[U-Boot-Users] Re : MV64460 Burst reads from BootCS

[U-Boot-Users] Re : MV64460 Burst reads from BootCS

[sebastien.douheret]

	Hi,
Marvell MV64460 only supports burst reads/writes on 32bits
width peripherals ;-(
I have the same issue with a Lauterbach probe. You must deny
burst on BootCS if this peripheral is not 32bits width !
This restriction also penalize boot process ! If your boot
flash is 16 or 8 bits witdh, you can't directly execute
instructions in flash. Because CPU like MPC74xx always load
instructions with burst reads.

Regards.
Sebastien.


> ------------------------------
> 
> Message: 5
> Date: Sun, 1 Oct 2006 20:19:12 +0200
> From: "Roel Loeffen" <roel.loeffen@prodrive.nl>
> Subject: [U-Boot-Users] MV64460 Burst reads from BootCS
> To: <u-boot-users@lists.sourceforge.net>
> Message-ID:
> 
<7826EB01AD927140A1DCF898FE63325D0D91B1@ExchangeSrv.Prodrive.nl>
> Content-Type: text/plain; charset="iso-8859-1"
> 
> Hi all,
> 
> We discover some serious problems accessing flash at BootCS 
> of the Marvell MV64460. It looks like burst reads make the 
> whole bridge hang. >From inside U-Boot as well as using only 
> the BDI2000 debugger.
> 
> Reading the flash with bursts enabled (which is default), 
> makes it sometimes impossible to access any 
> registers/dram/devices anymore, only zeros can be read.
> 
> It also freezes the CPU somehow the debugger cannot halt it 
> anymore with a normal halt (COP halt), it has to use COP 
> freeze to take over control again.
> 
> I cannot say this problem does not occur accessing other 
> spaces, so it can also be a problem at the PPC60x bus.
> 
> Has anyone over excpirianced this before?
> 
> Thanks in advance!
> 
> Roel




Acc?dez au courrier ?lectronique de La Poste 
sur www.laposte.net ou sur 3615 LAPOSTENET (0,34? TTC /mn) 
1 Giga de stockage gratuit ? Antispam et antivirus int?gr?s 

[U-Boot-Users] Re : MV64460 Burst reads from BootCS

[Jerry Van Baren]

sebastien.douheret wrote:
> 	Hi,
> Marvell MV64460 only supports burst reads/writes on 32bits
> width peripherals ;-(
> I have the same issue with a Lauterbach probe. You must deny
> burst on BootCS if this peripheral is not 32bits width !
> This restriction also penalize boot process ! If your boot
> flash is 16 or 8 bits witdh, you can't directly execute
> instructions in flash. Because CPU like MPC74xx always load
> instructions with burst reads.
> 
> Regards.
> Sebastien.

I don't believe the above is strictly correct.  The 74xx does burst 
reads only if the cache is enabled for that memory area, so you should 
be able to execute out of flash but would be unable to enable caching on 
the boot memory area.  After all, I presume the board boots out of flash.

gvb

[U-Boot-Users] Re : MV64460 Burst reads from BootCS

[joe secund]

Hi Roel

I've seen similar behavior using a MV64560 and power
PC processor using a Windriver ICE.

Basically I could not read the MV internal registers
correctly when I did not plug in a bootrom (leaving
empty socket). Was reading zeroes anywhere in memory.

Somehow putting in even an empty Flash device allowed
me to initialize the processor correctly and
read/write MV registers and burn the flash via ICE.

- Wim

 > ------------------------------
 >
 > Message: 5
 > Date: Sun, 1 Oct 2006 20:19:12 +0200
 > From: "Roel Loeffen" <roel.loeffen@pr...>
 > Subject: [U-Boot-Users] MV64460 Burst reads from
BootCS
 > To: <u-boot-users@li...>
 > Message-ID:
 >
 <7826EB01AD927140A1DCF898FE63325D0D91B1@Ex...>
 > Content-Type: text/plain; charset="iso-8859-1"
 >
 > Hi all,
 >
 > We discover some serious problems accessing flash
at BootCS
 > of the Marvell MV64460. It looks like burst reads
make the
 > whole bridge hang. >From inside U-Boot as well as
using only
 > the BDI2000 debugger.
 >
 > Reading the flash with bursts enabled (which is
default),
 > makes it sometimes impossible to access any
 > registers/dram/devices anymore, only zeros can be
read.
 >
 > It also freezes the CPU somehow the debugger cannot
halt it
 > anymore with a normal halt (COP halt), it has to
use COP
 > freeze to take over control again.
 >
 > I cannot say this problem does not occur accessing
other
 > spaces, so it can also be a problem at the PPC60x
bus.
 >
 > Has anyone over excpirianced this before?
 >
 > Thanks in advance!
 >
 > Roel


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