diff options
Diffstat (limited to 'packages/linux/nslu2-kernel/2.6.14/10-ixp4xx-copy-from.patch')
-rw-r--r-- | packages/linux/nslu2-kernel/2.6.14/10-ixp4xx-copy-from.patch | 6 |
1 files changed, 3 insertions, 3 deletions
diff --git a/packages/linux/nslu2-kernel/2.6.14/10-ixp4xx-copy-from.patch b/packages/linux/nslu2-kernel/2.6.14/10-ixp4xx-copy-from.patch index db71342ab0..f2e65ecd8f 100644 --- a/packages/linux/nslu2-kernel/2.6.14/10-ixp4xx-copy-from.patch +++ b/packages/linux/nslu2-kernel/2.6.14/10-ixp4xx-copy-from.patch @@ -12,12 +12,12 @@ #include <linux/reboot.h> -+/* On a little-endian IXP4XX system (tested on NSLU2) contrary to the -+ * Intel documentation LDRH/STRH appears to XOR the address with 10b. ++/* On a little-endian IXP4XX system (tested on NSLU2) an LDRH or STRH ++ * will flip the second address bit - i.e. XOR the address with 10b. + * This causes the cfi commands (sent to the command address, 0xAA for + * 16 bit flash) to fail. This is fixed here by XOR'ing the address + * before use with 10b. The cost of this is that the flash layout ends -+ * up with pdp-endiannes (on an LE syste), however this is not a problem ++ * up with pdp-endiannes (on an LE system), however this is not a problem + * as the access code consistently only accesses half words - so the + * endianness is not determinable on stuff which is written and read + * consistently in the little endian world. |