fix cache padding in device; fix spack package pmix option; fix READM…

…E typo;

README.md

@@ -81,7 +81,7 @@ make install
   - `ofi`: libfabrics, for all other networks (slingshot-11, ethernet, shared memory).
 - `LCI_FORCE_SERVER=ON/OFF`: Default value is `OFF`. If it is set to `ON`, 
   `LCI_SERVER` will not be treated as a hint but a requirement.
-- `LCI_WITH_LCT_ONLY=ON/OFF`: Whether to only build LCT (The Lightweight COmmunication Tools). 
+- `LCI_WITH_LCT_ONLY=ON/OFF`: Whether to only build LCT (The Lightweight Communication Tools). 
   Default is `OFF` (build both LCT and LCI).
 
 ## Run LCI applications

contrib/spack/packages/lci/package.py

@@ -74,7 +74,8 @@ def is_positive_int(val):
     variant('papi', default=False,
             description='Use PAPI to collect hardware counters')
     variant('gprof', default=False, description='Enable GPROF')
-    variant('enable-pmix', default=True, description='Enable PMIx as the process management backend')
+    variant('enable-pmix', default='auto', values=is_positive_int,
+            description='Enable PMIx as the process management backend')
 
     generator("ninja", "make", default="ninja")
 
@@ -110,9 +111,12 @@ def cmake_args(self):
             self.define_from_variant('LCI_ENABLE_SLOWDOWN', 'debug-slow'),
             self.define_from_variant('LCI_USE_PAPI', 'papi'),
             self.define_from_variant('LCI_USE_GPROF', 'gprof'),
-            self.define_from_variant('LCT_PMI_BACKEND_ENABLE_PMIX', 'enable-pmix'),
         ]
 
+        if self.spec.variants['enable-pmix'].value != 'auto':
+            arg = self.define_from_variant('LCT_PMI_BACKEND_ENABLE_PMIX', 'enable-pmix')
+            args.append(arg)
+
         if self.spec.variants['cache-line'].value != 'auto':
             arg = self.define_from_variant('LCI_CACHE_LINE', 'cache-line')
             args.append(arg)

lci/runtime/lcii.h

@@ -69,8 +69,8 @@ struct __attribute__((aligned(LCI_CACHE_LINE))) LCI_device_s {
   LCII_rcache_t rcache;                // 8B
   LCI_lbuffer_t heap;                  // 24B
   uintptr_t base_packet;               // 8B
-  LCIU_CACHE_PADDING(sizeof(LCIS_server_t) + 2 * sizeof(LCIS_endpoint_t) -
-                     sizeof(LCII_pool_t*) + sizeof(LCI_matchtable_t) -
+  LCIU_CACHE_PADDING(sizeof(LCIS_server_t) + 2 * sizeof(LCIS_endpoint_t) +
+                     sizeof(LCII_pool_t*) + sizeof(LCI_matchtable_t) +
                      sizeof(LCII_rcache_t*) + sizeof(LCI_lbuffer_t) +
                      sizeof(uintptr_t));
   // the following is shared by both progress threads and worker threads