Skip to contentSkip to conversation

High cpu load due to pdflush

DISCUSSION Bob Gill 1 participant 1 message Nov 3, 2007

lore
Share & tools
lore

Bob Gill

author
#0 lore raw

Hi. My computer is constipated. The load average when idle never goes below 100%. Top shows this:
top - 11:01:27 up 5 min, 2 users, load average: 2.63, 1.73, 0.76

Tasks:  91 total,   2 running,  89 sleeping,   0 stopped,   0 zombie
Cpu(s): 13.5%us,  4.4%sy,  1.6%ni,  9.9%id, 70.2%wa,  0.3%hi,  0.2%si,  0.0%st
Note that we are at 70% for the wait state, 9.9% idle.  Now when I run  ps aux | grep pdflush | grep -v grep  ...I get:
root       144  0.0  0.0      0     0 ?        S    10:56   0:00 [pdflush]
root       145  0.0  0.0      0     0 ?        D    10:56   0:00 [pdflush]

the second pdflush is dirty (and won't write). I haven't changed anything in /proc/sys/vm and am running the 2.6.24-rc1-git12 kernel on Ubuntu Gutsy. The original Gutsy kernel worked fine. The only way I can get around the problem is to go into a terminal and run:

Quoted 5 lines

while true
do
sync
sleep 10
done

...which will force writes and give me (from top)
top - 11:06:55 up 10 min, 3 users, load average: 0.42, 1.23, 0.85

Tasks:  93 total,   2 running,  91 sleeping,   0 stopped,   0 zombie
Cpu(s):  3.0%us,  0.9%sy,  0.0%ni, 95.6%id,  0.3%wa,  0.3%hi,  0.0%si,  0.0%st

(on the very same load as before and the load is still dropping).

The only real change I made to Ubuntu was in /etc/init.d/mountdevsubfs.sh where I uncommented:
# Magic to make /proc/bus/usb work
#
mkdir -p /dev/bus/usb/.usbfs
domount usbfs "" /dev/bus/usb/.usbfs -obusmode=0700,devmode=0600,listmode=0644 ln -s .usbfs/devices /dev/bus/usb/devices
mount --rbind /dev/bus/usb /proc/bus/usb

...so that usb would work with a custom kernel...

Oh, and btw, my cpu is an Intel P4.
Also, I applied the following patch and build/ran a kernel (but it did not reduce the cpu load on my system):

Quoted 11 lines

/> -------------------------- fs/jbd/transaction.c -----------------------------/ /> index cceaf57..d38e0d5 100644/
/> @@ -55,7 +55,7 @@ get_transaction(journal_t *journal, transaction_t *transaction)/ /> spin_lock_init(&transaction->t_handle_lock);/
/> /
/> /* Set up the commit timer for the new transaction. *//
/> - journal->j_commit_timer.expires = round_jiffies(transaction->t_expires);/ /> + journal->j_commit_timer.expires = transaction->t_expires;/
/> add_timer(&journal->j_commit_timer);/
/> /
/> J_ASSERT(journal->j_running_transaction == NULL);

/Any ideas? Please mail me if you need more information. Thanks in Advance, Bob