1 files changed, 14 insertions, 18 deletions

diff --git a/pcilib/lock.c b/pcilib/lock.c
index b99ecfc..b92b11d 100644
--- a/pcilib/lock.c
+++ b/pcilib/lock.c
@@ -7,6 +7,7 @@
 #include "lock.h"
 #include "pci.h"
 #include <stdio.h>
+
 /*
  * this function will take the lock for the semaphore pointed by semId
  */
@@ -21,6 +22,7 @@ void pcilib_lock(pcilib_lock_t *lock_ctx, pcilib_lock_flags_t flags, ...){
 	if(errno!=EOWNERDEAD && err!=0) pcilib_error("can't acquire lock %s, errno %i",(char*)(lock_ctx+sizeof(pcilib_lock_t)),errno);
 	/** if the lock haven't been acquired and errno==EOWNERDEAD, it means the previous application that got the lock crashed, we have to remake the lock "consistent" so*/
 	else if(errno==EOWNERDEAD){
+	  /* one question is "is pthread_mutex_consistent protected in case we call twice it?", it seems to not make any importance in fact regarding man pages, but we have to survey it in future applications*/
 		pthread_mutex_consistent(lock_ctx);
 		pthread_mutex_lock(lock_ctx);
 		if(err!=0) pcilib_error("can't acquire lock %s, errno %i",(char*)(lock_ctx+sizeof(pcilib_lock_t)),errno);
@@ -31,7 +33,7 @@ void pcilib_lock(pcilib_lock_t *lock_ctx, pcilib_lock_flags_t flags, ...){
 	if(errno!=EOWNERDEAD && err!=0) pcilib_error("can't acquire lock %s, errno %i",(char*)(lock_ctx+sizeof(pcilib_lock_t)),errno);
 	else if(errno==EOWNERDEAD){
 		pthread_mutex_consistent(lock_ctx);
-		pthread_mutex_lock(lock_ctx);
+		pthread_mutex_trylock(lock_ctx);
 		if(err!=0) pcilib_error("can't acquire lock %s, errno %i",(char*)(lock_ctx+sizeof(pcilib_lock_t)),errno);
 	}
 	}
@@ -58,43 +60,37 @@ void pcilib_unlock(pcilib_lock_t* lock_ctx){
 }
 
 /**
- * pcilib_init_lock 
  * this function initialize a new semaphore in the kernel if it's not already initialized given the key that permits to differentiate semaphores, and then return the integer that points to the semaphore that have been initialized or to a previously already initialized semaphore
- * @param[out] lock_ctx the pointer that will points to the semaphore for other functions
- * @param[in] keysem the integer that permits to define to what the semaphore is attached
  */
-pcilib_lock_t* pcilib_init_lock(pcilib_t *ctx, char* lock_id, ...){
+pcilib_lock_t* pcilib_init_lock(pcilib_t *ctx, pcilib_lock_init_flags_t flag, char* lock_id, ...){
 	int err;
 	pthread_mutexattr_t attr;
 	int i,j;
 	void* addr,*locks_addr;
+	char buffer[PCILIB_LOCK_SIZE-sizeof(pcilib_lock_t)];
+	/* here lock_id is the format string for vsprintf, the lock name will so be put in adding arguments*/
 	va_list pa;
 	va_start(pa,lock_id);
-	
-	char* temp;
-	temp=malloc((strlen(lock_id)+strlen("bank_register_"))*sizeof(char));
-	sprintf(temp,"bank_register_%s",lock_id);
-	
-	pcilib_lock_init_flags_t flag;
-	flag=va_arg(pa,pcilib_lock_init_flags_t);
+	err=vsprintf(buffer,lock_id,pa);
 	va_end(pa);
 
-	if(strlen(temp)>PCILIB_LOCK_SIZE-sizeof(pcilib_lock_t)) pcilib_error("the entered protocol name is too long");
+	if(err<0) pcilib_error("error in obtaining the lock name");
 	if(((PCILIB_MAX_NUMBER_LOCKS*PCILIB_LOCK_SIZE)%PCILIB_KMEM_PAGE_SIZE)!=0) pcilib_error("PCILIB_MAX_NUMBER_LOCKS*PCILIB_LOCK_SIZE should be a multiple of kmem page size");
 	
 	addr=pcilib_kmem_get_block_ua(ctx,ctx->locks_handle,0);
-	if((flag & LOCK_INIT)==0) pcilib_lock((pcilib_lock_t*)addr,MUTEX_LOCK);
+	if((flag & PCILIB_NO_LOCK)==0) pcilib_lock((pcilib_lock_t*)addr,MUTEX_LOCK);
 	/* we search for the given lock if it was already initialized*/
 	for(j=0;j<PCILIB_NUMBER_OF_LOCK_PAGES;j++){
 		i=0;
 		locks_addr=pcilib_kmem_get_block_ua(ctx,ctx->locks_handle,j);
 			while((*(char*)(locks_addr+i*PCILIB_LOCK_SIZE+sizeof(pcilib_lock_t))!=0) && (i<PCILIB_LOCKS_PER_PAGE)){
-		  if(strcmp(temp,(char*)(locks_addr+i*PCILIB_LOCK_SIZE+sizeof(pcilib_lock_t)))==0){
+		  if(strcmp(buffer,(char*)(locks_addr+i*PCILIB_LOCK_SIZE+sizeof(pcilib_lock_t)))==0){
 			    return (pcilib_lock_t*)(locks_addr+i*PCILIB_LOCK_SIZE);}
 		 i++;
 		}
 		if(i<PCILIB_LOCKS_PER_PAGE) break;
 	}
+	/* the kernel space could be full*/
 	if(i==PCILIB_LOCKS_PER_PAGE) pcilib_error("no more free space for a new lock\n");
 	/* if not, we create a new one*/
 	if((err= pthread_mutexattr_init(&attr))!=0) pcilib_error("can't initialize mutex attribute, errno %i",errno);
@@ -103,15 +99,15 @@ pcilib_lock_t* pcilib_init_lock(pcilib_t *ctx, char* lock_id, ...){
 		if((err=pthread_mutex_init((pcilib_lock_t*)(locks_addr+i*PCILIB_LOCK_SIZE),&attr))!=0) pcilib_error("can't set attributes to mutex, errno : %i",errno);
 		pthread_mutexattr_destroy(&attr);
 
-		strcpy((char*)(locks_addr+i*PCILIB_LOCK_SIZE+sizeof(pcilib_lock_t)),temp);
+		strcpy((char*)(locks_addr+i*PCILIB_LOCK_SIZE+sizeof(pcilib_lock_t)),buffer);
 
-		if((flag & LOCK_INIT)==0) pcilib_unlock((pcilib_lock_t*)addr);
+		if((flag & PCILIB_NO_LOCK)==0) pcilib_unlock((pcilib_lock_t*)addr);
 		return (pcilib_lock_t*)(locks_addr+i*PCILIB_LOCK_SIZE);
 	
 }
 
 /*
- * pcilib_lock_free does nothing for now, as we don't want to erase possible locks in the kernel memory (or at any erasing, we should rearrange the locks to have no free space between locks). A thing i think of would be to use chained lists in kernel memory, but how?
+ * we uninitialize a mutex and set its name to 0 pointed by lock_ctx with this function. setting name to is the real destroying operation, but we need to unitialize the lock to initialize it again after
  */
 void pcilib_free_lock(pcilib_lock_t *lock_ctx){
    int err;