Kernel uncompression error

Kernel uncompression error

[woody]

Hi, List

I just built a 3.2.x ARM kernel and tried to have it boot by a boot 
loader (vivi) on my s3c2410a ARM board.  The kernel was built with 
default configuration s3c2410.

... But, the following error printed on the console:

---------------------
Uncompressing Linux...

uncompression error

-- System halted
---------------------

What kinds of reasons can most possibly lead the error?  I searched on 
the net and found some answers, but I cannot well judge them:

1. There is a D-Cache patch should be applied, but the message was 
talking about 2.4.x kernel, so I guess if there really was a bug around 
here, it should already been fixed. Right?

2. Probably the kernel image (here is a zImage) is too big, then the 
uncompressed code was overwriting the original compressed part of the 
image.  This sounds a little relevant.  My boot loader was loading the 
kernel image to ram address 0x300008000, and my zImage is about 1.9M. 
Do you think this is the reason?

Thanks in advance!



-- 
woody
I can't go back to yesterday - because I was a different person then.

Kernel uncompression error

[Russell King - ARM Linux]

On Wed, Nov 14, 2012 at 10:23:55PM +0800, woody wrote:
> I just built a 3.2.x ARM kernel and tried to have it boot by a boot  
> loader (vivi) on my s3c2410a ARM board.  The kernel was built with  
> default configuration s3c2410.

I reported that S3C2410 is totally dead with 3.x kernels about a month
ago and got no response what so ever from anyone.  I guess S3C2410 is
just not cared for anymore.

When I get sufficient motivation, I'll try and track down what's going
wrong, but it means taking the old mailing list server (and now list
archive) offline for a while.

I don't think it's cache related; different image sizes seem to behave
differently - and when they _do_ successfully decompress, they don't
boot at all.

Kernel uncompression error

[Woody Wu]

2012-11-15 AM1:34?"Russell King - ARM Linux" <linux@arm.linux.org.uk>
>
> On Wed, Nov 14, 2012 at 10:23:55PM +0800, woody wrote:
> > I just built a 3.2.x ARM kernel and tried to have it boot by a boot
> > loader (vivi) on my s3c2410a ARM board.  The kernel was built with
> > default configuration s3c2410.
>
> I reported that S3C2410 is totally dead with 3.x kernels about a month
> ago and got no response what so ever from anyone.  I guess S3C2410 is
> just not cared for anymore.
>
> When I get sufficient motivation, I'll try and track down what's going
> wrong, but it means taking the old mailing list server (and now list
> archive) offline for a while.
>
> I don't think it's cache related; different image sizes seem to behave
> differently - and when they _do_ successfully decompress, they don't
> boot at all.

So, in your opinion, what version of kernel I should go for s3c2410?

BTW: what people are using after days of s4c2410? I can remember it's quit
popular at then.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.infradead.org/pipermail/linux-arm-kernel/attachments/20121115/8944a2ea/attachment-0001.html>

Kernel uncompression error

[woody]

On 11/15/2012 1:34 AM, Russell King - ARM Linux wrote:
> On Wed, Nov 14, 2012 at 10:23:55PM +0800, woody wrote:
>> I just built a 3.2.x ARM kernel and tried to have it boot by a boot
>> loader (vivi) on my s3c2410a ARM board.  The kernel was built with
>> default configuration s3c2410.
>
> I reported that S3C2410 is totally dead with 3.x kernels about a month
> ago and got no response what so ever from anyone.  I guess S3C2410 is
> just not cared for anymore.
>
> When I get sufficient motivation, I'll try and track down what's going
> wrong, but it means taking the old mailing list server (and now list
> archive) offline for a while.
>
> I don't think it's cache related; different image sizes seem to behave
> differently - and when they _do_ successfully decompress, they don't
> boot at all.

Now I switched from 3.2.x kernel to the latest stable 3.6.6. And, I did 
a lot of configuration to make the zImage size smaller, now it is 1.5M. 
After download it to the board, I got another error:

"Uncompressing Linux... done, booting the kernel."

Then the kernel halted.

I noticed, the dots (...) in the kernel log seems very short, is this a 
clue or just normal?



-- 
woody
I can't go back to yesterday - because I was a different person then.

Kernel uncompression error

[Russell King - ARM Linux]

On Thu, Nov 15, 2012 at 01:56:47PM +0800, woody wrote:
> On 11/15/2012 1:34 AM, Russell King - ARM Linux wrote:
>> On Wed, Nov 14, 2012 at 10:23:55PM +0800, woody wrote:
>>> I just built a 3.2.x ARM kernel and tried to have it boot by a boot
>>> loader (vivi) on my s3c2410a ARM board.  The kernel was built with
>>> default configuration s3c2410.
>>
>> I reported that S3C2410 is totally dead with 3.x kernels about a month
>> ago and got no response what so ever from anyone.  I guess S3C2410 is
>> just not cared for anymore.
>>
>> When I get sufficient motivation, I'll try and track down what's going
>> wrong, but it means taking the old mailing list server (and now list
>> archive) offline for a while.
>>
>> I don't think it's cache related; different image sizes seem to behave
>> differently - and when they _do_ successfully decompress, they don't
>> boot at all.
>
> Now I switched from 3.2.x kernel to the latest stable 3.6.6. And, I did  
> a lot of configuration to make the zImage size smaller, now it is 1.5M.  
> After download it to the board, I got another error:
>
> "Uncompressing Linux... done, booting the kernel."
>
> Then the kernel halted.
>
> I noticed, the dots (...) in the kernel log seems very short, is this a  
> clue or just normal?

You've not told me anything I don't already know...  As I said, it needs
investigation as to what's going on.  I was rather hoping that the Samsung
people would look at it, but alas there seems to be zero interest from
them.

As for the shorter dots, that's normal now with the multi-format
decompressor.

Kernel uncompression error

[Russell King - ARM Linux]

On Thu, Nov 15, 2012 at 07:35:30AM +0800, Woody Wu wrote:
> ? 2012-11-15 AM1:34?"Russell King - ARM Linux" <linux@arm.linux.org.uk>???
> >
> > On Wed, Nov 14, 2012 at 10:23:55PM +0800, woody wrote:
> > > I just built a 3.2.x ARM kernel and tried to have it boot by a boot
> > > loader (vivi) on my s3c2410a ARM board.  The kernel was built with
> > > default configuration s3c2410.
> >
> > I reported that S3C2410 is totally dead with 3.x kernels about a month
> > ago and got no response what so ever from anyone.  I guess S3C2410 is
> > just not cared for anymore.
> >
> > When I get sufficient motivation, I'll try and track down what's going
> > wrong, but it means taking the old mailing list server (and now list
> > archive) offline for a while.
> >
> > I don't think it's cache related; different image sizes seem to behave
> > differently - and when they _do_ successfully decompress, they don't
> > boot at all.
> 
> So, in your opinion, what version of kernel I should go for s3c2410?

lists.arm.linux.org.uk currently runs 2.6.32.8.  That's the last working
kernel I currently know of.

As I said, I tried several 3.x kernels, all of them failed.

Kernel uncompression error

[Arnaud Patard (Rtp)]

Russell King - ARM Linux <linux@arm.linux.org.uk> writes:

> On Thu, Nov 15, 2012 at 01:56:47PM +0800, woody wrote:
>> On 11/15/2012 1:34 AM, Russell King - ARM Linux wrote:
>>> On Wed, Nov 14, 2012 at 10:23:55PM +0800, woody wrote:
>>>> I just built a 3.2.x ARM kernel and tried to have it boot by a boot
>>>> loader (vivi) on my s3c2410a ARM board.  The kernel was built with
>>>> default configuration s3c2410.
>>>
>>> I reported that S3C2410 is totally dead with 3.x kernels about a month
>>> ago and got no response what so ever from anyone.  I guess S3C2410 is
>>> just not cared for anymore.
>>>
>>> When I get sufficient motivation, I'll try and track down what's going
>>> wrong, but it means taking the old mailing list server (and now list
>>> archive) offline for a while.
>>>
>>> I don't think it's cache related; different image sizes seem to behave
>>> differently - and when they _do_ successfully decompress, they don't
>>> boot at all.
>>
>> Now I switched from 3.2.x kernel to the latest stable 3.6.6. And, I did  
>> a lot of configuration to make the zImage size smaller, now it is 1.5M.  
>> After download it to the board, I got another error:
>>
>> "Uncompressing Linux... done, booting the kernel."
>>
>> Then the kernel halted.
>>
>> I noticed, the dots (...) in the kernel log seems very short, is this a  
>> clue or just normal?
>
> You've not told me anything I don't already know...  As I said, it needs
> investigation as to what's going on.  I was rather hoping that the Samsung
> people would look at it, but alas there seems to be zero interest from
> them.

I guess that a lot of people don't care anymore about 2410 or can't
help. For instance, the only serial port available on my h1940 is the IR
one and I don't have a device to receive the ir data anymore so this
won't help. I won't even talk about gta01, which has no support in
mainline. It may also be the usual -ENOTIME error.


Arnaud

Kernel uncompression error

[Vasily Khoruzhick]

On Wed, Nov 14, 2012 at 8:34 PM, Russell King - ARM Linux
<linux@arm.linux.org.uk> wrote:
> On Wed, Nov 14, 2012 at 10:23:55PM +0800, woody wrote:
>> I just built a 3.2.x ARM kernel and tried to have it boot by a boot
>> loader (vivi) on my s3c2410a ARM board.  The kernel was built with
>> default configuration s3c2410.
>
> I reported that S3C2410 is totally dead with 3.x kernels about a month
> ago and got no response what so ever from anyone.  I guess S3C2410 is
> just not cared for anymore.

It was working at least for 3.1 (tested on h1940 machine),
can you give a link to your post? I still care about s3c2410 and s3c244x :)

> When I get sufficient motivation, I'll try and track down what's going
> wrong, but it means taking the old mailing list server (and now list
> archive) offline for a while.
>
> I don't think it's cache related; different image sizes seem to behave
> differently - and when they _do_ successfully decompress, they don't
> boot at all.

Regards
Vasily

Kernel uncompression error

[Russell King - ARM Linux]

On Thu, Nov 15, 2012 at 02:29:40PM +0300, Vasily Khoruzhick wrote:
> On Wed, Nov 14, 2012 at 8:34 PM, Russell King - ARM Linux
> <linux@arm.linux.org.uk> wrote:
> > On Wed, Nov 14, 2012 at 10:23:55PM +0800, woody wrote:
> >> I just built a 3.2.x ARM kernel and tried to have it boot by a boot
> >> loader (vivi) on my s3c2410a ARM board.  The kernel was built with
> >> default configuration s3c2410.
> >
> > I reported that S3C2410 is totally dead with 3.x kernels about a month
> > ago and got no response what so ever from anyone.  I guess S3C2410 is
> > just not cared for anymore.
> 
> It was working at least for 3.1 (tested on h1940 machine),
> can you give a link to your post? I still care about s3c2410 and s3c244x :)

Subject: s3c2410 bast (== lists.arm.linux.org.uk) - appears to be broken
        with 3.1.8...

Kernel uncompression error

[Vasily Khoruzhick]

On Thu, Nov 15, 2012 at 2:22 PM, Arnaud Patard
<arnaud.patard@rtp-net.org> wrote:
>
> I guess that a lot of people don't care anymore about 2410 or can't
> help. For instance, the only serial port available on my h1940 is the IR
> one and I don't have a device to receive the ir data anymore so this
> won't help. I won't even talk about gta01, which has no support in
> mainline. It may also be the usual -ENOTIME error.

Well, I've s3c2440 and s3c2442 (they're close enough to s3c2410)
machines with serial port, I can take a look.

Regards
Vasily

Kernel uncompression error

[Woody Wu]

2012-11-15 PM5:34?"Russell King - ARM Linux" <linux@arm.linux.org.uk>
>
> On Thu, Nov 15, 2012 at 07:35:30AM +0800, Woody Wu wrote:
> > ? 2012-11-15 AM1:34?"Russell King - ARM Linux" <linux@arm.linux.org.uk>
wrote:
> > >
> > > On Wed, Nov 14, 2012 at 10:23:55PM +0800, woody wrote:
> > > > I just built a 3.2.x ARM kernel and tried to have it boot by a boot
> > > > loader (vivi) on my s3c2410a ARM board.  The kernel was built with
> > > > default configuration s3c2410.
> > >
> > > I reported that S3C2410 is totally dead with 3.x kernels about a month
> > > ago and got no response what so ever from anyone.  I guess S3C2410 is
> > > just not cared for anymore.
> > >
> > > When I get sufficient motivation, I'll try and track down what's going
> > > wrong, but it means taking the old mailing list server (and now list
> > > archive) offline for a while.
> > >
> > > I don't think it's cache related; different image sizes seem to behave
> > > differently - and when they _do_ successfully decompress, they don't
> > > boot at all.
> >
> > So, in your opinion, what version of kernel I should go for s3c2410?
>
> lists.arm.linux.org.uk currently runs 2.6.32.8.  That's the last working
> kernel I currently know of.

For the 2.6 kernel, I just found a working one, that 2.6.36. From 2.6.7 on,
even the compiling cant passed, an unresolved function call to a LED
backlight related function. But 2.6.36 work, passed compiling,
uncompressing, load and run.

> As I said, I tried several 3.x kernels, all of them failed.

By far, I probably met the same kind of error with you for the 3.x kernel.
Uncompressing done, then halt. I really hope someone told me there is a
working 3.2+ kernel since newer kernels usually have better driver surpport.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.infradead.org/pipermail/linux-arm-kernel/attachments/20121116/2b45d9de/attachment.html>

[PATCH] S3C24XX: Fix console baud rate guessing

[Domenico Andreoli]

From: Domenico Andreoli <domenico.andreoli@linux.com>

On Wed, Nov 14, 2012 at 05:34:53PM +0000, Russell King - ARM Linux wrote:
> On Wed, Nov 14, 2012 at 10:23:55PM +0800, woody wrote:
> > I just built a 3.2.x ARM kernel and tried to have it boot by a boot  
> > loader (vivi) on my s3c2410a ARM board.  The kernel was built with  
> > default configuration s3c2410.
> 
> I reported that S3C2410 is totally dead with 3.x kernels about a month
> ago and got no response what so ever from anyone.  I guess S3C2410 is
> just not cared for anymore.
> 
> When I get sufficient motivation, I'll try and track down what's going
> wrong, but it means taking the old mailing list server (and now list
> archive) offline for a while.

I found something also on my S3C2440. The board doesn't die but prints
garbage on the console. Indeed the baud rate guessing is broken (it's ok
if specified manually on the cmdline),
	
commit c3310fbbeb9db6967900ed22eb3d0bd0bb0e892c
Author: Thomas Abraham <thomas.abraham@linaro.org>
Date:   Mon Oct 24 11:47:57 2011 +0200

    serial: samsung: remove all uses of get_clksrc and set_clksrc
    
    With clkdev based clock lookup support, the clock set and get operation
    using clock names communicated between the samsung uart driver and the
    SoC specific extension can be removed.
    
    In addition to that, for each platform specific extension, add the
    default clock selection, number of clock options for uart baud generator,
    clock selection bit mask and shift values which is required by the
    clkdev support in samsung uart driver.
    
    The default clock selection value 'def_clk_sel' specifies the default clock
    to be used as the source clock for baud rate generator in case the platform
    code does not specify the same.
    
    Cc: Ben Dooks <ben-linux@fluff.org>
    Signed-off-by: Thomas Abraham <thomas.abraham@linaro.org>
    Signed-off-by: Kukjin Kim <kgene.kim@samsung.com>


To restore previous functionality I had to change the default UCON clock
back to UCLK.

Signed-off-by: Domenico Andreoli <domenico.andreoli@linux.com>
--
diff --git a/drivers/tty/serial/samsung.c b/drivers/tty/serial/samsung.c
index 7f04717..316b703 100644
--- a/drivers/tty/serial/samsung.c
+++ b/drivers/tty/serial/samsung.c
@@ -1572,7 +1572,7 @@ static struct s3c24xx_serial_drv_data s3c2440_serial_drv_data = {
                .tx_fifofull    = S3C2440_UFSTAT_TXFULL,
                .tx_fifomask    = S3C2440_UFSTAT_TXMASK,
                .tx_fifoshift   = S3C2440_UFSTAT_TXSHIFT,
-               .def_clk_sel    = S3C2410_UCON_CLKSEL2,
+               .def_clk_sel    = S3C2410_UCON_CLKSEL1,
                .num_clks       = 4,
                .clksel_mask    = S3C2412_UCON_CLKMASK,
                .clksel_shift   = S3C2412_UCON_CLKSHIFT,





The additional condition is that board file does not specify a 

the clocks being properly configured. 

 Something that doesn't happen if
the baud rate is specified onthe command line.

detection and goes unnoticed if you manually specify it on the cmdline
(like I do).

On my S3C2440 

I've a few observation

[PATCH] S3C24XX: Fix console baud rate guessing

[Domenico Andreoli]

From: Domenico Andreoli <domenico.andreoli@linux.com>

On Wed, Nov 14, 2012 at 05:34:53PM +0000, Russell King - ARM Linux wrote:
> On Wed, Nov 14, 2012 at 10:23:55PM +0800, woody wrote:
> > I just built a 3.2.x ARM kernel and tried to have it boot by a boot  
> > loader (vivi) on my s3c2410a ARM board.  The kernel was built with  
> > default configuration s3c2410.
> 
> I reported that S3C2410 is totally dead with 3.x kernels about a month
> ago and got no response what so ever from anyone.  I guess S3C2410 is
> just not cared for anymore.
> 
> When I get sufficient motivation, I'll try and track down what's going
> wrong, but it means taking the old mailing list server (and now list
> archive) offline for a while.

I found something also on my S3C2440. The board doesn't die but prints
garbage on the console. Indeed the baud rate guessing is broken (it's ok
if specified manually on the cmdline),
	
commit c3310fbbeb9db6967900ed22eb3d0bd0bb0e892c
Author: Thomas Abraham <thomas.abraham@linaro.org>
Date:   Mon Oct 24 11:47:57 2011 +0200

    serial: samsung: remove all uses of get_clksrc and set_clksrc
    
    With clkdev based clock lookup support, the clock set and get operation
    using clock names communicated between the samsung uart driver and the
    SoC specific extension can be removed.
    
    In addition to that, for each platform specific extension, add the
    default clock selection, number of clock options for uart baud generator,
    clock selection bit mask and shift values which is required by the
    clkdev support in samsung uart driver.
    
    The default clock selection value 'def_clk_sel' specifies the default clock
    to be used as the source clock for baud rate generator in case the platform
    code does not specify the same.
    
    Cc: Ben Dooks <ben-linux@fluff.org>
    Signed-off-by: Thomas Abraham <thomas.abraham@linaro.org>
    Signed-off-by: Kukjin Kim <kgene.kim@samsung.com>


To restore previous functionality I had to change the default UCON clock
back to UCLK.

Signed-off-by: Domenico Andreoli <domenico.andreoli@linux.com>
--
diff --git a/drivers/tty/serial/samsung.c b/drivers/tty/serial/samsung.c
index 7f04717..316b703 100644
--- a/drivers/tty/serial/samsung.c
+++ b/drivers/tty/serial/samsung.c
@@ -1572,7 +1572,7 @@ static struct s3c24xx_serial_drv_data s3c2440_serial_drv_data = {
                .tx_fifofull    = S3C2440_UFSTAT_TXFULL,
                .tx_fifomask    = S3C2440_UFSTAT_TXMASK,
                .tx_fifoshift   = S3C2440_UFSTAT_TXSHIFT,
-               .def_clk_sel    = S3C2410_UCON_CLKSEL2,
+               .def_clk_sel    = S3C2410_UCON_CLKSEL1,
                .num_clks       = 4,
                .clksel_mask    = S3C2412_UCON_CLKMASK,
                .clksel_shift   = S3C2412_UCON_CLKSHIFT,





The additional condition is that board file does not specify a 

the clocks being properly configured. 

 Something that doesn't happen if
the baud rate is specified onthe command line.

detection and goes unnoticed if you manually specify it on the cmdline
(like I do).

On my S3C2440 

I've a few observation

Kernel uncompression error

[Woody Wu]

On 2012-11-15, Russell King - ARM Linux <linux@arm.linux.org.uk> wrote:
> On Thu, Nov 15, 2012 at 07:35:30AM +0800, Woody Wu wrote:
>> > On Wed, Nov 14, 2012 at 10:23:55PM +0800, woody wrote:
>> > > I just built a 3.2.x ARM kernel and tried to have it boot by a boot
>> > > loader (vivi) on my s3c2410a ARM board.  The kernel was built with
>> > > default configuration s3c2410.
>> >
>> > I reported that S3C2410 is totally dead with 3.x kernels about a month
>> > ago and got no response what so ever from anyone.  I guess S3C2410 is
>> > just not cared for anymore.
>> >
>> > When I get sufficient motivation, I'll try and track down what's going
>> > wrong, but it means taking the old mailing list server (and now list
>> > archive) offline for a while.
>> >
>> > I don't think it's cache related; different image sizes seem to behave
>> > differently - and when they _do_ successfully decompress, they don't
>> > boot at all.
>> 
>> So, in your opinion, what version of kernel I should go for s3c2410?
>
> lists.arm.linux.org.uk currently runs 2.6.32.8.  That's the last working
> kernel I currently know of.
>
> As I said, I tried several 3.x kernels, all of them failed.
>

I just found 3.4.19 works.

-- 
woody
I can't go back to yesterday - because I was a different person then.