lkml.org 
[lkml]   [2017]   [Mar]   [30]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH] x86, asm-generic: add KASAN instrumentation to bitops

* Dmitry Vyukov <dvyukov@google.com> wrote:

> +#include <asm-generic/bitops-instrumented.h>
> +
> #include <asm-generic/bitops/find.h>
>
> #include <asm-generic/bitops/sched.h>
> diff --git a/include/asm-generic/bitops-instrumented.h b/include/asm-generic/bitops-instrumented.h
> new file mode 100644
> index 000000000000..01d02113fc7e
> --- /dev/null
> +++ b/include/asm-generic/bitops-instrumented.h
> @@ -0,0 +1,53 @@
> +/* See atomic-instrumented.h for explanation. */
> +#ifndef _LINUX_BITOPS_INSTRUMENTED_H
> +#define _LINUX_BITOPS_INSTRUMENTED_H
> +
> +#include <linux/kasan-checks.h>
> +
> +#define ADDR(nr, addr) ((void *)(addr) + ((nr) >> 3))
> +
> +#define INSTR_VOID(func) \
> +static __always_inline void func(long nr, volatile unsigned long *addr) \
> +{ \
> + kasan_check_write(ADDR(nr, addr), 1); \
> + arch_##func(nr, addr); \
> +}
> +
> +#define INSTR_BOOL(func) \
> +static __always_inline bool func(long nr, volatile unsigned long *addr) \
> +{ \
> + kasan_check_write(ADDR(nr, addr), 1); \
> + return arch_##func(nr, addr); \
> +}

Is the 'ADDR()' construct going to result in any extra inlined instructions in an
instrumented kernel? If yes, why not do it inside the KASAN callback to reduce
inlining overhead?

> +INSTR_VOID(set_bit);
> +INSTR_VOID(__set_bit);
> +INSTR_VOID(clear_bit);
> +INSTR_VOID(__clear_bit);
> +INSTR_VOID(clear_bit_unlock);
> +INSTR_VOID(__clear_bit_unlock);
> +INSTR_VOID(change_bit);
> +INSTR_VOID(__change_bit);
> +
> +INSTR_BOOL(test_and_set_bit);
> +INSTR_BOOL(test_and_set_bit_lock);
> +INSTR_BOOL(__test_and_set_bit);
> +INSTR_BOOL(test_and_clear_bit);
> +INSTR_BOOL(__test_and_clear_bit);
> +INSTR_BOOL(test_and_change_bit);
> +INSTR_BOOL(__test_and_change_bit);
> +#ifdef clear_bit_unlock_is_negative_byte
> +INSTR_BOOL(clear_bit_unlock_is_negative_byte);
> +#endif
> +
> +static bool test_bit(int nr, const volatile unsigned long *addr)
> +{
> + kasan_check_read(ADDR(nr, addr), 1);
> + return arch_test_bit(nr, addr);
> +}

Same objections as to the atomic primitives: don't hide function signatures behind
CPP complexity for marginal line count reduction.

Thanks,

Ingo

\
 
 \ /
  Last update: 2017-03-30 09:04    [W:0.826 / U:0.192 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site