[PATCH], Add PowerPC ISA 3.0 vector d-form addressing
Michael Meissner
meissner@linux.vnet.ibm.com
Thu May 5 18:05:00 GMT 2016
On Wed, May 04, 2016 at 11:16:52AM -0500, Segher Boessenkool wrote:
> Hi Mike,
>
> On Tue, May 03, 2016 at 06:39:55PM -0400, Michael Meissner wrote:
> > With this patch, I enable -mlra if the user did not specify either -mlra or
> > -mno-lra on the command line, and -mcpu=power9 or -mpower9-dform-vector were
> > used. I also enabled -mvsx-timode if LRA was used, which also is a RELOAD
> > issue, that works with LRA.
>
> I don't like enabling LRA if the user didn't ask for it; it is a bit too
> surprising. What do you do if there is -mno-lra explicitly? You can just
> do the same if no-lra is implicit?
Ok.
> > * doc/md.texi (wO constraint): Likewise.
>
> Everything is "likewise", that isn't very helpful. Writing big changelogs
> is annoying, I totally agree, but please try a bit harder.
>
> > --- gcc/config/rs6000/rs6000.opt (.../svn+ssh://meissner@gcc.gnu.org/svn/gcc/trunk/gcc/config/rs6000) (revision 235831)
> > +++ gcc/config/rs6000/rs6000.opt (.../gcc/config/rs6000) (working copy)
> > @@ -470,8 +470,8 @@ Target RejectNegative Joined UInteger Va
> > -mlong-double-<n> Specify size of long double (64 or 128 bits).
> >
> > mlra
> > -Target Report Var(rs6000_lra_flag) Init(0) Save
> > -Use LRA instead of reload.
> > +Target Undocumented Mask(LRA) Var(rs6000_isa_flags)
> > +Use the LRA register allocator instead of the reload register allocator.
>
> It wasn't "undocumented" before? Why the change to a mask bit btw?
It was always meant to be undocumented, but I changed to be similar to
before. I am trying to change all of the random switches that set a word to be
an option mask, so I made that part of the change in these next patches. I did
remove setting it for -mcpu=power9.
> > +mpower9-dform-scalar
> > +Target Report Mask(P9_DFORM_SCALAR) Var(rs6000_isa_flags)
> > +Use/do not use scalar register+offset memory instructions added in ISA 3.0.
> > +
> > +mpower9-dform-vector
> > +Target Report Mask(P9_DFORM_VECTOR) Var(rs6000_isa_flags)
> > +Use/do not use vector register+offset memory instructions added in ISA 3.0.
> > +
> > mpower9-dform
> > -Target Undocumented Mask(P9_DFORM) Var(rs6000_isa_flags)
> > -Use/do not use vector and scalar instructions added in ISA 3.0.
> > +Target Report Var(TARGET_P9_DFORM_BOTH) Init(-1) Save
> > +Use/do not use register+offset memory instructions added in ISA 3.0.
>
> These should probably all be undocumented, though (they're not something
> users should use).
I will make -mpower9-dform public (which I thought it was, but evidently I
missed adding the documentation for GCC 6), but I will make the -scalar and
-vector forms private.
> > +/* Return true if the ADDR is an acceptiable address for a quad memory
> ^ spelling
Ok.
> > + if (((addr_mask & RELOAD_REG_QUAD_OFFSET) == 0)
> > + || !quad_address_p (addr, mode, false))
>
Here is the latest version of the patch. Like before, it bootstraps and has no
regressions. Is it ok to apply to the trunk?
[gcc]
2016-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
* config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Use
-mpower9-dform-scalar instead of -mpower9-dform. Add note not to
include -mpower9-dform-vector until we switch over to LRA.
(POWERPC_MASKS): Add -mlra. Split -mpower9-dform into two
switches, -mpower9-dform-scalar and -mpower9-dform-vector.
* config/rs6000/rs6000.opt (-mlra): Switch to being an option mask
bit instead of being a separate word. Split -mpower9-dform into
two switches, -mpower9-dform-scalar and -mpower9-dform-vector.
* config/rs6000/rs6000.c (RELOAD_REG_QUAD_OFFSET): New addr_mask
for the register class supporting 128-bit quad word memory
offsets.
(mode_supports_vsx_dform_quad): Helper function to return if the
register class uses quad word memory offsets.
(rs6000_debug_addr_mask): Add support for quad word memory
offsets.
(rs6000_debug_reg_global): Use TARGET_LRA instead of calling the
lra_p target hook.
(rs6000_setup_reg_addr_masks): If ISA 3.0 vector d-form
instructions are enabled, set up the appropriate addr_masks for
128-bit types.
(rs6000_init_hard_regno_mode_ok): wb constraint is now based on
-mpower9-dform-scalar, instead of -mpower9-dform.
(rs6000_option_override_internal): Split -mpower9-dform into two
switches, -mpower9-dform-scalar and -mpower9-dform-vector. The
-mpower9-dform switch sets or clears both. If we are not using the
LRA register allocator, do not enable -mpower9-dform-vector by
default. If we are using LRA, enable -mpower9-dform-vector and
-mvsx-timode if it is appropriate. Issue a warning if either
-mpower9-dform-vector or -mvsx-timode are explicitly used without
enabling LRA.
(quad_address_offset_p): New helper function to return if the
offset is legal for quad word memory instructions.
(quad_address_p): New function to determin if GPR or vector
register quad word memory addresses are legal.
(mem_operand_gpr): Validate quad word address offsets.
(reg_offset_addressing_ok_p): Add support for ISA 3.0 vector
d-form (register + offset) instructions.
(offsettable_ok_by_alignment): Likewise.
(rs6000_legitimate_offset_address_p): Likewise.
(legitimate_lo_sum_address_p): Likewise.
(rs6000_legitimize_address): Likewise.
(rs6000_legitimize_reload_address): Add more debug statements for
-mdebug=addr.
(rs6000_legitimate_address_p): Add support for ISA 3.0 vector
d-form instructions.
(rs6000_secondary_reload_memory): Add support for ISA 3.0 vector
d-form instructions. Distinguish different cases in debug output.
(rs6000_secondary_reload_inner): Add support for ISA 3.0 vector
d-form instructions.
(rs6000_preferred_reload_class): Likewise.
(rs6000_output_move_128bit): Add support for ISA 3.0 d-form
instructions. If ISA 3.0 is available, generate lxvx/stxvx instead
of the ISA 2.06 indexed memory instructions.
(rs6000_emit_prologue): If we have ISA 3.0 d-form instructions,
use them to save/restore the saved vector registers instead of
using Altivec instructions.
(rs6000_emit_epilogue): Likewise.
(rs6000_lra_p): Use TARGET_LRA instead of the old option word.
(rs6000_opt_masks): Split -mpower9-dform into
-mpower9-dform-scalar and -mpower9-dform-vector.
(rs6000_print_options_internal): Print -mno-<switch> if <switch>
was not selected.
* config/rs6000/constraints.md (wO constraint): New constraint for
ISA 3.0 vector d-form support.
* config/rs6000/predicates.md (quad_memory_operand): Move most of
the code into quad_address_p and call it to share code with
vsx_quad_dform_memory_operand.
(vsx_quad_dform_memory_operand): New predicate for ISA 3.0 vector
d-form support.
* config/rs6000/rs6000-protos.h (quad_address_p): Add declaration.
* config/rs6000/rs6000.md (p9_vecload_<mode>): Delete hack to emit
ISA 3.0 vector indexed memory instructions, and fold the code into
the normal mov<mode> patterns.
(p9_vecstore_<mode>): Likewise.
(vsx_mov<mode>): Add support for ISA 3.0 vector d-form
instructions.
(vsx_movti_64bit): Likewise.
(vsx_movti_32bit): Likewise.
* doc/invoke.texi (RS/6000 and PowerPC Options): Add documentation
for -mpower9-dform.
* doc/md.texi (wO constraint): Document wO constraint.
[gcc/testsuite]
2016-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
* gcc.target/powerpc/p8vector-int128-1.c: Add -mlra to silence
warning when using -mvsx-timode.
* gcc.target/powerpc/pr68805.c: Likewise.
* gcc.target/powerpc/dform-3.c: New test for ISA 3.0 vector d-form
support.
* gcc.target/powerpc/dform-1.c: Add -mlra option.
* gcc.target/powerpc/dform-2.c: Likewise.
--
Michael Meissner, IBM
IBM, M/S 2506R, 550 King Street, Littleton, MA 01460-6245, USA
email: meissner@linux.vnet.ibm.com, phone: +1 (978) 899-4797
-------------- next part --------------
Index: gcc/config/rs6000/rs6000-cpus.def
===================================================================
--- gcc/config/rs6000/rs6000-cpus.def (.../svn+ssh://meissner@gcc.gnu.org/svn/gcc/trunk/gcc/config/rs6000) (revision 235831)
+++ gcc/config/rs6000/rs6000-cpus.def (.../gcc/config/rs6000) (working copy)
@@ -60,13 +60,14 @@
| OPTION_MASK_UPPER_REGS_SF)
/* Add ISEL back into ISA 3.0, since it is supposed to be a win. Do not add
- P9_DFORM or P9_MINMAX until they are fully debugged. */
+ P9_MINMAX until the hardware that supports it is available. Do not add
+ P9_DFORM_VECTOR until LRA is the default register allocator. */
#define ISA_3_0_MASKS_SERVER (ISA_2_7_MASKS_SERVER \
| OPTION_MASK_FLOAT128_HW \
| OPTION_MASK_ISEL \
| OPTION_MASK_MODULO \
| OPTION_MASK_P9_FUSION \
- | OPTION_MASK_P9_DFORM \
+ | OPTION_MASK_P9_DFORM_SCALAR \
| OPTION_MASK_P9_VECTOR)
#define POWERPC_7400_MASK (OPTION_MASK_PPC_GFXOPT | OPTION_MASK_ALTIVEC)
@@ -94,6 +95,7 @@
| OPTION_MASK_FPRND \
| OPTION_MASK_HTM \
| OPTION_MASK_ISEL \
+ | OPTION_MASK_LRA \
| OPTION_MASK_MFCRF \
| OPTION_MASK_MFPGPR \
| OPTION_MASK_MODULO \
@@ -101,7 +103,8 @@
| OPTION_MASK_NO_UPDATE \
| OPTION_MASK_P8_FUSION \
| OPTION_MASK_P8_VECTOR \
- | OPTION_MASK_P9_DFORM \
+ | OPTION_MASK_P9_DFORM_SCALAR \
+ | OPTION_MASK_P9_DFORM_VECTOR \
| OPTION_MASK_P9_FUSION \
| OPTION_MASK_P9_MINMAX \
| OPTION_MASK_P9_VECTOR \
Index: gcc/config/rs6000/rs6000.opt
===================================================================
--- gcc/config/rs6000/rs6000.opt (.../svn+ssh://meissner@gcc.gnu.org/svn/gcc/trunk/gcc/config/rs6000) (revision 235831)
+++ gcc/config/rs6000/rs6000.opt (.../gcc/config/rs6000) (working copy)
@@ -470,7 +470,7 @@ Target RejectNegative Joined UInteger Va
-mlong-double-<n> Specify size of long double (64 or 128 bits).
mlra
-Target Report Var(rs6000_lra_flag) Init(0) Save
+Target Report Mask(LRA) Var(rs6000_isa_flags)
Use LRA instead of reload.
msched-costly-dep=
@@ -609,9 +609,17 @@ mpower9-vector
Target Report Mask(P9_VECTOR) Var(rs6000_isa_flags)
Use/do not use vector and scalar instructions added in ISA 3.0.
+mpower9-dform-scalar
+Target Undocumented Mask(P9_DFORM_SCALAR) Var(rs6000_isa_flags)
+Use/do not use scalar register+offset memory instructions added in ISA 3.0.
+
+mpower9-dform-vector
+Target Undocumented Mask(P9_DFORM_VECTOR) Var(rs6000_isa_flags)
+Use/do not use vector register+offset memory instructions added in ISA 3.0.
+
mpower9-dform
-Target Undocumented Mask(P9_DFORM) Var(rs6000_isa_flags)
-Use/do not use vector and scalar instructions added in ISA 3.0.
+Target Report Var(TARGET_P9_DFORM_BOTH) Init(-1) Save
+Use/do not use register+offset memory instructions added in ISA 3.0.
mpower9-minmax
Target Undocumented Mask(P9_MINMAX) Var(rs6000_isa_flags)
Index: gcc/config/rs6000/rs6000.c
===================================================================
--- gcc/config/rs6000/rs6000.c (.../svn+ssh://meissner@gcc.gnu.org/svn/gcc/trunk/gcc/config/rs6000) (revision 235831)
+++ gcc/config/rs6000/rs6000.c (.../gcc/config/rs6000) (working copy)
@@ -452,6 +452,7 @@ typedef unsigned char addr_mask_type;
#define RELOAD_REG_PRE_INCDEC 0x10 /* PRE_INC/PRE_DEC valid. */
#define RELOAD_REG_PRE_MODIFY 0x20 /* PRE_MODIFY valid. */
#define RELOAD_REG_AND_M16 0x40 /* AND -16 addressing. */
+#define RELOAD_REG_QUAD_OFFSET 0x80 /* quad offset is limited. */
/* Register type masks based on the type, of valid addressing modes. */
struct rs6000_reg_addr {
@@ -499,6 +500,16 @@ mode_supports_vmx_dform (machine_mode mo
return ((reg_addr[mode].addr_mask[RELOAD_REG_VMX] & RELOAD_REG_OFFSET) != 0);
}
+/* Return true if we have D-form addressing in VSX registers. This addressing
+ is more limited than normal d-form addressing in that the offset must be
+ aligned on a 16-byte boundary. */
+static inline bool
+mode_supports_vsx_dform_quad (machine_mode mode)
+{
+ return ((reg_addr[mode].addr_mask[RELOAD_REG_ANY] & RELOAD_REG_QUAD_OFFSET)
+ != 0);
+}
+
/* Target cpu costs. */
@@ -2108,7 +2119,9 @@ rs6000_debug_addr_mask (addr_mask_type m
else if (keep_spaces)
*p++ = ' ';
- if ((mask & RELOAD_REG_OFFSET) != 0)
+ if ((mask & RELOAD_REG_QUAD_OFFSET) != 0)
+ *p++ = 'O';
+ else if ((mask & RELOAD_REG_OFFSET) != 0)
*p++ = 'o';
else if (keep_spaces)
*p++ = ' ';
@@ -2645,7 +2658,7 @@ rs6000_debug_reg_global (void)
if (TARGET_LINK_STACK)
fprintf (stderr, DEBUG_FMT_S, "link_stack", "true");
- if (targetm.lra_p ())
+ if (TARGET_LRA)
fprintf (stderr, DEBUG_FMT_S, "lra", "true");
if (TARGET_P8_FUSION)
@@ -2781,17 +2794,31 @@ rs6000_setup_reg_addr_masks (void)
}
/* GPR and FPR registers can do REG+OFFSET addressing, except
- possibly for SDmode. ISA 3.0 (i.e. power9) adds D-form
- addressing for scalars to altivec registers. */
+ possibly for SDmode. ISA 3.0 (i.e. power9) adds D-form addressing
+ for 64-bit scalars and 32-bit SFmode to altivec registers. */
if ((addr_mask != 0) && !indexed_only_p
&& msize <= 8
&& (rc == RELOAD_REG_GPR
- || rc == RELOAD_REG_FPR
- || (rc == RELOAD_REG_VMX
- && TARGET_P9_DFORM
- && (m2 == DFmode || m2 == SFmode))))
+ || ((msize == 8 || m2 == SFmode)
+ && (rc == RELOAD_REG_FPR
+ || (rc == RELOAD_REG_VMX
+ && TARGET_P9_DFORM_SCALAR)))))
addr_mask |= RELOAD_REG_OFFSET;
+ /* VSX registers can do REG+OFFSET addresssing if ISA 3.0
+ instructions are enabled. The offset for 128-bit VSX registers is
+ only 12-bits. While GPRs can handle the full offset range, VSX
+ registers can only handle the restricted range. */
+ else if ((addr_mask != 0) && !indexed_only_p
+ && msize == 16 && TARGET_P9_DFORM_VECTOR
+ && (ALTIVEC_OR_VSX_VECTOR_MODE (m2)
+ || (m2 == TImode && TARGET_VSX_TIMODE)))
+ {
+ addr_mask |= RELOAD_REG_OFFSET;
+ if (rc == RELOAD_REG_FPR || rc == RELOAD_REG_VMX)
+ addr_mask |= RELOAD_REG_QUAD_OFFSET;
+ }
+
/* VMX registers can do (REG & -16) and ((REG+REG) & -16)
addressing on 128-bit types. */
if (rc == RELOAD_REG_VMX && msize == 16
@@ -3114,7 +3141,7 @@ rs6000_init_hard_regno_mode_ok (bool glo
}
/* Support for new D-form instructions. */
- if (TARGET_P9_DFORM)
+ if (TARGET_P9_DFORM_SCALAR)
rs6000_constraints[RS6000_CONSTRAINT_wb] = ALTIVEC_REGS;
/* Support for ISA 3.0 (power9) vectors. */
@@ -3987,7 +4014,8 @@ rs6000_option_override_internal (bool gl
/* For the newer switches (vsx, dfp, etc.) set some of the older options,
unless the user explicitly used the -mno-<option> to disable the code. */
- if (TARGET_P9_VECTOR || TARGET_MODULO || TARGET_P9_DFORM || TARGET_P9_MINMAX)
+ if (TARGET_P9_VECTOR || TARGET_MODULO || TARGET_P9_DFORM_SCALAR
+ || TARGET_P9_DFORM_VECTOR || TARGET_P9_DFORM_BOTH > 0 || TARGET_P9_MINMAX)
rs6000_isa_flags |= (ISA_3_0_MASKS_SERVER & ~rs6000_isa_flags_explicit);
else if (TARGET_P8_VECTOR || TARGET_DIRECT_MOVE || TARGET_CRYPTO)
rs6000_isa_flags |= (ISA_2_7_MASKS_SERVER & ~rs6000_isa_flags_explicit);
@@ -4201,26 +4229,49 @@ rs6000_option_override_internal (bool gl
&& !(rs6000_isa_flags_explicit & OPTION_MASK_TOC_FUSION))
rs6000_isa_flags |= OPTION_MASK_TOC_FUSION;
+ /* -mpower9-dform turns on both -mpower9-dform-scalar and
+ -mpower9-dform-vector. There are currently problems if
+ -mpower9-dform-vector instructions are enabled when we use the RELOAD
+ register allocator. */
+ if (TARGET_P9_DFORM_BOTH > 0)
+ {
+ if (!(rs6000_isa_flags_explicit & OPTION_MASK_P9_DFORM_VECTOR)
+ && TARGET_LRA)
+ rs6000_isa_flags |= OPTION_MASK_P9_DFORM_VECTOR;
+
+ if (!(rs6000_isa_flags_explicit & OPTION_MASK_P9_DFORM_SCALAR))
+ rs6000_isa_flags |= OPTION_MASK_P9_DFORM_SCALAR;
+ }
+ else if (TARGET_P9_DFORM_BOTH == 0)
+ {
+ if (!(rs6000_isa_flags_explicit & OPTION_MASK_P9_DFORM_VECTOR))
+ rs6000_isa_flags &= ~OPTION_MASK_P9_DFORM_VECTOR;
+
+ if (!(rs6000_isa_flags_explicit & OPTION_MASK_P9_DFORM_SCALAR))
+ rs6000_isa_flags &= ~OPTION_MASK_P9_DFORM_SCALAR;
+ }
+
/* ISA 3.0 D-form instructions require p9-vector and upper-regs. */
- if (TARGET_P9_DFORM && !TARGET_P9_VECTOR)
+ if ((TARGET_P9_DFORM_SCALAR || TARGET_P9_DFORM_VECTOR) && !TARGET_P9_VECTOR)
{
if (rs6000_isa_flags_explicit & OPTION_MASK_P9_VECTOR)
error ("-mpower9-dform requires -mpower9-vector");
- rs6000_isa_flags &= ~OPTION_MASK_P9_DFORM;
+ rs6000_isa_flags &= ~(OPTION_MASK_P9_DFORM_SCALAR
+ | OPTION_MASK_P9_DFORM_VECTOR);
}
- if (TARGET_P9_DFORM && !TARGET_UPPER_REGS_DF)
+ if (TARGET_P9_DFORM_SCALAR && !TARGET_UPPER_REGS_DF)
{
if (rs6000_isa_flags_explicit & OPTION_MASK_UPPER_REGS_DF)
error ("-mpower9-dform requires -mupper-regs-df");
- rs6000_isa_flags &= ~OPTION_MASK_P9_DFORM;
+ rs6000_isa_flags &= ~OPTION_MASK_P9_DFORM_SCALAR;
}
- if (TARGET_P9_DFORM && !TARGET_UPPER_REGS_SF)
+ if (TARGET_P9_DFORM_SCALAR && !TARGET_UPPER_REGS_SF)
{
if (rs6000_isa_flags_explicit & OPTION_MASK_UPPER_REGS_SF)
error ("-mpower9-dform requires -mupper-regs-sf");
- rs6000_isa_flags &= ~OPTION_MASK_P9_DFORM;
+ rs6000_isa_flags &= ~OPTION_MASK_P9_DFORM_SCALAR;
}
/* ISA 3.0 vector instructions include ISA 2.07. */
@@ -4231,6 +4282,47 @@ rs6000_option_override_internal (bool gl
rs6000_isa_flags &= ~OPTION_MASK_P9_VECTOR;
}
+ /* There have been bugs with both -mvsx-timode and -mpower9-dform-vector that
+ don't show up with -mlra, but do show up with -mno-lra. Given -mlra will
+ become the default once PR 69847 is fixed, turn off the options with
+ problems by default if -mno-lra was used, and warn if the user explicitly
+ asked for the option.
+
+ Enable -mpower9-dform-vector by default if LRA and other power9 options.
+ Enable -mvsx-timode by default if LRA and VSX. */
+ if (!TARGET_LRA)
+ {
+ if (TARGET_VSX_TIMODE)
+ {
+ if ((rs6000_isa_flags_explicit & OPTION_MASK_VSX_TIMODE) != 0)
+ warning (0, "-mvsx-timode might need -mlra");
+
+ else
+ rs6000_isa_flags &= ~OPTION_MASK_VSX_TIMODE;
+ }
+
+ if (TARGET_P9_DFORM_VECTOR)
+ {
+ if ((rs6000_isa_flags_explicit & OPTION_MASK_P9_DFORM_VECTOR) != 0)
+ warning (0, "-mpower9-dform-vector might need -mlra");
+
+ else
+ rs6000_isa_flags &= ~OPTION_MASK_P9_DFORM_VECTOR;
+ }
+ }
+
+ else
+ {
+ if (TARGET_VSX && !TARGET_VSX_TIMODE
+ && (rs6000_isa_flags_explicit & OPTION_MASK_VSX_TIMODE) == 0)
+ rs6000_isa_flags |= OPTION_MASK_VSX_TIMODE;
+
+ if (TARGET_VSX && TARGET_P9_VECTOR && !TARGET_P9_DFORM_VECTOR
+ && TARGET_P9_DFORM_SCALAR && TARGET_P9_DFORM_BOTH < 0
+ && (rs6000_isa_flags_explicit & OPTION_MASK_P9_DFORM_VECTOR) == 0)
+ rs6000_isa_flags |= OPTION_MASK_P9_DFORM_VECTOR;
+ }
+
/* Set -mallow-movmisalign to explicitly on if we have full ISA 2.07
support. If we only have ISA 2.06 support, and the user did not specify
the switch, leave it set to -1 so the movmisalign patterns are enabled,
@@ -6915,6 +7007,59 @@ direct_move_p (rtx op0, rtx op1)
return false;
}
+/* Return true if the OFFSET is valid for the quad address instructions that
+ use d-form (register + offset) addressing. */
+
+static inline bool
+quad_address_offset_p (HOST_WIDE_INT offset)
+{
+ return (IN_RANGE (offset, -32768, 32767) && ((offset) & 0xf) == 0);
+}
+
+/* Return true if the ADDR is an acceptable address for a quad memory
+ operation of mode MODE (either LQ/STQ for general purpose registers, or
+ LXV/STXV for vector registers under ISA 3.0. GPR_P is true if this address
+ is intended for LQ/STQ. If it is false, the address is intended for the ISA
+ 3.0 LXV/STXV instruction. */
+
+bool
+quad_address_p (rtx addr, machine_mode mode, bool gpr_p)
+{
+ rtx op0, op1;
+
+ if (GET_MODE_SIZE (mode) != 16)
+ return false;
+
+ if (gpr_p)
+ {
+ if (!TARGET_QUAD_MEMORY && !TARGET_SYNC_TI)
+ return false;
+
+ /* LQ/STQ can handle indirect addresses. */
+ if (base_reg_operand (addr, Pmode))
+ return true;
+ }
+
+ else
+ {
+ if (!mode_supports_vsx_dform_quad (mode))
+ return false;
+ }
+
+ if (GET_CODE (addr) != PLUS)
+ return false;
+
+ op0 = XEXP (addr, 0);
+ if (!base_reg_operand (op0, Pmode))
+ return false;
+
+ op1 = XEXP (addr, 1);
+ if (!CONST_INT_P (op1))
+ return false;
+
+ return quad_address_offset_p (INTVAL (op1));
+}
+
/* Return true if this is a load or store quad operation. This function does
not handle the atomic quad memory instructions. */
@@ -7007,6 +7152,10 @@ mem_operand_gpr (rtx op, machine_mode mo
if (TARGET_POWERPC64 && (offset & 3) != 0)
return false;
+ if (mode_supports_vsx_dform_quad (mode)
+ && !quad_address_offset_p (offset))
+ return false;
+
extra = GET_MODE_SIZE (mode) - UNITS_PER_WORD;
if (extra < 0)
extra = 0;
@@ -7036,13 +7185,14 @@ reg_offset_addressing_ok_p (machine_mode
case TImode:
case TFmode:
case KFmode:
- /* AltiVec/VSX vector modes. Only reg+reg addressing is valid. While
- TImode is not a vector mode, if we want to use the VSX registers to
- move it around, we need to restrict ourselves to reg+reg addressing.
- Similarly for IEEE 128-bit floating point that is passed in a single
- vector register. */
+ /* AltiVec/VSX vector modes. Only reg+reg addressing was valid until the
+ ISA 3.0 vector d-form addressing mode was added. While TImode is not
+ a vector mode, if we want to use the VSX registers to move it around,
+ we need to restrict ourselves to reg+reg addressing. Similarly for
+ IEEE 128-bit floating point that is passed in a single vector
+ register. */
if (VECTOR_MEM_ALTIVEC_OR_VSX_P (mode))
- return false;
+ return mode_supports_vsx_dform_quad (mode);
break;
case V4HImode:
@@ -7109,6 +7259,11 @@ offsettable_ok_by_alignment (rtx op, HOS
if (GET_CODE (op) != SYMBOL_REF)
return false;
+ /* ISA 3.0 vector d-form addressing is restricted, don't allow
+ SYMBOL_REF. */
+ if (mode_supports_vsx_dform_quad (mode))
+ return false;
+
dsize = GET_MODE_SIZE (mode);
decl = SYMBOL_REF_DECL (op);
if (!decl)
@@ -7263,6 +7418,9 @@ rs6000_legitimate_offset_address_p (mach
return false;
if (!INT_REG_OK_FOR_BASE_P (XEXP (x, 0), strict))
return false;
+ if (mode_supports_vsx_dform_quad (mode))
+ return (virtual_stack_registers_memory_p (x)
+ || quad_address_p (x, mode, false));
if (!reg_offset_addressing_ok_p (mode))
return virtual_stack_registers_memory_p (x);
if (legitimate_constant_pool_address_p (x, mode, strict || lra_in_progress))
@@ -7401,6 +7559,9 @@ legitimate_lo_sum_address_p (machine_mod
return false;
if (!INT_REG_OK_FOR_BASE_P (XEXP (x, 0), strict))
return false;
+ /* quad word addresses are restricted, and we can't use LO_SUM. */
+ if (mode_supports_vsx_dform_quad (mode))
+ return false;
/* Restrict addressing for DI because of our SUBREG hackery. */
if (TARGET_E500_DOUBLE && GET_MODE_SIZE (mode) > UNITS_PER_WORD)
return false;
@@ -7412,7 +7573,7 @@ legitimate_lo_sum_address_p (machine_mod
if (DEFAULT_ABI == ABI_V4 && flag_pic)
return false;
- /* LRA don't use LEGITIMIZE_RELOAD_ADDRESS as it usually calls
+ /* LRA doesn't use LEGITIMIZE_RELOAD_ADDRESS as it usually calls
push_reload from reload pass code. LEGITIMIZE_RELOAD_ADDRESS
recognizes some LO_SUM addresses as valid although this
function says opposite. In most cases, LRA through different
@@ -7466,7 +7627,8 @@ rs6000_legitimize_address (rtx x, rtx ol
{
unsigned int extra;
- if (!reg_offset_addressing_ok_p (mode))
+ if (!reg_offset_addressing_ok_p (mode)
+ || mode_supports_vsx_dform_quad (mode))
{
if (virtual_stack_registers_memory_p (x))
return x;
@@ -8177,6 +8339,11 @@ rs6000_legitimize_reload_address (rtx x,
&& GET_CODE (XEXP (XEXP (x, 0), 1)) == CONST_INT
&& GET_CODE (XEXP (x, 1)) == CONST_INT)
{
+ if (TARGET_DEBUG_ADDR)
+ {
+ fprintf (stderr, "\nlegitimize_reload_address push_reload #1:\n");
+ debug_rtx (x);
+ }
push_reload (XEXP (x, 0), NULL_RTX, &XEXP (x, 0), NULL,
BASE_REG_CLASS, GET_MODE (x), VOIDmode, 0, 0,
opnum, (enum reload_type) type);
@@ -8188,6 +8355,11 @@ rs6000_legitimize_reload_address (rtx x,
if (GET_CODE (x) == LO_SUM
&& GET_CODE (XEXP (x, 0)) == HIGH)
{
+ if (TARGET_DEBUG_ADDR)
+ {
+ fprintf (stderr, "\nlegitimize_reload_address push_reload #2:\n");
+ debug_rtx (x);
+ }
push_reload (XEXP (x, 0), NULL_RTX, &XEXP (x, 0), NULL,
BASE_REG_CLASS, Pmode, VOIDmode, 0, 0,
opnum, (enum reload_type) type);
@@ -8220,6 +8392,11 @@ rs6000_legitimize_reload_address (rtx x,
{
rtx hi = gen_rtx_HIGH (Pmode, copy_rtx (x));
x = gen_rtx_LO_SUM (Pmode, hi, x);
+ if (TARGET_DEBUG_ADDR)
+ {
+ fprintf (stderr, "\nlegitimize_reload_address push_reload #3:\n");
+ debug_rtx (x);
+ }
push_reload (XEXP (x, 0), NULL_RTX, &XEXP (x, 0), NULL,
BASE_REG_CLASS, Pmode, VOIDmode, 0, 0,
opnum, (enum reload_type) type);
@@ -8257,6 +8434,11 @@ rs6000_legitimize_reload_address (rtx x,
GEN_INT (high)),
GEN_INT (low));
+ if (TARGET_DEBUG_ADDR)
+ {
+ fprintf (stderr, "\nlegitimize_reload_address push_reload #4:\n");
+ debug_rtx (x);
+ }
push_reload (XEXP (x, 0), NULL_RTX, &XEXP (x, 0), NULL,
BASE_REG_CLASS, GET_MODE (x), VOIDmode, 0, 0,
opnum, (enum reload_type) type);
@@ -8317,6 +8499,11 @@ rs6000_legitimize_reload_address (rtx x,
x = gen_rtx_LO_SUM (GET_MODE (x),
gen_rtx_HIGH (Pmode, x), x);
+ if (TARGET_DEBUG_ADDR)
+ {
+ fprintf (stderr, "\nlegitimize_reload_address push_reload #5:\n");
+ debug_rtx (x);
+ }
push_reload (XEXP (x, 0), NULL_RTX, &XEXP (x, 0), NULL,
BASE_REG_CLASS, Pmode, VOIDmode, 0, 0,
opnum, (enum reload_type) type);
@@ -8350,9 +8537,16 @@ rs6000_legitimize_reload_address (rtx x,
{
x = create_TOC_reference (x, NULL_RTX);
if (TARGET_CMODEL != CMODEL_SMALL)
- push_reload (XEXP (x, 0), NULL_RTX, &XEXP (x, 0), NULL,
- BASE_REG_CLASS, Pmode, VOIDmode, 0, 0,
- opnum, (enum reload_type) type);
+ {
+ if (TARGET_DEBUG_ADDR)
+ {
+ fprintf (stderr, "\nlegitimize_reload_address push_reload #6:\n");
+ debug_rtx (x);
+ }
+ push_reload (XEXP (x, 0), NULL_RTX, &XEXP (x, 0), NULL,
+ BASE_REG_CLASS, Pmode, VOIDmode, 0, 0,
+ opnum, (enum reload_type) type);
+ }
*win = 1;
return x;
}
@@ -8408,6 +8602,7 @@ static bool
rs6000_legitimate_address_p (machine_mode mode, rtx x, bool reg_ok_strict)
{
bool reg_offset_p = reg_offset_addressing_ok_p (mode);
+ bool quad_offset_p = mode_supports_vsx_dform_quad (mode);
/* If this is an unaligned stvx/ldvx type address, discard the outer AND. */
if (VECTOR_MEM_ALTIVEC_P (mode)
@@ -8427,15 +8622,26 @@ rs6000_legitimate_address_p (machine_mod
return 1;
if (virtual_stack_registers_memory_p (x))
return 1;
- if (reg_offset_p && legitimate_small_data_p (mode, x))
- return 1;
- if (reg_offset_p
- && legitimate_constant_pool_address_p (x, mode,
+
+ /* Handle restricted vector d-form offsets in ISA 3.0. */
+ if (quad_offset_p)
+ {
+ if (quad_address_p (x, mode, false))
+ return 1;
+ }
+
+ else if (reg_offset_p)
+ {
+ if (legitimate_small_data_p (mode, x))
+ return 1;
+ if (legitimate_constant_pool_address_p (x, mode,
reg_ok_strict || lra_in_progress))
- return 1;
- if (reg_offset_p && reg_addr[mode].fused_toc && GET_CODE (x) == UNSPEC
- && XINT (x, 1) == UNSPEC_FUSION_ADDIS)
- return 1;
+ return 1;
+ if (reg_addr[mode].fused_toc && GET_CODE (x) == UNSPEC
+ && XINT (x, 1) == UNSPEC_FUSION_ADDIS)
+ return 1;
+ }
+
/* For TImode, if we have load/store quad and TImode in VSX registers, only
allow register indirect addresses. This will allow the values to go in
either GPRs or VSX registers without reloading. The vector types would
@@ -8474,7 +8680,8 @@ rs6000_legitimate_address_p (machine_mod
&& legitimate_indexed_address_p (XEXP (x, 1), reg_ok_strict)))
&& rtx_equal_p (XEXP (XEXP (x, 1), 0), XEXP (x, 0)))
return 1;
- if (reg_offset_p && legitimate_lo_sum_address_p (mode, x, reg_ok_strict))
+ if (reg_offset_p && !quad_offset_p
+ && legitimate_lo_sum_address_p (mode, x, reg_ok_strict))
return 1;
return 0;
}
@@ -18368,13 +18575,23 @@ rs6000_secondary_reload_memory (rtx addr
}
}
+ else if ((addr_mask & RELOAD_REG_QUAD_OFFSET) != 0
+ && CONST_INT_P (plus_arg1))
+ {
+ if (!quad_address_offset_p (INTVAL (plus_arg1)))
+ {
+ extra_cost = 1;
+ type = "vector d-form offset";
+ }
+ }
+
/* Make sure the register class can handle offset addresses. */
else if (rs6000_legitimate_offset_address_p (mode, addr, false, true))
{
if ((addr_mask & RELOAD_REG_OFFSET) == 0)
{
extra_cost = 1;
- type = "offset";
+ type = "offset #2";
}
}
@@ -18387,7 +18604,14 @@ rs6000_secondary_reload_memory (rtx addr
break;
case LO_SUM:
- if (!legitimate_lo_sum_address_p (mode, addr, false))
+ /* Quad offsets are restricted and can't handle normal addresses. */
+ if ((addr_mask & RELOAD_REG_QUAD_OFFSET) != 0)
+ {
+ extra_cost = -1;
+ type = "vector d-form lo_sum";
+ }
+
+ else if (!legitimate_lo_sum_address_p (mode, addr, false))
{
fail_msg = "bad LO_SUM";
extra_cost = -1;
@@ -18404,8 +18628,17 @@ rs6000_secondary_reload_memory (rtx addr
case CONST:
case SYMBOL_REF:
case LABEL_REF:
- type = "address";
- extra_cost = rs6000_secondary_reload_toc_costs (addr_mask);
+ if ((addr_mask & RELOAD_REG_QUAD_OFFSET) != 0)
+ {
+ extra_cost = -1;
+ type = "vector d-form lo_sum #2";
+ }
+
+ else
+ {
+ type = "address";
+ extra_cost = rs6000_secondary_reload_toc_costs (addr_mask);
+ }
break;
/* TOC references look like offsetable memory. */
@@ -18416,6 +18649,12 @@ rs6000_secondary_reload_memory (rtx addr
extra_cost = -1;
}
+ else if ((addr_mask & RELOAD_REG_QUAD_OFFSET) != 0)
+ {
+ extra_cost = -1;
+ type = "vector d-form lo_sum #3";
+ }
+
else if ((addr_mask & RELOAD_REG_OFFSET) == 0)
{
extra_cost = 1;
@@ -19046,6 +19285,16 @@ rs6000_secondary_reload_inner (rtx reg,
}
}
+ else if (mode_supports_vsx_dform_quad (mode) && CONST_INT_P (op1))
+ {
+ if (((addr_mask & RELOAD_REG_QUAD_OFFSET) == 0)
+ || !quad_address_p (addr, mode, false))
+ {
+ emit_insn (gen_rtx_SET (scratch, addr));
+ new_addr = scratch;
+ }
+ }
+
/* Make sure the register class can handle offset addresses. */
else if (rs6000_legitimate_offset_address_p (mode, addr, false, true))
{
@@ -19076,6 +19325,13 @@ rs6000_secondary_reload_inner (rtx reg,
}
}
+ /* Quad offsets are restricted and can't handle normal addresses. */
+ else if (mode_supports_vsx_dform_quad (mode))
+ {
+ emit_insn (gen_rtx_SET (scratch, addr));
+ new_addr = scratch;
+ }
+
/* Make sure the register class can handle offset addresses. */
else if (legitimate_lo_sum_address_p (mode, addr, false))
{
@@ -19296,7 +19552,8 @@ rs6000_preferred_reload_class (rtx x, en
}
/* D-form addressing can easily reload the value. */
- if (mode_supports_vmx_dform (mode))
+ if (mode_supports_vmx_dform (mode)
+ || mode_supports_vsx_dform_quad (mode))
return rclass;
/* If this is a scalar floating point value and we don't have D-form
@@ -19731,8 +19988,16 @@ rs6000_output_move_128bit (rtx operands[
else if (TARGET_VSX && dest_vsx_p)
{
- if (mode == V16QImode || mode == V8HImode || mode == V4SImode)
+ if (mode_supports_vsx_dform_quad (mode)
+ && quad_address_p (XEXP (src, 0), mode, false))
+ return "lxv %x0,%1";
+
+ else if (TARGET_P9_VECTOR)
+ return "lxvx %x0,%y1";
+
+ else if (mode == V16QImode || mode == V8HImode || mode == V4SImode)
return "lxvw4x %x0,%y1";
+
else
return "lxvd2x %x0,%y1";
}
@@ -19761,8 +20026,16 @@ rs6000_output_move_128bit (rtx operands[
else if (TARGET_VSX && src_vsx_p)
{
- if (mode == V16QImode || mode == V8HImode || mode == V4SImode)
+ if (mode_supports_vsx_dform_quad (mode)
+ && quad_address_p (XEXP (dest, 0), mode, false))
+ return "stxv %x1,%0";
+
+ else if (TARGET_P9_VECTOR)
+ return "stxvx %x1,%y0";
+
+ else if (mode == V16QImode || mode == V8HImode || mode == V4SImode)
return "stxvw4x %x1,%y0";
+
else
return "stxvd2x %x1,%y0";
}
@@ -26198,25 +26471,37 @@ rs6000_emit_prologue (void)
if (info->vrsave_mask & ALTIVEC_REG_BIT (i))
{
rtx areg, savereg, mem;
- int offset;
+ HOST_WIDE_INT offset;
offset = (info->altivec_save_offset + frame_off
+ 16 * (i - info->first_altivec_reg_save));
savereg = gen_rtx_REG (V4SImode, i);
- NOT_INUSE (0);
- areg = gen_rtx_REG (Pmode, 0);
- emit_move_insn (areg, GEN_INT (offset));
-
- /* AltiVec addressing mode is [reg+reg]. */
- mem = gen_frame_mem (V4SImode,
- gen_rtx_PLUS (Pmode, frame_reg_rtx, areg));
-
- /* Rather than emitting a generic move, force use of the stvx
- instruction, which we always want. In particular we don't
- want xxpermdi/stxvd2x for little endian. */
- insn = emit_insn (gen_altivec_stvx_v4si_internal (mem, savereg));
+ if (TARGET_P9_DFORM_VECTOR && quad_address_offset_p (offset))
+ {
+ mem = gen_frame_mem (V4SImode,
+ gen_rtx_PLUS (Pmode, frame_reg_rtx,
+ GEN_INT (offset)));
+ insn = emit_insn (gen_rtx_SET (mem, savereg));
+ areg = NULL_RTX;
+ }
+ else
+ {
+ NOT_INUSE (0);
+ areg = gen_rtx_REG (Pmode, 0);
+ emit_move_insn (areg, GEN_INT (offset));
+
+ /* AltiVec addressing mode is [reg+reg]. */
+ mem = gen_frame_mem (V4SImode,
+ gen_rtx_PLUS (Pmode, frame_reg_rtx, areg));
+
+ /* Rather than emitting a generic move, force use of the stvx
+ instruction, which we always want on ISA 2.07 (power8) systems.
+ In particular we don't want xxpermdi/stxvd2x for little
+ endian. */
+ insn = emit_insn (gen_altivec_stvx_v4si_internal (mem, savereg));
+ }
rs6000_frame_related (insn, frame_reg_rtx, sp_off - frame_off,
areg, GEN_INT (offset));
@@ -26936,23 +27221,35 @@ rs6000_emit_epilogue (int sibcall)
for (i = info->first_altivec_reg_save; i <= LAST_ALTIVEC_REGNO; ++i)
if (info->vrsave_mask & ALTIVEC_REG_BIT (i))
{
- rtx addr, areg, mem, reg;
+ rtx addr, areg, mem, insn;
+ rtx reg = gen_rtx_REG (V4SImode, i);
+ HOST_WIDE_INT offset
+ = (info->altivec_save_offset + frame_off
+ + 16 * (i - info->first_altivec_reg_save));
- areg = gen_rtx_REG (Pmode, 0);
- emit_move_insn
- (areg, GEN_INT (info->altivec_save_offset
- + frame_off
- + 16 * (i - info->first_altivec_reg_save)));
+ if (TARGET_P9_DFORM_VECTOR && quad_address_offset_p (offset))
+ {
+ mem = gen_frame_mem (V4SImode,
+ gen_rtx_PLUS (Pmode, frame_reg_rtx,
+ GEN_INT (offset)));
+ insn = gen_rtx_SET (reg, mem);
+ }
+ else
+ {
+ areg = gen_rtx_REG (Pmode, 0);
+ emit_move_insn (areg, GEN_INT (offset));
- /* AltiVec addressing mode is [reg+reg]. */
- addr = gen_rtx_PLUS (Pmode, frame_reg_rtx, areg);
- mem = gen_frame_mem (V4SImode, addr);
+ /* AltiVec addressing mode is [reg+reg]. */
+ addr = gen_rtx_PLUS (Pmode, frame_reg_rtx, areg);
+ mem = gen_frame_mem (V4SImode, addr);
+
+ /* Rather than emitting a generic move, force use of the
+ lvx instruction, which we always want. In particular we
+ don't want lxvd2x/xxpermdi for little endian. */
+ insn = gen_altivec_lvx_v4si_internal (reg, mem);
+ }
- reg = gen_rtx_REG (V4SImode, i);
- /* Rather than emitting a generic move, force use of the
- lvx instruction, which we always want. In particular
- we don't want lxvd2x/xxpermdi for little endian. */
- (void) emit_insn (gen_altivec_lvx_v4si_internal (reg, mem));
+ (void) emit_insn (insn);
}
}
@@ -27139,23 +27436,35 @@ rs6000_emit_epilogue (int sibcall)
for (i = info->first_altivec_reg_save; i <= LAST_ALTIVEC_REGNO; ++i)
if (info->vrsave_mask & ALTIVEC_REG_BIT (i))
{
- rtx addr, areg, mem, reg;
+ rtx addr, areg, mem, insn;
+ rtx reg = gen_rtx_REG (V4SImode, i);
+ HOST_WIDE_INT offset
+ = (info->altivec_save_offset + frame_off
+ + 16 * (i - info->first_altivec_reg_save));
- areg = gen_rtx_REG (Pmode, 0);
- emit_move_insn
- (areg, GEN_INT (info->altivec_save_offset
- + frame_off
- + 16 * (i - info->first_altivec_reg_save)));
+ if (TARGET_P9_DFORM_VECTOR && quad_address_offset_p (offset))
+ {
+ mem = gen_frame_mem (V4SImode,
+ gen_rtx_PLUS (Pmode, frame_reg_rtx,
+ GEN_INT (offset)));
+ insn = gen_rtx_SET (reg, mem);
+ }
+ else
+ {
+ areg = gen_rtx_REG (Pmode, 0);
+ emit_move_insn (areg, GEN_INT (offset));
- /* AltiVec addressing mode is [reg+reg]. */
- addr = gen_rtx_PLUS (Pmode, frame_reg_rtx, areg);
- mem = gen_frame_mem (V4SImode, addr);
+ /* AltiVec addressing mode is [reg+reg]. */
+ addr = gen_rtx_PLUS (Pmode, frame_reg_rtx, areg);
+ mem = gen_frame_mem (V4SImode, addr);
+
+ /* Rather than emitting a generic move, force use of the
+ lvx instruction, which we always want. In particular we
+ don't want lxvd2x/xxpermdi for little endian. */
+ insn = gen_altivec_lvx_v4si_internal (reg, mem);
+ }
- reg = gen_rtx_REG (V4SImode, i);
- /* Rather than emitting a generic move, force use of the
- lvx instruction, which we always want. In particular
- we don't want lxvd2x/xxpermdi for little endian. */
- (void) emit_insn (gen_altivec_lvx_v4si_internal (reg, mem));
+ (void) emit_insn (insn);
}
}
@@ -34311,7 +34620,7 @@ rs6000_libcall_value (machine_mode mode)
static bool
rs6000_lra_p (void)
{
- return rs6000_lra_flag;
+ return TARGET_LRA;
}
/* Given FROM and TO register numbers, say whether this elimination is allowed.
@@ -34672,7 +34981,8 @@ static struct rs6000_opt_mask const rs60
{ "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-dform", OPTION_MASK_P9_DFORM, false, true },
+ { "power9-dform-scalar", OPTION_MASK_P9_DFORM_SCALAR, false, true },
+ { "power9-dform-vector", OPTION_MASK_P9_DFORM_VECTOR, false, true },
{ "power9-fusion", OPTION_MASK_P9_FUSION, false, true },
{ "power9-minmax", OPTION_MASK_P9_MINMAX, false, true },
{ "power9-vector", OPTION_MASK_P9_VECTOR, false, true },
@@ -35305,7 +35615,9 @@ rs6000_print_options_internal (FILE *fil
size_t i;
size_t start_column = 0;
size_t cur_column;
- size_t max_column = 76;
+ size_t max_column = 120;
+ size_t prefix_len = strlen (prefix);
+ size_t comma_len = 0;
const char *comma = "";
if (indent)
@@ -35323,27 +35635,45 @@ rs6000_print_options_internal (FILE *fil
cur_column = start_column;
for (i = 0; i < num_elements; i++)
{
- if ((flags & opts[i].mask) != 0)
+ bool invert = opts[i].invert;
+ const char *name = opts[i].name;
+ const char *no_str = "";
+ HOST_WIDE_INT mask = opts[i].mask;
+ size_t len = comma_len + prefix_len + strlen (name);
+
+ if (!invert)
{
- const char *no_str = rs6000_opt_masks[i].invert ? "no-" : "";
- size_t len = (strlen (comma)
- + strlen (prefix)
- + strlen (no_str)
- + strlen (rs6000_opt_masks[i].name));
+ if ((flags & mask) == 0)
+ {
+ no_str = "no-";
+ len += sizeof ("no-") - 1;
+ }
+
+ flags &= ~mask;
+ }
- cur_column += len;
- if (cur_column > max_column)
+ else
+ {
+ if ((flags & mask) != 0)
{
- fprintf (stderr, ", \\\n%*s", (int)start_column, "");
- cur_column = start_column + len;
- comma = "";
+ no_str = "no-";
+ len += sizeof ("no-") - 1;
}
- fprintf (file, "%s%s%s%s", comma, prefix, no_str,
- rs6000_opt_masks[i].name);
- flags &= ~ opts[i].mask;
- comma = ", ";
+ flags |= mask;
}
+
+ cur_column += len;
+ if (cur_column > max_column)
+ {
+ fprintf (stderr, ", \\\n%*s", (int)start_column, "");
+ cur_column = start_column + len;
+ comma = "";
+ }
+
+ fprintf (file, "%s%s%s%s", comma, prefix, no_str, name);
+ comma = ", ";
+ comma_len = sizeof (", ") - 1;
}
fputs ("\n", file);
Index: gcc/config/rs6000/constraints.md
===================================================================
--- gcc/config/rs6000/constraints.md (.../svn+ssh://meissner@gcc.gnu.org/svn/gcc/trunk/gcc/config/rs6000) (revision 235831)
+++ gcc/config/rs6000/constraints.md (.../gcc/config/rs6000) (working copy)
@@ -156,6 +156,11 @@ (define_constraint "wL"
(and (match_test "TARGET_DIRECT_MOVE_128")
(match_test "(ival == VECTOR_ELEMENT_MFVSRLD_64BIT)"))))
+;; ISA 3.0 vector d-form addresses
+(define_memory_constraint "wO"
+ "Memory operand suitable for the ISA 3.0 vector d-form instructions."
+ (match_operand 0 "vsx_quad_dform_memory_operand"))
+
;; Lq/stq validates the address for load/store quad
(define_memory_constraint "wQ"
"Memory operand suitable for the load/store quad instructions"
Index: gcc/config/rs6000/predicates.md
===================================================================
--- gcc/config/rs6000/predicates.md (.../svn+ssh://meissner@gcc.gnu.org/svn/gcc/trunk/gcc/config/rs6000) (revision 235831)
+++ gcc/config/rs6000/predicates.md (.../gcc/config/rs6000) (working copy)
@@ -698,48 +698,25 @@ (define_predicate "offsettable_mem_opera
(define_predicate "quad_memory_operand"
(match_code "mem")
{
- rtx addr, op0, op1;
- int ret;
-
if (!TARGET_QUAD_MEMORY && !TARGET_SYNC_TI)
- ret = 0;
-
- else if (!memory_operand (op, mode))
- ret = 0;
-
- else if (GET_MODE_SIZE (GET_MODE (op)) != 16)
- ret = 0;
-
- else if (MEM_ALIGN (op) < 128)
- ret = 0;
-
- else
- {
- addr = XEXP (op, 0);
- if (int_reg_operand (addr, Pmode))
- ret = 1;
+ return false;
- else if (GET_CODE (addr) != PLUS)
- ret = 0;
+ if (GET_MODE_SIZE (mode) != 16 || !MEM_P (op) || MEM_ALIGN (op) < 128)
+ return false;
- else
- {
- op0 = XEXP (addr, 0);
- op1 = XEXP (addr, 1);
- ret = (int_reg_operand (op0, Pmode)
- && GET_CODE (op1) == CONST_INT
- && IN_RANGE (INTVAL (op1), -32768, 32767)
- && (INTVAL (op1) & 15) == 0);
- }
- }
+ return quad_address_p (XEXP (op, 0), mode, true);
+})
- if (TARGET_DEBUG_ADDR)
- {
- fprintf (stderr, "\nquad_memory_operand, ret = %s\n", ret ? "true" : "false");
- debug_rtx (op);
- }
+;; Return 1 if the operand is suitable for load/store to vector registers with
+;; d-form addressing (register+offset), which was added in ISA 3.0.
+;; Unlike quad_memory_operand, we do not have to check for alignment.
+(define_predicate "vsx_quad_dform_memory_operand"
+ (match_code "mem")
+{
+ if (!TARGET_P9_DFORM_VECTOR || !MEM_P (op) || GET_MODE_SIZE (mode) != 16)
+ return false;
- return ret;
+ return quad_address_p (XEXP (op, 0), mode, false);
})
;; Return 1 if the operand is an indexed or indirect memory operand.
Index: gcc/config/rs6000/rs6000-protos.h
===================================================================
--- gcc/config/rs6000/rs6000-protos.h (.../svn+ssh://meissner@gcc.gnu.org/svn/gcc/trunk/gcc/config/rs6000) (revision 235831)
+++ gcc/config/rs6000/rs6000-protos.h (.../gcc/config/rs6000) (working copy)
@@ -86,6 +86,7 @@ extern int registers_ok_for_quad_peep (r
extern int mems_ok_for_quad_peep (rtx, rtx);
extern bool gpr_or_gpr_p (rtx, rtx);
extern bool direct_move_p (rtx, rtx);
+extern bool quad_address_p (rtx, machine_mode, bool);
extern bool quad_load_store_p (rtx, rtx);
extern bool fusion_gpr_load_p (rtx, rtx, rtx, rtx);
extern void expand_fusion_gpr_load (rtx *);
Index: gcc/config/rs6000/vsx.md
===================================================================
--- gcc/config/rs6000/vsx.md (.../svn+ssh://meissner@gcc.gnu.org/svn/gcc/trunk/gcc/config/rs6000) (revision 235831)
+++ gcc/config/rs6000/vsx.md (.../gcc/config/rs6000) (working copy)
@@ -301,24 +301,6 @@ (define_c_enum "unspec"
UNSPEC_VSX_XVCVDPUXDS
])
-;; VSX (P9) moves
-
-(define_insn "*p9_vecload_<mode>"
- [(set (match_operand:VSX_M2 0 "vsx_register_operand" "=<VSa>")
- (match_operand:VSX_M2 1 "memory_operand" "Z"))]
- "TARGET_P9_VECTOR"
- "lxvx %x0,%y1"
- [(set_attr "type" "vecload")
- (set_attr "length" "4")])
-
-(define_insn "*p9_vecstore_<mode>"
- [(set (match_operand:VSX_M2 0 "memory_operand" "=Z")
- (match_operand:VSX_M2 1 "vsx_register_operand" "<VSa>"))]
- "TARGET_P9_VECTOR"
- "stxvx %x1,%y0"
- [(set_attr "type" "vecstore")
- (set_attr "length" "4")])
-
;; VSX moves
;; The patterns for LE permuted loads and stores come before the general
@@ -788,8 +770,8 @@ (define_split
"")
(define_insn "*vsx_mov<mode>"
- [(set (match_operand:VSX_M 0 "nonimmediate_operand" "=Z,<VSr>,<VSr>,?Z,?<VSa>,?<VSa>,r,we,wQ,?&r,??Y,??r,??r,<VSr>,?<VSa>,*r,v,wZ,v")
- (match_operand:VSX_M 1 "input_operand" "<VSr>,Z,<VSr>,<VSa>,Z,<VSa>,we,b,r,wQ,r,Y,r,j,j,j,W,v,wZ"))]
+ [(set (match_operand:VSX_M 0 "nonimmediate_operand" "=ZwO,<VSr>,<VSr>,?ZwO,?<VSa>,?<VSa>,r,we,wQ,?&r,??Y,??r,??r,<VSr>,?<VSa>,*r,v,wZ,v")
+ (match_operand:VSX_M 1 "input_operand" "<VSr>,ZwO,<VSr>,<VSa>,ZwO,<VSa>,we,b,r,wQ,r,Y,r,j,j,j,W,v,wZ"))]
"VECTOR_MEM_VSX_P (<MODE>mode)
&& (register_operand (operands[0], <MODE>mode)
|| register_operand (operands[1], <MODE>mode))"
@@ -803,8 +785,8 @@ (define_insn "*vsx_mov<mode>"
;; use of TImode is for unions. However for plain data movement, slightly
;; favor the vector loads
(define_insn "*vsx_movti_64bit"
- [(set (match_operand:TI 0 "nonimmediate_operand" "=Z,wa,wa,wa,r,we,v,v,wZ,wQ,&r,Y,r,r,?r")
- (match_operand:TI 1 "input_operand" "wa,Z,wa,O,we,b,W,wZ,v,r,wQ,r,Y,r,n"))]
+ [(set (match_operand:TI 0 "nonimmediate_operand" "=ZwO,wa,wa,wa,r,we,v,v,wZ,wQ,&r,Y,r,r,?r")
+ (match_operand:TI 1 "input_operand" "wa,ZwO,wa,O,we,b,W,wZ,v,r,wQ,r,Y,r,n"))]
"TARGET_POWERPC64 && VECTOR_MEM_VSX_P (TImode)
&& (register_operand (operands[0], TImode)
|| register_operand (operands[1], TImode))"
@@ -815,8 +797,8 @@ (define_insn "*vsx_movti_64bit"
(set_attr "length" "4,4,4,4,8,4,16,4,4,8,8,8,8,8,8")])
(define_insn "*vsx_movti_32bit"
- [(set (match_operand:TI 0 "nonimmediate_operand" "=Z,wa,wa,wa,v, v,wZ,Q,Y,????r,????r,????r,r")
- (match_operand:TI 1 "input_operand" "wa, Z,wa, O,W,wZ, v,r,r, Q, Y, r,n"))]
+ [(set (match_operand:TI 0 "nonimmediate_operand" "=ZwO,wa,wa,wa,v,v,wZ,Q,Y,????r,????r,????r,r")
+ (match_operand:TI 1 "input_operand" "wa,ZwO,wa,O,W,wZ,v,r,r,Q,Y,r,n"))]
"! TARGET_POWERPC64 && VECTOR_MEM_VSX_P (TImode)
&& (register_operand (operands[0], TImode)
|| register_operand (operands[1], TImode))"
Index: gcc/doc/invoke.texi
===================================================================
--- gcc/doc/invoke.texi (.../svn+ssh://meissner@gcc.gnu.org/svn/gcc/trunk/gcc/doc) (revision 235831)
+++ gcc/doc/invoke.texi (.../gcc/doc) (working copy)
@@ -1004,7 +1004,8 @@ See RS/6000 and PowerPC Options.
-mupper-regs-df -mno-upper-regs-df -mupper-regs-sf -mno-upper-regs-sf @gol
-mupper-regs -mno-upper-regs -mmodulo -mno-modulo @gol
-mfloat128 -mno-float128 -mfloat128-hardware -mno-float128-hardware @gol
--mpower9-fusion -mno-mpower9-fusion -mpower9-vector -mno-power9-vector}
+-mpower9-fusion -mno-mpower9-fusion -mpower9-vector -mno-power9-vector @gol
+-mpower9-dform -mno-power9-dform}
@emph{RX Options}
@gccoptlist{-m64bit-doubles -m32bit-doubles -fpu -nofpu@gol
@@ -19909,7 +19910,7 @@ following options:
-msimple-fpu -mstring -mmulhw -mdlmzb -mmfpgpr -mvsx @gol
-mcrypto -mdirect-move -mpower8-fusion -mpower8-vector @gol
-mquad-memory -mquad-memory-atomic -mmodulo -mfloat128 -mfloat128-hardware @gol
--mpower9-fusion -mpower9-vector}
+-mpower9-fusion -mpower9-vector -mpower9-dform}
The particular options set for any particular CPU varies between
compiler versions, depending on what setting seems to produce optimal
@@ -20182,10 +20183,19 @@ processors.
@opindex mpower9-vector
@opindex mno-power9-vector
Generate code that uses (does not use) the vector and scalar
-instructions that were added in version 2.07 of the PowerPC ISA. Also
+instructions that were added in version 3.0 of the PowerPC ISA. Also
enable the use of built-in functions that allow more direct access to
the vector instructions.
+@item -mpower9-dform
+@itemx -mno-power9-dform
+@opindex mpower9-dform
+@opindex mno-power9-dform
+Enable (disable) scalar d-form (register + offset) memory instructions
+to load/store traditional Altivec registers. If the @var{LRA} register
+allocator is enabled, also enable (disable) vector d-form memory
+instructions.
+
@item -mfloat-gprs=@var{yes/single/double/no}
@itemx -mfloat-gprs
@opindex mfloat-gprs
Index: gcc/doc/md.texi
===================================================================
--- gcc/doc/md.texi (.../svn+ssh://meissner@gcc.gnu.org/svn/gcc/trunk/gcc/doc) (revision 235831)
+++ gcc/doc/md.texi (.../gcc/doc) (working copy)
@@ -3224,6 +3224,9 @@ Memory operand suitable for TOC fusion m
Int constant that is the element number that the MFVSRLD instruction
targets.
+@item wO
+A memory operand suitable for the ISA 3.0 vector d-form instructions.
+
@item wQ
A memory address that will work with the @code{lq} and @code{stq}
instructions.
Index: gcc/testsuite/gcc.target/powerpc/p8vector-int128-1.c
===================================================================
--- gcc/testsuite/gcc.target/powerpc/p8vector-int128-1.c (.../svn+ssh://meissner@gcc.gnu.org/svn/gcc/trunk/gcc/testsuite/gcc.target/powerpc) (revision 235831)
+++ gcc/testsuite/gcc.target/powerpc/p8vector-int128-1.c (.../gcc/testsuite/gcc.target/powerpc) (working copy)
@@ -2,7 +2,7 @@
/* { dg-skip-if "" { powerpc*-*-darwin* } { "*" } { "" } } */
/* { dg-require-effective-target powerpc_p8vector_ok } */
/* { dg-skip-if "do not override -mcpu" { powerpc*-*-* } { "-mcpu=*" } { "-mcpu=power8" } } */
-/* { dg-options "-mcpu=power8 -O3 -mvsx-timode" } */
+/* { dg-options "-mcpu=power8 -O3 -mvsx-timode -mlra" } */
#include <altivec.h>
Index: gcc/testsuite/gcc.target/powerpc/dform-3.c
===================================================================
--- gcc/testsuite/gcc.target/powerpc/dform-3.c (.../svn+ssh://meissner@gcc.gnu.org/svn/gcc/trunk/gcc/testsuite/gcc.target/powerpc) (revision 0)
+++ gcc/testsuite/gcc.target/powerpc/dform-3.c (.../gcc/testsuite/gcc.target/powerpc) (revision 235838)
@@ -0,0 +1,39 @@
+/* { dg-do compile { target { powerpc*-*-* && lp64 } } } */
+/* { dg-require-effective-target powerpc_p9vector_ok } */
+/* { dg-skip-if "do not override -mcpu" { powerpc*-*-* } { "-mcpu=*" } { "-mcpu=power9" } } */
+/* { dg-options "-mcpu=power9 -mpower9-dform -O2 -mlra" } */
+
+#ifndef TYPE
+#define TYPE vector double
+#endif
+
+struct foo {
+ TYPE a, b, c, d;
+};
+
+/* Test whether ISA 3.0 vector d-form instructions are implemented. */
+void
+add (struct foo *p)
+{
+ p->b = p->c + p->d;
+}
+
+/* Make sure we don't use direct moves to get stuff into GPR registers. */
+void
+gpr (struct foo *p)
+{
+ TYPE x = p->c;
+
+ __asm__ (" # reg = %0" : "+r" (x));
+
+ p->b = x;
+}
+
+/* { dg-final { scan-assembler "lxv " } } */
+/* { dg-final { scan-assembler "stxv " } } */
+/* { dg-final { scan-assembler-not "lxvx " } } */
+/* { dg-final { scan-assembler-not "stxvx " } } */
+/* { dg-final { scan-assembler-not "mfvsrd " } } */
+/* { dg-final { scan-assembler-not "mfvsrld " } } */
+/* { dg-final { scan-assembler "l\[dq\] " } } */
+/* { dg-final { scan-assembler "st\[dq\] " } } */
Index: gcc/testsuite/gcc.target/powerpc/pr68805.c
===================================================================
--- gcc/testsuite/gcc.target/powerpc/pr68805.c (.../svn+ssh://meissner@gcc.gnu.org/svn/gcc/trunk/gcc/testsuite/gcc.target/powerpc) (revision 235831)
+++ gcc/testsuite/gcc.target/powerpc/pr68805.c (.../gcc/testsuite/gcc.target/powerpc) (working copy)
@@ -1,6 +1,6 @@
/* { dg-do compile { target powerpc64le-*-* } } */
/* { dg-skip-if "do not override -mcpu" { powerpc*-*-* } { "-mcpu=*" } { "-mcpu=power8" } } */
-/* { dg-options "-O2 -mvsx-timode -mcpu=power8" } */
+/* { dg-options "-O2 -mvsx-timode -mcpu=power8 -mlra" } */
typedef struct bar {
void *a;
Index: gcc/testsuite/gcc.target/powerpc/dform-1.c
===================================================================
--- gcc/testsuite/gcc.target/powerpc/dform-1.c (.../svn+ssh://meissner@gcc.gnu.org/svn/gcc/trunk/gcc/testsuite/gcc.target/powerpc) (revision 235831)
+++ gcc/testsuite/gcc.target/powerpc/dform-1.c (.../gcc/testsuite/gcc.target/powerpc) (working copy)
@@ -1,7 +1,7 @@
/* { dg-do compile { target { powerpc*-*-* && lp64 } } } */
/* { dg-require-effective-target powerpc_p9vector_ok } */
/* { dg-skip-if "do not override -mcpu" { powerpc*-*-* } { "-mcpu=*" } { "-mcpu=power9" } } */
-/* { dg-options "-mcpu=power9 -mpower9-dform -O2" } */
+/* { dg-options "-mcpu=power9 -mpower9-dform -O2 -mlra" } */
#ifndef TYPE
#define TYPE double
Index: gcc/testsuite/gcc.target/powerpc/dform-2.c
===================================================================
--- gcc/testsuite/gcc.target/powerpc/dform-2.c (.../svn+ssh://meissner@gcc.gnu.org/svn/gcc/trunk/gcc/testsuite/gcc.target/powerpc) (revision 235831)
+++ gcc/testsuite/gcc.target/powerpc/dform-2.c (.../gcc/testsuite/gcc.target/powerpc) (working copy)
@@ -1,7 +1,7 @@
/* { dg-do compile { target { powerpc*-*-* && lp64 } } } */
/* { dg-require-effective-target powerpc_p9vector_ok } */
/* { dg-skip-if "do not override -mcpu" { powerpc*-*-* } { "-mcpu=*" } { "-mcpu=power9" } } */
-/* { dg-options "-mcpu=power9 -mpower9-dform -O2" } */
+/* { dg-options "-mcpu=power9 -mpower9-dform -O2 -mlra" } */
#ifndef TYPE
#define TYPE float
More information about the Gcc-patches
mailing list