X-Git-Url: http://git.vrable.net/?a=blobdiff_plain;f=store.cc;h=57ee11ba163ded18af96b2a9206dd2b63508f7eb;hb=cac644d993d130efd8d29539de6557b18c9b737e;hp=70be12664a4553f4271623b95845e8bee1b08e35;hpb=6fc3c86d0c9fc87ae993577c9954dbdca725a64d;p=cumulus.git

diff --git a/store.cc b/store.cc
index 70be126..57ee11b 100644
--- a/store.cc
+++ b/store.cc
@@ -10,10 +10,13 @@
 #include <stdio.h>
 #include <sys/types.h>
 #include <sys/stat.h>
+#include <sys/resource.h>
+#include <sys/wait.h>
 #include <unistd.h>
 #include <fcntl.h>
 #include <time.h>
 
+#include <algorithm>
 #include <list>
 #include <set>
 #include <string>
@@ -22,25 +25,101 @@
 #include "store.h"
 #include "ref.h"
 
+using std::max;
 using std::list;
 using std::set;
 using std::string;
 
+static char *const filter_program[] = {"bzip2", "-c", NULL};
+
+static void cloexec(int fd)
+{
+    long flags = fcntl(fd, F_GETFD);
+
+    if (flags < 0)
+        return;
+
+    fcntl(fd, F_SETFD, flags | FD_CLOEXEC);
+}
+
 Tarfile::Tarfile(const string &path, const string &segment)
     : size(0),
       segment_name(segment)
 {
-    if (tar_open(&t, (char *)path.c_str(), NULL, O_WRONLY | O_CREAT, 0600,
-                 TAR_VERBOSE | TAR_GNU) == -1)
+    real_fd = open(path.c_str(), O_WRONLY | O_CREAT, 0600);
+    if (real_fd < 0)
+        throw IOException("Error opening output file");
+
+    filter_fd = spawn_filter(real_fd);
+
+    if (tar_fdopen(&t, filter_fd, (char *)path.c_str(), NULL,
+                   O_WRONLY | O_CREAT, 0600, TAR_VERBOSE | TAR_GNU) == -1)
         throw IOException("Error opening Tarfile");
 }
 
 Tarfile::~Tarfile()
 {
+    /* Close the tar file... */
     tar_append_eof(t);
 
     if (tar_close(t) != 0)
         throw IOException("Error closing Tarfile");
+
+    /* ...and wait for filter process to finish. */
+    int status;
+    waitpid(filter_pid, &status, 0);
+
+    if (!WIFEXITED(status) || WEXITSTATUS(status) != 0) {
+        throw IOException("Filter process error");
+    }
+
+    close(real_fd);
+}
+
+/* Launch a child process which can act as a filter (compress, encrypt, etc.)
+ * on the TAR output.  The file descriptor to which output should be written
+ * must be specified; the return value is the file descriptor which will be
+ * attached to the standard input of the filter program. */
+int Tarfile::spawn_filter(int fd_out)
+{
+    int fds[2];
+
+    /* Create a pipe for communicating with the filter process. */
+    if (pipe(fds) < 0) {
+        throw IOException("Unable to create pipe for filter");
+    }
+
+    /* Create a child process which can exec() the filter program. */
+    filter_pid = fork();
+    if (filter_pid < 0)
+        throw IOException("Unable to fork filter process");
+
+    if (filter_pid > 0) {
+        /* Parent process */
+        close(fds[0]);
+        cloexec(fds[1]);
+    } else {
+        /* Child process.  Rearrange file descriptors.  stdin is fds[0], stdout
+         * is fd_out, stderr is unchanged. */
+        close(fds[1]);
+
+        if (dup2(fds[0], 0) < 0)
+            exit(1);
+        close(fds[0]);
+
+        if (dup2(fd_out, 1) < 0)
+            exit(1);
+        close(fd_out);
+
+        /* Exec the filter program. */
+        execvp(filter_program[0], filter_program);
+
+        /* Should not reach here except for error cases. */
+        fprintf(stderr, "Could not exec filter: %m\n");
+        exit(1);
+    }
+
+    return fds[1];
 }
 
 void Tarfile::write_object(int id, const char *data, size_t len)
@@ -91,6 +170,26 @@ void Tarfile::internal_write_object(const string &path,
     size += blocks * T_BLOCKSIZE;
 }
 
+/* Estimate the size based on the size of the actual output file on disk.
+ * However, it might be the case that the filter program is buffering all its
+ * data, and might potentially not write a single byte until we have closed
+ * our end of the pipe.  If we don't do so until we see data written, we have
+ * a problem.  So, arbitrarily pick an upper bound on the compression ratio
+ * that the filter will achieve (128:1), and return a size estimate which is
+ * the larger of a) bytes actually seen written to disk, and b) input
+ * bytes/128. */
+size_t Tarfile::size_estimate()
+{
+    struct stat statbuf;
+
+    if (fstat(real_fd, &statbuf) == 0)
+        return max((int64_t)statbuf.st_size, (int64_t)(size / 128));
+
+    /* Couldn't stat the file on disk, so just return the actual number of
+     * bytes, before compression. */
+    return size;
+}
+
 static const size_t SEGMENT_SIZE = 4 * 1024 * 1024;
 
 ObjectReference TarSegmentStore::write_object(const char *data, size_t len,
@@ -105,7 +204,7 @@ ObjectReference TarSegmentStore::write_object(const char *data, size_t len,
 
         segment->name = generate_uuid();
 
-        string filename = path + "/" + segment->name + ".tar";
+        string filename = path + "/" + segment->name + ".tar.bz2";
         segment->file = new Tarfile(filename, segment->name);
 
         segment->count = 0;