Re: [PATCH v2 0/8] kcsan: Compound read-write instrumentation

2020-07-24 Thread Paul E. McKenney
On Fri, Jul 24, 2020 at 10:39:20AM +0200, Peter Zijlstra wrote:
> On Fri, Jul 24, 2020 at 09:00:00AM +0200, Marco Elver wrote:
> 
> > Marco Elver (8):
> >   kcsan: Support compounded read-write instrumentation
> >   objtool, kcsan: Add __tsan_read_write to uaccess whitelist
> >   kcsan: Skew delay to be longer for certain access types
> >   kcsan: Add missing CONFIG_KCSAN_IGNORE_ATOMICS checks
> >   kcsan: Test support for compound instrumentation
> >   instrumented.h: Introduce read-write instrumentation hooks
> >   asm-generic/bitops: Use instrument_read_write() where appropriate
> >   locking/atomics: Use read-write instrumentation for atomic RMWs
> 
> Looks good to me,
> 
> Acked-by: Peter Zijlstra (Intel) 

Applied with ack, thank you both!

Thanx, Paul


Re: [PATCH v2 0/8] kcsan: Compound read-write instrumentation

2020-07-24 Thread Peter Zijlstra
On Fri, Jul 24, 2020 at 09:00:00AM +0200, Marco Elver wrote:

> Marco Elver (8):
>   kcsan: Support compounded read-write instrumentation
>   objtool, kcsan: Add __tsan_read_write to uaccess whitelist
>   kcsan: Skew delay to be longer for certain access types
>   kcsan: Add missing CONFIG_KCSAN_IGNORE_ATOMICS checks
>   kcsan: Test support for compound instrumentation
>   instrumented.h: Introduce read-write instrumentation hooks
>   asm-generic/bitops: Use instrument_read_write() where appropriate
>   locking/atomics: Use read-write instrumentation for atomic RMWs

Looks good to me,

Acked-by: Peter Zijlstra (Intel) 


[PATCH v2 0/8] kcsan: Compound read-write instrumentation

2020-07-24 Thread Marco Elver
This series adds support for enabling compounded read-write
instrumentation, if supported by the compiler (Clang 12 will be the
first compiler to support the feature). The new instrumentation is
emitted for sets of memory accesses in the same basic block to the same
address with at least one read appearing before a write. These typically
result from compound operations such as ++, --, +=, -=, |=, &=, etc. but
also equivalent forms such as "var = var + 1".

We can then benefit from improved performance (fewer instrumentation
calls) and better reporting for such accesses. In addition, existing
explicit instrumentation via instrumented.h was updated to use explicit
read-write instrumentation where appropriate, so we can also benefit
from the better report generation.

v2:
* Fix CC_HAS_TSAN_COMPOUND_READ_BEFORE_WRITE: s/--param -tsan/--param tsan/
* Add some {} for readability.
* Rewrite commit message of 'kcsan: Skew delay to be longer for certain
  access types'.
* Update comment for gen-atomic-instrumented.sh.

Marco Elver (8):
  kcsan: Support compounded read-write instrumentation
  objtool, kcsan: Add __tsan_read_write to uaccess whitelist
  kcsan: Skew delay to be longer for certain access types
  kcsan: Add missing CONFIG_KCSAN_IGNORE_ATOMICS checks
  kcsan: Test support for compound instrumentation
  instrumented.h: Introduce read-write instrumentation hooks
  asm-generic/bitops: Use instrument_read_write() where appropriate
  locking/atomics: Use read-write instrumentation for atomic RMWs

 include/asm-generic/atomic-instrumented.h | 330 +-
 .../asm-generic/bitops/instrumented-atomic.h  |   6 +-
 .../asm-generic/bitops/instrumented-lock.h|   2 +-
 .../bitops/instrumented-non-atomic.h  |   6 +-
 include/linux/instrumented.h  |  30 ++
 include/linux/kcsan-checks.h  |  45 ++-
 kernel/kcsan/core.c   |  51 ++-
 kernel/kcsan/kcsan-test.c |  65 +++-
 kernel/kcsan/report.c |   4 +
 lib/Kconfig.kcsan |   5 +
 scripts/Makefile.kcsan|   2 +-
 scripts/atomic/gen-atomic-instrumented.sh |  21 +-
 tools/objtool/check.c |   5 +
 13 files changed, 354 insertions(+), 218 deletions(-)

-- 
2.28.0.rc0.142.g3c755180ce-goog