On 10/14/06, atoka wotsa <atokawotsa@xxxxxxxxxxx> wrote:
hi everyone,
- I compiled Linux kernel 2.6.18 with hugetlbfs support on fedora core 5 and
the compilation was successful.
Did you reboot into the 2.6.18 kernel at this point?
- I then runned the following commands to mount the hugetlbfs filesystem and
set the numbers of hugepages in the system.
- First i tried cat /proc/meminfo command and i got
HugePages_Total: 0
HugePages_Free: 0
HugePages_Rsvd: 0
Hugepagesize: 2048 kB
- Then cat /proc/sys/vm/nr_hugepages (indicates the
current number of configured hugetlb) command and i got 0 as output.
- So i tried mounting the filesystem of type hugetlbfs by
using the command mount none /mnt/huge -t hugetlbfs and it got
mounted.
- Then i used echo 10 > /proc/sys/vm/nr_hugepages
to configure 10 hugepages in the system. and again tried cat
/proc/sys/vm/nr_hugepages and i got 1 as output.
where it should have be 10.
How much memory do you have in your system? nr_hugepages is a best
effort thing, as hugepages *must* be contiguous in physical memory. If
you waited a while after boot to allocate 10 hugepages, I can almost
*guarantee* you won't have sufficient contiguous memory, unless you
have lots of RAM (measured in GBs, most likely :) Consider booting the
kernel with the hugepages= parameter to get around this, as you'll get
the hugepages allocated as early as possible relative to the system's
boot time.
- i then tried to copy any temporary file to the mounted
directory but it
signaled "insufficient memory".
How big was the temporary file?
Thanks,
Nish
--
Kernelnewbies: Help each other learn about the Linux kernel.
Archive: http://mail.nl.linux.org/kernelnewbies/
FAQ: http://kernelnewbies.org/faq/
