Skip to contentSkip to conversation

cross compiling fw_printenv on big endian

U-Boot Applied Jeff Angielski 4 participants 9 messages Mar 16, 2010 – Mar 21, 2010

lore
Share & tools
lore
#0 lore raw

I seem to be having a problem with fw_printenv on my PowerPC 85xx target whereby I constantly get CRC32 errors:

# fw_printenv
Warning: Bad CRC, using default environment

I tracked this down to the fact that u-boot/lib_generic/crc32.c is getting compiled with the __LITTLE_ENDIAN defined even though I have CROSS_COMPILE setup correctly [e.g. I can build u-boot.bin just fine].

Is there a special way to cross compile fw_printenv so that the __LITTLE_ENDIAN is not defined? Are there other PowerPC users that are building fw_printenv?

I'd rather do the "proper" thing than resort of hacking the crc32.c file.

Signature · 3 lines
Jeff Angielski
The PTR Group
www.theptrgroup.com
#1 lore raw

Hi Jeff,

Quoted 9 lines

I seem to be having a problem with fw_printenv on my PowerPC 85xx target whereby I constantly get CRC32 errors:

# fw_printenv
Warning: Bad CRC, using default environment

I tracked this down to the fact that u-boot/lib_generic/crc32.c is getting compiled with the __LITTLE_ENDIAN defined even though I have CROSS_COMPILE setup correctly [e.g. I can build u-boot.bin just fine].

This makes little sense to me, as then the crc version used in U-Boot would also use a wrong endianness. Are you sure this happens? Las time I used the tool it worked for me without any such hacks.

Cheers
Detlev

Signature · 6 lines
The limits of my language stand for the limits of my world.
                                        -- Ludwig Wittgenstein
--
DENX Software Engineering GmbH,      MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich,  Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-40 Fax: (+49)-8142-66989-80 Email: dzu at denx.de
#2 lore raw
Quoted 9 lines

Hi Jeff,

Quoted 6 lines

I seem to be having a problem with fw_printenv on my PowerPC 85xx target whereby I constantly get CRC32 errors:

# fw_printenv
Warning: Bad CRC, using default environment
Quoted 7 lines
Quoted 3 lines

I tracked this down to the fact that u-boot/lib_generic/crc32.c is getting compiled with the __LITTLE_ENDIAN defined even though I have CROSS_COMPILE setup correctly [e.g. I can build u-boot.bin just fine].

This makes little sense to me, as then the crc version used in U-Boot would also use a wrong endianness. Are you sure this happens? Las time I used the tool it worked for me without any such hacks.

hmm, I recently discovered that normal user space headers always define both __LITTLE_ENDIAN and __BIG_ENDIAN so therefore a

# ifdef __LITTLE_ENDIAN
#  define DO_CRC(x) crc = tab[(crc ^ (x)) & 255] ^ (crc >> 8)
# else
#  define DO_CRC(x) crc = tab[((crc >> 24) ^ (x)) & 255] ^ (crc << 8)
# endif

Wont work. One have to use
#if __BYTE_ORDER == __LITTLE_ENDIAN
instead.

Problem is that I don't think u-boot #defines __BYTE_ORDER so that would have to be added too for all archs:
#define __BYTE_ORDER __LITTLE_ENDIAN
or
#define __BYTE_ORDER __BIG_ENDIAN

#3 lore raw

Dear Joakim Tjernlund,

Quoted 9 lines · In message <OFF4AB0804.BE309218-ONC12576E9.003A69DF-C12576E9.003B4D86@transmode…

hmm, I recently discovered that normal user space headers always define both __LITTLE_ENDIAN and __BIG_ENDIAN so therefore a

# ifdef __LITTLE_ENDIAN
#  define DO_CRC(x) crc = tab[(crc ^ (x)) & 255] ^ (crc >> 8)
# else
#  define DO_CRC(x) crc = tab[((crc >> 24) ^ (x)) & 255] ^ (crc << 8)
# endif
Quoted 3 lines

Wont work. One have to use
#if __BYTE_ORDER == __LITTLE_ENDIAN
instead.

Wenn...

3ee8c120 (Joakim Tjernlund                 2009-11-19 13:44:16 +0100 166) # ifdef __LITTLE_ENDIAN
3ee8c120 (Joakim Tjernlund                 2009-11-19 13:44:16 +0100 167) #  define DO_CRC(x) crc = tab[(crc ^ (x)) & 255] ^ (crc >> 8)
3ee8c120 (Joakim Tjernlund                 2009-11-19 13:44:16 +0100 168) # else
3ee8c120 (Joakim Tjernlund                 2009-11-19 13:44:16 +0100 169) #  define DO_CRC(x) crc = tab[((crc >> 24) ^ (x)) & 255] ^ (crc << 8)
3ee8c120 (Joakim Tjernlund                 2009-11-19 13:44:16 +0100 170) # endif

commit 3ee8c12071f0e3bdda25125b63c9d3fd54a7c9d8

Author: Joakim Tjernlund <[email protected]>
Date:   Thu Nov 19 13:44:16 2009 +0100

crc32: Impl. linux optimized crc32()

Ported over the more efficient linux crc32() function. A quick comparsion on ppc:
After changing the old crc32 to do 4 bytes in the inner loop to be able to compare with new version one can note:
- old inner loop has 61 insn, new has 19 insn.
- new crc32 does one 32 bit load of data to crc while the old does four 8 bits loads.
- size is bit bigger for the new crc32:
1392(old) 1416(new) of text. The is because the new version shares code with crc32_no_comp() instead of duplicating code.
- about 33% faster on ppc:
New > crc 0 0xfffffff -> 39 secs
Old > crc 0 0xfffffff -> 60 secs

Looks as if this were your very own commit. Do you have a fix in the works?

Best regards,

Wolfgang Denk

Signature · 8 lines
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
The manager will be continually amazed that policies he took for com-
mon knowledge are totally unknown by some member of his  team.  Since
his fundamental job is to keep everybody going in the same direction,
his chief daily task will be communication, not decision-making.
                              - Fred Brooks, "The Mythical Man Month"
#4 lore raw

Wolfgang Denk <[email protected]> wrote on 2010/03/17 12:57:31:

Quoted 18 lines

Dear Joakim Tjernlund,

In message <OFF4AB0804.BE309218-ONC12576E9.003A69DF-C12576E9.

Quoted 9 lines · 003B4D86 at transmode.se> you wrote:

hmm, I recently discovered that normal user space headers always define both __LITTLE_ENDIAN and __BIG_ENDIAN so therefore a

# ifdef __LITTLE_ENDIAN
#  define DO_CRC(x) crc = tab[(crc ^ (x)) & 255] ^ (crc >> 8)
# else
#  define DO_CRC(x) crc = tab[((crc >> 24) ^ (x)) & 255] ^ (crc << 8)
# endif
Quoted 3 lines

Wont work. One have to use
#if __BYTE_ORDER == __LITTLE_ENDIAN
instead.

Quoted 8 lines

Wenn...

Looks as if this were your very own commit. Do you have a fix in the works?

I know, but I don't have anything ATM. I am too busy debugging serious customer problems.

Jocke

#5 lore raw
Quoted 22 lines · Joakim Tjernlund

Wolfgang Denk <[email protected]> wrote on 2010/03/17 12:57:31:

Quoted 20 lines

Dear Joakim Tjernlund,

In message <OFF4AB0804.BE309218-ONC12576E9.003A69DF-C12576E9.

Quoted 11 lines · 003B4D86 at transmode.se> you wrote:

hmm, I recently discovered that normal user space headers always define both __LITTLE_ENDIAN and __BIG_ENDIAN so therefore a

# ifdef __LITTLE_ENDIAN
#  define DO_CRC(x) crc = tab[(crc ^ (x)) & 255] ^ (crc >> 8)
# else
#  define DO_CRC(x) crc = tab[((crc >> 24) ^ (x)) & 255] ^ (crc << 8)
# endif

Wont work. One have to use
#if __BYTE_ORDER == __LITTLE_ENDIAN
instead.

Wenn...

Quoted 7 lines
Quoted 2 lines

Looks as if this were your very own commit. Do you have a fix in the works?

I know, but I don't have anything ATM. I am too busy debugging serious customer problems.

Jocke

This appears to work for me on my big endian PowerPC target. Perhaps somebody with a little endian target can verify it does not break their env tools.

lib_generic/crc32.c +0 -0
index 468b397..27335a3 100644
@@ -163,7 +163,7 @@ const uint32_t * ZEXPORT get_crc_table()
  #endif
 
  /* 

========================================================================= */ -# ifdef __LITTLE_ENDIAN
+# if __BYTE_ORDER == __LITTLE_ENDIAN
# define DO_CRC(x) crc = tab[(crc ^ (x)) & 255] ^ (crc >> 8)

  # else
  #  define DO_CRC(x) crc = tab[((crc >> 24) ^ (x)) & 255] ^ (crc << 8)

Jeff Angielski
The PTR Group
www.theptrgroup.com

#6 lore raw

Jeff Angielski <[email protected]> wrote on 2010/03/17 16:10:50:

Quoted 45 lines
Quoted 22 lines · Joakim Tjernlund

Wolfgang Denk <[email protected]> wrote on 2010/03/17 12:57:31:

Quoted 20 lines

Dear Joakim Tjernlund,

In message <OFF4AB0804.BE309218-ONC12576E9.003A69DF-C12576E9.

Quoted 11 lines · 003B4D86 at transmode.se> you wrote:

hmm, I recently discovered that normal user space headers always define both __LITTLE_ENDIAN and __BIG_ENDIAN so therefore a

# ifdef __LITTLE_ENDIAN
#  define DO_CRC(x) crc = tab[(crc ^ (x)) & 255] ^ (crc >> 8)
# else
#  define DO_CRC(x) crc = tab[((crc >> 24) ^ (x)) & 255] ^ (crc << 8)
# endif

Wont work. One have to use
#if __BYTE_ORDER == __LITTLE_ENDIAN
instead.

Wenn...

Quoted 7 lines
Quoted 2 lines

Looks as if this were your very own commit. Do you have a fix in the works?

I know, but I don't have anything ATM. I am too busy debugging serious customer problems.

Jocke

This appears to work for me on my big endian PowerPC target. Perhaps somebody with a little endian target can verify it does not break their env tools.

lib_generic/crc32.c +0 -0
index 468b397..27335a3 100644
@@ -163,7 +163,7 @@ const uint32_t * ZEXPORT get_crc_table()
  #endif
 
  /*

========================================================================= */

lib_generic/crc32.c

Quoted 5 lines
lib_generic/crc32.c +1 -1
-# ifdef __LITTLE_ENDIAN
+# if __BYTE_ORDER == __LITTLE_ENDIAN
  #  define DO_CRC(x) crc = tab[(crc ^ (x)) & 255] ^ (crc >> 8)
  # else
  #  define DO_CRC(x) crc = tab[((crc >> 24) ^ (x)) & 255] ^ (crc << 8)

I THINK this will work. Looking@include/linux/byteorder/big_endian.h it says:
#define __BYTE_ORDER __BIG_ENDIAN
so it seems like the __BYTE_ORDER logic is in place in u-boot. Someone with a LE CPU should confirm this.

Mind doing a proper patch?

Jocke

#7 lore raw
Quoted 53 lines · Joakim Tjernlund

Jeff Angielski <[email protected]> wrote on 2010/03/17 16:10:50:

Quoted 40 lines
Quoted 20 lines · Joakim Tjernlund

Wolfgang Denk <[email protected]> wrote on 2010/03/17 12:57:31:

Quoted 19 lines

Dear Joakim Tjernlund,

In message <OFF4AB0804.BE309218-ONC12576E9.003A69DF-C12576E9.

Quoted 11 lines · 003B4D86 at transmode.se> you wrote:

hmm, I recently discovered that normal user space headers always define both __LITTLE_ENDIAN and __BIG_ENDIAN so therefore a

# ifdef __LITTLE_ENDIAN
#  define DO_CRC(x) crc = tab[(crc ^ (x)) & 255] ^ (crc >> 8)
# else
#  define DO_CRC(x) crc = tab[((crc >> 24) ^ (x)) & 255] ^ (crc << 8)
# endif

Wont work. One have to use
#if __BYTE_ORDER == __LITTLE_ENDIAN
instead.

Wenn...

Quoted 6 lines
Quoted 2 lines

Looks as if this were your very own commit. Do you have a fix in the works?

I know, but I don't have anything ATM. I am too busy debugging serious customer problems.

Jocke

This appears to work for me on my big endian PowerPC target. Perhaps somebody with a little endian target can verify it does not break their env tools.

lib_generic/crc32.c +0 -0
index 468b397..27335a3 100644
@@ -163,7 +163,7 @@ const uint32_t * ZEXPORT get_crc_table()
  #endif
 
  /*

========================================================================= */

lib_generic/crc32.c

Quoted 5 lines
lib_generic/crc32.c +1 -1
-# ifdef __LITTLE_ENDIAN
+# if __BYTE_ORDER == __LITTLE_ENDIAN
  #  define DO_CRC(x) crc = tab[(crc ^ (x)) & 255] ^ (crc >> 8)
  # else
  #  define DO_CRC(x) crc = tab[((crc >> 24) ^ (x)) & 255] ^ (crc << 8)

I THINK this will work. Looking at include/linux/byteorder/big_endian.h it says:
#define __BYTE_ORDER __BIG_ENDIAN
so it seems like the __BYTE_ORDER logic is in place in u-boot. Someone with a LE CPU should confirm this.

Quoted 2 lines

Mind doing a proper patch?

See attached file.

Signature · 10 lines
Jeff Angielski
The PTR Group
www.theptrgroup.com
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0001-env-fix-endian-ordering-in-crc-table.patch
Type: text/x-patch
Size: 0 bytes
Desc: not available
Url : http://lists.denx.de/pipermail/u-boot/attachments/20100317/27c25dd9/attachment.bin 
#8 lore raw

Dear Jeff Angielski,

Quoted 10 lines · In message <[email protected]> you wrote:

Subject: [PATCH] env: fix endian ordering in crc table

The crc table was being built as little endian for big endian targets. This would cause fw_printenv to always fail with "Warning: Bad CRC, using default environment" messages.

---
 lib_generic/crc32.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

Applied, thanks.

Best regards,

Wolfgang Denk

Signature · 5 lines
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
Mirrors should reflect a little before throwing back images.
- Jean Cocteau