[ECOS] JFFS2 no space left on device error

[ECOS] JFFS2 no space left on device error

[Edgar Grimberg]

Hi,

I have loaded the jffs2 image provided in the eCos tree onto my board 
(EB40A-like) using redboot and the serial port. I made an FIS entry 
named JFFS2 with this image. (no errors so far)
After that, using the BDI2000 debugger, I have programmed the jffs2_1 
test onto the board.
It seems that the mounting of the jffs2 image is working OK (at least no 
errors on the serial), I get some directory listing (perhaps not OK, 
considering the file structure from the support/jffs2 folder):

<INFO>: reading directory /
<INFO>: entry              . [mode 016f0001 ino 00000001 nlink 1 size 0]
<INFO>: entry             .. [mode 016f0001 ino 00000001 nlink 1 size 0]

Creating a file gives me:

<INFO>: create file /foo size 202
<FAIL>: open() returned -1 No space left on device
<FAIL>: write() returned -1 Bad file handle

and the bad file handle message continues.

Any hints?

Regards,
Edgar


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] JFFS2 no space left on device error

[Gary Thomas]

On Sun, 2005-07-31 at 19:37 +0300, Edgar Grimberg wrote:
> Hi,
> 
> I have loaded the jffs2 image provided in the eCos tree onto my board 
> (EB40A-like) using redboot and the serial port. I made an FIS entry 
> named JFFS2 with this image. (no errors so far)
> After that, using the BDI2000 debugger, I have programmed the jffs2_1 
> test onto the board.
> It seems that the mounting of the jffs2 image is working OK (at least no 
> errors on the serial), I get some directory listing (perhaps not OK, 
> considering the file structure from the support/jffs2 folder):
> 
> <INFO>: reading directory /
> <INFO>: entry              . [mode 016f0001 ino 00000001 nlink 1 size 0]
> <INFO>: entry             .. [mode 016f0001 ino 00000001 nlink 1 size 0]
> 
> Creating a file gives me:
> 
> <INFO>: create file /foo size 202
> <FAIL>: open() returned -1 No space left on device
> <FAIL>: write() returned -1 Bad file handle
> 
> and the bad file handle message continues.
> 
> Any hints?

When you created the image (in RedBoot), did you make it larger
than the actual JFFS2 image that you created?  In other words,
JFFS2 needs additional FLASH space within the RedBoot (FIS)
image in order to create new files (or even rewrite/update
existing ones).

-- 
------------------------------------------------------------
Gary Thomas                 |  Consulting for the
MLB Associates              |    Embedded world
------------------------------------------------------------


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] JFFS2 no space left on device error

[Edgar Grimberg]

>When you created the image (in RedBoot), did you make it larger
>than the actual JFFS2 image that you created?  In other words,
>JFFS2 needs additional FLASH space within the RedBoot (FIS)
>image in order to create new files (or even rewrite/update
>existing ones).
>
>  
>

Hello Gary,

Thanks for the response. I am describing the exact steps that I 
followed, maybe this will clear up the problems that I have:

1. the ecos.ecc file from my build tree contains:

cdl_component CYGNUM_IO_FLASH_BLOCK_CFG_STATIC_1 {
    # Flavor: bool
    user_value 0
    # value_source user
    # Default value: 1
};
cdl_component CYGNUM_IO_FLASH_BLOCK_CFG_FIS_1 {
    # Flavor: bool
    user_value 1
    # value_source user
    # Default value: 0
};
# >
# Name of FIS entry
#
cdl_component CYGDAT_IO_FLASH_BLOCK_FIS_NAME_1 {
    # Flavor: data
    user_value "\"jffs2\""
    # value_source user
    # Default value: "\"jffs2\""
};

2. Create the jffs2 image file using:

 mkfs.jffs2 --pad=0x20000 -e 0x10000 -r ./etc -o ./jffs2.img

3. Upload the image file to flash, using RedBoot and FIS:

RedBoot> load -m ymodem -b %{FREEMEMLO} -r
CRaw file loaded 0x20007000-0x20026fff, assumed entry at 0x20007000
xyzModem - CRC mode, 1026(SOH)/0(STX)/0(CAN) packets, 5 retries
RedBoot> fis create -b 0x20007000 -l 0x40000 -s 0x30000 -f 0x10b0000 JFFS2
... Erase from 0x010b0000-0x010f0000: ....
... Program from 0x20007000-0x20037000 at 0x010b0000: ...
... Erase from 0x013f0000-0x01400000: .
... Program from 0x20070000-0x20080000 at 0x013f0000: .
RedBoot> fis list
Name              FLASH addr  Mem addr    Length      Entry point
RedBoot           0x01000000  0x01000000  0x00020000  0x00000000
JFFS2             0x010B0000  0x010B0000  0x00040000  0x20007000
FIS directory     0x013F0000  0x013F0000  0x0000F000  0x00000000
RedBoot config    0x013FF000  0x013FF000  0x00001000  0x00000000

4. Program the first test for JFFS2 from ecos, using the bdi2000 
debugger, taking care not to erase the FIS directory or the jffs2 images 
from flash. The test is programmed at 0x1000000, just over redboot 
(maybe this is a problem?)

5. Run the test (jffs2_1) and get the following:

<INFO>: reading directory /
<INFO>: entry              . [mode 016f0001 ino 00000001 nlink 1 size 0]
<INFO>: entry             .. [mode 016f0001 ino 00000001 nlink 1 size 0]
<INFO>: create file /foo size 202
<FAIL>: open() returned -1 No space left on device
<FAIL>: write() returned -1 Bad file handle
<FAIL>: write() returned -1 Bad file handle

Am I doing something wrong?

Regards,
Edgar


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] JFFS2 no space left on device error

[Andrew Lunn]

On Mon, Aug 01, 2005 at 03:17:48PM +0300, Edgar Grimberg wrote:
> 
> >When you created the image (in RedBoot), did you make it larger
> >than the actual JFFS2 image that you created?  In other words,
> >JFFS2 needs additional FLASH space within the RedBoot (FIS)
> >image in order to create new files (or even rewrite/update
> >existing ones).
> >
> > 
> >
> 
> Hello Gary,
> 
> Thanks for the response. I am describing the exact steps that I 
> followed, maybe this will clear up the problems that I have:
> 
> 1. the ecos.ecc file from my build tree contains:
> 
> cdl_component CYGNUM_IO_FLASH_BLOCK_CFG_STATIC_1 {
>    # Flavor: bool
>    user_value 0
>    # value_source user
>    # Default value: 1
> };
> cdl_component CYGNUM_IO_FLASH_BLOCK_CFG_FIS_1 {
>    # Flavor: bool
>    user_value 1
>    # value_source user
>    # Default value: 0
> };
> # >
> # Name of FIS entry
> #
> cdl_component CYGDAT_IO_FLASH_BLOCK_FIS_NAME_1 {
>    # Flavor: data
>    user_value "\"jffs2\""
>    # value_source user
>    # Default value: "\"jffs2\""
> };

You actually called it JFFS2 in FIS, not jffs2. I'm not sure if this
matters.....
 
> 2. Create the jffs2 image file using:
> 
> mkfs.jffs2 --pad=0x20000 -e 0x10000 -r ./etc -o ./jffs2.img
> 
> 3. Upload the image file to flash, using RedBoot and FIS:
> 
> RedBoot> load -m ymodem -b %{FREEMEMLO} -r
> CRaw file loaded 0x20007000-0x20026fff, assumed entry at 0x20007000
> xyzModem - CRC mode, 1026(SOH)/0(STX)/0(CAN) packets, 5 retries
> RedBoot> fis create -b 0x20007000 -l 0x40000 -s 0x30000 -f 0x10b0000 JFFS2
> ... Erase from 0x010b0000-0x010f0000: ....
> ... Program from 0x20007000-0x20037000 at 0x010b0000: ...
> ... Erase from 0x013f0000-0x01400000: .
> ... Program from 0x20070000-0x20080000 at 0x013f0000: .
> RedBoot> fis list
> Name              FLASH addr  Mem addr    Length      Entry point
> RedBoot           0x01000000  0x01000000  0x00020000  0x00000000
> JFFS2             0x010B0000  0x010B0000  0x00040000  0x20007000
> FIS directory     0x013F0000  0x013F0000  0x0000F000  0x00000000
> RedBoot config    0x013FF000  0x013FF000  0x00001000  0x00000000
> 
> 4. Program the first test for JFFS2 from ecos, using the bdi2000 
> debugger, taking care not to erase the FIS directory or the jffs2 images 
> from flash. The test is programmed at 0x1000000, just over redboot 
> (maybe this is a problem?)

Err, actually yes it could be. To read the FIS directory is calls into
Redboot. If you have stomped over redboot im supprised the test even
runs. Looking at your FIS directory you have 0x01020000 - 0x10AFFFF
free. Try linking jffs2_1 to run at that address and put Redboot back
at 0x01000000.
 
> 5. Run the test (jffs2_1) and get the following:
> 
> <INFO>: reading directory /
> <INFO>: entry              . [mode 016f0001 ino 00000001 nlink 1 size 0]
> <INFO>: entry             .. [mode 016f0001 ino 00000001 nlink 1 size 0]

This does not seem right. There should be other entries. Try 

jffs2dump -content ./jffs2.img

and see if etc, passwd etc is in the image.

        Andrew

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] JFFS2 no space left on device error

[Edgar Grimberg]

Hello,

Thanks for the hints, Andrew.
One more thing I didn't specified, I have taken the mkfs.jffs2 from the 
CVS.

$ mkfs.jffs2 --version
mkfs.jffs2: revision 1.43

I did as you told me, as follows...

>You actually called it JFFS2 in FIS, not jffs2. I'm not sure if this
>matters.....
>  
>
RedBoot> fis list
Name              FLASH addr  Mem addr    Length      Entry point
RedBoot           0x01000000  0x01000000  0x00020000  0x00000000
appl              0x01020000  0x01020000  0x00030000  0x20007000
jffs2             0x010B0000  0x010B0000  0x00040000  0x20007000
FIS directory     0x013F0000  0x013F0000  0x0000F000  0x00000000
RedBoot config    0x013FF000  0x013FF000  0x00001000  0x00000000

I have changed the name of the jffs FIS image, but if the application 
can mount the JFFS2 file system, I am not sure it matters. Anyway, I am 
consistent with the ecos.ecc file.

>
>
>Err, actually yes it could be. To read the FIS directory is calls into
>Redboot. If you have stomped over redboot im supprised the test even
>runs. Looking at your FIS directory you have 0x01020000 - 0x10AFFFF
>free. Try linking jffs2_1 to run at that address and put Redboot back
>at 0x01000000.
>  
>
Did that, in the fis list command results above, you can see the appl 
image that contains the binary image of jffs2_1 test (made binary with  
arm-elf-objcopy ./jffs2_1 ./jffs2_1.bin -O binary and uploaded with 
RedBoot). I have stared the test with go 0x1020000 commmand in redboot, 
unfortunatly the same result.

>This does not seem right. There should be other entries. Try 
>
>jffs2dump -content ./jffs2.img
>
>and see if etc, passwd etc is in the image.
>
>  
>
It is not right, indeed...

$ ./jffs2dump --content /tftpboot/jffs2.img
         Dirent     node at 0x0000000c, totlen 0x0000002b, #pino     1, 
version     0, #ino         2, nsize        3, name CVS
         Inode      node at 0x00000038, totlen 0x00000044, #ino      2, 
version     1, isize        0, csize        0, dsize        0, 
offset        0
         Dirent     node at 0x0000007c, totlen 0x0000002d, #pino     1, 
version     1, #ino         3, nsize        5, name group
         Inode      node at 0x000000ac, totlen 0x0000013b, #ino      3, 
version     1, isize      445, csize      247, dsize      445, 
offset        0
         Dirent     node at 0x000001e8, totlen 0x0000002e, #pino     1, 
version     2, #ino         4, nsize        6, name passwd
         Inode      node at 0x00000218, totlen 0x0000019d, #ino      4, 
version     1, isize      658, csize      345, dsize      658, 
offset        0
         Dirent     node at 0x000003b8, totlen 0x0000002f, #pino     2, 
version     3, #ino         5, nsize        7, name Entries
         Inode      node at 0x000003e8, totlen 0x00000086, #ino      5, 
version     1, isize       85, csize       66, dsize       85, 
offset        0
         Dirent     node at 0x00000470, totlen 0x00000032, #pino     2, 
version     4, #ino         6, nsize       10, name Repository
         Inode      node at 0x000004a4, totlen 0x00000075, #ino      6, 
version     1, isize       49, csize       49, dsize       49, 
offset        0
         Dirent     node at 0x0000051c, totlen 0x0000002c, #pino     2, 
version     5, #ino         7, nsize        4, name Root
         Inode      node at 0x00000548, totlen 0x00000073, #ino      7, 
version     1, isize       47, csize       47, dsize       47, 
offset        0

Thanks for your patience,
Edgar

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] JFFS2 no space left on device error

[Edgar Grimberg]

Hello,

I was wondering if I am affected by the bug in the arm-elf-gcc version 
3.2.x, with a different symptomatic than the one in 
http://ecos.sourceware.org/ml/ecos-patches/2003-08/msg00006.html
My gcc version shouts:

$ arm-elf-gcc --version
arm-elf-gcc (GCC) 3.2.1
Copyright (C) 2002 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.


Regards,
Edgar


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss