the TLB entry, both the previous TLB
> > > entry and
> > > the temp TLB entry point to the same physical address.
> > >
> > > I've checked it.
> >
> > I know what the code does, I wrote it :)
> >
> > - kumar
> >
> >
> >
> > _______________________________________________
> > Linuxppc-embedded mailing list
> > Linuxppc-embedded@ozlabs.org
> > https://ozlabs.org/mailman/listinfo/linuxppc-embedded
> >
> >
> > Here=92s a new way to find what you're looking for - Yahoo! Answers
>
>
>
> Here=92s a new way to find what you're looking for - Yahoo! Answers

RE: PCI IRQ -MPC8540

[Jin Zhengxiong-R64188]

>What kinda of interrupt do you want on IRQ_OUT/INTA?

>- k

On Jan 16, 2007, at 7:54 PM, agnel juni wrote:

> Yes IRQ_OUT is wired to INTA.
>
> Could you please advise on how to generate interrupt for this.

You can refer to a Freescale Linux BSP for MPC8540EVAL board. There's a =
pci agent driver for MPC8540, and the corresponding host driver for =
MPC8560. The agent driver use a message interrupt in MPC8540 and =
redirect the interrupt to IRQ_OUT. As the IRQ_OUT is wired to INTA, you =
can treat MPC8540 as a PCI device.

Jason Jin


> Thanks
> J-
>
> ----- Original Message ----
> From: Kumar Gala <galak@kernel.crashing.org>
> To: agnel juni <junijoseph@yahoo.co.in>
> Cc: linuxppc-embedded@ozlabs.org
> Sent: Tuesday, 16 January, 2007 5:29:01 PM
> Subject: Re: PCI IRQ -MPC8540
>
>
> On Jan 16, 2007, at 5:20 PM, agnel juni wrote:
>
> > Hello all:
> >
> > I would like to understand if there is a way to program the PCI=20
> > interrupt pin register of MPC8540. I did look for this in the=20
> > datasheet, but couldn't find any positive answer.
> >
> > Currently, I am working in a set-up where MPC8540 is an agent,=20
> > plugged into the PC host. I couldn't get an IRQ for the card when=20
> > plugged in some PCI slots, whereas in one of the slots I do see a=20
> > non-zero value.
> >
> > It would be great if you could help me understand the issue.
> >
> > I am sorry if the question is inapporpriate in the group. I am=20
> > desperate to solve the issue.
> >
> > Thanks for any help.
>
> You need to find out how the PCI edge interrupts are wired on you=20
> board than we can talk about how to generate an interrupt based on=20
> that.  I'd guess IRQ_OUT is wired to INTA, but it would be good to=20
> verify that.
>
> - k
>
> > J.Joseph
> >
> > ----- Original Message ----
> > From: Kumar Gala <galak@kernel.crashing.org>
> > To: Wang Matthew-R59995 <Qi.W@freescale.com>
> > Cc: linuxppc-embedded@ozlabs.org
> > Sent: Wednesday, 18 October, 2006 6:54:24 PM
> > Subject: Re: Linuxppc-embedded Digest, Vol 26, Issue 36
> >
> >
> > On Oct 18, 2006, at 8:10 PM, Wang Matthew-R59995 wrote:
> >
> > > Hi Kumar,
> > >
> > > Actually I do many trials about it. Vxwerks Bootrom is smaller
> than
> > > U-Boot. The key difference between Bootrom and U-boot is that some =

> > > source code of Bootrom is invisible to the users.
> > >
> > > Actually the rfi instruction which I point out is the first rfi=20
> > > instruction of Linux PowerPC bringup.
> > >
> > > Before that, it's TLB entry invalidation and temp TLB entry
> mapping.
> > >
> > > I check MMU setting carefully before coming Linux Kernel.
> > >
> > > I just want to know if other guys met similar scenario like me. I=20
> > > don't need the precise answer, just overall suggestion about it
> because I
> > > understand that not everyone has the same bootloader of mine, that =

> > > bootloader is actually a customized bootloader.
> >
> > I understand that, thus I was asking what exact problem you were=20
> > seeing to try and help.
> >
> > > Anyway thank you.
> > >
> > > R9 point to LR register, mask the high 20 bit of r9 and send to
> r7,
> > > and
> > > then add 24, which means stride 6 instructions for rfi instruction =

> > > execution.
> > >
> > > Of course, rfi can switch the TLB entry, both the previous TLB=20
> > > entry and the temp TLB entry point to the same physical address.
> > >
> > > I've checked it.
> >
> > I know what the code does, I wrote it :)
> >
> > - kumar
> >
> >
> >
> > _______________________________________________
> > Linuxppc-embedded mailing list
> > Linuxppc-embedded@ozlabs.org=20
> > https://ozlabs.org/mailman/listinfo/linuxppc-embedded
> >
> >
> > Here's a new way to find what you're looking for - Yahoo! Answers
>
>
>
> Here's a new way to find what you're looking for - Yahoo! Answers

_______________________________________________
Linuxppc-embedded mailing list
Linuxppc-embedded@ozlabs.org =
https://ozlabs.org/mailman/listinfo/linuxppc-embedded

Re: PCI IRQ -MPC8540

[agnel juni]

[-- Attachment #1: Type: text/plain, Size: 6281 bytes --]

Hello all:



Can someone provide me the pointer for the patch for u-boot-1.1.2 to set-up  MPC8540 as an agent correctly.

Thanks
J.Joseph



----- Original Message ----
From: Jin Zhengxiong-R64188 <Jason.Jin@freescale.com>
To: agnel juni <junijoseph@yahoo.co.in>
Sent: Thursday, 18 January, 2007 6:43:55 PM
Subject: RE: PCI IRQ -MPC8540

Message

 
DIV {
MARGIN:0px;}




The comments inline the mail, FYI... 

 

 Ok.

The issue is , for the agent 
driver to route the IRQ, if I understand right ... the PCI(being the host) BIOS 
should assign an IRQ value isn't?.
 

Do you means the 
IRQ value should be write to the 8540 pci config space? If the 8540 card was set 
as agent correctly(there's another patch for the 8540 u-boot to set the agent 
correctly), it's the responsibility of the host to settup it's PCI IRQ. The 
agent should just act like and normal PCI card, such as the ether card.the 
agent just need to generate a interrupt on the IRQ_OUT ping when 
needed.

 
This doesn't happen in our 
case.

When we plug the 8540 card as an agent to the PC host, only in one 
of the slots does the host PC BIOS 
assigns an IRQ value for the card. In others it doesn't.


Try to patch the 
8540 u-boot to set the agent correctly, you can also pay attention to the start 
up sequence of the agent and host, may be you need to stat up the agent u-boot 
first(I can't remember this clearly,just FYI)

  
I am just trying to understand the 
possible reason.

Thanks
J.Joseph



----- 
Original Message ----
From: Jin Zhengxiong-R64188 
<Jason.Jin@freescale.com>
To: Kumar Gala 
<galak@kernel.crashing.org>; agnel juni 
<junijoseph@yahoo.co.in>
Cc: linuxppc-embedded@ozlabs.org
Sent: 
Wednesday, 17 January, 2007 1:34:06 AM
Subject: RE: PCI IRQ -MPC8540




>What kinda of interrupt do you want on 
IRQ_OUT/INTA?

>- k

On Jan 16, 2007, at 7:54 PM, agnel juni 
wrote:

> Yes IRQ_OUT is wired to INTA.
>
> Could you 
please advise on how to generate interrupt for this.

You can refer to a 
Freescale Linux BSP for MPC8540EVAL board. There's a pci agent driver for 
MPC8540, and the corresponding host driver for MPC8560. The agent driver use a 
message interrupt in MPC8540 and redirect the interrupt to IRQ_OUT. As the 
IRQ_OUT is wired to INTA, you can treat MPC8540 as a PCI device.

Jason 
Jin


> Thanks
> J-
>
> ----- Original Message 
----
> From: Kumar Gala <galak@kernel.crashing.org>
> To: 
agnel juni <junijoseph@yahoo.co.in>
> Cc: 
linuxppc-embedded@ozlabs.org
> Sent: Tuesday, 16 January, 2007 5:29:01 
PM
> Subject: Re: PCI IRQ -MPC8540
>
>
> On Jan 16, 
2007, at 5:20 PM, agnel juni wrote:
>
> > Hello all:
> 
>
> > I would like to understand if there is a way to program the 
PCI 
> > interrupt pin register of MPC8540. I did look for this in the 

> > datasheet, but couldn't find any positive answer.
> 
>
> > Currently, I am working in a set-up where MPC8540 is an agent, 

> > plugged into the PC host. I couldn't get an IRQ for the card when 

> > plugged in some PCI slots, whereas in one of the slots I do see a 

> > non-zero value.
> >
> > It would be great if you 
could help me understand the issue.
> >
> > I am sorry if the 
question is inapporpriate in the group. I am 
> > desperate to solve 
the issue.
> >
> > Thanks for any help.
>
> You 
need to find out how the PCI edge interrupts are wired on you 
> board 
than we can talk about how to generate an interrupt based on 
> 
that.  I'd guess IRQ_OUT is wired to INTA, but it would be good to 

> verify that.
>
> - k
>
> > J.Joseph
> 
>
> > ----- Original Message ----
> > From: Kumar Gala 
<galak@kernel.crashing.org>
> > To: Wang Matthew-R59995 
<Qi.W@freescale.com>
> > Cc: linuxppc-embedded@ozlabs.org
> 
> Sent: Wednesday, 18 October, 2006 6:54:24 PM
> > Subject: Re: 
Linuxppc-embedded Digest, Vol 26, Issue 36
> >
> >
> 
> On Oct 18, 2006, at 8:10 PM, Wang Matthew-R59995 wrote:
> 
>
> > > Hi Kumar,
> > >
> > > Actually I 
do many trials about it. Vxwerks Bootrom is smaller
> than
> > 
> U-Boot. The key difference between Bootrom and U-boot is that some 
> 
> > source code of Bootrom is invisible to the users.
> > 
>
> > > Actually the rfi instruction which I point out is the 
first rfi 
> > > instruction of Linux PowerPC bringup.
> > 
>
> > > Before that, it's TLB entry invalidation and temp TLB 
entry
> mapping.
> > >
> > > I check MMU setting 
carefully before coming Linux Kernel.
> > >
> > > I just 
want to know if other guys met similar scenario like me. I 
> > > 
don't need the precise answer, just overall suggestion about it
> because 
I
> > > understand that not everyone has the same bootloader of 
mine, that 
> > > bootloader is actually a customized 
bootloader.
> >
> > I understand that, thus I was asking what 
exact problem you were 
> > seeing to try and help.
> 
>
> > > Anyway thank you.
> > >
> > > R9 
point to LR register, mask the high 20 bit of r9 and send to
> r7,
> 
> > and
> > > then add 24, which means stride 6 instructions 
for rfi instruction 
> > > execution.
> > >
> > 
> Of course, rfi can switch the TLB entry, both the previous TLB 
> 
> > entry and the temp TLB entry point to the same physical 
address.
> > >
> > > I've checked it.
> 
>
> > I know what the code does, I wrote it :)
> >
> 
> - kumar
> >
> >
> >
> > 
_______________________________________________
> > Linuxppc-embedded 
mailing list
> > Linuxppc-embedded@ozlabs.org 
> > https://ozlabs.org/mailman/listinfo/linuxppc-embedded
> 
>
> >
> > Here's a new way to find what you're looking for 
- Yahoo! Answers
>
>
>
> Here's a new way to find what 
you're looking for - Yahoo! 
Answers

_______________________________________________
Linuxppc-embedded 
mailing list
Linuxppc-embedded@ozlabs.org https://ozlabs.org/mailman/listinfo/linuxppc-embedded









Here¢s a new way to find what you're looking for - Yahoo! 
Answers 





		
__________________________________________________________
Yahoo! India Answers: Share what you know. Learn something new
http://in.answers.yahoo.com/

[-- Attachment #2: Type: text/html, Size: 9679 bytes --]