[Fix] Cherry pick MKL compile related PR in intel-xla (#2716)

Author: cboss6

WORKSPACE

@@ -32,7 +32,7 @@ load(
 #    curl -L https://github.com/openxla/xla/archive/<git hash>.tar.gz | sha256sum
 #    and update XLA_SHA256 with the result.
 
-XLA_COMMIT = "ba22d09548e4f087ff7899f5ac5aa918e291d591"
+XLA_COMMIT = "c3ddd5c135d5ae3a215dfc7f00f79646fdb69df2"
 
 new_git_repository(
     name = "intel_extension_for_openxla",

configure.py

@@ -758,6 +758,7 @@ def toolkit_exists(toolkit_path):
     mkl_path = os.path.join(home_path, 'mkl' + '/' + version + '/')
     environ_cp['ONEAPI_MKL_PATH'] = mkl_path    
   set_mkl_path(environ_cp)
+  mkl_path = environ_cp['ONEAPI_MKL_PATH']
   lib_path += ':' + '%slib/intel64' % (mkl_path)
   print('Configured oneMKL Toolkit path: %s\n' % (mkl_path))