On Sat, Jun 19, 2021 at 7:10 PM H.J. Lu <hjl.to...@gmail.com> wrote:
>
> 1. On some targets, like PowerPC, reference to ifunc function resolver
> must be non-local so that compiler will properly emit PLT call. Add
> TARGET_IFUNC_REF_LOCAL_OK to allow binding indirect function resolver
> locally for targets which don't require special PLT call sequence.
> 2. Add ix86_call_use_plt_p to call local ifunc function resolvers via
> PLT.
>
> gcc/
>
> PR target/51469
> PR target/83782
> * target.def (ifunc_ref_local_ok): Add a target hook.
> * varasm.c (default_binds_local_p_3): Force indirect function
> resolver non-local only if targetm.ifunc_ref_local_ok returns
> false.
> * config/i386/i386-expand.c (ix86_expand_call): Call
> ix86_call_use_plt_p to check if PLT should be used.
> * config/i386/i386-protos.h (ix86_call_use_plt_p): New.
> * config/i386/i386.c (output_pic_addr_const): Call
> ix86_call_use_plt_p to check if "@PLT" is needed.
> (ix86_call_use_plt_p): New.
> (TARGET_IFUNC_REF_LOCAL_OK): New.
> * doc/tm.texi.in: Add TARGET_IFUNC_REF_LOCAL_OK.
> * doc/tm.texi: Regenerated.
>
> gcc/testsuite/
>
> PR target/51469
> PR target/83782
> * gcc.target/i386/pr83782-1.c: New test.
> * gcc.target/i386/pr83782-2.c: Likewise.
> ---
> gcc/config/i386/i386-expand.c | 2 +-
> gcc/config/i386/i386-protos.h | 1 +
> gcc/config/i386/i386.c | 25 +++++++++++++++++++++-
> gcc/doc/tm.texi | 5 +++++
> gcc/doc/tm.texi.in | 2 ++
> gcc/target.def | 8 +++++++
> gcc/testsuite/gcc.target/i386/pr83782-1.c | 26 +++++++++++++++++++++++
> gcc/testsuite/gcc.target/i386/pr83782-2.c | 26 +++++++++++++++++++++++
> gcc/varasm.c | 3 ++-
> 9 files changed, 95 insertions(+), 3 deletions(-)
> create mode 100644 gcc/testsuite/gcc.target/i386/pr83782-1.c
> create mode 100644 gcc/testsuite/gcc.target/i386/pr83782-2.c
>
> diff --git a/gcc/config/i386/i386-expand.c b/gcc/config/i386/i386-expand.c
> index 8f4e4e4d884..229a765cb40 100644
> --- a/gcc/config/i386/i386-expand.c
> +++ b/gcc/config/i386/i386-expand.c
> @@ -8184,7 +8184,7 @@ ix86_expand_call (rtx retval, rtx fnaddr, rtx callarg1,
> rtx addr = XEXP (fnaddr, 0);
> if (flag_pic
> && GET_CODE (addr) == SYMBOL_REF
> - && !SYMBOL_REF_LOCAL_P (addr))
> + && ix86_call_use_plt_p (addr))
> {
> if (flag_plt
> && (SYMBOL_REF_DECL (addr) == NULL_TREE
> diff --git a/gcc/config/i386/i386-protos.h b/gcc/config/i386/i386-protos.h
> index e6ac9390777..2d1cd07a215 100644
> --- a/gcc/config/i386/i386-protos.h
> +++ b/gcc/config/i386/i386-protos.h
> @@ -152,6 +152,7 @@ extern void ix86_expand_sse_movcc (rtx, rtx, rtx, rtx);
> extern void ix86_expand_sse_unpack (rtx, rtx, bool, bool);
> extern bool ix86_expand_int_addcc (rtx[]);
> extern rtx_insn *ix86_expand_call (rtx, rtx, rtx, rtx, rtx, bool);
> +extern bool ix86_call_use_plt_p (rtx);
> extern void ix86_split_call_vzeroupper (rtx, rtx);
> extern void x86_initialize_trampoline (rtx, rtx, rtx);
> extern rtx ix86_zero_extend_to_Pmode (rtx);
> diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c
> index 7d0d4143bca..731a516b516 100644
> --- a/gcc/config/i386/i386.c
> +++ b/gcc/config/i386/i386.c
> @@ -11971,7 +11971,7 @@ output_pic_addr_const (FILE *file, rtx x, int code)
> assemble_name (file, name);
> }
> if (!TARGET_MACHO && !(TARGET_64BIT && TARGET_PECOFF)
> - && code == 'P' && ! SYMBOL_REF_LOCAL_P (x))
> + && code == 'P' && ix86_call_use_plt_p (x))
> fputs ("@PLT", file);
> break;
>
> @@ -15678,6 +15678,26 @@ ix86_zero_extend_to_Pmode (rtx exp)
> return force_reg (Pmode, convert_to_mode (Pmode, exp, 1));
> }
>
> +/* Return true if the function is called via PLT. */
> +
> +bool
> +ix86_call_use_plt_p (rtx call_op)
> +{
> + if (SYMBOL_REF_LOCAL_P (call_op))
> + {
> + if (SYMBOL_REF_DECL (call_op))
> + {
> + /* NB: All ifunc functions must be called via PLT. */
> + cgraph_node *node
> + = cgraph_node::get (SYMBOL_REF_DECL (call_op));
> + if (node && node->ifunc_resolver)
> + return true;
> + }
> + return false;
> + }
> + return true;
> +}
> +
> /* Return true if the function being called was marked with attribute
> "noplt" or using -fno-plt and we are compiling for non-PIC. We need
> to handle the non-PIC case in the backend because there is no easy
> @@ -24016,6 +24036,9 @@ ix86_run_selftests (void)
> #define TARGET_GET_MULTILIB_ABI_NAME \
> ix86_get_multilib_abi_name
>
> +#undef TARGET_IFUNC_REF_LOCAL_OK
> +#define TARGET_IFUNC_REF_LOCAL_OK hook_bool_void_true
> +
> static bool ix86_libc_has_fast_function (int fcode ATTRIBUTE_UNUSED)
> {
> #ifdef OPTION_GLIBC
> diff --git a/gcc/doc/tm.texi b/gcc/doc/tm.texi
> index 2a41ae5fba1..d327af03c80 100644
> --- a/gcc/doc/tm.texi
> +++ b/gcc/doc/tm.texi
> @@ -12346,6 +12346,11 @@ The support includes the assembler, linker and
> dynamic linker.
> The default value of this hook is based on target's libc.
> @end deftypefn
>
> +@deftypefn {Target Hook} bool TARGET_IFUNC_REF_LOCAL_OK (void)
> +Return true if it is OK to reference indirect function resolvers
> +locally. The default is to return false.
> +@end deftypefn
> +
> @deftypefn {Target Hook} {unsigned int} TARGET_ATOMIC_ALIGN_FOR_MODE
> (machine_mode @var{mode})
> If defined, this function returns an appropriate alignment in bits for an
> atomic object of machine_mode @var{mode}. If 0 is returned then the
> diff --git a/gcc/doc/tm.texi.in b/gcc/doc/tm.texi.in
> index f881cdabe9e..cc6f7994cb1 100644
> --- a/gcc/doc/tm.texi.in
> +++ b/gcc/doc/tm.texi.in
> @@ -8073,6 +8073,8 @@ and the associated definitions of those functions.
>
> @hook TARGET_HAS_IFUNC_P
>
> +@hook TARGET_IFUNC_REF_LOCAL_OK
> +
> @hook TARGET_ATOMIC_ALIGN_FOR_MODE
>
> @hook TARGET_ATOMIC_ASSIGN_EXPAND_FENV
> diff --git a/gcc/target.def b/gcc/target.def
> index c009671c583..2c8aa90afac 100644
> --- a/gcc/target.def
> +++ b/gcc/target.def
> @@ -2990,6 +2990,14 @@ The default value of this hook is based on target's
> libc.",
> bool, (void),
> default_has_ifunc_p)
>
> +/* True if it is OK to reference indirect function resolvers locally. */
> +DEFHOOK
> +(ifunc_ref_local_ok,
> + "Return true if it is OK to reference indirect function resolvers\n\
> +locally. The default is to return false.",
> + bool, (void),
> + hook_bool_void_false)
> +
> /* True if it is OK to do sibling call optimization for the specified
> call expression EXP. DECL will be the called function, or NULL if
> this is an indirect call. */
> diff --git a/gcc/testsuite/gcc.target/i386/pr83782-1.c
> b/gcc/testsuite/gcc.target/i386/pr83782-1.c
> new file mode 100644
> index 00000000000..f4c7370142a
> --- /dev/null
> +++ b/gcc/testsuite/gcc.target/i386/pr83782-1.c
> @@ -0,0 +1,26 @@
> +/* { dg-do compile } */
> +/* { dg-require-ifunc "" } */
> +/* { dg-options "-O2 -fpic" } */
> +
> +static void
> +my_foo (void)
> +{
> +}
> +
> +static void (*resolve_foo (void)) (void)
> +{
> + return my_foo;
> +}
> +
> +extern void foo (void) __attribute__((ifunc("resolve_foo"),
> visibility("hidden")));
> +
> +void *
> +bar(void)
> +{
> + return foo;
> +}
> +
> +/* { dg-final { scan-assembler {leal[ \t]foo@GOTOFF\(%[^,]*\),[ \t]%eax} {
> target ia32 } } } */
> +/* { dg-final { scan-assembler {leaq[ \t]foo\(%rip\),[ \t]%rax} { target { !
> ia32 } } } } */
> +/* { dg-final { scan-assembler-not "foo@GOT\\\(" { target ia32 } } } */
> +/* { dg-final { scan-assembler-not "foo@GOTPCREL\\\(" { target { ! ia32 } }
> } } */
> diff --git a/gcc/testsuite/gcc.target/i386/pr83782-2.c
> b/gcc/testsuite/gcc.target/i386/pr83782-2.c
> new file mode 100644
> index 00000000000..6c6528fff46
> --- /dev/null
> +++ b/gcc/testsuite/gcc.target/i386/pr83782-2.c
> @@ -0,0 +1,26 @@
> +/* { dg-do compile } */
> +/* { dg-require-ifunc "" } */
> +/* { dg-options "-O2 -fpic" } */
> +
> +static void
> +my_foo (void)
> +{
> +}
> +
> +static void (*resolve_foo (void)) (void)
> +{
> + return my_foo;
> +}
> +
> +static void foo (void) __attribute__((ifunc("resolve_foo")));
> +
> +void *
> +bar(void)
> +{
> + return foo;
> +}
> +
> +/* { dg-final { scan-assembler {leal[ \t]foo@GOTOFF\(%[^,]*\),[ \t]%eax} {
> target ia32 } } } */
> +/* { dg-final { scan-assembler {leaq[ \t]foo\(%rip\),[ \t]%rax} { target { !
> ia32 } } } } */
> +/* { dg-final { scan-assembler-not "foo@GOT\\\(" { target ia32 } } } */
> +/* { dg-final { scan-assembler-not "foo@GOTPCREL\\\(" { target { ! ia32 } }
> } } */
> diff --git a/gcc/varasm.c b/gcc/varasm.c
> index 53cf6dea3f3..0cba7ab0ecd 100644
> --- a/gcc/varasm.c
> +++ b/gcc/varasm.c
> @@ -7470,7 +7470,8 @@ default_binds_local_p_3 (const_tree exp, bool shlib,
> bool weak_dominate,
> FIXME: We can resolve the weakref case more curefuly by looking at the
> weakref alias. */
> if (lookup_attribute ("weakref", DECL_ATTRIBUTES (exp))
> - || (TREE_CODE (exp) == FUNCTION_DECL
> + || (!targetm.ifunc_ref_local_ok ()
> + && TREE_CODE (exp) == FUNCTION_DECL
> && cgraph_node::get (exp)
> && cgraph_node::get (exp)->ifunc_resolver))
> return false;
> --
> 2.31.1
>
PING.
--
H.J.
