After freeze_fs was revoked (from Jan Kara) pages's write-back completion
is deffered before unwritten conversion, so explicit flush_unwritten_io()
was removed here: c724585b62411
But we still may face deferred conversion for aio-dio case
# Trivial testcase
for ((i=0;i<60;i++));do fsfreeze -f /mnt
d_root_check() takes the following locks in wrong order:
write_seqlock(&rename_lock);
br_read_lock(&vfsmount_lock);
thus we may endup with deadlock.
While we could fix it by taking these locks in reverse
order, it would be better to just delete this function.
d_root_check() is a
it seem to be non intentionaly changed to tx in ms commit:
adc810900a703ee78fe88fd65e086d359fec04b2
ixgbe: Refactor busy poll socket code to address multiple issues
yield stats can be shown through ethtool:
ethtool -S enp129s0 | grep yield
https://jira.sw.ru/browse/PSBM-40330
Signed-off-by: Pave
In netpoll_send_skb_on_dev we do not trigger warn so irqs are disabled.
And in ixgbe_poll->ixgbe_qv_unlock_napi with irqs disabled we use
spin_unlock_bh, that is bug as inside it has local_irq_enable, witch
will enable hardware interrupts.
So port patch witch changes locking here to atomic_cmpxchg
Customer reports that free or top command inside Centos7 CT shows 0 memory used
free from procps-ng-3.3.10-3.el7 uses /proc/meminfo output and calculates
used = total - free - buffers - cache - slab
However kernel in some cases calculates "Cached" value in meminfo output by
using
the same formul