Unfortunately no one can be told what fun_plug is - you have to see it for yourself.
You are not logged in.
Hi,
I've tried to rebuild the ramdisk and compile it with the Dlink GPL sources.
I followed the instructions on curvedbrain.org, exept instead of
dd if=/dev/zero of=new-ramdisk.img bs=1k count=8k
I had to use count=10k (to reflect the size of the original Dlink ramdisk that came with the source).
But when I tried to rebuild the uRamdisk (with NO changes in the ramdisk whatsoever, just strip the original uRamdisk to .gz, gunzip, mount, make new-ramdisk.img, copy original Ramdisk to new-ramdisk.img, gzip new-ramdisk.img) it had grown bigger than the original and too big to merge into firmware.
So my question: Has anyone tried something similar, and how did you succeed?
Is there some kind of special compression needed (I used gzip -9), since the original ramdisk is smaller than the new ramdisk, but both contain exactly the same files?
Last edited by IcyK (2011-01-22 15:16:53)
Offline
Did you compare the loopmounted images using df? Maybe you resolved symlinks instead of copying them?
Offline
I compared them, the home made one is bigger.
But even the re-packed image.cfs is bigger, though I didn't change a thing!
These were the steps (left out some sudo's):
- split firmware using splitdns323fw
- remove the uboot header from initrd -> result: DNSramdisk.gz
- gunzip DNSramdisk.gz
- mount -o loop DNSramdisk ./ramdisk
- mount -o loop ./ramdisk/image.cfs ./image
- cp -R ./image/* ./tmpcramfs
- mkcramfs tmpcramfs new-image.cfs
ls -al ./ramdisk/image.cfs:
-rwx------ 1 root root 5881856 2009-12-18 03:22 ramdisk/image.cfs
ls -al ./new-image.cfs:
-rw-r--r-- 1 root root 6762496 2011-02-06 00:48 new-image.cfs
So ... I didn't change a thing, but the new image.cfs is about 1 MB bigger than the original one ...
Anyone who can shed a light on this?
Offline
If you now loopmount new-image.cfs also, then df -B 1 can tell you if the filecontents are the same.
Offline
Whow, huge difference:
/home/dick/dns/ramdisk/image.cfs
5898240 5898240 0 100% /home/dick/dns/image
/home/dick/dns/new-image.cfs
15593472 15593472 0 100% /home/dick/dns/new-imageIt also seems that some directories in the new image is a tiny bit bigger:
~/dns/image$ du * ~/dns/new-image$ du * 2396 bin 2396 bin 259 codepages 259 codepages 27 default 29 default 15 driver 16 driver 1 etc_codepage 1 etc_codepage 2 files 2 files 64 language 64 language 2681 lib 2681 lib 41 lltd 41 lltd 1305 LPRng 1305 LPRng 2586 samba 2587 samba 2524 sbin 2527 sbin 19 scsi 20 scsi 0 shared_name/CD 1 shared_name/CD 0 shared_name/CR 1 shared_name/CR 0 shared_name/HD 1 shared_name/HD 0 shared_name 4 shared_name 207 upnp 207 upnp 885 web 886 web 223 web_page/advanced 224 web_page/advanced 37 web_page/bt/img 38 web_page/bt/img 113 web_page/bt 114 web_page/bt 228 web_page/formatHD 229 web_page/formatHD 85 web_page/function 85 web_page/function 61 web_page/help 62 web_page/help 112 web_page/home 113 web_page/home 117 web_page/images 118 web_page/images 28 web_page/module 29 web_page/module 22 web_page/status 23 web_page/status 205 web_page/tools 206 web_page/tools 126 web_page/user 126 web_page/user 1347 web_page 1356 web_page
So I guess itś not resolved symlinks (in that case only a few directories would be significant bigger).
Thanks Mijzelf for the help so far ... (sounds funny in dutch ;-) )
Offline
I don't see any difference in the du outputs which excuses the difference in the df output. What is the du output of ./tmpcramfs?
Offline