]> git.openfabrics.org - ~shefty/rdma-dev.git/commit
powerpc: Fix unpaired probe_hcall_entry and probe_hcall_exit
authorLi Zhong <zhong@linux.vnet.ibm.com>
Sun, 18 Dec 2011 16:03:04 +0000 (16:03 +0000)
committerBenjamin Herrenschmidt <benh@kernel.crashing.org>
Tue, 3 Jan 2012 01:09:27 +0000 (12:09 +1100)
commite4f387d8db3ba3c2dae4d8bdfe7bb5f4fe1bcb0d
tree148f0524d6f8413dbc5ed09e9bf056e2592649e6
parent1bb0b7d21584b3f878e2bc880db62351ddee5185
powerpc: Fix unpaired probe_hcall_entry and probe_hcall_exit

Unpaired calling of probe_hcall_entry and probe_hcall_exit might happen
as following, which could cause incorrect preempt count.

__trace_hcall_entry => trace_hcall_entry -> probe_hcall_entry =>
get_cpu_var => preempt_disable

__trace_hcall_exit => trace_hcall_exit -> probe_hcall_exit =>
put_cpu_var => preempt_enable

where:
A => B and A -> B means A calls B, but
=> means A will call B through function name, and B will definitely be
called.
-> means A will call B through function pointer, so B might not be
called if the function pointer is not set.

So error happens when only one of probe_hcall_entry and probe_hcall_exit
get called during a hcall.

This patch tries to move the preempt count operations from
probe_hcall_entry and probe_hcall_exit to its callers.

Reported-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Li Zhong <zhong@linux.vnet.ibm.com>
Tested-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
CC: stable@kernel.org [v2.6.32+]
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
arch/powerpc/platforms/pseries/hvCall_inst.c
arch/powerpc/platforms/pseries/lpar.c