Re: [PATCH v2] testsuite: Split out known-misaligned V2QI store target [PR124342] [PR124619]

Tue, 29 Sep 2026 11:53:15 -0700 



On 2026-09-24 15:56, Richard Biener wrote:

On Thu, Sep 24, 2026 at 3:34 PM Torbjorn SVENSSON
<[email protected]> wrote:




On 2026-09-24 15:25, Richard Biener wrote:

On Thu, Sep 24, 2026 at 2:24 PM Torbjorn SVENSSON
<[email protected]> wrote:




On 2026-09-23 15:40, Richard Biener wrote:

On Wed, Sep 2, 2026 at 1:18 PM Torbjörn Svensson via Sourceware Forge
<[email protected]> wrote:


From: Torbjörn SVENSSON <[email protected]>

vect_slp_v2qi_store_unalign tests whether the vectorizer handles a V2QI
store through a byte-aligned base, but some -Wstringop-overflow tests
instead depend on a V2QI store at a known one-byte misalignment from a
defined object.

These two variants are not the same on armv7ve with -mfloat-abi=hard.
Add a separate effective target for the latter case.


Can you again explain why they are not the same?  I'm guessing it's
about knowing the misalignment vs. unknown misalignment but then
vect_slp_v2qi_store_misalign vs. _unalign is not obvious.  Maybe
_unknown_misalign vs. _known_misalign?  But I'm still curious
why a target choses to allow one but not the other (for stores there's
no dr_explicit_realign[_optimized])?


Unless I've completly missunderstood the sequence, it should stem from
https://gcc.gnu.org/git/?p=gcc.git;a=blob;f=gcc/config/arm/arm.cc;hb=refs/heads/trunk#l30984

After another bisect, I identified that the first failing commit is
r16-4314-g5e9eecc6686 (for r16-4315 and later, r16-4355-g53085a46e7f162
is also needed to build).

I think the reason for the regression is that the value of "align" is
not the same for TYPE_ALIGN_UNIT(type) and GET_MODE_UNIT_SIZE(mode).

In the v2qi tests, mode would match HImode, thus:

- GET_MODE_UNIT_SIZE(HImode) would return 2.
- TYPE_ALIGN_UNIT(type) would return 1 for char.

So, maybe my patch is not correct and instead the function should
simply be like this since NEON apparently accepts unknown alignment:

static bool
arm_builtin_support_vector_misalignment (machine_mode mode,
                                           int misalignment,
                                           bool is_packed,
                                           bool is_gather_scatter)
{
     if (TARGET_NEON && !BYTES_BIG_ENDIAN && unaligned_access)
       {
         HOST_WIDE_INT align = GET_MODE_UNIT_SIZE (mode);

         if (is_gather_scatter)
          return true;

         if (is_packed)
           return align == 1;

         /* NEON supports access regardless of the alignment.  */
         return true;
       }

     return default_builtin_support_vector_misalignment (mode, misalignment,
                                                        is_packed,
                                                        is_gather_scatter);
}


WDYT?


Yeah, the current code is

        /* If the misalignment is unknown, we should be able to handle the 
access
           so long as it is not to a member of a packed data structure.  */
        if (misalignment == -1)
          return true;

        /* Return true if the misalignment is a multiple of the natural 
alignment
           of the vector's element type.  This is probably always going to be
           true in practice, since we've already established that this isn't a
           packed access.  */
        return ((misalignment % align) == 0);

which says it accepts any misalignment (misalignment == -1) but not
a known misalignment, of say 3.  That's just odd.


Also, the check "align == 1" in the is_packed branch is probably not
doing the right thing since the value in "align" is the size(?) and not
the alignment.


yeah, a vector mode of size 1 probably does not exist.  Iff we
accept misalignment == -1 there's no reason to reject "is_packed".
is_packed just means we cannot reach alignment by peeling.
There's no good reason to check is_packed.


So, based on this, maybe the function body could be reduced to:

{
    if (TARGET_NEON && !BYTES_BIG_ENDIAN && unaligned_access)
      return true;

    return default_builtin_support_vector_misalignment (mode, misalignment,
                                                         is_packed,
                                                         is_gather_scatter);
}

I can do this change and do a regression test run on it to see if it causes
any regressions if you think this is better.


Looks reasonable.  Note the aarch64 variant does

   if (TARGET_SIMD && STRICT_ALIGNMENT)
     {
       if (is_gather_scatter)
         return true;

       /* Return if movmisalign pattern is not supported for this mode.  */
       if (optab_handler (movmisalign_optab, mode) == CODE_FOR_nothing)
         return false;

       /* Misalignment factor is unknown at compile time.  */
       if (misalignment == -1)
         return false;
     }
   return default_builtin_support_vector_misalignment (mode, misalignment,
                                                       is_packed,
                                                       is_gather_scatter);

so basically defaults for !STRICT_ALIGNMENT.  I suppose the equivalent
is unaligned_access
on the arm side.  Compared to your variant with 'return true' it falls
through to the
default (which tests for movmisalign support of the mode).  So if arm
can do that for all
vector modes a return true sounds good, if not, fallthru to the
default hook looks safer.


I tried to use the aarch64 approach, but replaced TARGET_SIMD with TARGET_NEON.
This works for the test in question, but it causes other regressions for the
vect-early-break.
Then I tried to simply return true for any TARGET_NEON, same result.

After having tried all these ways, I got lazy and simply asked AI if it was
possible to solve this without having the "type" parameter and the conclusion
it gave me was:

There is no generic way for this hook to exactly reproduce
`TYPE_ALIGN_UNIT(type)` from only `mode`, `misalignment`, and the flags: two
accesses can have the same hook arguments but different element-type alignment.
The broad NEON `return true` passes `Wstringop-overflow-2.c` but adds four
`vect-early-break` failures. The mode-unit rule with an HImode adjustment still
adds those failures; the full comparison is also affected by different non-GCC
components in the prebuilt baseline versus this build.


I think it's safe to say that this might be an unsolvable problem.
Should we consider to revert the "type" parameter removal in 
r16-4314-g5e9eecc6686
and r16-4355-g53085a46e7f162?

Kind regards,
Torbjörn






The arm hook should probably be a copy of the aarch64 one unless
MVE has different constraints


I have no idea if there is a difference for MWE and/or aarch64 regarding this
hook. Earnshaw might have insight on this...?


Kind regards,
Torbjörn





           PR testsuite/124342
           PR testsuite/124619

gcc/ChangeLog:

           * doc/sourcebuild.texi (vect_slp_v2qi_store_unalign): Document.
           (vect_slp_v2qi_store_misalign): Likewise.

gcc/testsuite/ChangeLog:

           * lib/target-supports.exp (check_vect_slp_store_usage): Add
           TEST_V2QI_MISALIGN.  Renamed TEST_V2QI to TEST_V2QI_ALIGN and
           TEST_V2QI_2 to TEST_V2QI_UNALIGN.
           (check_effective_target_vect_slp_v2qi_store_misalign): New proc.
           (check_effective_target_vect_slp_v2qi_store_align): Define
           TEST_V2QI_ALIGN.
           (check_effective_target_vect_slp_v2qi_store_unalign): Define
           TEST_V2QI_UNALIGN.
           * c-c++-common/Wstringop-overflow-2.c: Use
           check_effective_target_vect_slp_v2qi_store_misalign.
           * g++.dg/warn/Wstringop-overflow-3.C: Likewise.

Signed-off-by: Torbjörn SVENSSON <[email protected]>
---
Changes since v1:

- Renamed define TEST_V2QI to TEST_V2QI_ALIGN
- Renamed define TEST_V2QI_2 to TEST_V2QI_UNALIGN


I've tested this change on top of r17-3738-gaaa0f6b1e87bc6 and 
r16-9620-g04c3e1efd4917e.

Ok for trunk and releases/gcc-16?


--



This is a forge pull request published on the gcc-patches mailing list mailing 
list
as requested by Torbjörn Svensson via Sourceware Forge 
<[email protected]>.
Forge discussion: https://forge.sourceware.org/gcc/gcc/pulls/215

Get it locally using:
```
git fetch forge-upstream "+refs/versioned_pull/215/*:refs/versioned_pull/215/*"
git switch -c "pr-215-v2" "refs/versioned_pull/215/2/head"
```
Or, download the patch at: https://forge.sourceware.org/gcc/gcc/pulls/215.diff
Created on: 2026-08-10 17:42:02+00:00
Latest update: 2026-09-02 11:17:13+00:00
Changes: 4 changed files, 81 additions, 51 deletions
Head revision: azoff/gcc ref pr/Wstringop-overflow commit 
3f5592502b2ba71cb4088daeae109915f431b6a9
Base revision: gcc/gcc ref trunk commit 
aaa0f6b1e87bc62ea15da6eab431af6dcebf6414 r17-3738-gaaa0f6b1e87bc6
Merge base: aaa0f6b1e87bc62ea15da6eab431af6dcebf6414
Requested Reviewers:

Changed files:
- M: gcc/doc/sourcebuild.texi
- M: gcc/testsuite/c-c++-common/Wstringop-overflow-2.c
- M: gcc/testsuite/g++.dg/warn/Wstringop-overflow-3.C
- M: gcc/testsuite/lib/target-supports.exp

    gcc/doc/sourcebuild.texi                      |  8 +++
    .../c-c++-common/Wstringop-overflow-2.c       | 50 +++++++++----------
    .../g++.dg/warn/Wstringop-overflow-3.C        | 44 ++++++++--------
    gcc/testsuite/lib/target-supports.exp         | 30 +++++++++--
    4 files changed, 81 insertions(+), 51 deletions(-)

diff --git a/gcc/doc/sourcebuild.texi b/gcc/doc/sourcebuild.texi
index aad51550eba40..5215d5a9529d5 100644
--- a/gcc/doc/sourcebuild.texi
+++ b/gcc/doc/sourcebuild.texi
@@ -1946,6 +1946,14 @@ Target supports loop vectorization with partial vectors 
and
    Target supports vectorization of 2-byte char stores with 2-byte aligned
    address at plain @option{-O2}.

+@item vect_slp_v2qi_store_unalign
+Target supports vectorization of 2-byte char stores with an unaligned
+address at plain @option{-O2}.
+
+@item vect_slp_v2qi_store_misalign
+Target supports vectorization of 2-byte char stores at a known one-byte
+misaligned address at plain @option{-O2}.
+
    @item vect_slp_v4qi_store_align
    Target supports vectorization of 4-byte char stores with 4-byte aligned
    address at plain @option{-O2}.
diff --git a/gcc/testsuite/c-c++-common/Wstringop-overflow-2.c 
b/gcc/testsuite/c-c++-common/Wstringop-overflow-2.c
index 3f9171947c252..4b0e443f28852 100644
--- a/gcc/testsuite/c-c++-common/Wstringop-overflow-2.c
+++ b/gcc/testsuite/c-c++-common/Wstringop-overflow-2.c
@@ -19,7 +19,7 @@ struct Ax ax_;
    void gax_ (void)
    {
      ax_.a[0] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" }
-  ax_.a[1] = 1;                 // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_unalign } } }
+  ax_.a[1] = 1;                 // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_misalign } } }
      ax_.a[2] = 2;                 // { dg-warning "\\\[-Wstringop-overflow" }
    }

@@ -30,7 +30,7 @@ struct Ax ax0 = { 0 };
    void gax0 (void)
    {
      ax0.a[0] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" }
-  ax0.a[1] = 1;                 // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_unalign } } }
+  ax0.a[1] = 1;                 // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_misalign } } }
      ax0.a[2] = 2;                 // { dg-warning "\\\[-Wstringop-overflow" }
    }

@@ -41,7 +41,7 @@ struct Ax ax0_ = { 0, { } };
    void gax0_ (void)
    {
      ax0_.a[0] = 0;                // { dg-warning "\\\[-Wstringop-overflow" }
-  ax0_.a[1] = 1;                // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_unalign } } }
+  ax0_.a[1] = 1;                // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_misalign } } }
      ax0_.a[2] = 2;                // { dg-warning "\\\[-Wstringop-overflow" }
    }

@@ -51,8 +51,8 @@ struct Ax ax1 = { 1, { 0 } };

    void gax1 (void)
    {
-  ax1.a[0] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" "" { 
target { vect_slp_v2qi_store_unalign } } }
-  ax1.a[1] = 1;                 // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_unalign } } }
+  ax1.a[0] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" "" { 
target { vect_slp_v2qi_store_misalign } } }
+  ax1.a[1] = 1;                 // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_misalign } } }
      ax1.a[2] = 2;                 // { dg-warning "\\\[-Wstringop-overflow" }
    }

@@ -100,7 +100,7 @@ struct A0 a0_;
    void ga0_ (void)
    {
      a0_.a[0] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" }
-  a0_.a[1] = 1;                 // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_unalign } } }
+  a0_.a[1] = 1;                 // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_misalign } } }
      a0_.a[2] = 2;                 // { dg-warning "\\\[-Wstringop-overflow" }
    }

@@ -111,7 +111,7 @@ struct A0 a00 = { 0 };
    void ga00 (void)
    {
      a00.a[0] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" }
-  a00.a[1] = 1;                 // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_unalign } } }
+  a00.a[1] = 1;                 // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_misalign } } }
      a00.a[2] = 2;                 // { dg-warning "\\\[-Wstringop-overflow" }
    }

@@ -122,7 +122,7 @@ struct A0 a00_ = { 0, { } };
    void ga00_ (void)
    {
      a00_.a[0] = 0;                // { dg-warning "\\\[-Wstringop-overflow" }
-  a00_.a[1] = 1;                // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_unalign } } }
+  a00_.a[1] = 1;                // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_misalign } } }
      a00_.a[2] = 2;                // { dg-warning "\\\[-Wstringop-overflow" }
    }

@@ -166,13 +166,13 @@ struct A1 a1_;

    void ga1_ (void)
    {
-  a1_.a[0] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" "" { 
target { vect_slp_v2qi_store_unalign } } }
-  a1_.a[1] = 1;                 // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_unalign } } }
+  a1_.a[0] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" "" { 
target { vect_slp_v2qi_store_misalign } } }
+  a1_.a[1] = 1;                 // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_misalign } } }
      a1_.a[2] = 2;                 // { dg-warning "\\\[-Wstringop-overflow" }

      struct A1 a;
-  a.a[0] = 0;                   // { dg-warning "\\\[-Wstringop-overflow" "" { 
target { vect_slp_v2qi_store_unalign } } }
-  a.a[1] = 1;                   // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_unalign } } }
+  a.a[0] = 0;                   // { dg-warning "\\\[-Wstringop-overflow" "" { 
target { vect_slp_v2qi_store_misalign } } }
+  a.a[1] = 1;                   // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_misalign } } }
      a.a[2] = 2;                   // { dg-warning "\\\[-Wstringop-overflow" }
      sink (&a);
    }
@@ -183,8 +183,8 @@ struct A1 a1__ = { 0 };

    void ga1__ (void)
    {
-  a1__.a[0] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" "" 
{ target { vect_slp_v2qi_store_unalign } } }
-  a1__.a[1] = 1;                 // { dg-warning "\\\[-Wstringop-overflow" "" 
{ xfail { vect_slp_v2qi_store_unalign } } }
+  a1__.a[0] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" "" 
{ target { vect_slp_v2qi_store_misalign } } }
+  a1__.a[1] = 1;                 // { dg-warning "\\\[-Wstringop-overflow" "" 
{ xfail { vect_slp_v2qi_store_misalign } } }
      a1__.a[2] = 2;                 // { dg-warning "\\\[-Wstringop-overflow" }

      struct A1 a = { 1 };
@@ -200,8 +200,8 @@ struct A1 a1_0 = { 0, { } };

    void ga1_0_ (void)
    {
-  a1_0.a[0] = 0;                // { dg-warning "\\\[-Wstringop-overflow" "" { 
target { vect_slp_v2qi_store_unalign } } }
-  a1_0.a[1] = 1;                // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_unalign } } }
+  a1_0.a[0] = 0;                // { dg-warning "\\\[-Wstringop-overflow" "" { 
target { vect_slp_v2qi_store_misalign } } }
+  a1_0.a[1] = 1;                // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_misalign } } }
      a1_0.a[2] = 2;                // { dg-warning "\\\[-Wstringop-overflow" }

      struct A1 a = { 1, { } };
@@ -217,8 +217,8 @@ struct A1 a1_1 = { 0, { 1 } };

    void ga1_1 (void)
    {
-  a1_1.a[0] = 0;                // { dg-warning "\\\[-Wstringop-overflow" "" { 
target { vect_slp_v2qi_store_unalign } } }
-  a1_1.a[1] = 1;                // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_unalign } } }
+  a1_1.a[0] = 0;                // { dg-warning "\\\[-Wstringop-overflow" "" { 
target { vect_slp_v2qi_store_misalign } } }
+  a1_1.a[1] = 1;                // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_misalign } } }
      a1_1.a[2] = 2;                // { dg-warning "\\\[-Wstringop-overflow" }

      struct A1 a = { 0, { 1 } };   // { dg-warning "\\\[-Wstringop-overflow" 
"pr102706" { target { vect_slp_v4qi_store_align } } }
@@ -266,12 +266,12 @@ struct A1i a1i_;
    void ga1i_ (void)
    {
      a1i_.a[0] = 0;
-  a1i_.a[1] = 1;                // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_unalign } } }
+  a1i_.a[1] = 1;                // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_misalign } } }
      a1i_.a[2] = 2;                // { dg-warning "\\\[-Wstringop-overflow" }

      struct A1i a;
      a.a[0] = 1;
-  a.a[1] = 2;                   // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_unalign } } }
+  a.a[1] = 2;                   // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_misalign } } }
      a.a[2] = 3;                   // { dg-warning "\\\[-Wstringop-overflow" }
      sink (&a);
    }
@@ -283,7 +283,7 @@ struct A1i a1i__ = { 0 };
    void ga1i__ (void)
    {
      a1i__.a[0] = 0;
-  a1i__.a[1] = 1;                // { dg-warning "\\\[-Wstringop-overflow" "" 
{ xfail { vect_slp_v2qi_store_unalign } } }
+  a1i__.a[1] = 1;                // { dg-warning "\\\[-Wstringop-overflow" "" 
{ xfail { vect_slp_v2qi_store_misalign } } }
      a1i__.a[2] = 2;                // { dg-warning "\\\[-Wstringop-overflow" }

      struct A1i a = { 0 };
@@ -299,8 +299,8 @@ struct A1 a1i_0 = { 0, { } };

    void ga1i_0_ (void)
    {
-  a1i_0.a[0] = 0;               // { dg-warning "\\\[-Wstringop-overflow" "" { 
target { vect_slp_v2qi_store_unalign } } }
-  a1i_0.a[1] = 1;               // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_unalign } } }
+  a1i_0.a[0] = 0;               // { dg-warning "\\\[-Wstringop-overflow" "" { 
target { vect_slp_v2qi_store_misalign } } }
+  a1i_0.a[1] = 1;               // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_misalign } } }
      a1i_0.a[2] = 2;               // { dg-warning "\\\[-Wstringop-overflow" }

      struct A1 a = { 0, { } };
@@ -316,8 +316,8 @@ struct A1 a1i_1 = { 0, { 1 } };

    void ga1i_1 (void)
    {
-  a1i_1.a[0] = 0;               // { dg-warning "\\\[-Wstringop-overflow" "" { 
target { vect_slp_v2qi_store_unalign } } }
-  a1i_1.a[1] = 1;               // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_unalign } } }
+  a1i_1.a[0] = 0;               // { dg-warning "\\\[-Wstringop-overflow" "" { 
target { vect_slp_v2qi_store_misalign } } }
+  a1i_1.a[1] = 1;               // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_misalign } } }
      a1i_1.a[2] = 2;               // { dg-warning "\\\[-Wstringop-overflow" }

      struct A1 a = { 0, { 1 } };   // { dg-warning "\\\[-Wstringop-overflow" 
"pr102462" { target { vect_slp_v4qi_store_align } } }
diff --git a/gcc/testsuite/g++.dg/warn/Wstringop-overflow-3.C 
b/gcc/testsuite/g++.dg/warn/Wstringop-overflow-3.C
index e1d547fe4d40c..0e8523b8ed182 100644
--- a/gcc/testsuite/g++.dg/warn/Wstringop-overflow-3.C
+++ b/gcc/testsuite/g++.dg/warn/Wstringop-overflow-3.C
@@ -21,7 +21,7 @@ Ax ax_;
    NOIPA void gax_ ()
    {
      ax_.a[0] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" }
-  ax_.a[1] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_unalign } } }
+  ax_.a[1] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_misalign } } }
      ax_.a[2] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" }
    }

@@ -32,7 +32,7 @@ Ax ax0 = { 0 };
    NOIPA void gax0 ()
    {
      ax0.a[0] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" }
-  ax0.a[1] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_unalign } } }
+  ax0.a[1] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_misalign } } }
      ax0.a[2] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" }
    }

@@ -43,7 +43,7 @@ Ax ax0_ = { 0, { } };
    NOIPA void gax0_ ()
    {
      ax0_.a[0] = 0;                // { dg-warning "\\\[-Wstringop-overflow" }
-  ax0_.a[1] = 0;                // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_unalign } } }
+  ax0_.a[1] = 0;                // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_misalign } } }
      ax0_.a[2] = 0;                // { dg-warning "\\\[-Wstringop-overflow" }
    }

@@ -53,8 +53,8 @@ Ax ax1 = { 1, { 0 } };

    NOIPA void gax1 ()
    {
-  ax1.a[0] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" "" { 
target { vect_slp_v2qi_store_unalign } } }
-  ax1.a[1] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_unalign } } }
+  ax1.a[0] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" "" { 
target { vect_slp_v2qi_store_misalign } } }
+  ax1.a[1] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_misalign } } }
      ax1.a[2] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" }
    }

@@ -102,7 +102,7 @@ A0 a0_;
    NOIPA void ga0_ ()
    {
      a0_.a[0] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" }
-  a0_.a[1] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_unalign } } }
+  a0_.a[1] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_misalign } } }
      a0_.a[2] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" }
    }

@@ -113,7 +113,7 @@ A0 a00 = { 0 };
    NOIPA void ga00 ()
    {
      a00.a[0] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" }
-  a00.a[1] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_unalign } } }
+  a00.a[1] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_misalign } } }
      a00.a[2] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" }
    }

@@ -124,7 +124,7 @@ A0 a00_ = { 0, { } };
    NOIPA void ga00_ ()
    {
      a00_.a[0] = 0;                // { dg-warning "\\\[-Wstringop-overflow" }
-  a00_.a[1] = 0;                // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_unalign } } }
+  a00_.a[1] = 0;                // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_misalign } } }
      a00_.a[2] = 0;                // { dg-warning "\\\[-Wstringop-overflow" }
    }

@@ -168,8 +168,8 @@ A1 a1_;

    NOIPA void ga1_ ()
    {
-  a1_.a[0] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" "" { 
target { vect_slp_v2qi_store_unalign } } }
-  a1_.a[1] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_unalign } } }
+  a1_.a[0] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" "" { 
target { vect_slp_v2qi_store_misalign } } }
+  a1_.a[1] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_misalign } } }
      a1_.a[2] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" }
    }

@@ -179,8 +179,8 @@ A1 a1__ = { 0 };

    NOIPA void ga1__ ()
    {
-  a1__.a[0] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" "" 
{ target { vect_slp_v2qi_store_unalign } } }
-  a1__.a[1] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" "" 
{ xfail { vect_slp_v2qi_store_unalign } } }
+  a1__.a[0] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" "" 
{ target { vect_slp_v2qi_store_misalign } } }
+  a1__.a[1] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" "" 
{ xfail { vect_slp_v2qi_store_misalign } } }
      a1__.a[2] = 0;                 // { dg-warning "\\\[-Wstringop-overflow" }
    }

@@ -190,8 +190,8 @@ A1 a1_0 = { 0, { } };

    NOIPA void ga1_0_ ()
    {
-  a1_0.a[0] = 0;                // { dg-warning "\\\[-Wstringop-overflow" "" { 
target { vect_slp_v2qi_store_unalign } } }
-  a1_0.a[1] = 0;                // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_unalign } } }
+  a1_0.a[0] = 0;                // { dg-warning "\\\[-Wstringop-overflow" "" { 
target { vect_slp_v2qi_store_misalign } } }
+  a1_0.a[1] = 0;                // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_misalign } } }
      a1_0.a[2] = 0;                // { dg-warning "\\\[-Wstringop-overflow" }
    }

@@ -201,8 +201,8 @@ A1 a1_1 = { 0, { 1 } };

    NOIPA void ga1_1 ()
    {
-  a1_1.a[0] = 0;                // { dg-warning "\\\[-Wstringop-overflow" "" { 
target { vect_slp_v2qi_store_unalign } } }
-  a1_1.a[1] = 0;                // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_unalign } } }
+  a1_1.a[0] = 0;                // { dg-warning "\\\[-Wstringop-overflow" "" { 
target { vect_slp_v2qi_store_misalign } } }
+  a1_1.a[1] = 0;                // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_misalign } } }
      a1_1.a[2] = 0;                // { dg-warning "\\\[-Wstringop-overflow" }
    }

@@ -244,7 +244,7 @@ A1i a1i_;
    NOIPA void ga1i_ ()
    {
      a1i_.a[0] = 0;
-  a1i_.a[1] = 0;                // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_unalign } } }
+  a1i_.a[1] = 0;                // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_misalign } } }
      a1i_.a[2] = 0;                // { dg-warning "\\\[-Wstringop-overflow" }
    }

@@ -255,7 +255,7 @@ A1i a1i__ = { 0 };
    NOIPA void ga1i__ ()
    {
      a1i__.a[0] = 0;
-  a1i__.a[1] = 0;                // { dg-warning "\\\[-Wstringop-overflow" "" 
{ xfail { vect_slp_v2qi_store_unalign } } }
+  a1i__.a[1] = 0;                // { dg-warning "\\\[-Wstringop-overflow" "" 
{ xfail { vect_slp_v2qi_store_misalign } } }
      a1i__.a[2] = 0;                // { dg-warning "\\\[-Wstringop-overflow" }
    }

@@ -265,8 +265,8 @@ A1 a1i_0 = { 0, { } };

    NOIPA void ga1i_0_ ()
    {
-  a1i_0.a[0] = 0;               // { dg-warning "\\\[-Wstringop-overflow" "" { 
target { vect_slp_v2qi_store_unalign } } }
-  a1i_0.a[1] = 0;               // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_unalign } } }
+  a1i_0.a[0] = 0;               // { dg-warning "\\\[-Wstringop-overflow" "" { 
target { vect_slp_v2qi_store_misalign } } }
+  a1i_0.a[1] = 0;               // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_misalign } } }
      a1i_0.a[2] = 0;               // { dg-warning "\\\[-Wstringop-overflow" }
    }

@@ -276,8 +276,8 @@ A1 a1i_1 = { 0, { 1 } };

    NOIPA void ga1i_1 ()
    {
-  a1i_1.a[0] = 0;               // { dg-warning "\\\[-Wstringop-overflow" "" { 
target { vect_slp_v2qi_store_unalign } } }
-  a1i_1.a[1] = 0;               // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_unalign } } }
+  a1i_1.a[0] = 0;               // { dg-warning "\\\[-Wstringop-overflow" "" { 
target { vect_slp_v2qi_store_misalign } } }
+  a1i_1.a[1] = 0;               // { dg-warning "\\\[-Wstringop-overflow" "" { 
xfail { vect_slp_v2qi_store_misalign } } }
      a1i_1.a[2] = 0;               // { dg-warning "\\\[-Wstringop-overflow" }
    }

diff --git a/gcc/testsuite/lib/target-supports.exp 
b/gcc/testsuite/lib/target-supports.exp
index d467e9fa3b7fd..e752532298901 100644
--- a/gcc/testsuite/lib/target-supports.exp
+++ b/gcc/testsuite/lib/target-supports.exp
@@ -9696,7 +9696,7 @@ proc check_vect_slp_store_usage { pattern macro } {
           {
               *(AC4*)a = Ac4;
           }
-       #elif TEST_V2QI
+       #elif TEST_V2QI_ALIGN
           struct A2
           {
               char a[2];
@@ -9709,7 +9709,7 @@ proc check_vect_slp_store_usage { pattern macro } {
               a.a[1] = 4;
               sink (&a);
           }
-       #elif TEST_V2QI_2
+       #elif TEST_V2QI_UNALIGN
           extern char p[2];
           void
           foo3_2 ()
@@ -9717,6 +9717,19 @@ proc check_vect_slp_store_usage { pattern macro } {
               p[0] = 0;
               p[1] = 1;
           }
+       #elif TEST_V2QI_MISALIGN
+       struct A2
+       {
+           char n;
+           char a[2];
+       };
+       struct A2 p __attribute__ ((aligned (2)));
+       void
+       foo3_misalign ()
+       {
+           p.a[0] = 0;
+           p.a[1] = 1;
+       }
           #elif TEST_V4HI
           struct Ax
           {
@@ -9812,7 +9825,7 @@ proc check_vect_slp_store_usage { pattern macro } {
    # is the exact match of relative testcases
    proc check_effective_target_vect_slp_v2qi_store_align { } {
        set pattern {add new stmt: MEM <vector\(2\) char>}
-    set macro "TEST_V2QI"
+    set macro "TEST_V2QI_ALIGN"
        return [check_cached_effective_target vect_slp_v2qi_store_align {
           expr [check_vect_slp_store_usage $pattern $macro] }]
    }
@@ -9821,11 +9834,20 @@ proc check_effective_target_vect_slp_v2qi_store_align { 
} {
    # with unaligned address at plain O2.
    proc check_effective_target_vect_slp_v2qi_store_unalign { } {
        set pattern {add new stmt: MEM <vector\(2\) char>}
-    set macro "TEST_V2QI_2"
+    set macro "TEST_V2QI_UNALIGN"
        return [check_cached_effective_target vect_slp_v2qi_store_unalign {
           expr [check_vect_slp_store_usage $pattern $macro ] }]
    }

+# Return the true if target support vectorization of 2-byte char stores
+# with a known one-byte misaligned address at plain O2.
+proc check_effective_target_vect_slp_v2qi_store_misalign { } {
+    set pattern {add new stmt: MEM <vector\(2\) char>}
+    set macro "TEST_V2QI_MISALIGN"
+    return [check_cached_effective_target vect_slp_v2qi_store_misalign {
+       expr [check_vect_slp_store_usage $pattern $macro ] }]
+}
+
    # Return the true if target support vectorization of 4-byte char stores
    # with 4-byte aligned address at plain O2.
    # NB: This target should be removed after real issues are fixed for

Range-diff against v1:
1:  1a3f4342b3ffa ! 1:  3f5592502b2ba testsuite: Split out known-misaligned 
V2QI store target [PR124342] [PR124619]
       @@ Commit message
            gcc/testsuite/ChangeLog:

                    * lib/target-supports.exp (check_vect_slp_store_usage): Add
       -            TEST_V2QI_MISALIGN.
       +            TEST_V2QI_MISALIGN.  Renamed TEST_V2QI to TEST_V2QI_ALIGN 
and
       +            TEST_V2QI_2 to TEST_V2QI_UNALIGN.
                    (check_effective_target_vect_slp_v2qi_store_misalign): New 
proc.
       -            * c-c++-common/Wstringop-overflow-2.c: Use it.
       +            (check_effective_target_vect_slp_v2qi_store_align): Define
       +            TEST_V2QI_ALIGN.
       +            (check_effective_target_vect_slp_v2qi_store_unalign): Define
       +            TEST_V2QI_UNALIGN.
       +            * c-c++-common/Wstringop-overflow-2.c: Use
       +            check_effective_target_vect_slp_v2qi_store_misalign.
                    * g++.dg/warn/Wstringop-overflow-3.C: Likewise.

            Signed-off-by: Torbjörn SVENSSON <[email protected]>
       @@ gcc/testsuite/g++.dg/warn/Wstringop-overflow-3.C: A1 a1i_1 = { 0, { 1 
} };


         ## gcc/testsuite/lib/target-supports.exp ##
       +@@ gcc/testsuite/lib/target-supports.exp: proc 
check_vect_slp_store_usage { pattern macro } {
       +   {
       +       *(AC4*)a = Ac4;
       +   }
       +-  #elif TEST_V2QI
       ++  #elif TEST_V2QI_ALIGN
       +   struct A2
       +   {
       +       char a[2];
       +@@ gcc/testsuite/lib/target-supports.exp: proc 
check_vect_slp_store_usage { pattern macro } {
       +       a.a[1] = 4;
       +       sink (&a);
       +   }
       +-  #elif TEST_V2QI_2
       ++  #elif TEST_V2QI_UNALIGN
       +   extern char p[2];
       +   void
       +   foo3_2 ()
        @@ gcc/testsuite/lib/target-supports.exp: proc 
check_vect_slp_store_usage { pattern macro } {
               p[0] = 0;
               p[1] = 1;
       @@ gcc/testsuite/lib/target-supports.exp: proc 
check_vect_slp_store_usage { pattern
           #elif TEST_V4HI
           struct Ax
           {
       -@@ gcc/testsuite/lib/target-supports.exp: proc 
check_effective_target_vect_slp_v2qi_store_unalign { } {
       +@@ gcc/testsuite/lib/target-supports.exp: proc 
check_vect_slp_store_usage { pattern macro } {
       + # is the exact match of relative testcases
       + proc check_effective_target_vect_slp_v2qi_store_align { } {
       +     set pattern {add new stmt: MEM <vector\(2\) char>}
       +-    set macro "TEST_V2QI"
       ++    set macro "TEST_V2QI_ALIGN"
       +     return [check_cached_effective_target vect_slp_v2qi_store_align {
       +   expr [check_vect_slp_store_usage $pattern $macro] }]
       + }
       +@@ gcc/testsuite/lib/target-supports.exp: proc 
check_effective_target_vect_slp_v2qi_store_align { } {
       + # with unaligned address at plain O2.
       + proc check_effective_target_vect_slp_v2qi_store_unalign { } {
       +     set pattern {add new stmt: MEM <vector\(2\) char>}
       +-    set macro "TEST_V2QI_2"
       ++    set macro "TEST_V2QI_UNALIGN"
       +     return [check_cached_effective_target vect_slp_v2qi_store_unalign {
           expr [check_vect_slp_store_usage $pattern $macro ] }]
         }

--
2.54.0





























					Reply via email to