Redboot causes PCI abort in Linux Ixp425

Redboot causes PCI abort in Linux Ixp425

[Ravi modgekar]

Hi,
   I have an  IXP425 based board based on Gateworks
GW2345 board.This
board already had a redboot image programmed  through
which we could
easily load linux and work on it.
  Now I have compiled the redboot for our IXP based
board from the
Avila redboot sources.With this newly compiled redboot
I am able to
boot linux, but unfortunately the PCI   cribs and the
linux crashes
with PCI abort. The error message being "!!!
abort_handler addr =
0x40015000 ,isr = 0x0 , status = 0x22a0".
Interestingly the same Linux
image works fine with the old redboot ,for which we
don't have the
source code.
  Has anybody in this list faced this problem.Am I
missing
configuring something in the Redboot  which is causing
the PCI abort.I
have tried with completely disabling PCI interface ,
but I still see
this problem.

 any pointers !!!

Regds
Ravi

__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 

RE: [ECOS] Redboot causes PCI abort in Linux Ixp425

[Stephen Goadhouse]

Xscale processors, and perhaps other ARM-based processors, will get a
data abort exception while scanning the PCI bus.  This is normal.  To
scan the bus, the system should be trying to read every available PCI
configuration slot address to see if a device is there.  If no device is
there, you'll get a data abort exception because technically the PCI
transaction did not complete (no target to handle the PCI configuration
read).  Typically, this is handled by installing a data abort exception
handler prior to scanning the bus that gracefully handles these
exceptions during PCI bus scan.  Another alternative with the IXP425 may
be to setup its internal PCI bridge so that it returns 0xFFFFFFFF instead
of doing a data abort.  I have not looked at the ixp425 manual in a
while, but I am fairly certain that it has this ability.  I would expect
that's what your old RedBoot did to handle this so that Linux would not
have to.  Receiving 0xFFFFFFFF when reading non-existent PCI slots is
what happens in a PC environment and I'm fairly certain that is what
Linux typically expects.   


_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/  
_/  Stephen Goadhouse                 Work: (434) 978-2888 x254 _/
_/  Senior Staff Engineer  stephen.goadhouse@adiengineering.com _/
_/  ADI Engineering               http://www.adiengineering.com _/
_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/ 


> -----Original Message-----
> From: ecos-discuss-owner@ecos.sourceware.org 
> [mailto:ecos-discuss-owner@ecos.sourceware.org] On Behalf Of 
> Ravi modgekar
> Sent: Thursday, March 16, 2006 1:04 AM
> To: ecos-devel@sourceware.org; ecos-discuss@sourceware.org
> Subject: [ECOS] Redboot causes PCI abort in Linux Ixp425
> 
> Hi,
>    I have an  IXP425 based board based on Gateworks
> GW2345 board.This
> board already had a redboot image programmed  through which 
> we could easily load linux and work on it.
>   Now I have compiled the redboot for our IXP based board 
> from the Avila redboot sources.With this newly compiled 
> redboot I am able to
> boot linux, but unfortunately the PCI   cribs and the
> linux crashes
> with PCI abort. The error message being "!!!
> abort_handler addr =
> 0x40015000 ,isr = 0x0 , status = 0x22a0".
> Interestingly the same Linux
> image works fine with the old redboot ,for which we don't 
> have the source code.
>   Has anybody in this list faced this problem.Am I missing 
> configuring something in the Redboot  which is causing the 
> PCI abort.I have tried with completely disabling PCI 
> interface , but I still see this problem.
> 
>  any pointers !!!
> 
> Regds
> Ravi
> 
> __________________________________________________
> Do You Yahoo!?
> Tired of spam?  Yahoo! Mail has the best spam protection 
> around http://mail.yahoo.com 
> 
> --
> Before posting, please read the FAQ: 
> http://ecos.sourceware.org/fom/ecos
> and search the list archive: 
> http://ecos.sourceware.org/ml/ecos-discuss
> 
> 
> __________ NOD32 1.1445 (20060315) Information __________
> 
> This message was checked by NOD32 antivirus system.
> http://www.eset.com
> 
> 

RE: [ECOS] Redboot causes PCI abort in Linux Ixp425

[Ravi modgekar]

Hi Stephen,
   Data abort exception during PCI scan is ok.But in
my case, the abort does not happen while scanning the
bus , but it happens while accessing the device!!!!
   Basically Linux does the PCI enumeration at the
boot time and sets up the device properly.There is
absolutely no problem while scanning.The abort happens
only when you try to access the PCI device.Now when
you are trying to access the device , there is no
question of scanning the bus again.
  I am wondering , if I am missing something else in
the redboot .I also don't see any need to scan the PCI
bus in the redboot, since linux is re doing it for
me.what could be wrong.
  By the way, I think even the ADI coyote board does
not scan the PCI bus in Redboot.

Regds
Ravi
   

--- Stephen Goadhouse
<stephen.goadhouse@adiengineering.com> wrote:

> 
> Xscale processors, and perhaps other ARM-based
> processors, will get a
> data abort exception while scanning the PCI bus. 
> This is normal.  To
> scan the bus, the system should be trying to read
> every available PCI
> configuration slot address to see if a device is
> there.  If no device is
> there, you'll get a data abort exception because
> technically the PCI
> transaction did not complete (no target to handle
> the PCI configuration
> read).  Typically, this is handled by installing a
> data abort exception
> handler prior to scanning the bus that gracefully
> handles these
> exceptions during PCI bus scan.  Another alternative
> with the IXP425 may
> be to setup its internal PCI bridge so that it
> returns 0xFFFFFFFF instead
> of doing a data abort.  I have not looked at the
> ixp425 manual in a
> while, but I am fairly certain that it has this
> ability.  I would expect
> that's what your old RedBoot did to handle this so
> that Linux would not
> have to.  Receiving 0xFFFFFFFF when reading
> non-existent PCI slots is
> what happens in a PC environment and I'm fairly
> certain that is what
> Linux typically expects.   
> 
> 
>
_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/
>  
> _/  Stephen Goadhouse                 Work: (434)
> 978-2888 x254 _/
> _/  Senior Staff Engineer 
> stephen.goadhouse@adiengineering.com _/
> _/  ADI Engineering              
> http://www.adiengineering.com _/
>
_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/
> 
> 
> 
> > -----Original Message-----
> > From: ecos-discuss-owner@ecos.sourceware.org 
> > [mailto:ecos-discuss-owner@ecos.sourceware.org] On
> Behalf Of 
> > Ravi modgekar
> > Sent: Thursday, March 16, 2006 1:04 AM
> > To: ecos-devel@sourceware.org;
> ecos-discuss@sourceware.org
> > Subject: [ECOS] Redboot causes PCI abort in Linux
> Ixp425
> > 
> > Hi,
> >    I have an  IXP425 based board based on
> Gateworks
> > GW2345 board.This
> > board already had a redboot image programmed 
> through which 
> > we could easily load linux and work on it.
> >   Now I have compiled the redboot for our IXP
> based board 
> > from the Avila redboot sources.With this newly
> compiled 
> > redboot I am able to
> > boot linux, but unfortunately the PCI   cribs and
> the
> > linux crashes
> > with PCI abort. The error message being "!!!
> > abort_handler addr =
> > 0x40015000 ,isr = 0x0 , status = 0x22a0".
> > Interestingly the same Linux
> > image works fine with the old redboot ,for which
> we don't 
> > have the source code.
> >   Has anybody in this list faced this problem.Am I
> missing 
> > configuring something in the Redboot  which is
> causing the 
> > PCI abort.I have tried with completely disabling
> PCI 
> > interface , but I still see this problem.
> > 
> >  any pointers !!!
> > 
> > Regds
> > Ravi
> > 
> > __________________________________________________
> > Do You Yahoo!?
> > Tired of spam?  Yahoo! Mail has the best spam
> protection 
> > around http://mail.yahoo.com 
> > 
> > --
> > Before posting, please read the FAQ: 
> > http://ecos.sourceware.org/fom/ecos
> > and search the list archive: 
> > http://ecos.sourceware.org/ml/ecos-discuss
> > 
> > 
> > __________ NOD32 1.1445 (20060315) Information
> __________
> > 
> > This message was checked by NOD32 antivirus
> system.
> > http://www.eset.com
> > 
> > 
> 
> 


__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 

RE: [ECOS] Redboot causes PCI abort in Linux Ixp425

[Stephen Goadhouse]

Yes, you are correct about the ADI Coyote board.  Almost all of our other
platforms scan PCI devices in RedBoot and Linux makes use of the RedBoot
scan.  The opposite is true of Coyote where RedBoot does no scanning and
Linux does the scanning.  Because it's the exception for us, I often forget
that.

You may want to try creating a simple data abort handler that captures the
PCI status and the PC of the offending instruction.  There should be a data
abort handler for PCI scans in some eCos HALs that you can use as an example
for your case.  The PCI status may tell you that you are getting a master
abort or some other type of PCI error or it may have no error, in which case
it could be the address you are using to access the device (make sure it's a
proper virtual address that maps to the correct physical address) or perhaps
some register in the IXP425 is not setup correctly.  

Stephen Goadhouse  
Senior Staff Engineer  
ADI Engineering (http://www.adiengineering.com)
(434) 978-2888 x254

  

> -----Original Message-----
> From: Ravi modgekar [mailto:ravilm@yahoo.com] 
> Sent: Wednesday, March 22, 2006 9:21 AM
> To: Stephen Goadhouse; ecos-devel@sourceware.org; 
> ecos-discuss@sourceware.org
> Subject: RE: [ECOS] Redboot causes PCI abort in Linux Ixp425
> 
> Hi Stephen,
>    Data abort exception during PCI scan is ok.But in my case, 
> the abort does not happen while scanning the bus , but it 
> happens while accessing the device!!!!
>    Basically Linux does the PCI enumeration at the boot time 
> and sets up the device properly.There is absolutely no 
> problem while scanning.The abort happens only when you try to 
> access the PCI device.Now when you are trying to access the 
> device , there is no question of scanning the bus again.
>   I am wondering , if I am missing something else in the 
> redboot .I also don't see any need to scan the PCI bus in the 
> redboot, since linux is re doing it for me.what could be wrong.
>   By the way, I think even the ADI coyote board does not scan 
> the PCI bus in Redboot.
> 
> Regds
> Ravi
>    
> 
> --- Stephen Goadhouse
> <stephen.goadhouse@adiengineering.com> wrote:
> 
> > 
> > Xscale processors, and perhaps other ARM-based processors, 
> will get a 
> > data abort exception while scanning the PCI bus.
> > This is normal.  To
> > scan the bus, the system should be trying to read every 
> available PCI 
> > configuration slot address to see if a device is there.  If 
> no device 
> > is there, you'll get a data abort exception because technically the 
> > PCI transaction did not complete (no target to handle the PCI 
> > configuration read).  Typically, this is handled by 
> installing a data 
> > abort exception handler prior to scanning the bus that gracefully 
> > handles these exceptions during PCI bus scan.  Another alternative 
> > with the IXP425 may be to setup its internal PCI bridge so that it 
> > returns 0xFFFFFFFF instead of doing a data abort.  I have 
> not looked 
> > at the
> > ixp425 manual in a
> > while, but I am fairly certain that it has this ability.  I would 
> > expect that's what your old RedBoot did to handle this so 
> that Linux 
> > would not have to.  Receiving 0xFFFFFFFF when reading 
> non-existent PCI 
> > slots is what happens in a PC environment and I'm fairly 
> certain that 
> > is what
> > Linux typically expects.   
> > 
> > 
> >
> _/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/
> >  
> > _/  Stephen Goadhouse                 Work: (434)
> > 978-2888 x254 _/
> > _/  Senior Staff Engineer
> > stephen.goadhouse@adiengineering.com _/
> > _/  ADI Engineering              
> > http://www.adiengineering.com _/
> >
> _/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/
> > 
> > 
> > 
> > > -----Original Message-----
> > > From: ecos-discuss-owner@ecos.sourceware.org
> > > [mailto:ecos-discuss-owner@ecos.sourceware.org] On
> > Behalf Of
> > > Ravi modgekar
> > > Sent: Thursday, March 16, 2006 1:04 AM
> > > To: ecos-devel@sourceware.org;
> > ecos-discuss@sourceware.org
> > > Subject: [ECOS] Redboot causes PCI abort in Linux
> > Ixp425
> > > 
> > > Hi,
> > >    I have an  IXP425 based board based on
> > Gateworks
> > > GW2345 board.This
> > > board already had a redboot image programmed
> > through which
> > > we could easily load linux and work on it.
> > >   Now I have compiled the redboot for our IXP
> > based board
> > > from the Avila redboot sources.With this newly
> > compiled
> > > redboot I am able to
> > > boot linux, but unfortunately the PCI   cribs and
> > the
> > > linux crashes
> > > with PCI abort. The error message being "!!!
> > > abort_handler addr =
> > > 0x40015000 ,isr = 0x0 , status = 0x22a0".
> > > Interestingly the same Linux
> > > image works fine with the old redboot ,for which
> > we don't
> > > have the source code.
> > >   Has anybody in this list faced this problem.Am I
> > missing
> > > configuring something in the Redboot  which is
> > causing the
> > > PCI abort.I have tried with completely disabling
> > PCI
> > > interface , but I still see this problem.
> > > 
> > >  any pointers !!!
> > > 
> > > Regds
> > > Ravi
> > > 
> > > __________________________________________________
> > > Do You Yahoo!?
> > > Tired of spam?  Yahoo! Mail has the best spam
> > protection
> > > around http://mail.yahoo.com
> > > 
> > > --
> > > Before posting, please read the FAQ: 
> > > http://ecos.sourceware.org/fom/ecos
> > > and search the list archive: 
> > > http://ecos.sourceware.org/ml/ecos-discuss
> > > 
> > > 
> > > __________ NOD32 1.1445 (20060315) Information
> > __________
> > > 
> > > This message was checked by NOD32 antivirus
> > system.
> > > http://www.eset.com
> > > 
> > > 
> > 
> > 
> 
> 
> __________________________________________________
> Do You Yahoo!?
> Tired of spam?  Yahoo! Mail has the best spam protection 
> around http://mail.yahoo.com 

RE: [ECOS] Redboot causes PCI abort in Linux Ixp425

[Daly, Jeffrey]

The situation isn't quite clear to me from your original email.  Are you
saying you've got 2 boards, one GW2345 and one custom?  The GW2345 works
fine.  The custom one works fine with the GW2345 redboot and linux, but
doesn't work whith the custom redboot and linux?  Pls clarify.

Sounds to me like your PCI configuration during redboot is wrong (or at
least is different that what the original does).  You should dump the
PCI config of the GW2345 (from redboot, before booting linux) and dump
the PCi config of the custom board (again redboot, not linux) and post
that.

-----Original Message-----
From: ecos-devel-owner@ecos.sourceware.org
[mailto:ecos-devel-owner@ecos.sourceware.org] On Behalf Of Ravi modgekar
Sent: Wednesday, March 22, 2006 9:21 AM
To: Stephen Goadhouse; ecos-devel@sourceware.org;
ecos-discuss@sourceware.org
Subject: RE: [ECOS] Redboot causes PCI abort in Linux Ixp425

Hi Stephen,
   Data abort exception during PCI scan is ok.But in
my case, the abort does not happen while scanning the
bus , but it happens while accessing the device!!!!
   Basically Linux does the PCI enumeration at the
boot time and sets up the device properly.There is
absolutely no problem while scanning.The abort happens
only when you try to access the PCI device.Now when
you are trying to access the device , there is no
question of scanning the bus again.
  I am wondering , if I am missing something else in
the redboot .I also don't see any need to scan the PCI
bus in the redboot, since linux is re doing it for
me.what could be wrong.
  By the way, I think even the ADI coyote board does
not scan the PCI bus in Redboot.

Regds
Ravi
   

--- Stephen Goadhouse
<stephen.goadhouse@adiengineering.com> wrote:

> 
> Xscale processors, and perhaps other ARM-based
> processors, will get a
> data abort exception while scanning the PCI bus. 
> This is normal.  To
> scan the bus, the system should be trying to read
> every available PCI
> configuration slot address to see if a device is
> there.  If no device is
> there, you'll get a data abort exception because
> technically the PCI
> transaction did not complete (no target to handle
> the PCI configuration
> read).  Typically, this is handled by installing a
> data abort exception
> handler prior to scanning the bus that gracefully
> handles these
> exceptions during PCI bus scan.  Another alternative
> with the IXP425 may
> be to setup its internal PCI bridge so that it
> returns 0xFFFFFFFF instead
> of doing a data abort.  I have not looked at the
> ixp425 manual in a
> while, but I am fairly certain that it has this
> ability.  I would expect
> that's what your old RedBoot did to handle this so
> that Linux would not
> have to.  Receiving 0xFFFFFFFF when reading
> non-existent PCI slots is
> what happens in a PC environment and I'm fairly
> certain that is what
> Linux typically expects.   
> 
> 
>
_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/
>  
> _/  Stephen Goadhouse                 Work: (434)
> 978-2888 x254 _/
> _/  Senior Staff Engineer 
> stephen.goadhouse@adiengineering.com _/
> _/  ADI Engineering              
> http://www.adiengineering.com _/
>
_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/
> 
> 
> 
> > -----Original Message-----
> > From: ecos-discuss-owner@ecos.sourceware.org 
> > [mailto:ecos-discuss-owner@ecos.sourceware.org] On
> Behalf Of 
> > Ravi modgekar
> > Sent: Thursday, March 16, 2006 1:04 AM
> > To: ecos-devel@sourceware.org;
> ecos-discuss@sourceware.org
> > Subject: [ECOS] Redboot causes PCI abort in Linux
> Ixp425
> > 
> > Hi,
> >    I have an  IXP425 based board based on
> Gateworks
> > GW2345 board.This
> > board already had a redboot image programmed 
> through which 
> > we could easily load linux and work on it.
> >   Now I have compiled the redboot for our IXP
> based board 
> > from the Avila redboot sources.With this newly
> compiled 
> > redboot I am able to
> > boot linux, but unfortunately the PCI   cribs and
> the
> > linux crashes
> > with PCI abort. The error message being "!!!
> > abort_handler addr =
> > 0x40015000 ,isr = 0x0 , status = 0x22a0".
> > Interestingly the same Linux
> > image works fine with the old redboot ,for which
> we don't 
> > have the source code.
> >   Has anybody in this list faced this problem.Am I
> missing 
> > configuring something in the Redboot  which is
> causing the 
> > PCI abort.I have tried with completely disabling
> PCI 
> > interface , but I still see this problem.
> > 
> >  any pointers !!!
> > 
> > Regds
> > Ravi
> > 
> > __________________________________________________
> > Do You Yahoo!?
> > Tired of spam?  Yahoo! Mail has the best spam
> protection 
> > around http://mail.yahoo.com 
> > 
> > --
> > Before posting, please read the FAQ: 
> > http://ecos.sourceware.org/fom/ecos
> > and search the list archive: 
> > http://ecos.sourceware.org/ml/ecos-discuss
> > 
> > 
> > __________ NOD32 1.1445 (20060315) Information
> __________
> > 
> > This message was checked by NOD32 antivirus
> system.
> > http://www.eset.com
> > 
> > 
> 
> 


__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com