RE: [Ltib] RE: Reg ltib

[Makhija, Neha (IE10)]

Hi Stuart, Hi Vijay and Hi Nikam

I am using MCF5474 processor and with the help of ltib I have generated
the kernel image of 2.6.10 version. The same image is being used to
start my application. Also the board has NAND Flash of Samsung make. The
board takes 100 seconds to boot till the application.
The NAND Flash is changed from the Samsung make to ST Micro of same foot
print which made the boot up time from 100 to 180 seconds. No change has
been done at the kernel. While debugging this issue, some key
observations were:
- When the NAND Flash is being accessed with a stand alone code (NO OS
basically) both Samsung and ST Micro are taking identical time to read
the NAND Flash.
- The READY/BUSY signal was probed with the oscilloscope. With Samsung
very frequently the READY/BUSY pin was being toggled in comparison to
the ST Micro while OS was booting.

These 2 observations helped in the conclusion that the problem could be
with the way OS is trying to access the NAND Flash.
I have few questions like
- Is the NAND Flash being accessed like a poll device?
- Which parameter of the NAND Flash the kernel polls/checks, frequently
before issuing a read command? 

Thanks and Regards,
Neha Makhija.

-----Original Message-----
From: Vijay Nikam [mailto:address@hidden 
Sent: Tuesday, July 08, 2008 12:48 PM
To: Makhija, Neha (IE10)
Cc: Stuart Hughes; address@hidden; address@hidden
Subject: Re: [Ltib] RE: Reg ltib

Hi,

You should have the gdb package ... in the ltib installed directory
/dist/lfs-5.1 ... once you checked if it is there ... then execute
./ltib -p gdb

I tried it here and it is successful ... the log file is attached ...
I have MPC8313 board ...

Kind Regards,
Vijay Nikam

On 7/8/08, Vijay Nikam <address@hidden> wrote:
> Hi,
>
> The file gdb-6.0.tar.gz.md5 is available at gpp
> (http://www.bitshrine.org/gpp/) check it ...
>
> I think the file is not getting download from gpp because of your
> netwrok connection ... which is going through proxy server ... I mean
> even if you give correct proxy address, still it will not be able to
> download ... and thats why you need to have direct internet connection
> ... then only wget or yum works ...
>
> Anyways the second option you have is to modify the .ltibrc, disable
> gpp and ppp ... only lpp should be enabled ... then download the
> gdb-6.0.tar.gz.md5 and gdb-6.0.tar.gz from
> http://www.bitshrine.org/gpp and save it at the loaction where the
> file should be ... and then execute ./ltib -p <package name> in your
> case it is gdb ...
>
> I think this will do for your problem ...
>
> Kind Regards,
> Vijay Nikam
>
> On 7/8/08, Makhija, Neha (IE10) <address@hidden> wrote:
>> Hi Stuart/Vijay
>>
>> Thanks for the reply!!
>> I tried whatever was mentioned by you. Looks like it is not the
problem
>> with the proxy address mentioned in the .ltibrc file as after running
>> the command "./ltib -p gdb", I got the following error
>>
>> Processing: gdb
>> =================
>> Getting gdb-6.0.tar.gz.md5 from the Private Package Pool
>> /home/bna3/Neha/ltib
>>
>>>> As it is clearly visible that it was able to >>extract
>> gdb-6.0.tar.gz.md5 file from Global >>Package Pool
>>
>> Getting gdb-6.0.tar.gz.md5 from the Global Package Pool
>> /home/bna3/Neha/ltib
>> Getting gdb-6.0.tar.gz from the Private Package Pool
>> /home/bna3/Neha/ltib
>>
>>>> Here it is giving error.
>>
>> Getting gdb-6.0.tar.gz from the Global Package Pool
>> /home/bna3/Neha/ltib
>> Can't get: gdb-6.0.tar.gz at ./ltib line 515.
>> Died at ./ltib line 518.
>>
>> I think the file doesn't exist in the Global Package Pool.
>>
>> Please correct me if I am wrong!!
>>
>> Regards,
>> Neha
>>
>>
>> -----Original Message-----
>> From: Stuart Hughes [mailto:address@hidden
>> Sent: Monday, July 07, 2008 5:49 PM
>> To: Makhija, Neha (IE10)
>> Cc: address@hidden
>> Subject: RE: Reg ltib
>>
>> Hi Neha,
>>
>> On Mon, 2008-07-07 at 16:56 +0530, Makhija, Neha (IE10) wrote:
>>> Hi Stuart
>>>
>>> Thanks for the help!!
>>> I am the member of the ltib group. I have registered through Yahoo
id.
>> During office hours can't surf yahoo, so can't mail you from that id.
>>>
>>
>> Okay, I'll let those messages go through from that email to save you
>> getting everything twice.
>>
>>
>>> I tried whatever you suggested. I am getting the following error:
>>>
>>> Can't get: gdb-6.0.tar.gz at ./ltib line 515.
>>>
>>
>> You need to have Internet access, ltib will then be able to download
>> this.  If you are inside a corporate network, then you'll need to
enter
>> the proxy setting into the .ltibrc file.
>>
>>>
>>> Also what is " uclinux-dev" ??
>>>
>>
>> This is a mailing list for MMUless chips:
>> http://mailman.uclinux.org/mailman/listinfo/uclinux-dev
>>
>>
>>> Thanks !!
>>>
>>> Best Regards,
>>> Neha
>>
>>
>> Regards, Stuart
>>
>>
>>> -----Original Message-----
>>> From: Stuart Hughes [mailto:address@hidden
>>> Sent: Monday, July 07, 2008 4:45 PM
>>> To: Makhija, Neha (IE10)
>>> Cc: address@hidden
>>> Subject: RE: Reg ltib
>>>
>>> Hi Neha,
>>>
>>> If the problem is in the kernel then gdb won't help without kgdb on
>> the
>>> target (as you noted).
>>>
>>> If the gdb executable is dying with SEGV though you can re-build it.
>>>
>>> $ ./ltib -m config
>>>
>>> and select gdb in the package list (cross/gdbserver).  Then run:
>>>
>>> $ ./ltib -p gdb
>>>
>>> This will re-generate the cross gdb executable and place it under
the
>>> bin directory.  It that doesn't work, you'll need to contact
Freescale
>>> support for help.
>>>
>>> To actually debug the kernel, you'll need help from a kernel expert.
>> It
>>> it's an MMUless platform, try uclinux-dev.  Otherwise it may be
worth
>>> contacting Freescale support.
>>>
>>> Regards, Stuart
>>>
>>> On Mon, 2008-07-07 at 15:59 +0530, Makhija, Neha (IE10) wrote:
>>> > Hi Stuart
>>> >
>>> > I need your help!!
>>> > I have one embedded board with coldfire processor.
>>> >  Because of some reason in the application, the drivers (Written
by
>> our
>>> >  team) are getting closed (whatever are opened in the
application).
>> To
>>> >  debug the reason for kill of the application by the kernel, I
used
>> the
>>> >  gdbserver and gdbclient generated by the ltib. It showed
>> "SEGMENTATION
>>> >  KILL" and I am not able to debug further. I tried the gdb binary
>> image
>>> >  available in the ltib folder but it didn't work. Can you suggest
me
>> a
>>> >  way to debug the application through ltib ?? Secondly, To debug
the
>>> >  Kernel drivers, one can use kdb and kgdb. But I am not able to
find
>>> >  the two in the ltib folder. Is there any other way to debug the
>>> >  drivers through ltib?
>>> >
>>> > Thanks in advance!!
>>> > Neha
>>> >
>>> > -----Original Message-----
>>> > From: Stuart Hughes [mailto:address@hidden
>>> > Sent: Wednesday, November 28, 2007 3:56 PM
>>> > To: Makhija, Neha (IE10)
>>> > Cc: address@hidden
>>> > Subject: Re: Reg ltib
>>> >
>>> > Hi Neha,
>>> >
>>> > I'm not sure about the OOB data question, maybe others on this
list
>> or
>>> > even on the MTD mailing list know.
>>> >
>>> > As far as your other options go, there are techniques for
combining
>> the
>>> > kernel and rootfs into a single image (it's a general technique
that
>> you
>>> > can find on various mailing list).  However in my opinion for most
>> cases
>>> > you get more flexibility by having the kernel and rootfs in
separate
>>> > Flash partitions.
>>> >
>>> > Regards, Stuart
>>> >
>>> > On Wed, 2007-11-28 at 09:25 +0530, Makhija, Neha (IE10) wrote:
>>> > > Hi Stuart
>>> > >
>>> > >
>>> > >
>>> > > I am the member of the ltib mailing list. I have one doubt. I am
>>> > > having one kernel image which will take the file system through
>> NFS
>>> > > and will download it on the board's NAND flash. So I am having
two
>>> > > images one is kernel image and second one is root file system
>> image
>>> > > which is .jffs2. Now the requirement is that only one single
image
>>> > > should be there to program the empty flash, ie I need to
>> concatenate
>>> > > the two images. We need this because the final burning of the
NAND
>>> > > flash will be done by the 3rd party vendor, so they want only
>> image
>>> > > for the first time. To burn NAND flash, peedi programmer will be
>> used.
>>> > > Peedi programmer takes care of the bad blocks in the NAND flash.
>> What
>>> > > I want to know is, If I will concatenate the two images with the
>> help
>>> > > of batch command will it work? I mean will it update the OOB
table
>> of
>>> > > the bad block as done by linux kernel while loading root file
>> system
>>> > > image (JFFS2 image)?
>>> > >
>>> > >
>>> > >
>>> > > Basically I am not able to check where exactly the OOB table is
>>> > > getting updated for the root file system? Or Is there a way to
>>> > > generate the kernel and Root file system image altogether?
>>> > >
>>> > >
>>> > >
>>> > > Thanks in advance.
>>> > >
>>> > >
>>> > >
>>> > > Regards,
>>> > >
>>> > > Neha
>>> > >
>>> > >
>>> > >
>>> > >
>>>
>>
>>
>