xe: jit: gemm: rearrange runtime dim checks

This reverts commit d22e43f and splits
the runtime dimension checks to avoid undefined behavior.

src/gpu/intel/jit/gemm/xe_hp_systolic_gemm.cpp

@@ -67,16 +67,26 @@ status_t xe_hp_systolic_gemm_t::pd_t::init(impl::engine_t *engine) {
         c_zp_ = !attr()->zero_points_.has_default_values(DNNL_ARG_DST);
     }
 
-    VDISPATCH_GEMM_SC(
-            attr_.set_default_formats(dst_md(0)), VERBOSE_UNSUPPORTED_TAG);
-
-    VDISPATCH_GEMM(!use_nocopy(), VERBOSE_SKIP_PRIMITIVE_IMPL);
-
     // LIMITATIONS:
     // - batch is not supported for unpacked inputs.
     // - runtime dims are not supported
     bool limits_ok
             = !utils::one_of(DNNL_RUNTIME_DIM_VAL, d->m(), d->n(), d->k());
+
+    VDISPATCH_GEMM(limits_ok, VERBOSE_RUNTIMEDIM_UNSUPPORTED);
+
+    // Must check runtime dimensions before calling `set_default_formats` to
+    // avoid undefined behavior.
+    VDISPATCH_GEMM_SC(
+            set_default_formats(d->a_type()), VERBOSE_UNSUPPORTED_TAG);
+
+    VDISPATCH_GEMM_SC(
+            attr_.set_default_formats(dst_md(0)), VERBOSE_UNSUPPORTED_TAG);
+
+    VDISPATCH_GEMM(!use_nocopy(), VERBOSE_SKIP_PRIMITIVE_IMPL);
+
+    // `set_default_formats` determines a/b/c packing, so it must be called
+    // prior to this.
     if (!packed_a())
         limits_ok = limits_ok && (d->lda() != DNNL_RUNTIME_DIM_VAL)
                 && (d->batch() == 1);
@@ -85,10 +95,6 @@ status_t xe_hp_systolic_gemm_t::pd_t::init(impl::engine_t *engine) {
                 && (d->batch() == 1);
     if (!packed_c())
         limits_ok = limits_ok && (d->ldc() != DNNL_RUNTIME_DIM_VAL);
-    VDISPATCH_GEMM(limits_ok, VERBOSE_RUNTIMEDIM_UNSUPPORTED);
-
-    VDISPATCH_GEMM_SC(
-            set_default_formats(d->a_type()), VERBOSE_UNSUPPORTED_TAG);
 
     auto attr_skip_mask = smask_t::scales_runtime | smask_t::post_ops;
 
@@ -97,6 +103,7 @@ status_t xe_hp_systolic_gemm_t::pd_t::init(impl::engine_t *engine) {
     bool arch_ok = utils::one_of(arch, arch_t::xe_hp, arch_t::xe_hpg,
             arch_t::xe_hpc, arch_t::xe2, arch_t::xe3);
 
+    VDISPATCH_GEMM(limits_ok, VERBOSE_RUNTIMEDIM_UNSUPPORTED);
     VDISPATCH_GEMM((dt_float_ok || dt_int_ok), VERBOSE_UNSUPPORTED_DT_CFG);
     VDISPATCH_GEMM(arch_ok, VERBOSE_UNSUPPORTED_ARCH, "gpu");
     VDISPATCH_GEMM(