mpc8xx-2.2.13 booting problem

mpc8xx-2.2.13 booting problem

[Clinton Lam]

Hi,

I am new to the embeded linux. I am trying to port mpc8xx-2.2.13 kernel to
our custom board, called Touchstone. It is a reference design based on
mpc823, which has 32 Mbyte of SDRAM, and 4 Mbyte of flash.

So far, I've developed a network bootloader, which downloads the zImage to
the target. I am at the stage where I can unzip the kernel image to address
location 0x0, and jump to it. Then, inside the routine "start_kernel", I am
able to get the console up and have the linux banner partially displayed at
the cosole.

However, after some more kernel initialization in "start_kernel", the
program stuck in a routine called "sti()". I've debugged through this
routine (it is an assembly routine) , and it seems that it is looping
forever inside the routine. I know that the routine is trying to set the EE
bit in MSR, and also doing some thing with ppc_n_lost_interrupts, which I
don't quite understand.

Does anyone have any idea what is wrong with my port? Can someone explain to
me what is the purpose of this "sti()" routine, what are the possible cause
for it to hang?

Your response is greately appreciated!

Thanks in advance.

Clinton Lam

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

Re: mpc8xx-2.2.13 booting problem

[Alan Mimms]

Clinton,

This function enables interrupts.  Some bit of hardware is leaving an
interrupt asserted permanently, or else your port or your driver for some bit
of hardware is not turning off the interrupt request by servicing the interrupt
properly.  Or, possibly, your 8xx interrupt management is screwed up.

One of those explanations is it, just about guaranteed...

Luck.
a

On Thu, 03 Feb 2000, Clinton Lam wrote:
> Hi,
>
> I am new to the embeded linux. I am trying to port mpc8xx-2.2.13 kernel to
> our custom board, called Touchstone. It is a reference design based on
> mpc823, which has 32 Mbyte of SDRAM, and 4 Mbyte of flash.
>
> So far, I've developed a network bootloader, which downloads the zImage to
> the target. I am at the stage where I can unzip the kernel image to address
> location 0x0, and jump to it. Then, inside the routine "start_kernel", I am
> able to get the console up and have the linux banner partially displayed at
> the cosole.
>
> However, after some more kernel initialization in "start_kernel", the
> program stuck in a routine called "sti()". I've debugged through this
> routine (it is an assembly routine) , and it seems that it is looping
> forever inside the routine. I know that the routine is trying to set the EE
> bit in MSR, and also doing some thing with ppc_n_lost_interrupts, which I
> don't quite understand.
>
> Does anyone have any idea what is wrong with my port? Can someone explain to
> me what is the purpose of this "sti()" routine, what are the possible cause
> for it to hang?
>
> Your response is greately appreciated!
>
> Thanks in advance.
>
> Clinton Lam
--
Alan Mimms     Packet Engines, Inc.     Spokane, Washington [99214-0497]
  USA, Earth, Sol, Milky Way, The Local Group, Virgo Supercluster, U0
Despite the cost of living, have you noticed how popular it remains?
  -- Steven Wright?

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

RE: mpc8xx-2.2.13 booting problem

[Clinton Lam]

Alan,

After another 5 hours of debugging, I've narrowed the problem to the
decrement counter interrupt. For some unknown reasons, the interrupt never
gets clear. The system just keep looping and waiting for this interrupt to
be cleared.

In the file time.c (this is the file where all of the timer interrupt code
is), it mentions something about making an assumption that MBX board is
running off a 32.768KHz crystal. But the Touchstone board has a 3.57 MHz
crystal. So, something has to be changed to make this work! Anyone has any
ideas???

BTW, the network bootloader is modified from a comercial OS. We use it to
jump start our project. I don't think we can share the code with everyone,
sorry...

Regards
Clinton Lam

> -----Original Message-----
> From: Alan Mimms [mailto:alan@packetengines.com]
> Sent: Thursday, February 03, 2000 10:03 AM
> To: Clinton Lam; Linuxppc-Embedded (E-mail)
> Subject: Re: mpc8xx-2.2.13 booting problem
>
>
> Clinton,
>
> This function enables interrupts.  Some bit of hardware is leaving an
> interrupt asserted permanently, or else your port or your
> driver for some bit
> of hardware is not turning off the interrupt request by
> servicing the interrupt
> properly.  Or, possibly, your 8xx interrupt management is screwed up.
>
> One of those explanations is it, just about guaranteed...
>
> Luck.
> a
>
> On Thu, 03 Feb 2000, Clinton Lam wrote:
> > Hi,
> >
> > I am new to the embeded linux. I am trying to port
> mpc8xx-2.2.13 kernel to
> > our custom board, called Touchstone. It is a reference
> design based on
> > mpc823, which has 32 Mbyte of SDRAM, and 4 Mbyte of flash.
> >
> > So far, I've developed a network bootloader, which
> downloads the zImage to
> > the target. I am at the stage where I can unzip the kernel
> image to address
> > location 0x0, and jump to it. Then, inside the routine
> "start_kernel", I am
> > able to get the console up and have the linux banner
> partially displayed at
> > the cosole.
> >
> > However, after some more kernel initialization in
> "start_kernel", the
> > program stuck in a routine called "sti()". I've debugged
> through this
> > routine (it is an assembly routine) , and it seems that it
> is looping
> > forever inside the routine. I know that the routine is
> trying to set the EE
> > bit in MSR, and also doing some thing with
> ppc_n_lost_interrupts, which I
> > don't quite understand.
> >
> > Does anyone have any idea what is wrong with my port? Can
> someone explain to
> > me what is the purpose of this "sti()" routine, what are
> the possible cause
> > for it to hang?
> >
> > Your response is greately appreciated!
> >
> > Thanks in advance.
> >
> > Clinton Lam
--
Alan Mimms     Packet Engines, Inc.     Spokane, Washington [99214-0497]
  USA, Earth, Sol, Milky Way, The Local Group, Virgo Supercluster, U0
Despite the cost of living, have you noticed how popular it remains?
  -- Steven Wright?

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

RE: mpc8xx-2.2.13 booting problem

[Alan Mimms]

Sounds like you need to turn on some prescaling for the timebase/decrementer.
If you can't find the required stuff, maybe I can help, but I'm not all THAT
familiar with these cockamamie 8xx processors myself.  Perhaps someone else?

a

On
Thu, 03 Feb 2000, Clinton Lam wrote: > Alan,
>
> After another 5 hours of debugging, I've narrowed the problem to the
> decrement counter interrupt. For some unknown reasons, the interrupt never
> gets clear. The system just keep looping and waiting for this interrupt to
> be cleared.
>
> In the file time.c (this is the file where all of the timer interrupt code
> is), it mentions something about making an assumption that MBX board is
> running off a 32.768KHz crystal. But the Touchstone board has a 3.57 MHz
> crystal. So, something has to be changed to make this work! Anyone has any
> ideas???
>
> BTW, the network bootloader is modified from a comercial OS. We use it to
> jump start our project. I don't think we can share the code with everyone,
> sorry...
>
> Regards
> Clinton Lam
>
> > -----Original Message-----
> > From: Alan Mimms [mailto:alan@packetengines.com]
> > Sent: Thursday, February 03, 2000 10:03 AM
> > To: Clinton Lam; Linuxppc-Embedded (E-mail)
> > Subject: Re: mpc8xx-2.2.13 booting problem
> >
> >
> > Clinton,
> >
> > This function enables interrupts.  Some bit of hardware is leaving an
> > interrupt asserted permanently, or else your port or your
> > driver for some bit
> > of hardware is not turning off the interrupt request by
> > servicing the interrupt
> > properly.  Or, possibly, your 8xx interrupt management is screwed up.
> >
> > One of those explanations is it, just about guaranteed...
> >
> > Luck.
> > a
> >
> > On Thu, 03 Feb 2000, Clinton Lam wrote:
> > > Hi,
> > >
> > > I am new to the embeded linux. I am trying to port
> > mpc8xx-2.2.13 kernel to
> > > our custom board, called Touchstone. It is a reference
> > design based on
> > > mpc823, which has 32 Mbyte of SDRAM, and 4 Mbyte of flash.
> > >
> > > So far, I've developed a network bootloader, which
> > downloads the zImage to
> > > the target. I am at the stage where I can unzip the kernel
> > image to address
> > > location 0x0, and jump to it. Then, inside the routine
> > "start_kernel", I am
> > > able to get the console up and have the linux banner
> > partially displayed at
> > > the cosole.
> > >
> > > However, after some more kernel initialization in
> > "start_kernel", the
> > > program stuck in a routine called "sti()". I've debugged
> > through this
> > > routine (it is an assembly routine) , and it seems that it
> > is looping
> > > forever inside the routine. I know that the routine is
> > trying to set the EE
> > > bit in MSR, and also doing some thing with
> > ppc_n_lost_interrupts, which I
> > > don't quite understand.
> > >
> > > Does anyone have any idea what is wrong with my port? Can
> > someone explain to
> > > me what is the purpose of this "sti()" routine, what are
> > the possible cause
> > > for it to hang?
> > >
> > > Your response is greately appreciated!
> > >
> > > Thanks in advance.
> > >
> > > Clinton Lam
> --
> Alan Mimms     Packet Engines, Inc.     Spokane, Washington [99214-0497]
>   USA, Earth, Sol, Milky Way, The Local Group, Virgo Supercluster, U0
> Despite the cost of living, have you noticed how popular it remains?
>   -- Steven Wright?
--
Alan Mimms     Packet Engines, Inc.     Spokane, Washington [99214-0497]
  USA, Earth, Sol, Milky Way, The Local Group, Virgo Supercluster, U0
Despite the cost of living, have you noticed how popular it remains?
  -- Steven Wright?

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

Re: mpc8xx-2.2.13 booting problem

[Richard Hendricks]

You can either have the TB/DEC run from the oscclk/4, or from the
system clock/16.  I would assume Linux would use oscclk/4, or 8kHz,
and assume that value for its calculations.  You'll have to change
the funtion that calculates the timeout value.

Alan Mimms wrote:
>
> Sounds like you need to turn on some prescaling for the timebase/decrementer.
> If you can't find the required stuff, maybe I can help, but I'm not all THAT
> familiar with these cockamamie 8xx processors myself.  Perhaps someone else?
>
> a
>
> On
> Thu, 03 Feb 2000, Clinton Lam wrote: > Alan,
> >
> > After another 5 hours of debugging, I've narrowed the problem to the
> > decrement counter interrupt. For some unknown reasons, the interrupt never
> > gets clear. The system just keep looping and waiting for this interrupt to
> > be cleared.
> >
> > In the file time.c (this is the file where all of the timer interrupt code
> > is), it mentions something about making an assumption that MBX board is
> > running off a 32.768KHz crystal. But the Touchstone board has a 3.57 MHz
> > crystal. So, something has to be changed to make this work! Anyone has any
> > ideas???
> >
> > BTW, the network bootloader is modified from a comercial OS. We use it to
> > jump start our project. I don't think we can share the code with everyone,
> > sorry...
> >
> > Regards
> > Clinton Lam
> >
> > > -----Original Message-----
> > > From: Alan Mimms [mailto:alan@packetengines.com]
> > > Sent: Thursday, February 03, 2000 10:03 AM
> > > To: Clinton Lam; Linuxppc-Embedded (E-mail)
> > > Subject: Re: mpc8xx-2.2.13 booting problem
> > >
> > >
> > > Clinton,
> > >
> > > This function enables interrupts.  Some bit of hardware is leaving an
> > > interrupt asserted permanently, or else your port or your
> > > driver for some bit
> > > of hardware is not turning off the interrupt request by
> > > servicing the interrupt
> > > properly.  Or, possibly, your 8xx interrupt management is screwed up.
> > >
> > > One of those explanations is it, just about guaranteed...
> > >
> > > Luck.
> > > a
> > >
> > > On Thu, 03 Feb 2000, Clinton Lam wrote:
> > > > Hi,
> > > >
> > > > I am new to the embeded linux. I am trying to port
> > > mpc8xx-2.2.13 kernel to
> > > > our custom board, called Touchstone. It is a reference
> > > design based on
> > > > mpc823, which has 32 Mbyte of SDRAM, and 4 Mbyte of flash.
> > > >
> > > > So far, I've developed a network bootloader, which
> > > downloads the zImage to
> > > > the target. I am at the stage where I can unzip the kernel
> > > image to address
> > > > location 0x0, and jump to it. Then, inside the routine
> > > "start_kernel", I am
> > > > able to get the console up and have the linux banner
> > > partially displayed at
> > > > the cosole.
> > > >
> > > > However, after some more kernel initialization in
> > > "start_kernel", the
> > > > program stuck in a routine called "sti()". I've debugged
> > > through this
> > > > routine (it is an assembly routine) , and it seems that it
> > > is looping
> > > > forever inside the routine. I know that the routine is
> > > trying to set the EE
> > > > bit in MSR, and also doing some thing with
> > > ppc_n_lost_interrupts, which I
> > > > don't quite understand.
> > > >
> > > > Does anyone have any idea what is wrong with my port? Can
> > > someone explain to
> > > > me what is the purpose of this "sti()" routine, what are
> > > the possible cause
> > > > for it to hang?
> > > >
> > > > Your response is greately appreciated!
> > > >
> > > > Thanks in advance.
> > > >
> > > > Clinton Lam
> > --
> > Alan Mimms     Packet Engines, Inc.     Spokane, Washington [99214-0497]
> >   USA, Earth, Sol, Milky Way, The Local Group, Virgo Supercluster, U0
> > Despite the cost of living, have you noticed how popular it remains?
> >   -- Steven Wright?
> --
> Alan Mimms     Packet Engines, Inc.     Spokane, Washington [99214-0497]
>   USA, Earth, Sol, Milky Way, The Local Group, Virgo Supercluster, U0
> Despite the cost of living, have you noticed how popular it remains?
>   -- Steven Wright?

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