Add support for AIX systems:

* Memory leak in slurm_cred.c, added EVP_MD_CTX_cleanup().
* Pthread stack size too small on AIX. Resulting in stack corruption
  and ugly failure modes. Added slurm_attr_init to macros.h to
  explicitly set the stack size  for all pthreads.
* /dev/urandom not present on AIX, use rand() as needed instead in
  constructing a credential. Used in "srun --join".
* getsockopt(Socket, Level, SO_ERROR, &err, OptionLenght) sometime
  returns an error code of -1. This causes an assert failure in
  slurmd/io.c:_update_error_state().
* Function aliasing is not working on AIX. It is being turned off
  via a variable in config.h and "#if" logic in macros.h and
  slurm_xlator.h.
* dlopen failing if plugins reference any functions not present in
  caller. This may be fixed with the LDFLAG "-Wl,-bgcbypass=1000"
  being added for the slurm commands (avoid garbage collection of
  unused functions).
* read() is sometimes generates EAGAIN error, which was not handled
  in some places.
* vsnprintf() for string NULL is printing "" instead of "(null)"
  as produced by snprintf(). More format printing was added to log.c
  to produce more consistent log messages.
* poll() takes a timeout of -1 for unlimited rather than any negative
  number. Modify logic that was always multiplying by 1000 to convert
  usec to msec.
* getopt_long keyword table was not NULL terminated, resulting in
  segfault with invalid command-line argument in most commands.
* xmalloc module assert failures were not generating a core file.
  Changed "fatal();abort();" to "error();abort();".
* Change msg timeout from 3 sec to 5 sec. Running everything on single
  AIX node was very slow.

Author: Moe Jette

configure.ac

@@ -103,8 +103,12 @@ CFLAGS="$CFLAGS $PTHREAD_CFLAGS"
 LIBS="$PTHREAD_LIBS $LIBS"
 
 case "$host" in
-  *-*-aix*) LDFLAGS="$LDFLAGS -Wl,-brtl -Wl,-bexpall " ;;
+  *-*-aix*) LDFLAGS="$LDFLAGS -Wl,-brtl -Wl,-bexpall"
+            CMD_LDFLAGS="-Wl,-bgcbypass:1000" # keep all common functions 
+            AC_DEFINE(USE_ALIAS, 0, [Define slurm_ prefix function aliases for plusins]) ;;
+  *)        AC_DEFINE(USE_ALIAS, 1, [Define slurm_ prefix function aliases for plugins]) ;;
 esac
+AC_SUBST(CMD_LDFLAGS)
 
 AC_SLURM_SEMAPHORE
 

slurm/bnr.h

@@ -13,11 +13,11 @@ typedef int BNR_gid;
 #define BNR_SUCCESS 0
 #define BNR_ERROR   1
 
-int BNR_Init(BNR_gid *mygid);
-int BNR_Put(BNR_gid gid, char *attr, char *val);
-int BNR_Fence(BNR_gid gid);
-int BNR_Get(BNR_gid  gid, char *attr, char *val);
-int BNR_Finalize();
-int BNR_Rank(BNR_gid group, int *myrank);
-int BNR_Nprocs(BNR_gid group, int *nprocs);
+extern int BNR_Init(BNR_gid *mygid);
+extern int BNR_Put(BNR_gid gid, char *attr, char *val);
+extern int BNR_Fence(BNR_gid gid);
+extern int BNR_Get(BNR_gid  gid, char *attr, char *val);
+extern int BNR_Finalize();
+extern int BNR_Rank(BNR_gid group, int *myrank);
+extern int BNR_Nprocs(BNR_gid group, int *nprocs);
 

src/common/cbuf.c

@@ -1275,7 +1275,7 @@ cbuf_get_fd (void *dstbuf, int *psrcfd, int len)
 
     do {
         n = read(*psrcfd, dstbuf, len);
-    } while ((n < 0) && (errno == EINTR));
+    } while ((n < 0) && ((errno == EINTR) || (errno == EAGAIN)));
     return(n);
 }
 

src/common/log.c

@@ -414,6 +414,28 @@ static char *vxstrfmt(const char *fmt, va_list ap)
 				} else
 					xstrcat(buf, "%d");
 				break;
+			case 'u':
+				if (unprocessed == 0) {
+					snprintf(tmp, sizeof(tmp), "%u",
+						 va_arg(ap, int));
+					xstrcat(buf, tmp);
+				} else
+					xstrcat(buf, "%u");
+				break;
+			case 'l':
+				if ((unprocessed == 0) && (*(p+1) == 'u')) {
+					snprintf(tmp, sizeof(tmp), "%lu",
+						va_arg(ap, long unsigned));
+					xstrcat(buf, tmp);
+					p++;
+				} else if ((unprocessed==0) && (*(p+1)=='d')) {
+					snprintf(tmp, sizeof(tmp), "%ld",
+						va_arg(ap, long int));
+					xstrcat(buf, tmp);
+					p++;
+				} else
+					xstrcat(buf, "%l");
+				break; 
 			default:	/* try to handle the rest  */
 				xstrcatchar(buf, '%');
 				xstrcatchar(buf, *p);

src/common/macros.h

@@ -192,17 +192,47 @@ typedef enum {false, true} bool;
          }                                                                    \
      } _STMT_END
 
+#  ifdef PTHREAD_SCOPE_SYSTEM
+#  define slurm_attr_init(attr)                                               \
+     _STMT_START {                                                            \
+	if (pthread_attr_init(attr))                                          \
+		fatal("pthread_attr_init: %m");                               \
+	/* we want 1:1 threads if there is a choice */                        \
+	if (pthread_attr_setscope(attr, PTHREAD_SCOPE_SYSTEM))                \
+		error("pthread_attr_setscope: %m");                           \
+	if (pthread_attr_setstacksize(attr, 1024*1024))                       \
+		error("pthread_attr_setstacksize: %m");                       \
+     } _STMT_END
+#  else
+#  define slurm_attr_init(attr)                                               \
+     _STMT_START {                                                            \
+        if (pthread_attr_init(attr))                                          \
+                fatal("pthread_attr_init: %m");                               \
+        if (pthread_attr_setstacksize(attr, 1024*1024))                       \
+                error("pthread_attr_setstacksize: %m");                       \
+     } _STMT_END
+#  endif
+
 #else /* !WITH_PTHREADS */
 
 #  define slurm_mutex_init(mutex)
 #  define slurm_mutex_destroy(mutex)
 #  define slurm_mutex_lock(mutex)
 #  define slurm_mutex_unlock(mutex)
+#  define slurm_attr_init(attr)
 
 #endif /* WITH_PTHREADS */
 
 #ifndef strong_alias
-#  define strong_alias(name, aliasname) \
-   extern __typeof (name) aliasname __attribute ((alias (#name)))
+#  if USE_ALIAS
+#    define strong_alias(name, aliasname) \
+     extern __typeof (name) aliasname __attribute ((alias (#name)))
+#  else
+     /* dummy function definition,
+      * confirm "aliasname" is free and waste "name" */
+#    define strong_alias(name, aliasname) \
+     extern void aliasname(int name)
+#  endif
 #endif
+
 #endif /* !_MACROS_H */

src/common/slurm_cred.c

@@ -31,8 +31,10 @@
 
 #include <slurm/slurm_errno.h>
 
-#include <stdarg.h>
 #include <fcntl.h>
+#include <stdarg.h>
+#include <stdlib.h>
+#include <sys/time.h>
 
 /*
  * OpenSSL includes
@@ -395,14 +397,21 @@ slurm_cred_faker(slurm_cred_arg_t *arg)
 	cred->ctime  = time(NULL);
 	cred->siglen = SLURM_IO_KEY_SIZE;
 
-	if ((fd = open("/dev/urandom", O_RDONLY)) < 0)
-		error ("unable to open /dev/random: %m");
-
 	cred->signature = xmalloc(cred->siglen * sizeof(char));
-	read(fd, cred->signature, cred->siglen);
 
-	if (close(fd) < 0)
-		error ("close(/dev/random): %m");
+	if ((fd = open("/dev/urandom", O_RDONLY)) >= 0) {
+		read(fd, cred->signature, cred->siglen);
+		if (close(fd) < 0)
+		error ("close(/dev/urandom): %m");
+	} else {	/* Note: some systems lack this file */
+		unsigned int i;
+		struct timeval tv;
+		gettimeofday(&tv, NULL);
+		i = (unsigned int) (tv.tv_sec + tv.tv_usec);
+		srand((unsigned int) i);
+		for (i=0; i<cred->siglen; i++)
+			cred->signature[i] = (rand() & 0xff);
+	}
 
 	slurm_mutex_unlock(&cred->mutex);
 	return cred;
@@ -966,6 +975,7 @@ _slurm_cred_sign(slurm_cred_ctx_t ctx, slurm_cred_t cred)
 		rc = SLURM_ERROR;
 	}
 
+	EVP_MD_CTX_cleanup(&ectx);
 	free_buf(buffer);
 
 	return rc;

src/common/slurm_protocol_interface.h

@@ -65,7 +65,7 @@
 #include "src/common/pack.h"
 #include "src/common/slurm_protocol_common.h"
 
-#define SLURM_MESSAGE_TIMEOUT_MSEC_STATIC 3000
+#define SLURM_MESSAGE_TIMEOUT_MSEC_STATIC 5000
 
 /****************\
  **  Data Types  **

src/common/slurm_protocol_socket_implementation.c

@@ -252,7 +252,7 @@ int _slurm_send_timeout(slurm_fd fd, char *buf, size_t size,
 			goto done;
                 }
                 if ((rc = poll(&ufds, 1, timeout)) <= 0) {
-			if ((rc == 0) || (errno == EINTR)) 
+			if ((rc == 0) || (errno == EINTR) || (errno == EAGAIN)) 
  				continue;
 			else {
 				debug("_slurm_send_timeout at %d of %d, "
@@ -329,7 +329,7 @@ int _slurm_recv_timeout(slurm_fd fd, char *buffer, size_t size,
                 }
 
                 if ((rc = poll(&ufds, 1, timeout)) <= 0) {
-                        if ((errno == EINTR) || (rc == 0))
+                        if ((errno == EINTR) || (errno == EAGAIN) || (rc == 0))
                                 continue;
                         else {
 				debug("_slurm_recv_timeout at %d of %d, "

src/common/slurm_protocol_util.c

@@ -62,6 +62,7 @@ void init_header(header_t * header, slurm_msg_type_t msg_type,
 	header->version = SLURM_PROTOCOL_VERSION;
 	header->flags = flags;
 	header->msg_type = msg_type;
+	header->body_length = 0;	/* over-written later */
 }
 
 /*

src/common/slurm_xlator.h

@@ -10,7 +10,7 @@
  *
  *  All SLURM functions referenced from the switch and auth plugins should
  *  have aliases established. Functions not referenced from the plugins 
- *  need not be aliased.  
+ *  need not be aliased.
  *
  *  To use this header file:
  *  1. In the module containing the exported function code, establish an 
@@ -24,6 +24,8 @@
  *     and remove other slurm header files (they should all be in this header).
  *     This logic will have the plugin link only to the function names with 
  *     the "slurm_" prefix.
+ *
+ *  NOTE: Not all operating systems support this function aliasing (e.g. AIX). 
  *****************************************************************************
  *  Copyright (C) 2004 The Regents of the University of California.
  *  Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER).
@@ -51,6 +53,12 @@
 #ifndef __SLURM_XLATOR_H__
 #define __SLURM_XLATOR_H__
 
+#if HAVE_CONFIG_H
+#  include "config.h"
+#endif
+
+#if USE_ALIAS
+
 /* arg_desc.[ch] functions*/
 #define	arg_count		slurm_arg_count
 #define	arg_idx_by_name		slurm_arg_idx_by_name
@@ -233,6 +241,8 @@
 #define	xstrdup			slurm_xstrdup
 #define	xbasename		slurm_xbasename
 
+#endif /* USE_ALIAS */
+
 /* Include the function definitions after redefining their names. */
 #include "src/common/arg_desc.h"
 #include "src/common/bitstring.h"