x86, trace: Fix CR2 corruption when tracing page faults
authorJiri Olsa <jolsa@redhat.com>
Fri, 28 Feb 2014 16:05:26 +0000 (17:05 +0100)
committerH. Peter Anvin <hpa@linux.intel.com>
Wed, 5 Mar 2014 00:00:14 +0000 (16:00 -0800)
commit0ac09f9f8cd1fb028a48330edba6023d347d3cea
tree6c0552e39b416d90334dd488fe7ce82e3318bb46
parent3c0b566334c9942c89927bd1b5ad2f32baab81fa
x86, trace: Fix CR2 corruption when tracing page faults

The trace_do_page_fault function trigger tracepoint
and then handles the actual page fault.

This could lead to error if the tracepoint caused page
fault. The original cr2 value gets lost and the original
page fault handler kills current process with SIGSEGV.

This happens if you record page faults with callchain
data, the user part of it will cause tracepoint handler
to page fault:

  # perf record -g -e exceptions:page_fault_user ls

Fixing this by saving the original cr2 value
and using it after tracepoint handler is done.

v2: Moving the cr2 read before exception_enter, because
    it could trigger tracepoint as well.

Reported-by: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Reported-by: Vince Weaver <vincent.weaver@maine.edu>
Tested-by: Vince Weaver <vincent.weaver@maine.edu>
Acked-by: Steven Rostedt <rostedt@goodmis.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Seiji Aguchi <seiji.aguchi@hds.com>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
Link: http://lkml.kernel.org/r/alpine.DEB.2.10.1402211701380.6395@vincent-weaver-1.um.maine.edu
Link: http://lkml.kernel.org/r/20140228160526.GD1133@krava.brq.redhat.com
arch/x86/mm/fault.c