asan: Add --param=asan-kernel-mem-intrinsic-prefix= [PR108777]

  Hi!

While in the -fsanitize=address case libasan overloads memcpy, memset,
memmove and many other builtins, such that they are always instrumented,
Linux kernel for -fsanitize=kernel-address recently changed or is changing,
such that memcpy, memset and memmove actually aren't instrumented because
they are often used also from no_sanitize ("kernel-address") functions
and wants __{,hw,}asaN_{memcpy,memset,memmove} to be used instead
for the instrumented calls.  See e.g. the https://lkml.org/lkml/2023/2/9/1182
thread.  Without appropriate support on the compiler side, that will mean
any time a kernel-address instrumented function (most of them) calls
memcpy/memset/memmove, they will not be instrumented and thus won't catch
kernel bugs.  Apparently clang 15 has a param for this.

The following patch implements the same (except it is a usual GCC --param,
not -mllvm argument) on the GCC side.  I know this isn't a regression
bugfix, but given that -fsanitize=kernel-address has a single project that
uses it which badly wants this I think it would be worthwhile to make an
exception and get this into GCC 13 rather than waiting another year, it
won't affect non-kernel code, nor even the kernel unless the new parameter
is used.

Bootstrapped/regtested on x86_64-linux and i686-linux and Marco has tested
it on the kernel, ok for trunk?

2023-02-14  Jakub Jelinek  <jakub@redhat.com>

	PR sanitizer/108777
	* params.opt (-param=asan-kernel-mem-intrinsic-prefix=): New param.
	* asan.h (asan_memfn_rtl): Declare.
	* asan.cc (asan_memfn_rtls): New variable.
	(asan_memfn_rtl): New function.
	* builtins.cc (expand_builtin): If
	param_asan_kernel_mem_intrinsic_prefix and function is
	kernel-{,hw}address sanitized, emit calls to
	__{,hw}asan_{memcpy,memmove,memset} rather than
	{memcpy,memmove,memset}.  Use sanitize_flags_p (SANITIZE_ADDRESS)
	instead of flag_sanitize & SANITIZE_ADDRESS to check if
	asan_intercepted_p functions shouldn't be expanded inline.

	* gcc.dg/asan/pr108777-1.c: New test.
	* gcc.dg/asan/pr108777-2.c: New test.
	* gcc.dg/asan/pr108777-3.c: New test.
	* gcc.dg/asan/pr108777-4.c: New test.
	* gcc.dg/asan/pr108777-5.c: New test.
	* gcc.dg/asan/pr108777-6.c: New test.
	* gcc.dg/completion-3.c: Adjust expected multiline output.


	Jakub
  

On Tue, 14 Feb 2023, Jakub Jelinek wrote:

> Hi!
> 
> While in the -fsanitize=address case libasan overloads memcpy, memset,
> memmove and many other builtins, such that they are always instrumented,
> Linux kernel for -fsanitize=kernel-address recently changed or is changing,
> such that memcpy, memset and memmove actually aren't instrumented because
> they are often used also from no_sanitize ("kernel-address") functions
> and wants __{,hw,}asaN_{memcpy,memset,memmove} to be used instead
> for the instrumented calls.  See e.g. the https://lkml.org/lkml/2023/2/9/1182
> thread.  Without appropriate support on the compiler side, that will mean
> any time a kernel-address instrumented function (most of them) calls
> memcpy/memset/memmove, they will not be instrumented and thus won't catch
> kernel bugs.  Apparently clang 15 has a param for this.
> 
> The following patch implements the same (except it is a usual GCC --param,
> not -mllvm argument) on the GCC side.  I know this isn't a regression
> bugfix, but given that -fsanitize=kernel-address has a single project that
> uses it which badly wants this I think it would be worthwhile to make an
> exception and get this into GCC 13 rather than waiting another year, it
> won't affect non-kernel code, nor even the kernel unless the new parameter
> is used.
> 
> Bootstrapped/regtested on x86_64-linux and i686-linux and Marco has tested
> it on the kernel, ok for trunk?

OK.

Thanks,
Richard.

> 2023-02-14  Jakub Jelinek  <jakub@redhat.com>
> 
> 	PR sanitizer/108777
> 	* params.opt (-param=asan-kernel-mem-intrinsic-prefix=): New param.
> 	* asan.h (asan_memfn_rtl): Declare.
> 	* asan.cc (asan_memfn_rtls): New variable.
> 	(asan_memfn_rtl): New function.
> 	* builtins.cc (expand_builtin): If
> 	param_asan_kernel_mem_intrinsic_prefix and function is
> 	kernel-{,hw}address sanitized, emit calls to
> 	__{,hw}asan_{memcpy,memmove,memset} rather than
> 	{memcpy,memmove,memset}.  Use sanitize_flags_p (SANITIZE_ADDRESS)
> 	instead of flag_sanitize & SANITIZE_ADDRESS to check if
> 	asan_intercepted_p functions shouldn't be expanded inline.
> 
> 	* gcc.dg/asan/pr108777-1.c: New test.
> 	* gcc.dg/asan/pr108777-2.c: New test.
> 	* gcc.dg/asan/pr108777-3.c: New test.
> 	* gcc.dg/asan/pr108777-4.c: New test.
> 	* gcc.dg/asan/pr108777-5.c: New test.
> 	* gcc.dg/asan/pr108777-6.c: New test.
> 	* gcc.dg/completion-3.c: Adjust expected multiline output.
> 
> --- gcc/params.opt.jj	2023-02-10 19:04:58.289014706 +0100
> +++ gcc/params.opt	2023-02-13 16:19:50.411101775 +0100
> @@ -50,6 +50,10 @@ Enable asan store operations protection.
>  Common Joined UInteger Var(param_asan_instrumentation_with_call_threshold) Init(7000) Param Optimization
>  Use callbacks instead of inline code if number of accesses in function becomes greater or equal to this number.
>  
> +-param=asan-kernel-mem-intrinsic-prefix=
> +Common Joined UInteger Var(param_asan_kernel_mem_intrinsic_prefix) Init(0) IntegerRange(0, 1) Param Optimization
> +Prefix calls to memcpy, memset and memmove with __asan_ or __hwasan_ for -fsanitize=kernel-address or -fsanitize=kernel-hwaddress.
> +
>  -param=asan-memintrin=
>  Common Joined UInteger Var(param_asan_memintrin) Init(1) IntegerRange(0, 1) Param Optimization
>  Enable asan builtin functions protection.
> --- gcc/asan.h.jj	2023-01-02 09:32:26.721222635 +0100
> +++ gcc/asan.h	2023-02-13 16:45:14.475088159 +0100
> @@ -33,6 +33,7 @@ extern bool asan_expand_check_ifn (gimpl
>  extern bool asan_expand_mark_ifn (gimple_stmt_iterator *);
>  extern bool asan_expand_poison_ifn (gimple_stmt_iterator *, bool *,
>  				    hash_map<tree, tree> &);
> +extern rtx asan_memfn_rtl (tree);
>  
>  extern void hwasan_record_frame_init ();
>  extern void hwasan_record_stack_var (rtx, rtx, poly_int64, poly_int64);
> --- gcc/asan.cc.jj	2023-02-02 10:54:44.326473507 +0100
> +++ gcc/asan.cc	2023-02-13 16:52:16.711015256 +0100
> @@ -391,6 +391,46 @@ asan_memintrin (void)
>  }
>  
>  
> +/* Support for --param asan-kernel-mem-intrinsic-prefix=1.  */
> +static GTY(()) rtx asan_memfn_rtls[3];
> +
> +rtx
> +asan_memfn_rtl (tree fndecl)
> +{
> +  int i;
> +  const char *f, *p;
> +  char buf[sizeof ("__hwasan_memmove")];
> +
> +  switch (DECL_FUNCTION_CODE (fndecl))
> +    {
> +    case BUILT_IN_MEMCPY: i = 0; f = "memcpy"; break;
> +    case BUILT_IN_MEMSET: i = 1; f = "memset"; break;
> +    case BUILT_IN_MEMMOVE: i = 2; f = "memmove"; break;
> +    default: gcc_unreachable ();
> +    }
> +  if (asan_memfn_rtls[i] == NULL_RTX)
> +    {
> +      tree save_name = DECL_NAME (fndecl);
> +      tree save_assembler_name = DECL_ASSEMBLER_NAME (fndecl);
> +      rtx save_rtl = DECL_RTL (fndecl);
> +      if (flag_sanitize & SANITIZE_KERNEL_HWADDRESS)
> +	p = "__hwasan_";
> +      else
> +	p = "__asan_";
> +      strcpy (buf, p);
> +      strcat (buf, f);
> +      DECL_NAME (fndecl) = get_identifier (buf);
> +      DECL_ASSEMBLER_NAME_RAW (fndecl) = NULL_TREE;
> +      SET_DECL_RTL (fndecl, NULL_RTX);
> +      asan_memfn_rtls[i] = DECL_RTL (fndecl);
> +      DECL_NAME (fndecl) = save_name;
> +      DECL_ASSEMBLER_NAME_RAW (fndecl) = save_assembler_name;
> +      SET_DECL_RTL (fndecl, save_rtl);
> +    }
> +  return asan_memfn_rtls[i];
> +}
> +
> +
>  /* Checks whether section SEC should be sanitized.  */
>  
>  static bool
> --- gcc/builtins.cc.jj	2023-02-02 10:54:44.330473449 +0100
> +++ gcc/builtins.cc	2023-02-13 16:46:42.127826612 +0100
> @@ -7326,7 +7326,24 @@ expand_builtin (tree exp, rtx target, rt
>       by ASan.  */
>  
>    enum built_in_function fcode = DECL_FUNCTION_CODE (fndecl);
> -  if ((flag_sanitize & SANITIZE_ADDRESS) && asan_intercepted_p (fcode))
> +  if (param_asan_kernel_mem_intrinsic_prefix
> +      && sanitize_flags_p (SANITIZE_KERNEL_ADDRESS
> +			   | SANITIZE_KERNEL_HWADDRESS))
> +    switch (fcode)
> +      {
> +	rtx save_decl_rtl, ret;
> +      case BUILT_IN_MEMCPY:
> +      case BUILT_IN_MEMMOVE:
> +      case BUILT_IN_MEMSET:
> +	save_decl_rtl = DECL_RTL (fndecl);
> +	DECL_RTL (fndecl) = asan_memfn_rtl (fndecl);
> +	ret = expand_call (exp, target, ignore);
> +	DECL_RTL (fndecl) = save_decl_rtl;
> +	return ret;
> +      default:
> +	break;
> +      }
> +  if (sanitize_flags_p (SANITIZE_ADDRESS) && asan_intercepted_p (fcode))
>      return expand_call (exp, target, ignore);
>  
>    /* When not optimizing, generate calls to library functions for a certain
> --- gcc/testsuite/gcc.dg/asan/pr108777-1.c.jj	2023-02-13 17:49:23.139002484 +0100
> +++ gcc/testsuite/gcc.dg/asan/pr108777-1.c	2023-02-13 17:51:09.433479988 +0100
> @@ -0,0 +1,28 @@
> +/* PR sanitizer/108777 */
> +/* { dg-do compile } */
> +/* { dg-options "-O2 -fno-sanitize=all -fsanitize=kernel-address --param asan-kernel-mem-intrinsic-prefix=1" } */
> +/* { dg-final { scan-assembler "__asan_memcpy" } } */
> +/* { dg-final { scan-assembler "__asan_memset" } } */
> +/* { dg-final { scan-assembler "__asan_memmove" } } */
> +
> +extern void *memcpy (void *, const void *, __SIZE_TYPE__);
> +extern void *memmove (void *, const void *, __SIZE_TYPE__);
> +extern void *memset (void *, int, __SIZE_TYPE__);
> +
> +void
> +foo (void *p, void *q, int s)
> +{
> +  memcpy (p, q, s);
> +}
> +
> +void
> +bar (void *p, void *q, int s)
> +{
> +  memmove (p, q, s);
> +}
> +
> +void
> +baz (void *p, int c, int s)
> +{
> +  memset (p, c, s);
> +}
> --- gcc/testsuite/gcc.dg/asan/pr108777-2.c.jj	2023-02-13 17:49:27.215944098 +0100
> +++ gcc/testsuite/gcc.dg/asan/pr108777-2.c	2023-02-13 17:51:09.433479988 +0100
> @@ -0,0 +1,24 @@
> +/* PR sanitizer/108777 */
> +/* { dg-do compile } */
> +/* { dg-options "-O2 -fno-sanitize=all -fsanitize=kernel-address --param asan-kernel-mem-intrinsic-prefix=1" } */
> +/* { dg-final { scan-assembler "__asan_memcpy" } } */
> +/* { dg-final { scan-assembler "__asan_memset" } } */
> +/* { dg-final { scan-assembler "__asan_memmove" } } */
> +
> +void
> +foo (void *p, void *q, int s)
> +{
> +  __builtin_memcpy (p, q, s);
> +}
> +
> +void
> +bar (void *p, void *q, int s)
> +{
> +  __builtin_memmove (p, q, s);
> +}
> +
> +void
> +baz (void *p, int c, int s)
> +{
> +  __builtin_memset (p, c, s);
> +}
> --- gcc/testsuite/gcc.dg/asan/pr108777-3.c.jj	2023-02-13 17:49:30.683894408 +0100
> +++ gcc/testsuite/gcc.dg/asan/pr108777-3.c	2023-02-13 17:51:09.434479973 +0100
> @@ -0,0 +1,28 @@
> +/* PR sanitizer/108777 */
> +/* { dg-do compile } */
> +/* { dg-options "-O2 -fno-sanitize=all -fsanitize=kernel-address --param asan-kernel-mem-intrinsic-prefix=1" } */
> +/* { dg-final { scan-assembler-not "__asan_memcpy" } } */
> +/* { dg-final { scan-assembler-not "__asan_memset" } } */
> +/* { dg-final { scan-assembler-not "__asan_memmove" } } */
> +
> +extern void *memcpy (void *, const void *, __SIZE_TYPE__);
> +extern void *memmove (void *, const void *, __SIZE_TYPE__);
> +extern void *memset (void *, int, __SIZE_TYPE__);
> +
> +__attribute__((no_sanitize("kernel-address"))) void
> +foo (void *p, void *q, int s)
> +{
> +  memcpy (p, q, s);
> +}
> +
> +__attribute__((no_sanitize("kernel-address"))) void
> +bar (void *p, void *q, int s)
> +{
> +  memmove (p, q, s);
> +}
> +
> +__attribute__((no_sanitize("kernel-address"))) void
> +baz (void *p, int c, int s)
> +{
> +  memset (p, c, s);
> +}
> --- gcc/testsuite/gcc.dg/asan/pr108777-4.c.jj	2023-02-13 17:49:33.985847114 +0100
> +++ gcc/testsuite/gcc.dg/asan/pr108777-4.c	2023-02-13 17:51:09.434479973 +0100
> @@ -0,0 +1,24 @@
> +/* PR sanitizer/108777 */
> +/* { dg-do compile } */
> +/* { dg-options "-O2 -fno-sanitize=all -fsanitize=kernel-address --param asan-kernel-mem-intrinsic-prefix=1" } */
> +/* { dg-final { scan-assembler-not "__asan_memcpy" } } */
> +/* { dg-final { scan-assembler-not "__asan_memset" } } */
> +/* { dg-final { scan-assembler-not "__asan_memmove" } } */
> +
> +__attribute__((no_sanitize("kernel-address"))) void
> +foo (void *p, void *q, int s)
> +{
> +  __builtin_memcpy (p, q, s);
> +}
> +
> +__attribute__((no_sanitize("kernel-address"))) void
> +bar (void *p, void *q, int s)
> +{
> +  __builtin_memmove (p, q, s);
> +}
> +
> +__attribute__((no_sanitize("kernel-address"))) void
> +baz (void *p, int c, int s)
> +{
> +  __builtin_memset (p, c, s);
> +}
> --- gcc/testsuite/gcc.dg/asan/pr108777-5.c.jj	2023-02-13 17:49:37.195801146 +0100
> +++ gcc/testsuite/gcc.dg/asan/pr108777-5.c	2023-02-13 17:51:09.434479973 +0100
> @@ -0,0 +1,28 @@
> +/* PR sanitizer/108777 */
> +/* { dg-do compile } */
> +/* { dg-options "-O2 -fno-sanitize=all -fsanitize=kernel-address" } */
> +/* { dg-final { scan-assembler-not "__asan_memcpy" } } */
> +/* { dg-final { scan-assembler-not "__asan_memset" } } */
> +/* { dg-final { scan-assembler-not "__asan_memmove" } } */
> +
> +extern void *memcpy (void *, const void *, __SIZE_TYPE__);
> +extern void *memmove (void *, const void *, __SIZE_TYPE__);
> +extern void *memset (void *, int, __SIZE_TYPE__);
> +
> +void
> +foo (void *p, void *q, int s)
> +{
> +  memcpy (p, q, s);
> +}
> +
> +void
> +bar (void *p, void *q, int s)
> +{
> +  memmove (p, q, s);
> +}
> +
> +void
> +baz (void *p, int c, int s)
> +{
> +  memset (p, c, s);
> +}
> --- gcc/testsuite/gcc.dg/asan/pr108777-6.c.jj	2023-02-13 17:49:40.282756931 +0100
> +++ gcc/testsuite/gcc.dg/asan/pr108777-6.c	2023-02-13 17:51:09.434479973 +0100
> @@ -0,0 +1,24 @@
> +/* PR sanitizer/108777 */
> +/* { dg-do compile } */
> +/* { dg-options "-O2 -fno-sanitize=all -fsanitize=kernel-address" } */
> +/* { dg-final { scan-assembler-not "__asan_memcpy" } } */
> +/* { dg-final { scan-assembler-not "__asan_memset" } } */
> +/* { dg-final { scan-assembler-not "__asan_memmove" } } */
> +
> +void
> +foo (void *p, void *q, int s)
> +{
> +  __builtin_memcpy (p, q, s);
> +}
> +
> +void
> +bar (void *p, void *q, int s)
> +{
> +  __builtin_memmove (p, q, s);
> +}
> +
> +void
> +baz (void *p, int c, int s)
> +{
> +  __builtin_memset (p, c, s);
> +}
> --- gcc/testsuite/gcc.dg/completion-3.c.jj	2020-01-14 20:02:47.249602853 +0100
> +++ gcc/testsuite/gcc.dg/completion-3.c	2023-02-14 09:39:44.613203143 +0100
> @@ -7,6 +7,7 @@
>  --param=asan-instrument-reads=
>  --param=asan-instrument-writes=
>  --param=asan-instrumentation-with-call-threshold=
> +--param=asan-kernel-mem-intrinsic-prefix=
>  --param=asan-memintrin=
>  --param=asan-stack=
>  --param=asan-use-after-return=
> 
> 	Jakub
> 
>