diff options
| author | Ilya Leoshkevich <iii@linux.ibm.com> | 2024-03-21 01:47:49 +0300 |
|---|---|---|
| committer | Vasily Gorbik <gor@linux.ibm.com> | 2024-04-03 16:00:20 +0300 |
| commit | c9c260681f521e4ad9f9f4cc71fe35b978e06222 (patch) | |
| tree | a9a4ff0b4395d595204870a6856586b1daa2c7b3 /tools/perf/scripts/python | |
| parent | 01cac82ae02b43983173ea8e475a1c999edd25a6 (diff) | |
| download | linux-c9c260681f521e4ad9f9f4cc71fe35b978e06222.tar.xz | |
s390/preempt: mark all functions __always_inline
preempt_count-related functions are quite ubiquitous and may be called
by noinstr ones, introducing unwanted instrumentation. Here is one
example call chain:
irqentry_nmi_enter() # noinstr
lockdep_hardirqs_enabled()
this_cpu_read()
__pcpu_size_call_return()
this_cpu_read_*()
this_cpu_generic_read()
__this_cpu_generic_read_nopreempt()
preempt_disable_notrace()
__preempt_count_inc()
__preempt_count_add()
They are very small, so there are no significant downsides to
force-inlining them.
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Link: https://lore.kernel.org/r/20240320230007.4782-3-iii@linux.ibm.com
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Signed-off-by: Vasily Gorbik <gor@linux.ibm.com>
Diffstat (limited to 'tools/perf/scripts/python')
0 files changed, 0 insertions, 0 deletions