have a building kernel, can run old tests, but cannot run new ones. execv clearly needs work

kern/arch/mips/syscall/syscall.c

@@ -135,6 +135,8 @@ syscall(struct trapframe * tf)
 	 case SYS_fork:
 	  err = sys_fork(tf, (pid_t*)&retval);
 	  break;
+	 case SYS_execv:
+	  err = sys_execv((char *)tf->tf_a0,(userptr_t)tf->tf_a1);
  
 	default:
 	  kprintf("Unknown syscall %d\n", callno);

kern/compile/ASST2/.depend

@@ -374,7 +374,8 @@ proc_syscalls.o: ../../syscall/proc_syscalls.c ../../include/types.h \
   ../../include/setjmp.h includelinks/kern/machine/setjmp.h \
   ../../include/proc.h ../../include/limits.h ../../include/kern/limits.h \
   ../../include/synch.h ../../include/list.h ../../include/addrspace.h \
-  ../../include/vm.h includelinks/machine/vm.h ../../include/copyinout.h
+  ../../include/vm.h includelinks/machine/vm.h ../../include/copyinout.h \
+  ../../include/kern/fcntl.h
 runprogram.o: ../../syscall/runprogram.c ../../include/types.h \
   ../../include/kern/types.h includelinks/kern/machine/types.h \
   includelinks/machine/types.h ../../include/kern/errno.h \

kern/compile/ASST2/vers.c

@@ -1,3 +1,3 @@
 /* This file is automatically generated. Edits will be lost.*/
-const int buildversion = 123;
+const int buildversion = 125;
 const char buildconfig[] = "ASST2";

kern/compile/ASST2/version

@@ -1 +1 @@
-123
+125

kern/include/syscall.h

@@ -64,6 +64,7 @@ void sys__exit(int exitcode);
 int sys_getpid(pid_t *  retval);
 int sys_waitpid(pid_t pid, userptr_t status, int options, pid_t * retval);
 int sys_fork(struct trapframe * tf, int * retval);
+int sys_execv(const char * program, userptr_t args);
 
 #endif // UW
 

kern/syscall/proc_syscalls.c

@@ -13,6 +13,8 @@
 #include <addrspace.h>
 #include <copyinout.h>
 #include <synch.h>
+#include <kern/fcntl.h>
+
 
 void sys__exit(int exitcode)
 {
@@ -140,17 +142,19 @@ int sys_fork(struct trapframe * tf, int * retval)
 // args is an array of null terminated strings, with the last element being a null pointer so we don't overflow if iterating
 // since this is a userptr type, we need to use copyin and copyout to get data properly
 // hopefully the path in program is actually a full filepath
-int execv(const char * program, userptr_t args)
+int sys_execv(const char * program, userptr_t args)
 {
-  userptr_t ** temp = args;
+  userptr_t temp = args;
   int argcount = 0;
   char ** argv;
+  int errcode; // why don't we have errno!!! my beautful one-liners!!!
   
   // see how long args is
   while (1)
   {
     char * temp2;
-    if (copyin(temp, temp2, sizeof(char *))) return errno;
+    errcode = copyin((const userptr_t)temp, temp2, sizeof(char *));
+    if (errcode) return errcode;
     if (!(temp2)) break;
     temp += sizeof(char *);
     ++argcount;
@@ -164,17 +168,18 @@ int execv(const char * program, userptr_t args)
   temp = args;
   for (int i = 0; i < argcount; ++i)
   {
-    if (copyin(args, argv[i], sizeof(char *)))
+    errcode = copyin(temp, argv[i], sizeof(char *));
+    if (errcode)
     {
       kfree(argv);
-      return errno;
+      return errcode;
     }
     
     temp += sizeof(char *);
   }
   
   // allocate space for filepath
-  const char * filepath = kmalloc(sizeof(program));
+  char * filepath = kmalloc(sizeof(program));
   if (!(filepath))
   {
     kfree(argv);
@@ -183,63 +188,75 @@ int execv(const char * program, userptr_t args)
   
   // get kernel copy of filepath
   size_t useless;
-  if (copyinstr((const_userptr_t)program, filepath, strlen(program) + 1, &useless))
+  errcode = copyinstr((const_userptr_t)program, filepath, strlen(program) + 1, &useless);
+  if (errcode)
   {
     kfree(argv);
     kfree(filepath);
-    return errno;
+    return errcode;
   }
   
   // open program file
   struct vnode * v;
-  if (vfs_open(filepath, O_RDONLY, 0, &v))
+  errcode = vfs_open(filepath, O_RDONLY, 0, &v);
+  if (errcode)
   {
     kfree(argv);
     kfree(filepath);
-    return errno;
+    return errcode;
   }
   
   // create new address space
-	struct addrspace * as = as_create();
-	if (!(as))
-	{
-		vfs_close(v);
-		kfree(argv);
-		kfree(filepath);
-		return ENOMEM;
-	}
+  struct addrspace * as = as_create();
+  if (!(as))
+  {
+    vfs_close(v);
+    kfree(argv);
+    kfree(filepath);
+    return ENOMEM;
+  }
 
-	// Switch to the new as and activate. keep old in case we fail later
-	struct addrspace * oldas = curproc_getas();
-	curproc_setas(as);
-	as_activate();
+  // Switch to the new as and activate. keep old in case we fail later
+  struct addrspace * oldas = curproc_getas();
+  curproc_setas(as);
+  as_activate();
 
-	// Load santa's favourite slave
-	if (load_elf(v, &entrypoint))
-	{
-		vfs_close(v);
-		kfree(argv);
-		kfree(filepath);
-		curproc_setas(oldas);
-		as_destroy(as);
-		return errno;
-	}
+  vaddr_t entrypoint;
+  // Load santa's favourite slave
+  errcode = load_elf(v, &entrypoint);
+  if (errcode)
+  {
+    vfs_close(v);
+    kfree(argv);
+    kfree(filepath);
+    curproc_setas(oldas);
+    as_destroy(as);
+    return errcode;
+  }
 
-	// close the file now that we're done
-	vfs_close(v);
+  // close the file now that we're done
+  vfs_close(v);
   
   // create the new stack
   vaddr_t newstack;
-  if (as_define_stack(as, &newstack))
+  errcode = as_define_stack(as, &newstack);
+  if (errcode)
   {
     kfree(argv);
     kfree(filepath);
     curproc_setas(oldas);
     as_destroy(as);
-    return errno;
+    return errcode;
   }
   
   userptr_t userArgv;
   
   // need to copy data over into the new address space, as rn it is all in kernel heap
+
+  // delete old addrspace, enter new process
+  as_destroy(oldas);
+  enter_new_process(argcount, userArgv, newstack, entrypoint);
+  panic("Enter new process returned!\n");
+  return EINVAL;
 }
+