Merge branch 'master' of git+ssh://amsterdam.csail.mit.edu/home/am0/6.828/xv6

bio.c

@@ -80,6 +80,8 @@ bget(uint dev, uint sector)
   }
 
   // Not cached; recycle some non-busy and clean buffer.
+  // "clean" because B_DIRTY and !B_BUSY means log.c
+  // hasn't yet committed the changes to the buffer.
   for(b = bcache.head.prev; b != &bcache.head; b = b->prev){
     if((b->flags & B_BUSY) == 0 && (b->flags & B_DIRTY) == 0){
       b->dev = dev;

defs.h

@@ -81,8 +81,8 @@ void            microdelay(int);
 // log.c
 void            initlog(void);
 void            log_write(struct buf*);
-void            begin_trans();
+void            begin_op();
-void            commit_trans();
+void            end_op();
 
 // mp.c
 extern int      ismp;

exec.c

@@ -18,9 +18,9 @@ exec(char *path, char **argv)
   struct proghdr ph;
   pde_t *pgdir, *oldpgdir;
 
-  begin_trans();
+  begin_op();
   if((ip = namei(path)) == 0){
-    commit_trans();
+    end_op();
     return -1;
   }
   ilock(ip);
@@ -50,7 +50,7 @@ exec(char *path, char **argv)
       goto bad;
   }
   iunlockput(ip);
-  commit_trans();
+  end_op();
   ip = 0;
 
   // Allocate two pages at the next page boundary.
@@ -101,7 +101,7 @@ exec(char *path, char **argv)
     freevm(pgdir);
   if(ip){
     iunlockput(ip);
-    commit_trans();
+    end_op();
   }
   return -1;
 }

file.c

@@ -72,9 +72,9 @@ fileclose(struct file *f)
   if(ff.type == FD_PIPE)
     pipeclose(ff.pipe, ff.writable);
   else if(ff.type == FD_INODE){
-    begin_trans();
+    begin_op();
     iput(ff.ip);
-    commit_trans();
+    end_op();
   }
 }
 
@@ -136,12 +136,12 @@ filewrite(struct file *f, char *addr, int n)
       if(n1 > max)
         n1 = max;
 
-      begin_trans();
+      begin_op();
       ilock(f->ip);
       if ((r = writei(f->ip, addr + i, f->off, n1)) > 0)
         f->off += r;
       iunlock(f->ip);
-      commit_trans();
+      end_op();
 
       if(r < 0)
         break;

log.c

@@ -5,18 +5,19 @@
 #include "fs.h"
 #include "buf.h"
 
-// Simple logging. Each file system system call
+// Simple logging that allows concurrent FS system calls.
-// should be surrounded with begin_trans() and commit_trans() calls.
 //
-// The log holds at most one transaction at a time. Commit forces
+// A log transaction contains the updates of multiple FS system
-// the log (with commit record) to disk, then installs the affected
+// calls. The logging system only commits when there are
-// blocks to disk, then erases the log. begin_trans() ensures that
+// no FS system calls active. Thus there is never
-// only one system call can be in a transaction; others must wait.
+// any reasoning required about whether a commit might
+// write an uncommitted system call's updates to disk.
 //
-// Allowing only one transaction at a time means that the file
+// A system call should call begin_op()/end_op() to mark
-// system code doesn't have to worry about the possibility of
+// its start and end. Usually begin_op() just increments
-// one transaction reading a block that another one has modified,
+// the count of in-progress FS system calls and returns.
-// for example an i-node block.
+// But if it thinks the log is close to running out, it
+// sleeps until the last outstanding end_op() commits.
 //
 // The log is a physical re-do log containing disk blocks.
 // The on-disk log format:
@@ -38,13 +39,15 @@ struct log {
   struct spinlock lock;
   int start;
   int size;
-  int busy; // a transaction is active
+  int outstanding; // how many FS sys calls are executing.
+  int committing;  // in commit(), please wait.
   int dev;
   struct logheader lh;
 };
 struct log log;
 
 static void recover_from_log(void);
+static void commit();
 
 void
 initlog(void)
@@ -117,36 +120,88 @@ recover_from_log(void)
   write_head(); // clear the log
 }
 
+// called at the start of each FS system call.
 void
-begin_trans(void)
+begin_op(void)
 {
   acquire(&log.lock);
-  while (log.busy) {
+  while(1){
+    if(log.committing){
       sleep(&log, &log.lock);
-  }
+    } else if(log.lh.n + (log.outstanding+1)*MAXOPBLOCKS > LOGSIZE){
-  log.busy = 1;
+      // this op might exhaust log space; wait for commit.
+      sleep(&log, &log.lock);
+    } else {
+      log.outstanding += 1;
       release(&log.lock);
+      break;
+    }
+  }
 }
 
+// called at the end of each FS system call.
+// commits if this was the last outstanding operation.
 void
-commit_trans(void)
+end_op(void)
+{
+  int do_commit = 0;
+
+  acquire(&log.lock);
+  log.outstanding -= 1;
+  if(log.committing)
+    panic("log.committing");
+  if(log.outstanding == 0){
+    do_commit = 1;
+    log.committing = 1;
+  } else {
+    // begin_op() may be waiting for log space.
+    wakeup(&log);
+  }
+  release(&log.lock);
+
+  if(do_commit){
+    // call commit w/o holding locks, since not allowed
+    // to sleep with locks.
+    commit();
+    acquire(&log.lock);
+    log.committing = 0;
+    wakeup(&log);
+    release(&log.lock);
+  }
+}
+
+// Copy modified blocks from cache to log.
+static void 
+write_log(void)
+{
+  int tail;
+
+  for (tail = 0; tail < log.lh.n; tail++) {
+    struct buf *to = bread(log.dev, log.start+tail+1); // log block
+    struct buf *from = bread(log.dev, log.lh.sector[tail]); // cache block
+    memmove(to->data, from->data, BSIZE);
+    bwrite(to);  // write the log
+    brelse(from); 
+    brelse(to);
+  }
+}
+
+static void
+commit()
 {
   if (log.lh.n > 0) {
+    write_log();     // Write modified blocks from cache to log
     write_head();    // Write header to disk -- the real commit
     install_trans(); // Now install writes to home locations
     log.lh.n = 0; 
     write_head();    // Erase the transaction from the log
   }
-  
-  acquire(&log.lock);
-  log.busy = 0;
-  wakeup(&log);
-  release(&log.lock);
 }
 
 // Caller has modified b->data and is done with the buffer.
-// Append the block to the log and record the block number, 
+// Record the block number and pin in the cache with B_DIRTY.
-// but don't write the log header (which would commit the write).
+// commit()/write_log() will do the disk write.
+//
 // log_write() replaces bwrite(); a typical use is:
 //   bp = bread(...)
 //   modify bp->data[]
@@ -159,21 +214,17 @@ log_write(struct buf *b)
 
   if (log.lh.n >= LOGSIZE || log.lh.n >= log.size - 1)
     panic("too big a transaction");
-  if (!log.busy)
+  if (log.outstanding < 1)
-    panic("write outside of trans");
+    panic("log_write outside of trans");
 
   for (i = 0; i < log.lh.n; i++) {
-    if (log.lh.sector[i] == b->sector)   // log absorbtion?
+    if (log.lh.sector[i] == b->sector)   // log absorbtion
       break;
   }
   log.lh.sector[i] = b->sector;
-  struct buf *lbuf = bread(b->dev, log.start+i+1);
-  memmove(lbuf->data, b->data, BSIZE);
-  bwrite(lbuf);
-  brelse(lbuf);
   if (i == log.lh.n)
     log.lh.n++;
-  b->flags |= B_DIRTY; // XXX prevent eviction
+  b->flags |= B_DIRTY; // prevent eviction
 }
 
 //PAGEBREAK!

mkfs.c

@@ -13,7 +13,7 @@
 
 #define static_assert(a, b) do { switch (0) case 0: case (a): ; } while (0)
 
-int nblocks = 985;
+int nblocks = (995-LOGSIZE);
 int nlog = LOGSIZE;
 int ninodes = 200;
 int size = 1024;

param.h

@@ -3,10 +3,11 @@
 #define NCPU          8  // maximum number of CPUs
 #define NOFILE       16  // open files per process
 #define NFILE       100  // open files per system
-#define NBUF         10  // size of disk block cache
 #define NINODE       50  // maximum number of active i-nodes
 #define NDEV         10  // maximum major device number
 #define ROOTDEV       1  // device number of file system root disk
 #define MAXARG       32  // max exec arguments
-#define LOGSIZE      10  // max data sectors in on-disk log
+#define MAXOPBLOCKS  10  // max # of blocks any FS op writes
+#define LOGSIZE      (MAXOPBLOCKS*3)  // max data sectors in on-disk log
+#define NBUF         (MAXOPBLOCKS*3)  // size of disk block cache (>= LOGSIZE)
 

proc.c

@@ -186,9 +186,9 @@ exit(void)
     }
   }
 
-  begin_trans();
+  begin_op();
   iput(proc->cwd);
-  commit_trans();
+  end_op();
   proc->cwd = 0;
 
   acquire(&ptable.lock);

sysfile.c

@@ -121,16 +121,16 @@ sys_link(void)
   if(argstr(0, &old) < 0 || argstr(1, &new) < 0)
     return -1;
 
-  begin_trans();
+  begin_op();
   if((ip = namei(old)) == 0){
-    commit_trans();
+    end_op();
     return -1;
   }
 
   ilock(ip);
   if(ip->type == T_DIR){
     iunlockput(ip);
-    commit_trans();
+    end_op();
     return -1;
   }
 
@@ -148,7 +148,7 @@ sys_link(void)
   iunlockput(dp);
   iput(ip);
 
-  commit_trans();
+  end_op();
 
   return 0;
 
@@ -157,7 +157,7 @@ bad:
   ip->nlink--;
   iupdate(ip);
   iunlockput(ip);
-  commit_trans();
+  end_op();
   return -1;
 }
 
@@ -189,9 +189,9 @@ sys_unlink(void)
   if(argstr(0, &path) < 0)
     return -1;
 
-  begin_trans();
+  begin_op();
   if((dp = nameiparent(path, name)) == 0){
-    commit_trans();
+    end_op();
     return -1;
   }
 
@@ -225,13 +225,13 @@ sys_unlink(void)
   iupdate(ip);
   iunlockput(ip);
 
-  commit_trans();
+  end_op();
 
   return 0;
 
 bad:
   iunlockput(dp);
-  commit_trans();
+  end_op();
   return -1;
 }
 
@@ -291,23 +291,23 @@ sys_open(void)
   if(argstr(0, &path) < 0 || argint(1, &omode) < 0)
     return -1;
 
-  begin_trans();
+  begin_op();
 
   if(omode & O_CREATE){
     ip = create(path, T_FILE, 0, 0);
     if(ip == 0){
-      commit_trans();
+      end_op();
       return -1;
     }
   } else {
     if((ip = namei(path)) == 0){
-      commit_trans();
+      end_op();
       return -1;
     }
     ilock(ip);
     if(ip->type == T_DIR && omode != O_RDONLY){
       iunlockput(ip);
-      commit_trans();
+      end_op();
       return -1;
     }
   }
@@ -316,11 +316,11 @@ sys_open(void)
     if(f)
       fileclose(f);
     iunlockput(ip);
-    commit_trans();
+    end_op();
     return -1;
   }
   iunlock(ip);
-  commit_trans();
+  end_op();
 
   f->type = FD_INODE;
   f->ip = ip;
@@ -336,13 +336,13 @@ sys_mkdir(void)
   char *path;
   struct inode *ip;
 
-  begin_trans();
+  begin_op();
   if(argstr(0, &path) < 0 || (ip = create(path, T_DIR, 0, 0)) == 0){
-    commit_trans();
+    end_op();
     return -1;
   }
   iunlockput(ip);
-  commit_trans();
+  end_op();
   return 0;
 }
 
@@ -354,16 +354,16 @@ sys_mknod(void)
   int len;
   int major, minor;
   
-  begin_trans();
+  begin_op();
   if((len=argstr(0, &path)) < 0 ||
      argint(1, &major) < 0 ||
      argint(2, &minor) < 0 ||
      (ip = create(path, T_DEV, major, minor)) == 0){
-    commit_trans();
+    end_op();
     return -1;
   }
   iunlockput(ip);
-  commit_trans();
+  end_op();
   return 0;
 }
 
@@ -373,20 +373,20 @@ sys_chdir(void)
   char *path;
   struct inode *ip;
 
-  begin_trans();
+  begin_op();
   if(argstr(0, &path) < 0 || (ip = namei(path)) == 0){
-    commit_trans();
+    end_op();
     return -1;
   }
   ilock(ip);
   if(ip->type != T_DIR){
     iunlockput(ip);
-    commit_trans();
+    end_op();
     return -1;
   }
   iunlock(ip);
   iput(proc->cwd);
-  commit_trans();
+  end_op();
   proc->cwd = ip;
   return 0;
 }

usertests.c

@@ -512,18 +512,20 @@ sharedfd(void)
   }
 }
 
-// two processes write two different files at the same
+// four processes write different files at the same
 // time, to test block allocation.
 void
-twofiles(void)
+fourfiles(void)
 {
-  int fd, pid, i, j, n, total;
+  int fd, pid, i, j, n, total, pi;
+  char *names[] = { "f0", "f1", "f2", "f3" };
   char *fname;
 
-  printf(1, "twofiles test\n");
+  printf(1, "fourfiles test\n");
 
-  unlink("f1");
+  for(pi = 0; pi < 4; pi++){
-  unlink("f2");
+    fname = names[pi];
+    unlink(fname);
 
     pid = fork();
     if(pid < 0){
@@ -531,32 +533,35 @@ twofiles(void)
       exit();
     }
 
-  fname = pid ? "f1" : "f2";
+    if(pid == 0){
       fd = open(fname, O_CREATE | O_RDWR);
       if(fd < 0){
         printf(1, "create failed\n");
         exit();
       }
       
-  memset(buf, pid?'p':'c', 512);
+      memset(buf, '0'+pi, 512);
       for(i = 0; i < 12; i++){
         if((n = write(fd, buf, 500)) != 500){
           printf(1, "write failed %d\n", n);
           exit();
         }
       }
-  close(fd);
-  if(pid)
-    wait();
-  else
       exit();
+    }
+  }
+
+  for(pi = 0; pi < 4; pi++){
+    wait();
+  }
 
   for(i = 0; i < 2; i++){
-    fd = open(i?"f1":"f2", 0);
+    fname = names[i];
+    fd = open(fname, 0);
     total = 0;
     while((n = read(fd, buf, sizeof(buf))) > 0){
       for(j = 0; j < n; j++){
-        if(buf[j] != (i?'p':'c')){
+        if(buf[j] != '0'+i){
           printf(1, "wrong char\n");
           exit();
         }
@@ -568,30 +573,31 @@ twofiles(void)
       printf(1, "wrong length %d\n", total);
       exit();
     }
+    unlink(fname);
   }
 
-  unlink("f1");
+  printf(1, "fourfiles ok\n");
-  unlink("f2");
-
-  printf(1, "twofiles ok\n");
 }
 
-// two processes create and delete different files in same directory
+// four processes create and delete different files in same directory
 void
 createdelete(void)
 {
   enum { N = 20 };
-  int pid, i, fd;
+  int pid, i, fd, pi;
   char name[32];
 
   printf(1, "createdelete test\n");
+
+  for(pi = 0; pi < 4; pi++){
     pid = fork();
     if(pid < 0){
       printf(1, "fork failed\n");
       exit();
     }
 
-  name[0] = pid ? 'p' : 'c';
+    if(pid == 0){
+      name[0] = 'p' + pi;
       name[2] = '\0';
       for(i = 0; i < N; i++){
         name[1] = '0' + i;
@@ -609,27 +615,18 @@ createdelete(void)
           }
         }
       }
-
-  if(pid==0)
       exit();
-  else
+    }
+  }
+
+  for(pi = 0; pi < 4; pi++){
     wait();
-
-  for(i = 0; i < N; i++){
-    name[0] = 'p';
-    name[1] = '0' + i;
-    fd = open(name, 0);
-    if((i == 0 || i >= N/2) && fd < 0){
-      printf(1, "oops createdelete %s didn't exist\n", name);
-      exit();
-    } else if((i >= 1 && i < N/2) && fd >= 0){
-      printf(1, "oops createdelete %s did exist\n", name);
-      exit();
   }
-    if(fd >= 0)
-      close(fd);
 
-    name[0] = 'c';
+  name[0] = name[1] = name[2] = 0;
+  for(i = 0; i < N; i++){
+    for(pi = 0; pi < 4; pi++){
+      name[0] = 'p' + pi;
       name[1] = '0' + i;
       fd = open(name, 0);
       if((i == 0 || i >= N/2) && fd < 0){
@@ -642,13 +639,14 @@ createdelete(void)
       if(fd >= 0)
         close(fd);
     }
+  }
 
   for(i = 0; i < N; i++){
-    name[0] = 'p';
+    for(pi = 0; pi < 4; pi++){
+      name[0] = 'p' + i;
       name[1] = '0' + i;
       unlink(name);
-    name[0] = 'c';
+    }
-    unlink(name);
   }
 
   printf(1, "createdelete ok\n");
@@ -1716,6 +1714,12 @@ main(int argc, char *argv[])
   }
   close(open("usertests.ran", O_CREATE));
 
+  createdelete();
+  linkunlink();
+  concreate();
+  fourfiles();
+  sharedfd();
+
   bigargtest();
   bigwrite();
   bigargtest();
@@ -1741,18 +1745,12 @@ main(int argc, char *argv[])
   fourteen();
   bigfile();
   subdir();
-  concreate();
-  linkunlink();
   linktest();
   unlinkread();
-  createdelete();
-  twofiles();
-  sharedfd();
   dirfile();
   iref();
   forktest();
   bigdir(); // slow
-
   exectest();
 
   exit();