[LEDE-DEV] [RFC 7/7] ramips: set blocksize for dch-m225/dir-860l-b1

Mathias Kresin dev at kresin.me
Tue Sep 13 22:41:21 PDT 2016


14.09.2016 02:02, Russell Senior:
>> diff --git a/target/linux/ramips/image/mt7621.mk b/target/linux/ramips/image/mt7621.mk
>> index e602dfb..acc17c3 100644
>> --- a/target/linux/ramips/image/mt7621.mk
>> +++ b/target/linux/ramips/image/mt7621.mk
>> @@ -65,15 +65,17 @@ TARGET_DEVICES += wsr-1166
>>
>>  define Device/dir-860l-b1
>>    DTS := DIR-860L-B1
>> +  BLOCKSIZE := 4k
>>    IMAGES += factory.bin
>>    KERNEL := kernel-bin | patch-dtb | relocate-kernel | lzma | uImage lzma
>>    IMAGE_SIZE := $(ralink_default_fw_size_16M)
>>    IMAGE/sysupgrade.bin := \
>> -	append-kernel | pad-offset 65536 64 | append-rootfs | \
>> +	append-kernel | pad-offset $$$$(BLOCKSIZE) 64 | append-rootfs | \
>>  	seama -m "dev=/dev/mtdblock/2" -m "type=firmware" | \
>>  	pad-rootfs | check-size $$$$(IMAGE_SIZE)
>>    IMAGE/factory.bin := \
>> -	append-kernel | pad-offset 65536 64 | append-rootfs | pad-rootfs -x 64 | \
>> +	append-kernel | pad-offset $$$$(BLOCKSIZE) 64 | \
>> +	append-rootfs | pad-rootfs -x 64 | \
>>  	seama -m "dev=/dev/mtdblock/2" -m "type=firmware" | \
>>  	seama-seal -m "signature=wrgac13_dlink.2013gui_dir860lb" | \
>>  	check-size $$$$(IMAGE_SIZE)
>> --
>> 2.7.4
>
> This patch broke dir-860l-b1.  Changing BLOCKSIZE for dir-860l-b1 to 64k makes it work again.

Thanks for the report.

To my knowledge the D-Link DIR-860L rev B1 has a MX25L12805D flash chip 
which supports 4K sector size. But I've missed to include the required 
kernel config changes. This was fixed with 
925e63e71fa6691f3fa374f71c77d0a91c7e7539. Would you please make sure 
that this commit is in your tree.

Please provide a full bootlog and the output of 'cat /proc/mtd' from an 
image which has the blocksize reverted to 64k. Might be possible that 
it's related to the seama header. I've checked the seama mtd splitter 
code before committing the change but might have missed something.

Mathias



More information about the Lede-dev mailing list