Add code in the proxy cleaner component to iterate over new inodes.

[bluesky.git] / bluesky / log.c

diff --git a/bluesky/log.c b/bluesky/log.c
index 3fb9ef0..2e5c920 100644 (file)
--- a/bluesky/log.c
+++ b/bluesky/log.c
@@ -41,6 +41,22 @@
 #define HEADER_MAGIC 0x676f4c0a
 #define FOOTER_MAGIC 0x2e435243
 
+static size_t readbuf(int fd, char *buf, size_t len)
+{
+    size_t total_bytes = 0;
+    while (len > 0) {
+        ssize_t bytes = read(fd, buf, len);
+        if (bytes < 0 && errno == EINTR)
+            continue;
+        g_assert(bytes >= 0);
+        if (bytes == 0)
+            break;
+        buf += bytes;
+        len -= bytes;
+    }
+    return total_bytes;
+}
+
 static void writebuf(int fd, const char *buf, size_t len)
 {
     while (len > 0) {
@@ -567,6 +583,57 @@ static void cloudlog_fetch_start(BlueSkyCacheFile *cachefile)
     bluesky_store_async_unref(async);
 }
 
+/* Map and return a read-only version of a byte range from a cached file.  The
+ * CacheFile object must be locked. */
+BlueSkyRCStr *bluesky_cachefile_map_raw(BlueSkyCacheFile *cachefile,
+                                        off_t offset, size_t size)
+{
+    cachefile->atime = bluesky_get_current_time();
+
+    /* Easy case: the needed data is already in memory */
+    if (cachefile->addr != NULL && offset + size <= cachefile->len)
+        return bluesky_string_new_from_mmap(cachefile, offset, size);
+
+    int fd = openat(cachefile->log->dirfd, cachefile->filename, O_RDONLY);
+    if (fd < 0) {
+        fprintf(stderr, "Error opening logfile %s: %m\n",
+                cachefile->filename);
+        return NULL;
+    }
+
+    off_t length = lseek(fd, 0, SEEK_END);
+    if (offset + size > length) {
+        close(fd);
+        return NULL;
+    }
+
+    /* File is not mapped in memory.  Map the entire file in, then return a
+     * pointer to just the required data. */
+    if (cachefile->addr == NULL) {
+        cachefile->addr = (const char *)mmap(NULL, length, PROT_READ,
+                                             MAP_SHARED, fd, 0);
+        cachefile->len = length;
+        g_atomic_int_inc(&cachefile->refcount);
+
+        close(fd);
+        return bluesky_string_new_from_mmap(cachefile, offset, size);
+    }
+
+    /* Otherwise, the file was mapped in but doesn't cover the data we need.
+     * This shouldn't happen much, if at all, but if it does just read the data
+     * we need directly from the file.  We lose memory-management benefits of
+     * using mmapped data, but otherwise this works. */
+    char *buf = g_malloc(size);
+    size_t actual_size = readbuf(fd, buf, size);
+    close(fd);
+    if (actual_size != size) {
+        g_free(buf);
+        return NULL;
+    } else {
+        return bluesky_string_new(buf, size);
+    }
+}
+
 /* The arguments are mostly straightforward.  log_dir is -1 for access from the
  * journal, and non-negative for access to a cloud log segment.  map_data
  * should be TRUE for the case that are mapping just the data of an item where
@@ -575,7 +642,6 @@ static void cloudlog_fetch_start(BlueSkyCacheFile *cachefile)
 BlueSkyRCStr *bluesky_log_map_object(BlueSkyCloudLog *item, gboolean map_data)
 {
     BlueSkyFS *fs = item->fs;
-    BlueSkyLog *log = fs->log;
     BlueSkyCacheFile *map = NULL;
     BlueSkyRCStr *str = NULL;
     int location = 0;
@@ -655,24 +721,6 @@ BlueSkyRCStr *bluesky_log_map_object(BlueSkyCloudLog *item, gboolean map_data)
         }
     }
 
-    if (map->addr == NULL) {
-        int fd = openat(log->dirfd, map->filename, O_RDONLY);
-
-        if (fd < 0) {
-            fprintf(stderr, "Error opening logfile %s: %m\n", map->filename);
-            goto exit2;
-        }
-
-        off_t length = lseek(fd, 0, SEEK_END);
-        map->addr = (const char *)mmap(NULL, length, PROT_READ, MAP_SHARED,
-                                       fd, 0);
-        map->len = length;
-
-        g_atomic_int_inc(&map->refcount);
-
-        close(fd);
-    }
-
     if (map_data) {
         if (location == CLOUDLOG_JOURNAL)
             file_offset += sizeof(struct log_header);
@@ -681,8 +729,7 @@ BlueSkyRCStr *bluesky_log_map_object(BlueSkyCloudLog *item, gboolean map_data)
 
         file_size = item->data_size;
     }
-    str = bluesky_string_new_from_mmap(map, file_offset, file_size);
-    map->atime = bluesky_get_current_time();
+    str = bluesky_cachefile_map_raw(map, file_offset, file_size);
 
 exit2:
     bluesky_cachefile_unref(map);