Request: kernel initrd support

[retux]

Would you please consider including this option in stock kernel build?

CONFIG_BLK_DEV_INITRD=y

initrd is a very useful feature, i.e. for using uuid ID in cmdline.txt.

Thanks!

[popcornmix]

Can you determine the memory cost of this option?
i.e. after booting from default kernel what does
free -h
report? After enabling that option what does it report?

[retux]

kernel using initrd:

$ free -h
total used free shared buffers cached
Mem: 438M 52M 385M 0B 6,6M 25M
-/+ buffers/cache: 21M 417M
Swap: 99M 0B 99M

Reported by dmesg:

[ 1.049995] Trying to unpack rootfs image as initramfs...
[ 1.501765] Freeing initrd memory: 3456K

kernel not using initrd:

$ zcat /proc/config.gz | grep -i initrd

CONFIG_BLK_DEV_INITRD is not set

pi@raspberrypi ~ $ dmesg | grep -i initr
pi@raspberrypi ~ $

$ free -h
total used free shared buffers cached
Mem: 438M 51M 386M 0B 7.9M 23M
-/+ buffers/cache: 20M 417M
Swap: 99M 0B 99M

It seems there's no big differences in terms of memory usage.

[popcornmix]

Can you test next firmware branch - should be added.
sudo BRANCH=next rpi-update

[retux]

cool! I'll test it as soon as possible.
I tested and everything initrd related works just fine.
Thanks a lot!

[evilpower]

This was the only reason why I was compiling my own kernels for the past year. With initrd enabled and a simple initramfs we can set up root fs as btrfs or f2fs or anything else that's built in the kernel with a breeze. I hope this gets carried over to the master branch as well. Thank you!

[retux]

lurch,
Yes, it does. I'm using UUID through cmdline.txt, but in order to use it is required to create an initial ram disk (initrd). That's why adding initrd support to the stock kernel has been very useful.
Now initrd support is included in the stock kernel using initrd and all the features it allows is much easier.

[lurch]

In that case, maybe both this bug and your other one I linked to ( raspberrypi/firmware#209 ) should both be closed? ;-)

[retux]

You're right. Sorry I just closed both issues.
Regards,
Matías

El 06/11/13 21:43, Andrew Scheller escribió:

In that case, maybe both this bug and your other one I linked to (
raspberrypi/firmware#209
raspberrypi/firmware#209 ) should both be
closed? ;-)

—
Reply to this email directly or view it on GitHub
#392 (comment).