cap on rootfs

cap on rootfs

[glphvgacs]

hello,
so after reading
Documentation/filesystems/ramfs-rootfs-initramfs.txt
and 
Documentation/filesystems/tmpfs.txt

i thought i would put everything in /* into initrd since i have 8G of
ram sitting around with pretty much no use.  crazy? maybe.
but i must have hit an old cap on how much one can shove into initramfs
which is not cool, IMHO. here are related sizes:

$ ls -lh /boot/initramfs.cpio.gz:
1.2G Mar  5 16:59 /boot/initramfs.cpio.gz

$ df -h:
Filesystem      Size  Used Avail Use% Mounted on
rootfs          3.3G  3.3G     0 100% /
devtmpfs        3.4G     0  3.4G   0% /dev


note that df -h on rootfs is not even consistent; last boot i had 2.8G!
in both cases it violates what Documentations say, which is 50% of the
availabe ram.

now services can't write to for example /var/tmp/ since we're out of
space. remounting / will *fix* it:

# mount -o remount,size=90% /

$ df -h:
Filesystem      Size  Used Avail Use% Mounted on
rootfs          7.0G  3.3G  3.8G  47% /

but that's no solution.

so my question:
where is this set in kernel? is there a kernel param i can pass to
kernel to give it the go ahead to use 90% of the availabe ram? i have
btw tried rootflags=size=90% thinking that it might work like it does in
fstab for tmpfs, no luck. or is there a CONFIG_* that sets this? i
didn't try messing with CONFIG_BLK_DEV_RAM_SIZE since that seemed to be
for ramdisk and in any case too small to be the one being used, in my
case:
CONFIG_BLK_DEV_RAM_SIZE=4096

kernel version btw:
3.14.0-rc5 #1 SMP PREEMPT x86_64 Intel(R) Core(TM) i7-4770 CPU @ 3.40GHz


thanks

Re: cap on rootfs

[glphvgacs]

one fstab "solution"?
tmpfs / tmpfs defaults,remount,noatime,size=90% 0 0