[Clang][SYCL][LIT] Fix lb_sm_90.cpp when nvptx target is not enabled (#14289)

sarnex · web-flow · commit ddca45fbdfd7 · 2024-06-26T14:25:07.000Z
This fails for me constantly in `check-clang` locally because I don't
build nvptx. I verified this is UNSUPPORTED when nvptx is not built and
runs and passes when it is built.

Signed-off-by: Sarnie, Nick &lt;nick.sarnie@intel.com&gt;
diff --git a/clang/test/SemaSYCL/lb_sm_90.cpp b/clang/test/SemaSYCL/lb_sm_90.cpp
@@ -1,3 +1,4 @@
+// REQUIRES: nvptx-registered-target
 // RUN: %clang_cc1 -internal-isystem %S/Inputs %s -triple nvptx64-nvidia-cuda -target-cpu sm_90 -fsycl-is-device -Wno-c++23-extensions -verify -S -o %t
 
 // Maximum work groups per multi-processor, mapped to maxclusterrank PTX

Original file line number	Diff line number	Diff line change
`@@ -1,3 +1,4 @@`
	`1`	`+// REQUIRES: nvptx-registered-target`
`1`	`2`	`// RUN: %clang_cc1 -internal-isystem %S/Inputs %s -triple nvptx64-nvidia-cuda -target-cpu sm_90 -fsycl-is-device -Wno-c++23-extensions -verify -S -o %t`
`2`	`3`
`3`	`4`	`// Maximum work groups per multi-processor, mapped to maxclusterrank PTX`