Re: [PATCH V5] Use reg mode to move sub blocks for parameters and returns
Hi,
On 2023-06-05 00:59, Jeff Law wrote:
On 5/9/23 07:43, Jiufu Guo wrote:
Thanks for point out this! Yes, BLKmode rtx may not always be a MEM.
MEM_SIZE is only ok for MEM after the it's known size is computed.
Here MEM_SIZE is fine just because it is an stack rtx corresponding
to the type of parameter and returns which has been computed.
I updated the patch to resolve the conflicts with the trunk, and
retest bootstrap&testsuite, and then updated the patch a new version.
And this version pass bootstrap and regtest on ppc64{,le}, x86_64.
The major change is 'move_sub_blocks' only handles the case when
the block size can be move by same submode, or say (size % sub_size)
is 0. If no objection, I would committed the new version.
BR,
Jeff (Jiufu)
gcc/ChangeLog:
* cfgexpand.cc (expand_used_vars): Update to mark
DECL_USEDBY_RETURN_P
for returns.
* expr.cc (move_sub_blocks): New function.
(expand_assignment): Update assignment code about returns/parameters.
* function.cc (assign_parm_setup_block): Update to mark
DECL_REGS_TO_STACK_P for parameter.
* tree-core.h (struct tree_decl_common): Add comment.
* tree.h (DECL_USEDBY_RETURN_P): New define.
(DECL_REGS_TO_STACK_P): New define.
gcc/testsuite/ChangeLog:
* gcc.target/powerpc/pr65421-1.c: New test.
* gcc.target/powerpc/pr65421-2.c: New test.
I don't think this was ever explicitly ACK'd. OK for the trunk.
Thanks so much! And sorry for the late reply.
I'm trying to investigate another patch that may fix other PRs and also
could
handle this issue. So, I may suspend this for the new patch.
BR,
Jeff (Jiufu Guo)
jeff
Re: [PATCH V5] Use reg mode to move sub blocks for parameters and returns
On 5/9/23 07:43, Jiufu Guo wrote:
Thanks for point out this! Yes, BLKmode rtx may not always be a MEM.
MEM_SIZE is only ok for MEM after the it's known size is computed.
Here MEM_SIZE is fine just because it is an stack rtx corresponding
to the type of parameter and returns which has been computed.
I updated the patch to resolve the conflicts with the trunk, and
retest bootstrap&testsuite, and then updated the patch a new version.
And this version pass bootstrap and regtest on ppc64{,le}, x86_64.
The major change is 'move_sub_blocks' only handles the case when
the block size can be move by same submode, or say (size % sub_size)
is 0. If no objection, I would committed the new version.
BR,
Jeff (Jiufu)
gcc/ChangeLog:
* cfgexpand.cc (expand_used_vars): Update to mark DECL_USEDBY_RETURN_P
for returns.
* expr.cc (move_sub_blocks): New function.
(expand_assignment): Update assignment code about returns/parameters.
* function.cc (assign_parm_setup_block): Update to mark
DECL_REGS_TO_STACK_P for parameter.
* tree-core.h (struct tree_decl_common): Add comment.
* tree.h (DECL_USEDBY_RETURN_P): New define.
(DECL_REGS_TO_STACK_P): New define.
gcc/testsuite/ChangeLog:
* gcc.target/powerpc/pr65421-1.c: New test.
* gcc.target/powerpc/pr65421-2.c: New test.
I don't think this was ever explicitly ACK'd. OK for the trunk.
jeff
Re: [PATCH V5] Use reg mode to move sub blocks for parameters and returns
Hi,
Jeff Law writes:
> On 5/3/23 23:49, guojiufu wrote:
>> Hi,
>>
>> On 2023-05-01 03:00, Jeff Law wrote:
>>> On 3/16/23 21:39, Jiufu Guo wrote:
Hi,
When assigning a parameter to a variable, or assigning a variable to
return value with struct type, and the parameter/return is passed
through registers.
For this kind of case, it would be better to use the nature mode of
the registers to move the content for the assignment.
As the example code (like code in PR65421):
typedef struct SA {double a[3];} A;
A ret_arg_pt (A *a) {return *a;} // on ppc64le, expect only 3 lfd(s)
A ret_arg (A a) {return a;} // just empty fun body
void st_arg (A a, A *p) {*p = a;} //only 3 stfd(s)
Comparing with previous version:
https://gcc.gnu.org/pipermail/gcc-patches/2023-January/609394.html
This version refine code to eliminated reductant code in the sub
routine "move_sub_blocks".
Bootstrap and regtest pass on ppc64{,le}.
Is this ok for trunk?
>> ...
>>
diff --git a/gcc/expr.cc b/gcc/expr.cc
index 15be1c8db99..97a7be9542e 100644
--- a/gcc/expr.cc
+++ b/gcc/expr.cc
@@ -5559,6 +5559,41 @@ mem_ref_refers_to_non_mem_p (tree ref)
return non_mem_decl_p (base);
}
+/* Sub routine of expand_assignment, invoked when assigning from a
+ parameter or assigning to a return val on struct type which may
+ be passed through registers. The mode of register is used to
+ move the content for the assignment.
+
+ This routine generates code for expression FROM which is BLKmode,
+ and move the generated content to TO_RTX by su-blocks in SUB_MODE. */
+
+static void
+move_sub_blocks (rtx to_rtx, tree from, machine_mode sub_mode)
+{
+ gcc_assert (MEM_P (to_rtx));
+
+ HOST_WIDE_INT size = MEM_SIZE (to_rtx).to_constant ();
>>> Consider the case of a BLKmode return value. Isn't TO_RTX in this
>>> case a BLKmode object?
>>
>> Thanks for this question!
>>
>> Yes, the mode of TO_RTX is BLKmode.
>> As we know, when the function returns via registers, the mode of
>> the `return-rtx` could also be BLKmode. This patch is going to
>> improve these kinds of cases.
>>
>> For example:
>> ```
>> typedef struct FLOATS
>> {
>> double a[3];
>> } FLOATS;
>> FLOATS ret_arg_pt (FLOATS *a){return *a;}
>> ```
>>
>> D.3952 = *a_2(D); //this patch enhance this assignment
>> return D.3952;
>>
>> The mode is BLKmode for the rtx of `D.3952` is BLKmode, and the
>> rtx for "DECL_RESULT(current_function_decl)". And the DECL_RESULT
>> represents the return registers.
> I didn't think MEM_SIZE worked for BLKmode. BUt looking at its
> definition, it's pulling the size out of the attributes rather than
> from the mode. SO I guess there's a reasonable chance it's going to
> work :-)
Thanks for point out this! Yes, BLKmode rtx may not always be a MEM.
MEM_SIZE is only ok for MEM after the it's known size is computed.
Here MEM_SIZE is fine just because it is an stack rtx corresponding
to the type of parameter and returns which has been computed.
I updated the patch to resolve the conflicts with the trunk, and
retest bootstrap&testsuite, and then updated the patch a new version.
And this version pass bootstrap and regtest on ppc64{,le}, x86_64.
The major change is 'move_sub_blocks' only handles the case when
the block size can be move by same submode, or say (size % sub_size)
is 0. If no objection, I would committed the new version.
BR,
Jeff (Jiufu)
gcc/ChangeLog:
* cfgexpand.cc (expand_used_vars): Update to mark DECL_USEDBY_RETURN_P
for returns.
* expr.cc (move_sub_blocks): New function.
(expand_assignment): Update assignment code about returns/parameters.
* function.cc (assign_parm_setup_block): Update to mark
DECL_REGS_TO_STACK_P for parameter.
* tree-core.h (struct tree_decl_common): Add comment.
* tree.h (DECL_USEDBY_RETURN_P): New define.
(DECL_REGS_TO_STACK_P): New define.
gcc/testsuite/ChangeLog:
* gcc.target/powerpc/pr65421-1.c: New test.
* gcc.target/powerpc/pr65421-2.c: New test.
---
gcc/cfgexpand.cc | 14 +
gcc/expr.cc | 62
gcc/function.cc | 3 +
gcc/tree-core.h | 4 +-
gcc/tree.h | 9 +++
gcc/testsuite/gcc.target/powerpc/pr65421-1.c | 6 ++
gcc/testsuite/gcc.target/powerpc/pr65421-2.c | 33 +++
7 files changed, 130 insertions(+), 1 deletion(-)
create mode 100644 gcc/testsuite/gcc.target/powerpc/pr65421-1.c
create mode 100644 gcc/testsuite/gcc.target/powerpc/pr65421-2.c
diff --git a/gcc/cfgexpand.cc b/gcc/cfgexpand.cc
index 1a1b26b1c6c..7b6a2216492 100644
--- a/gcc/cfgexpand.cc
+++ b/gcc/cfgexpand.cc
@@ -2158,6 +2158,20 @@ expand_used_vars (bi
Re: [PATCH V5] Use reg mode to move sub blocks for parameters and returns
On 5/3/23 23:49, guojiufu wrote:
Hi,
On 2023-05-01 03:00, Jeff Law wrote:
On 3/16/23 21:39, Jiufu Guo wrote:
Hi,
When assigning a parameter to a variable, or assigning a variable to
return value with struct type, and the parameter/return is passed
through registers.
For this kind of case, it would be better to use the nature mode of
the registers to move the content for the assignment.
As the example code (like code in PR65421):
typedef struct SA {double a[3];} A;
A ret_arg_pt (A *a) {return *a;} // on ppc64le, expect only 3 lfd(s)
A ret_arg (A a) {return a;} // just empty fun body
void st_arg (A a, A *p) {*p = a;} //only 3 stfd(s)
Comparing with previous version:
https://gcc.gnu.org/pipermail/gcc-patches/2023-January/609394.html
This version refine code to eliminated reductant code in the sub
routine "move_sub_blocks".
Bootstrap and regtest pass on ppc64{,le}.
Is this ok for trunk?
...
diff --git a/gcc/expr.cc b/gcc/expr.cc
index 15be1c8db99..97a7be9542e 100644
--- a/gcc/expr.cc
+++ b/gcc/expr.cc
@@ -5559,6 +5559,41 @@ mem_ref_refers_to_non_mem_p (tree ref)
return non_mem_decl_p (base);
}
+/* Sub routine of expand_assignment, invoked when assigning from a
+ parameter or assigning to a return val on struct type which may
+ be passed through registers. The mode of register is used to
+ move the content for the assignment.
+
+ This routine generates code for expression FROM which is BLKmode,
+ and move the generated content to TO_RTX by su-blocks in
SUB_MODE. */
+
+static void
+move_sub_blocks (rtx to_rtx, tree from, machine_mode sub_mode)
+{
+ gcc_assert (MEM_P (to_rtx));
+
+ HOST_WIDE_INT size = MEM_SIZE (to_rtx).to_constant ();
Consider the case of a BLKmode return value. Isn't TO_RTX in this
case a BLKmode object?
Thanks for this question!
Yes, the mode of TO_RTX is BLKmode.
As we know, when the function returns via registers, the mode of
the `return-rtx` could also be BLKmode. This patch is going to
improve these kinds of cases.
For example:
```
typedef struct FLOATS
{
double a[3];
} FLOATS;
FLOATS ret_arg_pt (FLOATS *a){return *a;}
```
D.3952 = *a_2(D); //this patch enhance this assignment
return D.3952;
The mode is BLKmode for the rtx of `D.3952` is BLKmode, and the
rtx for "DECL_RESULT(current_function_decl)". And the DECL_RESULT
represents the return registers.
I didn't think MEM_SIZE worked for BLKmode. BUt looking at its
definition, it's pulling the size out of the attributes rather than from
the mode. SO I guess there's a reasonable chance it's going to work :-)
OK for the trunk.
jeff
Re: [PATCH V5] Use reg mode to move sub blocks for parameters and returns
Hi, On 2023-05-01 23:52, Segher Boessenkool wrote: Hi! On Fri, Mar 17, 2023 at 11:39:52AM +0800, Jiufu Guo wrote: gcc/testsuite/ChangeLog: * gcc.target/powerpc/pr65421-1.c: New test. * gcc.target/powerpc/pr65421.c: New test. Please name the tests something else? -1.c and -2.c maybe. Or something more inspired. Just not something that makes the less important of the (so far) two testcases look more important than it is. Right! Thanks for pointing out this! BR, Jeff (Jiufu) The testcases are fine otherwise, thanks! Segher
Re: [PATCH V5] Use reg mode to move sub blocks for parameters and returns
Hi,
On 2023-05-01 03:00, Jeff Law wrote:
On 3/16/23 21:39, Jiufu Guo wrote:
Hi,
When assigning a parameter to a variable, or assigning a variable to
return value with struct type, and the parameter/return is passed
through registers.
For this kind of case, it would be better to use the nature mode of
the registers to move the content for the assignment.
As the example code (like code in PR65421):
typedef struct SA {double a[3];} A;
A ret_arg_pt (A *a) {return *a;} // on ppc64le, expect only 3 lfd(s)
A ret_arg (A a) {return a;} // just empty fun body
void st_arg (A a, A *p) {*p = a;} //only 3 stfd(s)
Comparing with previous version:
https://gcc.gnu.org/pipermail/gcc-patches/2023-January/609394.html
This version refine code to eliminated reductant code in the sub
routine "move_sub_blocks".
Bootstrap and regtest pass on ppc64{,le}.
Is this ok for trunk?
...
diff --git a/gcc/expr.cc b/gcc/expr.cc
index 15be1c8db99..97a7be9542e 100644
--- a/gcc/expr.cc
+++ b/gcc/expr.cc
@@ -5559,6 +5559,41 @@ mem_ref_refers_to_non_mem_p (tree ref)
return non_mem_decl_p (base);
}
+/* Sub routine of expand_assignment, invoked when assigning from a
+ parameter or assigning to a return val on struct type which may
+ be passed through registers. The mode of register is used to
+ move the content for the assignment.
+
+ This routine generates code for expression FROM which is BLKmode,
+ and move the generated content to TO_RTX by su-blocks in SUB_MODE.
*/
+
+static void
+move_sub_blocks (rtx to_rtx, tree from, machine_mode sub_mode)
+{
+ gcc_assert (MEM_P (to_rtx));
+
+ HOST_WIDE_INT size = MEM_SIZE (to_rtx).to_constant ();
Consider the case of a BLKmode return value. Isn't TO_RTX in this
case a BLKmode object?
Thanks for this question!
Yes, the mode of TO_RTX is BLKmode.
As we know, when the function returns via registers, the mode of
the `return-rtx` could also be BLKmode. This patch is going to
improve these kinds of cases.
For example:
```
typedef struct FLOATS
{
double a[3];
} FLOATS;
FLOATS ret_arg_pt (FLOATS *a){return *a;}
```
D.3952 = *a_2(D); //this patch enhance this assignment
return D.3952;
The mode is BLKmode for the rtx of `D.3952` is BLKmode, and the
rtx for "DECL_RESULT(current_function_decl)". And the DECL_RESULT
represents the return registers.
BR,
Jeff (Jiufu)
It looks pretty good at this point.
jeff
Re: [PATCH V5] Use reg mode to move sub blocks for parameters and returns
Hi! On Fri, Mar 17, 2023 at 11:39:52AM +0800, Jiufu Guo wrote: > gcc/testsuite/ChangeLog: > > * gcc.target/powerpc/pr65421-1.c: New test. > * gcc.target/powerpc/pr65421.c: New test. Please name the tests something else? -1.c and -2.c maybe. Or something more inspired. Just not something that makes the less important of the (so far) two testcases look more important than it is. The testcases are fine otherwise, thanks! Segher
Re: [PATCH V5] Use reg mode to move sub blocks for parameters and returns
On 3/16/23 21:39, Jiufu Guo wrote:
Hi,
When assigning a parameter to a variable, or assigning a variable to
return value with struct type, and the parameter/return is passed
through registers.
For this kind of case, it would be better to use the nature mode of
the registers to move the content for the assignment.
As the example code (like code in PR65421):
typedef struct SA {double a[3];} A;
A ret_arg_pt (A *a) {return *a;} // on ppc64le, expect only 3 lfd(s)
A ret_arg (A a) {return a;} // just empty fun body
void st_arg (A a, A *p) {*p = a;} //only 3 stfd(s)
Comparing with previous version:
https://gcc.gnu.org/pipermail/gcc-patches/2023-January/609394.html
This version refine code to eliminated reductant code in the sub
routine "move_sub_blocks".
Bootstrap and regtest pass on ppc64{,le}.
Is this ok for trunk?
BR,
Jeff (Jiufu)
PR target/65421
gcc/ChangeLog:
* cfgexpand.cc (expand_used_vars): Update to mark DECL_USEDBY_RETURN_P
for returns.
* expr.cc (move_sub_blocks): New function.
(expand_assignment): Update assignment code about returns/parameters.
* function.cc (assign_parm_setup_block): Update to mark
DECL_REGS_TO_STACK_P for parameter.
* tree-core.h (struct tree_decl_common): Add comment.
* tree.h (DECL_USEDBY_RETURN_P): New define.
(DECL_REGS_TO_STACK_P): New define.
gcc/testsuite/ChangeLog:
* gcc.target/powerpc/pr65421-1.c: New test.
* gcc.target/powerpc/pr65421.c: New test.
---
gcc/cfgexpand.cc | 14 +
gcc/expr.cc | 61
gcc/function.cc | 3 +
gcc/tree-core.h | 4 +-
gcc/tree.h | 9 +++
gcc/testsuite/gcc.target/powerpc/pr65421-1.c | 6 ++
gcc/testsuite/gcc.target/powerpc/pr65421.c | 33 +++
7 files changed, 129 insertions(+), 1 deletion(-)
create mode 100644 gcc/testsuite/gcc.target/powerpc/pr65421-1.c
create mode 100644 gcc/testsuite/gcc.target/powerpc/pr65421.c
diff --git a/gcc/expr.cc b/gcc/expr.cc
index 15be1c8db99..97a7be9542e 100644
--- a/gcc/expr.cc
+++ b/gcc/expr.cc
@@ -5559,6 +5559,41 @@ mem_ref_refers_to_non_mem_p (tree ref)
return non_mem_decl_p (base);
}
+/* Sub routine of expand_assignment, invoked when assigning from a
+ parameter or assigning to a return val on struct type which may
+ be passed through registers. The mode of register is used to
+ move the content for the assignment.
+
+ This routine generates code for expression FROM which is BLKmode,
+ and move the generated content to TO_RTX by su-blocks in SUB_MODE. */
+
+static void
+move_sub_blocks (rtx to_rtx, tree from, machine_mode sub_mode)
+{
+ gcc_assert (MEM_P (to_rtx));
+
+ HOST_WIDE_INT size = MEM_SIZE (to_rtx).to_constant ();
Consider the case of a BLKmode return value. Isn't TO_RTX in this case
a BLKmode object?
It looks pretty good at this point.
jeff
[PATCH V5] Use reg mode to move sub blocks for parameters and returns
Hi,
When assigning a parameter to a variable, or assigning a variable to
return value with struct type, and the parameter/return is passed
through registers.
For this kind of case, it would be better to use the nature mode of
the registers to move the content for the assignment.
As the example code (like code in PR65421):
typedef struct SA {double a[3];} A;
A ret_arg_pt (A *a) {return *a;} // on ppc64le, expect only 3 lfd(s)
A ret_arg (A a) {return a;} // just empty fun body
void st_arg (A a, A *p) {*p = a;} //only 3 stfd(s)
Comparing with previous version:
https://gcc.gnu.org/pipermail/gcc-patches/2023-January/609394.html
This version refine code to eliminated reductant code in the sub
routine "move_sub_blocks".
Bootstrap and regtest pass on ppc64{,le}.
Is this ok for trunk?
BR,
Jeff (Jiufu)
PR target/65421
gcc/ChangeLog:
* cfgexpand.cc (expand_used_vars): Update to mark DECL_USEDBY_RETURN_P
for returns.
* expr.cc (move_sub_blocks): New function.
(expand_assignment): Update assignment code about returns/parameters.
* function.cc (assign_parm_setup_block): Update to mark
DECL_REGS_TO_STACK_P for parameter.
* tree-core.h (struct tree_decl_common): Add comment.
* tree.h (DECL_USEDBY_RETURN_P): New define.
(DECL_REGS_TO_STACK_P): New define.
gcc/testsuite/ChangeLog:
* gcc.target/powerpc/pr65421-1.c: New test.
* gcc.target/powerpc/pr65421.c: New test.
---
gcc/cfgexpand.cc | 14 +
gcc/expr.cc | 61
gcc/function.cc | 3 +
gcc/tree-core.h | 4 +-
gcc/tree.h | 9 +++
gcc/testsuite/gcc.target/powerpc/pr65421-1.c | 6 ++
gcc/testsuite/gcc.target/powerpc/pr65421.c | 33 +++
7 files changed, 129 insertions(+), 1 deletion(-)
create mode 100644 gcc/testsuite/gcc.target/powerpc/pr65421-1.c
create mode 100644 gcc/testsuite/gcc.target/powerpc/pr65421.c
diff --git a/gcc/cfgexpand.cc b/gcc/cfgexpand.cc
index 1a1b26b1c6c..eda4d85d140 100644
--- a/gcc/cfgexpand.cc
+++ b/gcc/cfgexpand.cc
@@ -2158,6 +2158,20 @@ expand_used_vars (bitmap forced_stack_vars)
frame_phase = off ? align - off : 0;
}
+ /* Collect VARs on returns. */
+ if (DECL_RESULT (current_function_decl))
+{
+ edge_iterator ei;
+ edge e;
+ FOR_EACH_EDGE (e, ei, EXIT_BLOCK_PTR_FOR_FN (cfun)->preds)
+ if (greturn *ret = safe_dyn_cast (last_stmt (e->src)))
+ {
+ tree val = gimple_return_retval (ret);
+ if (val && VAR_P (val))
+ DECL_USEDBY_RETURN_P (val) = 1;
+ }
+}
+
/* Set TREE_USED on all variables in the local_decls. */
FOR_EACH_LOCAL_DECL (cfun, i, var)
TREE_USED (var) = 1;
diff --git a/gcc/expr.cc b/gcc/expr.cc
index 15be1c8db99..97a7be9542e 100644
--- a/gcc/expr.cc
+++ b/gcc/expr.cc
@@ -5559,6 +5559,41 @@ mem_ref_refers_to_non_mem_p (tree ref)
return non_mem_decl_p (base);
}
+/* Sub routine of expand_assignment, invoked when assigning from a
+ parameter or assigning to a return val on struct type which may
+ be passed through registers. The mode of register is used to
+ move the content for the assignment.
+
+ This routine generates code for expression FROM which is BLKmode,
+ and move the generated content to TO_RTX by su-blocks in SUB_MODE. */
+
+static void
+move_sub_blocks (rtx to_rtx, tree from, machine_mode sub_mode)
+{
+ gcc_assert (MEM_P (to_rtx));
+
+ HOST_WIDE_INT size = MEM_SIZE (to_rtx).to_constant ();
+ HOST_WIDE_INT sub_size = GET_MODE_SIZE (sub_mode).to_constant ();
+ HOST_WIDE_INT len = size / sub_size;
+ gcc_assert (size % sub_size == 0);
+
+ push_temp_slots ();
+
+ rtx from_rtx = expand_expr (from, NULL_RTX, GET_MODE (to_rtx),
EXPAND_NORMAL);
+ for (int i = 0; i < len; i++)
+{
+ rtx temp = gen_reg_rtx (sub_mode);
+ rtx src = adjust_address (from_rtx, sub_mode, sub_size * i);
+ rtx dest = adjust_address (to_rtx, sub_mode, sub_size * i);
+ emit_move_insn (temp, src);
+ emit_move_insn (dest, temp);
+}
+
+ preserve_temp_slots (to_rtx);
+ pop_temp_slots ();
+ return;
+}
+
/* Expand an assignment that stores the value of FROM into TO. If NONTEMPORAL
is true, try generating a nontemporal store. */
@@ -6045,6 +6080,32 @@ expand_assignment (tree to, tree from, bool nontemporal)
return;
}
+ /* If it is assigning from a struct param which may be passed via registers,
+ it would be better to use the register's mode to move sub-blocks for the
+ assignment. */
+ if (TREE_CODE (from) == PARM_DECL && mode == BLKmode
+ && DECL_REGS_TO_STACK_P (from))
+{
+ rtx parm = DECL_INCOMING_RTL (from);
+ machine_mode sub_mode
+ = REG_P (parm) ? word_mode : GET_MODE (XEXP (XVECEXP (parm, 0, 0), 0));
+ move_sub_blocks (to_rtx, from, sub_mode);
+
