btl/sm: rewrite of fast box (per-peer receive buffers)

I was investigating possibly lost btl/sm messages and realized that the code is
difficult to follow and it is not always clear what I was attempting to do. It
is not clear if there is a problem but the rewrite is worth committing.

This change does the following:

 - Seperate the fast box metadata out from the fast box receive data. These
   parts are logically separate so there was no need to keep adjusting the
   offset based on the metadata (start of buffer was offset 64, now 0).

 - Use modulo-math instead of toggling an extra bit to determine full vs
   empty. To keep this fast the modulo is done with bitwise-and with a
   mask and the fast box size has been limited to a power of two. This
   change simplifies the math and only has one special case to cover (
   end overflow-- end less than start).

 - General cleanup of the code overall to improve readability.

Signed-off-by: Nathan Hjelm <[email protected]>

Author: hjelmn

opal/mca/btl/sm/btl_sm_component.c

@@ -21,7 +21,7 @@
  * Copyright (c) 2018      Amazon.com, Inc. or its affiliates.  All Rights reserved.
  * Copyright (c) 2018      Triad National Security, LLC. All rights
  *                         reserved.
- * Copyright (c) 2019-2021 Google, Inc. All rights reserved.
+ * Copyright (c) 2019-2025 Google, Inc. All rights reserved.
  * Copyright (c) 2021      Nanook Consulting.  All rights reserved.
  * Copyright (c) 2022      IBM Corporation.  All rights reserved.
  * Copyright (c) 2022      Computer Architecture and VLSI Systems (CARV)
@@ -36,6 +36,7 @@
 
 #include "opal/mca/btl/base/btl_base_error.h"
 #include "opal/mca/threads/mutex.h"
+#include "opal/util/bit_ops.h"
 #include "opal/util/output.h"
 #include "opal/util/printf.h"
 
@@ -182,7 +183,7 @@ static int mca_btl_sm_component_register(void)
 
     mca_btl_sm_component.fbox_size = 4096;
     (void) mca_base_component_var_register(&mca_btl_sm_component.super.btl_version, "fbox_size",
-                                           "Size of per-peer fast transfer buffers (default: 4k)",
+                                           "Size of per-peer fast transfer buffers. Must be a power of two (default: 4k)",
                                            MCA_BASE_VAR_TYPE_UNSIGNED_INT, NULL, 0,
                                            MCA_BASE_VAR_FLAG_SETTABLE, OPAL_INFO_LVL_5,
                                            MCA_BASE_VAR_SCOPE_LOCAL,
@@ -324,8 +325,10 @@ mca_btl_sm_component_init(int *num_btls, bool enable_progress_threads, bool enab
         component->segment_size = (2 << 20);
     }
 
-    component->fbox_size = (component->fbox_size + MCA_BTL_SM_FBOX_ALIGNMENT_MASK)
-                           & ~MCA_BTL_SM_FBOX_ALIGNMENT_MASK;
+    if (component->fbox_size & (component->fbox_size - 1)) {
+        BTL_VERBOSE(("fast box size must be a power of two, rounding up to next power of two."));
+        component->fbox_size = opal_next_poweroftwo_inclusive(component->fbox_size);
+    }
 
     if (component->segment_size > (1ul << MCA_BTL_SM_OFFSET_BITS)) {
         component->segment_size = 2ul << MCA_BTL_SM_OFFSET_BITS;