diff mbox series

[V3] Eliminate power8 fusion options, use power8 tuning, PR target/102059

Message ID Yk9635GNa/fcaOqI@toto.the-meissners.org
State New
Headers show
Series [V3] Eliminate power8 fusion options, use power8 tuning, PR target/102059 | expand

Commit Message

Michael Meissner April 7, 2022, 11:59 p.m. UTC 
Eliminate power8 fusion options, use power8 tuning, PR target/102059

This is V3 of the patch.  Compared to V2 of the patch, it changed some of
the comments based on the feedback.  Since -mpower8-fusion-sign was an
undocumented option, I removed some of the wording about its removal.
I removed an older comment that was appropriate when they were written for
the power7 days, but it is not longer appropriate now.

The power8 fusion support used to be set automatically when -mcpu=power8 or
-mtune=power8 was used, and it was cleared for other cpu's.  However, if you
used the target attribute or target #pragma to change the default cpu type or
tuning, you would get an error that a target specifiction option mismatch
occurred.

This occurred because the rs6000_can_inline_p function just compares the ISA
bits between the called inline function and the caller.  If the ISA flags of
the called function is not a subset of the ISA flags of the caller, we won't do
the inlinging.  When a power9 or power10 function inlines a function that is
explicitly compiled for power8, the power8 function has the power8 fusion bits
set and the power9 or power10 functions do not have the fusion bits set.

This code removes the -mpower8-fusion option.  It also removes the
undocumented -mpower8-fusion-sign option.  It only enables power8 fusion
if we are tuning for a power8.  Power8 sign fusion is only enabled if we
are tuning for a power8 and we have -O3 optimization or higher.

I left the -mno-power8-fusion in rs6000.opt and made it so it doesn't
issue a warning.  If the user explicitly used -mpower8-fusion, then they
will get a warning that the swtich has been removed.

Similarly, I left in the pragma target and attribute target support for
power8-fusion, but using it doesn't do anything now.  This is because I
believe the customer who encountered this problem now is explicitly
setting the no-power8-fusion option in the pragma or attribute to avoid
the warning.

I have tested this patch on a little endian power10 system.  I have tested
previous versions on little endian power9 and big endian power8 systems.
Can I apply this patch to the master branch?

I will want to backport the patch to GCC 10 and GCC 11 in a few days.
Note this patch can't be used directly for those backports due to the
other changes for PR target/102059 that have gone into the master branch
but were not back ported.

2022-04-07   Michael Meissner  <meissner@linux.ibm.com>

gcc/
	PR target/102059
	* config/rs6000/rs6000-cpus.def (OTHER_FUSION_MASKS): Delete.
	(ISA_3_0_MASKS_SERVER): Don't clear the fusion masks.
	(POWERPC_MASKS): Remove OPTION_MASK_P8_FUSION.
	* config/rs6000/rs6000.cc (rs6000_option_override_internal):
	Delete code that set the power8 fusion options automatically.
	(rs6000_opt_masks): Allow #pragma target and attribute target to set
	power8-fusion, but these no longer represent an option that the
	user can set.
	(rs6000_print_options_internal): Skip printing nop options.
	* config/rs6000/rs6000.h (TARGET_P8_FUSION): New macro.
	(TARGET_P8_FUSION_SIGN): Likewise.
	(MASK_P8_FUSION): Delete.
	* config/rs6000/rs6000.opt (-mpower8-fusion): Recognize the option but
	ignore the no form and warn that the option was removed for the regular
	form.
	(-mpower8-fusion-sign): Warn that the option has been removed.
	* doc/invoke.texi (RS/6000 and PowerPC Options): Delete
	-mpower8-fusion.

gcc/testsuite/
	PR target/102059
	* gcc.dg/lto/pr102059-1_0.c: Remove -mno-power8-fusion.
	* gcc.dg/lto/pr102059-2_0.c: Likewise.
	* gcc.target/powerpc/pr102059-3.c: Likewise.
	* gcc.target/powerpc/pr102059-4.c: New test.

---
 gcc/config/rs6000/rs6000-cpus.def             | 21 +++-----
 gcc/config/rs6000/rs6000.cc                   | 48 +++++--------------
 gcc/config/rs6000/rs6000.h                    | 14 +++++-
 gcc/config/rs6000/rs6000.opt                  | 17 +++++--
 gcc/doc/invoke.texi                           | 13 +----
 gcc/testsuite/gcc.dg/lto/pr102059-1_0.c       |  2 +-
 gcc/testsuite/gcc.dg/lto/pr102059-2_0.c       |  2 +-
 gcc/testsuite/gcc.target/powerpc/pr102059-3.c |  2 +-
 gcc/testsuite/gcc.target/powerpc/pr102059-4.c | 23 +++++++++
 9 files changed, 71 insertions(+), 71 deletions(-)
 create mode 100644 gcc/testsuite/gcc.target/powerpc/pr102059-4.c

Comments

Segher Boessenkool April 11, 2022, 10:11 p.m. UTC | #1 
On Thu, Apr 07, 2022 at 07:59:27PM -0400, Michael Meissner wrote:
> I have tested this patch on a little endian power10 system.  I have tested
> previous versions on little endian power9 and big endian power8 systems.

Please test on at least p8 as well.

> I will want to backport the patch to GCC 10 and GCC 11 in a few days.
> Note this patch can't be used directly for those backports due to the
> other changes for PR target/102059 that have gone into the master branch
> but were not back ported.

Then send a new patch for there.  If you cannot backport, you cannot
backport.

> 	(rs6000_opt_masks): Allow #pragma target and attribute target to set
> 	power8-fusion, but these no longer represent an option that the
> 	user can set.

Do you have evidence that anything used this in the wild?  Nothing
should have.  Please try to remove this in GCC 13 at least.

> 	(rs6000_print_options_internal): Skip printing nop options.

It is not clear what you mean here (not in the code either).  Please
rephrase.

> 	* config/rs6000/rs6000.opt (-mpower8-fusion): Recognize the option but
> 	ignore the no form and warn that the option was removed for the regular
> 	form.

If it is ignored, you should not warn for it.

> 	(-mpower8-fusion-sign): Warn that the option has been removed.

And no one could correctly have used this ever, so don't warn anything
about this in any case.

> -/* For now, don't provide an embedded version of ISA 2.07.  Do not set power8
> -   fusion here, instead set it in rs6000.cc if we are tuning for a power8
> -   system.  */

You should keep part of this comment.  Maybe just "We don't implement an
embedded version of ISA 2.07 ."?

> +  /* The Power8 fusion option was removed.  We ignore using it in #pragma and
> +     attribute target.  Users may have used the options to suppress errors if
> +     they declare an inline function to be specifically power8 and the function
> +     was included by power9 or power10 which turned off the power8 fusion
> +     support.  */
> +  { "power8-fusion",		0,				false, true  },

This sounds very hypothetical.  Does this actually happen?  If so,
rephrase the comment; if not, remove the whole thing.

> @@ -24687,6 +24657,10 @@ rs6000_print_options_internal (FILE *file,
>        HOST_WIDE_INT mask = opts[i].mask;
>        size_t len = comma_len + prefix_len + strlen (name);
>  
> +      /* Don't print NOP options.  */
> +      if (!mask)
> +	continue;

It is not clear at all what "NOP options" means.  I can read the code
to understand the comment, but that makes the code less readable than
without comment.  Please fix this.

> +/* Power8 has special fusion operations that are enabled if we are tuning for
> +   power8.  This used to be settable with an option (-mpower8-fusion), but that
> +   option has been removed.  */
> +#define TARGET_P8_FUSION	(rs6000_tune == PROCESSOR_POWER8)

That comment does not belong here.

> +/* Power8 fusion does not fuse loads with sign extends.  If we are doing higher
> +   optimization levels, split loads with sign extension to loads with zero
> +   extension and an explicit sign extend operation, so that the zero extending
> +   load can be fused.  */
> +#define TARGET_P8_FUSION_SIGN	(TARGET_P8_FUSION			\
> +				 && optimize_function_for_speed_p (cfun) \
> +				 && optimize >= 3)

This should not depend on -O3 or sneaky ways to disable optimising for
speed.  It does not have to anyway.  Just remove this whole thing, use
TARGET_P8_FUSION instead everywhere?

> +; The -mpower8-fusion option existed in the past, but it has been removed.
> +; Some users in the past needed to use the -mno-power8-fusion option when they
> +; had inline functions that were specified as generating power8 code and the
> +; functions were included by power9 or power10 function.  Using
> +; -mno-power8-fusion prevented an error in this case.  We allow the
> +; -mno-power8-fusion option without a warning.

Please don't say all these things.  It should say what all our other
removed options say: a single half-line comment, and nothing more.

>  mpower8-fusion
> -Target Mask(P8_FUSION) Var(rs6000_isa_flags)
> -Fuse certain integer operations together for better performance on power8.
> +Target RejectNegative Undocumented WarnRemoved

> +mno-power8-fusion
> +Target RejectNegative Undocumented

It also should have "Ignore".

The "no-" variant (the always active one!) goes first.  Both share that
one short comment.

> +; The -mpower8-fusion-sign option was an undocumented option that modified the
> +; -mpower8-fusion option.  It has been removed.
>  mpower8-fusion-sign
> -Target Undocumented Mask(P8_FUSION_SIGN) Var(rs6000_isa_flags)
> -Allow sign extension in fusion operations.
> +Target Undocumented WarnRemoved

Don't WarnRemoved.  Do Ignore.  Just remove the whole thing please.

> --- /dev/null
> +++ b/gcc/testsuite/gcc.target/powerpc/pr102059-4.c
> @@ -0,0 +1,23 @@
> +/* { dg-do compile } */
> +/* { dg-options "-O2 -mdejagnu-cpu=power10" } */
> +/* { dg-require-effective-target power10_ok } */

These last two lines the other way around: first state requirements, and
only then the options that those are for.

Please resend with those things fixed.  Thanks!


Segher
diff mbox series

Patch

diff --git a/gcc/config/rs6000/rs6000-cpus.def b/gcc/config/rs6000/rs6000-cpus.def
index 963947f6939..9db0c62a3ee 100644
--- a/gcc/config/rs6000/rs6000-cpus.def
+++ b/gcc/config/rs6000/rs6000-cpus.def
@@ -43,9 +43,6 @@ 
 				 | OPTION_MASK_ALTIVEC			\
 				 | OPTION_MASK_VSX)
 
-/* For now, don't provide an embedded version of ISA 2.07.  Do not set power8
-   fusion here, instead set it in rs6000.cc if we are tuning for a power8
-   system.  */
 #define ISA_2_7_MASKS_SERVER	(ISA_2_6_MASKS_SERVER			\
 				 | OPTION_MASK_P8_VECTOR		\
 				 | OPTION_MASK_CRYPTO			\
@@ -54,19 +51,14 @@ 
 				 | OPTION_MASK_QUAD_MEMORY		\
 				 | OPTION_MASK_QUAD_MEMORY_ATOMIC)
 
-/* ISA masks setting fusion options.  */
-#define OTHER_FUSION_MASKS	(OPTION_MASK_P8_FUSION			\
-				 | OPTION_MASK_P8_FUSION_SIGN)
-
 /* Add ISEL back into ISA 3.0, since it is supposed to be a win.  Do not add
    FLOAT128_HW here until we are ready to make -mfloat128 on by default.  */
-#define ISA_3_0_MASKS_SERVER	((ISA_2_7_MASKS_SERVER			\
-				  | OPTION_MASK_ISEL			\
-				  | OPTION_MASK_MODULO			\
-				  | OPTION_MASK_P9_MINMAX		\
-				  | OPTION_MASK_P9_MISC			\
-				  | OPTION_MASK_P9_VECTOR)		\
-				 & ~OTHER_FUSION_MASKS)
+#define ISA_3_0_MASKS_SERVER	(ISA_2_7_MASKS_SERVER			\
+				 | OPTION_MASK_ISEL			\
+				 | OPTION_MASK_MODULO			\
+				 | OPTION_MASK_P9_MINMAX		\
+				 | OPTION_MASK_P9_MISC			\
+				 | OPTION_MASK_P9_VECTOR)
 
 /* Support for the IEEE 128-bit floating point hardware requires a lot of the
    VSX instructions that are part of ISA 3.0.  */
@@ -140,7 +132,6 @@ 
 				 | OPTION_MASK_MODULO			\
 				 | OPTION_MASK_MULHW			\
 				 | OPTION_MASK_NO_UPDATE		\
-				 | OPTION_MASK_P8_FUSION		\
 				 | OPTION_MASK_P8_VECTOR		\
 				 | OPTION_MASK_P9_MINMAX		\
 				 | OPTION_MASK_P9_MISC			\
diff --git a/gcc/config/rs6000/rs6000.cc b/gcc/config/rs6000/rs6000.cc
index ceaddafd33b..fac81f0cc93 100644
--- a/gcc/config/rs6000/rs6000.cc
+++ b/gcc/config/rs6000/rs6000.cc
@@ -4040,41 +4040,6 @@  rs6000_option_override_internal (bool global_init_p)
       && optimize_function_for_speed_p (cfun))
     rs6000_isa_flags |= OPTION_MASK_SAVE_TOC_INDIRECT;
 
-  /* Enable power8 fusion if we are tuning for power8, even if we aren't
-     generating power8 instructions.  Power9 does not optimize power8 fusion
-     cases.  */
-  if (!(rs6000_isa_flags_explicit & OPTION_MASK_P8_FUSION))
-    {
-      if (processor_target_table[tune_index].processor == PROCESSOR_POWER8)
-	rs6000_isa_flags |= OPTION_MASK_P8_FUSION;
-      else
-	rs6000_isa_flags &= ~OPTION_MASK_P8_FUSION;
-    }
-
-  /* Setting additional fusion flags turns on base fusion.  */
-  if (!TARGET_P8_FUSION && TARGET_P8_FUSION_SIGN)
-    {
-      if (rs6000_isa_flags_explicit & OPTION_MASK_P8_FUSION)
-	{
-	  if (TARGET_P8_FUSION_SIGN)
-	    error ("%qs requires %qs", "-mpower8-fusion-sign",
-		   "-mpower8-fusion");
-
-	  rs6000_isa_flags &= ~OPTION_MASK_P8_FUSION;
-	}
-      else
-	rs6000_isa_flags |= OPTION_MASK_P8_FUSION;
-    }
-
-  /* Power8 does not fuse sign extended loads with the addis.  If we are
-     optimizing at high levels for speed, convert a sign extended load into a
-     zero extending load, and an explicit sign extension.  */
-  if (TARGET_P8_FUSION
-      && !(rs6000_isa_flags_explicit & OPTION_MASK_P8_FUSION_SIGN)
-      && optimize_function_for_speed_p (cfun)
-      && optimize >= 3)
-    rs6000_isa_flags |= OPTION_MASK_P8_FUSION_SIGN;
-
   /* ISA 3.0 vector instructions include ISA 2.07.  */
   if (TARGET_P9_VECTOR && !TARGET_P8_VECTOR)
     {
@@ -24010,8 +23975,6 @@  static struct rs6000_opt_mask const rs6000_opt_masks[] =
   { "pcrel-opt",		OPTION_MASK_PCREL_OPT,		false, true  },
   { "popcntb",			OPTION_MASK_POPCNTB,		false, true  },
   { "popcntd",			OPTION_MASK_POPCNTD,		false, true  },
-  { "power8-fusion",		OPTION_MASK_P8_FUSION,		false, true  },
-  { "power8-fusion-sign",	OPTION_MASK_P8_FUSION_SIGN,	false, true  },
   { "power8-vector",		OPTION_MASK_P8_VECTOR,		false, true  },
   { "power9-minmax",		OPTION_MASK_P9_MINMAX,		false, true  },
   { "power9-misc",		OPTION_MASK_P9_MISC,		false, true  },
@@ -24051,6 +24014,13 @@  static struct rs6000_opt_mask const rs6000_opt_masks[] =
 #endif
   { "soft-float",		OPTION_MASK_SOFT_FLOAT,		false, false },
   { "string",			0,				false, false },
+
+  /* The Power8 fusion option was removed.  We ignore using it in #pragma and
+     attribute target.  Users may have used the options to suppress errors if
+     they declare an inline function to be specifically power8 and the function
+     was included by power9 or power10 which turned off the power8 fusion
+     support.  */
+  { "power8-fusion",		0,				false, true  },
 };
 
 /* Builtin mask mapping for printing the flags.  */
@@ -24687,6 +24657,10 @@  rs6000_print_options_internal (FILE *file,
       HOST_WIDE_INT mask = opts[i].mask;
       size_t len = comma_len + prefix_len + strlen (name);
 
+      /* Don't print NOP options.  */
+      if (!mask)
+	continue;
+
       if (!invert)
 	{
 	  if ((flags & mask) == 0)
diff --git a/gcc/config/rs6000/rs6000.h b/gcc/config/rs6000/rs6000.h
index 523256a5c9d..f89f215d57a 100644
--- a/gcc/config/rs6000/rs6000.h
+++ b/gcc/config/rs6000/rs6000.h
@@ -504,6 +504,19 @@  extern int rs6000_vector_align[];
 #define TARGET_MINMAX	(TARGET_HARD_FLOAT && TARGET_PPC_GFXOPT		\
 			 && (TARGET_P9_MINMAX || !flag_trapping_math))
 
+/* Power8 has special fusion operations that are enabled if we are tuning for
+   power8.  This used to be settable with an option (-mpower8-fusion), but that
+   option has been removed.  */
+#define TARGET_P8_FUSION	(rs6000_tune == PROCESSOR_POWER8)
+
+/* Power8 fusion does not fuse loads with sign extends.  If we are doing higher
+   optimization levels, split loads with sign extension to loads with zero
+   extension and an explicit sign extend operation, so that the zero extending
+   load can be fused.  */
+#define TARGET_P8_FUSION_SIGN	(TARGET_P8_FUSION			\
+				 && optimize_function_for_speed_p (cfun) \
+				 && optimize >= 3)
+
 /* In switching from using target_flags to using rs6000_isa_flags, the options
    machinery creates OPTION_MASK_<xxx> instead of MASK_<xxx>.  For now map
    OPTION_MASK_<xxx> back into MASK_<xxx>.  */
@@ -517,7 +530,6 @@  extern int rs6000_vector_align[];
 #define MASK_FLOAT128_KEYWORD		OPTION_MASK_FLOAT128_KEYWORD
 #define MASK_FLOAT128_HW		OPTION_MASK_FLOAT128_HW
 #define MASK_FPRND			OPTION_MASK_FPRND
-#define MASK_P8_FUSION			OPTION_MASK_P8_FUSION
 #define MASK_HARD_FLOAT			OPTION_MASK_HARD_FLOAT
 #define MASK_HTM			OPTION_MASK_HTM
 #define MASK_ISEL			OPTION_MASK_ISEL
diff --git a/gcc/config/rs6000/rs6000.opt b/gcc/config/rs6000/rs6000.opt
index 4931d781c4e..95ef25d48c6 100644
--- a/gcc/config/rs6000/rs6000.opt
+++ b/gcc/config/rs6000/rs6000.opt
@@ -474,13 +474,22 @@  Save the TOC in the prologue for indirect calls rather than inline.
 mvsx-timode
 Target RejectNegative Undocumented Ignore
 
+; The -mpower8-fusion option existed in the past, but it has been removed.
+; Some users in the past needed to use the -mno-power8-fusion option when they
+; had inline functions that were specified as generating power8 code and the
+; functions were included by power9 or power10 function.  Using
+; -mno-power8-fusion prevented an error in this case.  We allow the
+; -mno-power8-fusion option without a warning.
 mpower8-fusion
-Target Mask(P8_FUSION) Var(rs6000_isa_flags)
-Fuse certain integer operations together for better performance on power8.
+Target RejectNegative Undocumented WarnRemoved
+
+mno-power8-fusion
+Target RejectNegative Undocumented
 
+; The -mpower8-fusion-sign option was an undocumented option that modified the
+; -mpower8-fusion option.  It has been removed.
 mpower8-fusion-sign
-Target Undocumented Mask(P8_FUSION_SIGN) Var(rs6000_isa_flags)
-Allow sign extension in fusion operations.
+Target Undocumented WarnRemoved
 
 mpower8-vector
 Target Mask(P8_VECTOR) Var(rs6000_isa_flags)
diff --git a/gcc/doc/invoke.texi b/gcc/doc/invoke.texi
index 1a51759e6e4..7c21779e3b2 100644
--- a/gcc/doc/invoke.texi
+++ b/gcc/doc/invoke.texi
@@ -1266,8 +1266,7 @@  See RS/6000 and PowerPC Options.
 -mveclibabi=@var{type}  -mfriz  -mno-friz @gol
 -mpointers-to-nested-functions  -mno-pointers-to-nested-functions @gol
 -msave-toc-indirect  -mno-save-toc-indirect @gol
--mpower8-fusion  -mno-mpower8-fusion  -mpower8-vector  -mno-power8-vector @gol
--mcrypto  -mno-crypto  -mhtm  -mno-htm @gol
+-mpower8-vector  -mno-power8-vector -mcrypto  -mno-crypto  -mhtm  -mno-htm @gol
 -mquad-memory  -mno-quad-memory @gol
 -mquad-memory-atomic  -mno-quad-memory-atomic @gol
 -mcompat-align-parm  -mno-compat-align-parm @gol
@@ -28355,7 +28354,7 @@  following options:
 -mpopcntb  -mpopcntd  -mpowerpc64 @gol
 -mpowerpc-gpopt  -mpowerpc-gfxopt @gol
 -mmulhw  -mdlmzb  -mmfpgpr  -mvsx @gol
--mcrypto  -mhtm  -mpower8-fusion  -mpower8-vector @gol
+-mcrypto  -mhtm  -mpower8-vector @gol
 -mquad-memory  -mquad-memory-atomic  -mfloat128 @gol
 -mfloat128-hardware -mprefixed -mpcrel -mmma @gol
 -mrop-protect}
@@ -28470,14 +28469,6 @@  Enable (disable) the use of the built-in functions that allow direct
 access to the Hardware Transactional Memory (HTM) instructions that
 were added in version 2.07 of the PowerPC ISA.
 
-@item -mpower8-fusion
-@itemx -mno-power8-fusion
-@opindex mpower8-fusion
-@opindex mno-power8-fusion
-Generate code that keeps (does not keeps) some integer operations
-adjacent so that the instructions can be fused together on power8 and
-later processors.
-
 @item -mpower8-vector
 @itemx -mno-power8-vector
 @opindex mpower8-vector
diff --git a/gcc/testsuite/gcc.dg/lto/pr102059-1_0.c b/gcc/testsuite/gcc.dg/lto/pr102059-1_0.c
index e1004f5cfbf..b215b701097 100644
--- a/gcc/testsuite/gcc.dg/lto/pr102059-1_0.c
+++ b/gcc/testsuite/gcc.dg/lto/pr102059-1_0.c
@@ -1,7 +1,7 @@ 
 /* { dg-lto-do link } */
 /* { dg-skip-if "power10 and above only" { ! { power10_ok } } } */
 /* -Wno-attributes suppresses always_inline warnings.  */
-/* { dg-lto-options { "-O2 -mdejagnu-cpu=power8 -flto -Wno-attributes -mno-power8-fusion" } } */
+/* { dg-lto-options { "-O2 -mdejagnu-cpu=power8 -flto -Wno-attributes" } } */
 
 int __attribute__ ((always_inline))
 foo1 (int *b)
diff --git a/gcc/testsuite/gcc.dg/lto/pr102059-2_0.c b/gcc/testsuite/gcc.dg/lto/pr102059-2_0.c
index ebb0a7b8fa1..ffab23ab7e1 100644
--- a/gcc/testsuite/gcc.dg/lto/pr102059-2_0.c
+++ b/gcc/testsuite/gcc.dg/lto/pr102059-2_0.c
@@ -1,6 +1,6 @@ 
 /* { dg-lto-do link } */
 /* { dg-skip-if "power10 and above only" { ! { power10_ok } } } */
-/* { dg-lto-options { "-O2 -mdejagnu-cpu=power8 -mno-power8-fusion -flto -fdump-ipa-inline" } } */
+/* { dg-lto-options { "-O2 -mdejagnu-cpu=power8 -flto -fdump-ipa-inline" } } */
 
 int
 foo1 (int *b)
diff --git a/gcc/testsuite/gcc.target/powerpc/pr102059-3.c b/gcc/testsuite/gcc.target/powerpc/pr102059-3.c
index 21c982d93f0..0cb3a4cb9f9 100644
--- a/gcc/testsuite/gcc.target/powerpc/pr102059-3.c
+++ b/gcc/testsuite/gcc.target/powerpc/pr102059-3.c
@@ -1,5 +1,5 @@ 
 /* { dg-do compile } */
-/* { dg-options "-O2 -mdejagnu-cpu=power8 -mno-power8-fusion -fdump-tree-einline-optimized" } */
+/* { dg-options "-O2 -mdejagnu-cpu=power8 -fdump-tree-einline-optimized" } */
 
 /* Like pr102059-1.c, to verify the inlining still happens
    even without always_inline attribute.  */
diff --git a/gcc/testsuite/gcc.target/powerpc/pr102059-4.c b/gcc/testsuite/gcc.target/powerpc/pr102059-4.c
new file mode 100644
index 00000000000..5fe66f8af4b
--- /dev/null
+++ b/gcc/testsuite/gcc.target/powerpc/pr102059-4.c
@@ -0,0 +1,23 @@ 
+/* { dg-do compile } */
+/* { dg-options "-O2 -mdejagnu-cpu=power10" } */
+/* { dg-require-effective-target power10_ok } */
+
+/* Verify that power10 can explicity include functions compiled for power8.
+   The issue was -mcpu=power8 enables -mpower8-fusion, but -mcpu=power9 or
+   -mcpu=power10 do not set power8-fusion by default.  Thus when doing this
+   compilation, they would get an error that the inline function failed in its
+   inlining due to having incompatible options.  */
+
+static inline int __attribute__ ((always_inline,target("cpu=power8")))
+foo (int *b)
+{
+  *b += 10;
+  return *b;
+}
+
+int
+bar (int *a)
+{
+  *a = foo (a);
+  return 0;
+}