dokan: simple case insensitive emulation

While CephFS is case sensitive, Windows software commonly assume
that the filesystem is case insensitive.

In order to unblock certain use cases, a short term solution is to
simply normalize paths when creating or accessing files or
directories.

This change adds two ceph-dokan parameters:

* --case-insensitive: if set, paths are normalized
* --force-lowercase: normalized paths are converted to lowercase
                     instead of uppercase

This trivial implementation has some limitations:

* the original case is not preserved
  * we could later on use xattr to store the original name
* can't access existing files that have a different case
* handled at ceph-dokan level

The advantage is that it's simple, shouldn't impact performance
and doesn't require libcephfs or mds changes.

In the future, we may conider a more robust implementation
at the mds and/or libcephfs level.

Signed-off-by: Lucian Petrut <lpetrut@cloudbasesolutions.com>

3 files changed, 50 insertions, 7 deletions

diff --git a/src/dokan/ceph_dokan.cc b/src/dokan/ceph_dokan.cc
index 9e115222cab..6459ea261bf 100644
--- a/src/dokan/ceph_dokan.cc
+++ b/src/dokan/ceph_dokan.cc
@@ -77,9 +77,26 @@ typedef struct {
 static_assert(sizeof(fd_context) <= 8,
               "fd_context exceeds DOKAN_FILE_INFO.Context size.");
 
-string get_path(LPCWSTR path_w) {
+string get_path(LPCWSTR path_w, bool normalize_case=true) {
   string path = to_string(path_w);
   replace(path.begin(), path.end(), '\\', '/');
+
+  if (normalize_case && !g_cfg->case_sensitive) {
+    if (g_cfg->convert_to_uppercase) {
+      std::transform(
+        path.begin(), path.end(), path.begin(),
+        [](unsigned char c){
+          return std::toupper(c);
+        });
+    } else {
+      std::transform(
+        path.begin(), path.end(), path.begin(),
+        [](unsigned char c){
+          return std::tolower(c);
+        });
+    }
+  }
+
   return path;
 }
 
@@ -543,6 +560,11 @@ static NTSTATUS WinCephFindFiles(
     return cephfs_errno_to_ntstatus_map(ret);
   }
 
+  // TODO: retrieve the original case (e.g. using xattr) if configured
+  // to do so.
+  // TODO: provide aliases when case insensitive mounts cause collisions.
+  // For example, when having test.txt and Test.txt, the latter becomes
+  // TEST~1.txt
   WIN32_FIND_DATAW findData;
   int count = 0;
   while (1) {
@@ -794,14 +816,18 @@ static NTSTATUS WinCephGetVolumeInformation(
 {
   g_cfg->win_vol_name.copy(VolumeNameBuffer, VolumeNameSize);
   *VolumeSerialNumber = g_cfg->win_vol_serial;
-
   *MaximumComponentLength = g_cfg->max_path_len;
 
-  *FileSystemFlags = FILE_CASE_SENSITIVE_SEARCH |
-            FILE_CASE_PRESERVED_NAMES |
-            FILE_SUPPORTS_REMOTE_STORAGE |
-            FILE_UNICODE_ON_DISK |
-            FILE_PERSISTENT_ACLS;
+  *FileSystemFlags =
+    FILE_SUPPORTS_REMOTE_STORAGE |
+    FILE_UNICODE_ON_DISK |
+    FILE_PERSISTENT_ACLS;
+
+  if (g_cfg->case_sensitive) {
+    *FileSystemFlags |=
+      FILE_CASE_SENSITIVE_SEARCH |
+      FILE_CASE_PRESERVED_NAMES;
+  }
 
   wcscpy(FileSystemNameBuffer, L"Ceph");
   return 0;
diff --git a/src/dokan/ceph_dokan.h b/src/dokan/ceph_dokan.h
index 5957d4dead1..fe48aa45814 100644
--- a/src/dokan/ceph_dokan.h
+++ b/src/dokan/ceph_dokan.h
@@ -36,6 +36,14 @@ struct Config {
   unsigned long max_path_len = 256;
   mode_t file_mode = 0755;
   mode_t dir_mode = 0755;
+
+  bool case_sensitive = true;
+  // Convert new file paths to upper case in case of case insensitive mounts.
+  // Visual Studio recommends normalizing to uppercase in order to avoid
+  // locale issues (CA1308).
+  bool convert_to_uppercase = true;
+  // TODO: consider adding an option to preserve the original case.
+  // It could be stored using an extended attribute.
 };
 
 extern Config *g_cfg;
diff --git a/src/dokan/options.cc b/src/dokan/options.cc
index 1ed90ef9d34..705e1117ca9 100644
--- a/src/dokan/options.cc
+++ b/src/dokan/options.cc
@@ -45,6 +45,11 @@ Map options:
   --max-path-len              The value of the maximum path length. Default: 256.
   --file-mode                 The access mode to be used when creating files.
   --dir-mode                  The access mode to be used when creating directories.
+  --case-insensitive          Emulate a case insensitive filesystem by normalizing
+                              paths. The original case is NOT preserved. Existing
+                              paths with a different case cannot be accessed.
+  --force-lowercase           Use lowercase when normalizing paths. Uppercase is
+                              used by default.
 
 Unmap options:
   -l [ --mountpoint ] arg     mountpoint (path or drive letter) (e.g -l x).
@@ -196,6 +201,10 @@ int parse_args(
         *err_msg << "ceph-dokan: Invalid argument for operation-timeout";
         return -EINVAL;
       }
+    } else if (ceph_argparse_flag(args, i, "--case-insensitive", (char *)NULL)) {
+      cfg->case_sensitive = false;
+    } else if (ceph_argparse_flag(args, i, "--force-lowercase", (char *)NULL)) {
+      cfg->convert_to_uppercase = false;
     } else {
       ++i;
     }