A solution for a particular "Magic bitmask 0x1985 not found" error

A solution for a particular "Magic bitmask 0x1985 not found" error

[Chuck Meade]

Hello,

Regarding this error message sequence at boot time:

...
jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00040000: 0x2003 instead
jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00040004: 0x000c instead
jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00040008: 0xdc6d instead
jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00080000: 0x2003 instead
jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00080004: 0x000c instead
jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00080008: 0xdc6d instead
...

I had this happen for a long time each time I booted Linux.
I finally found a solution which stopped it, and maybe this
will help anyone else out there who is getting it.  I have
seen it mentioned in the archives, so I know I am not the 
first to get this error "triplet".

The solution was to simply make the jffs2 filesystem with a
newer version of mkfs.jffs2.  I had been using version 1.9
(see output of your "mkfs.jffs2 --version") when I got the
errors, and it stopped when I began using version 1.35.  So
it looks like this message is due to compatibility issues
between the jffs2 support in your kernel and the version of
mkfs.jffs2 that you use.

One thing of note is that the filesystem did work OK after
spewing all these msgs, they were just a nuisance -- I guess
that's why I lived with them for a while before seeking the
answer. :)

Chuck Meade

Re: A solution for a particular "Magic bitmask 0x1985 not found" error

[RobertS]

Chuck Meade wrote:

>Hello,
>
>Regarding this error message sequence at boot time:
>
>...
>jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00040000: 0x2003 instead
>jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00040004: 0x000c instead
>jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00040008: 0xdc6d instead
>jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00080000: 0x2003 instead
>jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00080004: 0x000c instead
>jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00080008: 0xdc6d instead
>...
>
>I had this happen for a long time each time I booted Linux.
>I finally found a solution which stopped it, and maybe this
>will help anyone else out there who is getting it.  I have
>seen it mentioned in the archives, so I know I am not the 
>first to get this error "triplet".
>
>The solution was to simply make the jffs2 filesystem with a
>newer version of mkfs.jffs2.  I had been using version 1.9
>(see output of your "mkfs.jffs2 --version") when I got the
>errors, and it stopped when I began using version 1.35.  So
>it looks like this message is due to compatibility issues
>between the jffs2 support in your kernel and the version of
>mkfs.jffs2 that you use.
>
>One thing of note is that the filesystem did work OK after
>spewing all these msgs, they were just a nuisance -- I guess
>that's why I lived with them for a while before seeking the
>answer. :)
>
>Chuck Meade
>
>
>______________________________________________________
>Linux MTD discussion mailing list
>http://lists.infradead.org/mailman/listinfo/linux-mtd/
>
>  
>
It seems that we are having similiar experiences. I have also 
experienced a similiar problem. The addresses were different but the 
values returned were the same.  When I updated to the latest snapshot of 
the mkfs.jffs2 utility and started with a fresh file system, the 
messages went away but only briefly.  After I had been exercising the 
file system, the messages returned again when powering up.

I investigated further and came to the conclusion that the file system 
was unable to "reclaim" the first node on an erase block. In my case 
these are the nodes at addresses on a multiple of 0x20000.  I verified 
this by completing filling my file system, then erasing the entire 
contents, then rebooting the system.  When I did this, the message: 
jffs2_scan_eraseblock(): Magic bitmask 0x1985 not .... at 0x.....20000: 
0x2003 instead ... occurred for every eraseblock boundary in the mounted 
partition.

This problem is just another non-critical item on my open issues list. I 
will be returning to it at a later date. I expect that your problem has 
just been masked by the creation of the new file system. It may be 
worthwhile to test for this problem in a manner similiar to mine.

Re: A solution for a particular "Magic bitmask 0x1985 not found" error

[Brian T.]

I have seen this also, but it was caused by uploading the jffs2 image to the
unit using xmodem, which annoyingly pads the last (128 - (SIZE % 128)) of
the file being uploaded with 0x1A.  I fixed this by padding the end of the
file with 0xFF prior to upload to make it look erased instead.

Funny, we actually just found this 1/2 hour ago.

Regards,

-Brian


> Chuck Meade wrote:
>
> >Hello,
> >
> >Regarding this error message sequence at boot time:
> >
> >...
> >jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00040000:
0x2003 instead
> >jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00040004:
0x000c instead
> >jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00040008:
0xdc6d instead
> >jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00080000:
0x2003 instead
> >jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00080004:
0x000c instead
> >jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00080008:
0xdc6d instead
> >...
> >
> >I had this happen for a long time each time I booted Linux.
> >I finally found a solution which stopped it, and maybe this
> >will help anyone else out there who is getting it.  I have
> >seen it mentioned in the archives, so I know I am not the
> >first to get this error "triplet".
> >
> >The solution was to simply make the jffs2 filesystem with a
> >newer version of mkfs.jffs2.  I had been using version 1.9
> >(see output of your "mkfs.jffs2 --version") when I got the
> >errors, and it stopped when I began using version 1.35.  So
> >it looks like this message is due to compatibility issues
> >between the jffs2 support in your kernel and the version of
> >mkfs.jffs2 that you use.
> >
> >One thing of note is that the filesystem did work OK after
> >spewing all these msgs, they were just a nuisance -- I guess
> >that's why I lived with them for a while before seeking the
> >answer. :)
> >
> >Chuck Meade
> >
> >
> >______________________________________________________
> >Linux MTD discussion mailing list
> >http://lists.infradead.org/mailman/listinfo/linux-mtd/
> >
> >
> >
> It seems that we are having similiar experiences. I have also
> experienced a similiar problem. The addresses were different but the
> values returned were the same.  When I updated to the latest snapshot of
> the mkfs.jffs2 utility and started with a fresh file system, the
> messages went away but only briefly.  After I had been exercising the
> file system, the messages returned again when powering up.
>
> I investigated further and came to the conclusion that the file system
> was unable to "reclaim" the first node on an erase block. In my case
> these are the nodes at addresses on a multiple of 0x20000.  I verified
> this by completing filling my file system, then erasing the entire
> contents, then rebooting the system.  When I did this, the message:
> jffs2_scan_eraseblock(): Magic bitmask 0x1985 not .... at 0x.....20000:
> 0x2003 instead ... occurred for every eraseblock boundary in the mounted
> partition.
>
> This problem is just another non-critical item on my open issues list. I
> will be returning to it at a later date. I expect that your problem has
> just been masked by the creation of the new file system. It may be
> worthwhile to test for this problem in a manner similiar to mine.
>

Re: A solution for a particular "Magic bitmask 0x1985 not found" error

[David Woodhouse]

On Fri, 23 May 2003, Brian T. wrote:

> I have seen this also, but it was caused by uploading the jffs2 image to the
> unit using xmodem, which annoyingly pads the last (128 - (SIZE % 128)) of
> the file being uploaded with 0x1A.  I fixed this by padding the end of the
> file with 0xFF prior to upload to make it look erased instead.

That's why I added the '-p' option to mkfs.jffs2 many moons ago. It's not 
quite the same as the others are seeing though -- if you look closer at 
their messages, the words on the flash that JFFS2 is complaining about 
aren't all 0x1A1A and aren't at the end of the image; they're at the 
beginning.

-- 
dwmw2

RE: A solution for a particular "Magic bitmask 0x1985 not found"error

[Chuck Meade]

> -----Original Message-----
> From: RobertS [mailto:RobertS@visi.com]
> Sent: Friday, May 23, 2003 9:03 AM
> To: Chuck Meade
> Cc: linux-mtd@lists.infradead.org
> Subject: Re: A solution for a particular "Magic bitmask 0x1985 not
> found"error
> 
> 
> Chuck Meade wrote:
> 
> >Hello,
> >
> >Regarding this error message sequence at boot time:
> >
> >...
> >jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00040000: 0x2003 instead
> >jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00040004: 0x000c instead
> >jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00040008: 0xdc6d instead
> >jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00080000: 0x2003 instead
> >jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00080004: 0x000c instead
> >jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x00080008: 0xdc6d instead
> >...
> >
> >I had this happen for a long time each time I booted Linux.
> >I finally found a solution which stopped it, and maybe this
> >will help anyone else out there who is getting it.  I have
> >seen it mentioned in the archives, so I know I am not the 
> >first to get this error "triplet".
> >
> >The solution was to simply make the jffs2 filesystem with a
> >newer version of mkfs.jffs2.  I had been using version 1.9
> >(see output of your "mkfs.jffs2 --version") when I got the
> >errors, and it stopped when I began using version 1.35.  So
> >it looks like this message is due to compatibility issues
> >between the jffs2 support in your kernel and the version of
> >mkfs.jffs2 that you use.
> >
> >One thing of note is that the filesystem did work OK after
> >spewing all these msgs, they were just a nuisance -- I guess
> >that's why I lived with them for a while before seeking the
> >answer. :)
> >
> >Chuck Meade
> >
> >
> >______________________________________________________
> >Linux MTD discussion mailing list
> >http://lists.infradead.org/mailman/listinfo/linux-mtd/
> >
> >  
> >
> It seems that we are having similiar experiences. I have also 
> experienced a similiar problem. The addresses were different but the 
> values returned were the same.  When I updated to the latest snapshot of 
> the mkfs.jffs2 utility and started with a fresh file system, the 
> messages went away but only briefly.  After I had been exercising the 
> file system, the messages returned again when powering up.
> 
> I investigated further and came to the conclusion that the file system 
> was unable to "reclaim" the first node on an erase block. In my case 
> these are the nodes at addresses on a multiple of 0x20000.  I verified 
> this by completing filling my file system, then erasing the entire 
> contents, then rebooting the system.  When I did this, the message: 
> jffs2_scan_eraseblock(): Magic bitmask 0x1985 not .... at 0x.....20000: 
> 0x2003 instead ... occurred for every eraseblock boundary in the mounted 
> partition.
> 
> This problem is just another non-critical item on my open issues list. I 
> will be returning to it at a later date. I expect that your problem has 
> just been masked by the creation of the new file system. It may be 
> worthwhile to test for this problem in a manner similiar to mine.

Yes I think that is exactly what happened.  After working with the
filesystem a bit, the messages are back. :(

Well I am back to the drawing board.  If anyone knows what causes the
0x2003, 0x000c, 0xdc6d triplet (see above for the exact error msg format)
it would be good to know.

Chuck