[PATCH 0/5] MR1756: ntdll: Various performance improvements to hash_short_file_name.

[Rémi Bernon (＠rbernon)]

The function is called a lot, I believe something like this would improve its performance in general.

[PATCH 1/5] ntdll: Use invalid char lookup table in is_invalid_dos_char.

[Rémi Bernon]

From: Rémi Bernon

---
 dlls/ntdll/unix/file.c | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/dlls/ntdll/unix/file.c b/dlls/ntdll/unix/file.c
index 957d9f3b801..b1b09e80e5a 100644
--- a/dlls/ntdll/unix/file.c
+++ b/dlls/ntdll/unix/file.c
@@ -250,9 +250,12 @@ static pthread_mutex_t mnt_mutex = PTHREAD_MUTEX_INITIALIZER;
 /* check if a given Unicode char is OK in a DOS short name */
 static inline BOOL is_invalid_dos_char( WCHAR ch )
 {
-    static const WCHAR invalid_chars[] = { INVALID_DOS_CHARS,'~','.',0 };
-    if (ch > 0x7f) return TRUE;
-    return wcschr( invalid_chars, ch ) != NULL;
+    static const char is_invalid[0x7f] =
+    {
+        ['*'] = 1, ['?'] = 1, ['<'] = 1, ['>'] = 1, ['|'] = 1, ['"'] = 1, ['+'] = 1, ['='] = 1,
+        [','] = 1, [';'] = 1, ['['] = 1, [']'] = 1, [' '] = 1, ['~'] = 1, ['.'] = 1,
+    };
+    return ch <= 0x7f ? is_invalid[ch] : TRUE;
 }

/* check if the device can be a mounted volume */

[PATCH 2/5] ntdll: Use invalid char lookup table in is_legal_8dot3_name.

[Rémi Bernon]

From: Rémi Bernon

Instead of costly wcschr.
---
 dlls/ntdll/unix/file.c | 14 +++++++++++---
 1 file changed, 11 insertions(+), 3 deletions(-)

diff --git a/dlls/ntdll/unix/file.c b/dlls/ntdll/unix/file.c
index b1b09e80e5a..097346b475d 100644
--- a/dlls/ntdll/unix/file.c
+++ b/dlls/ntdll/unix/file.c
@@ -258,6 +258,16 @@ static inline BOOL is_invalid_dos_char( WCHAR ch )
     return ch <= 0x7f ? is_invalid[ch] : TRUE;
 }

+static inline BOOL is_invalid_8dot3_char( WCHAR ch )
+{
+    static const char is_invalid[0x7f] =
+    {
+        ['*'] = 1, ['?'] = 1, ['<'] = 1, ['>'] = 1, ['|'] = 1, ['"'] = 1, ['+'] = 1, ['='] = 1,
+        [','] = 1, [';'] = 1, ['['] = 1, [']'] = 1, [' '] = 1, [':'] = 1, ['/'] = 1, ['\'] = 1,
+    };
+    return ch <= 0x7f ? is_invalid[ch] : TRUE;
+}
+
 /* check if the device can be a mounted volume */
 static inline BOOL is_valid_mounted_device( const struct stat *st )
 {
@@ -1461,7 +1471,6 @@ static BOOLEAN match_filename( const WCHAR *name, int length, const UNICODE_STRI
  */
 static BOOLEAN is_legal_8dot3_name( const WCHAR *name, int len )
 {
-    static const WCHAR invalid_chars[] = { INVALID_DOS_CHARS,':','/','\',0 };
     int i, dot = -1;

if (len > 12) return FALSE;
@@ -1471,8 +1480,7 @@ static BOOLEAN is_legal_8dot3_name( const WCHAR *name, int len )

for (i = 0; i < len; i++)
     {
-        if (name[i] > 0x7f) return FALSE;
-        if (wcschr( invalid_chars, name[i] )) return FALSE;
+        if (is_invalid_8dot3_char( name[i] )) return FALSE;
         if (name[i] == '.')
         {
             if (dot != -1) return FALSE;

[PATCH 3/5] ntdll: Use invalid char lookup table in nt_to_unix_file_name_no_root.

[Rémi Bernon]

From: Rémi Bernon

To lookup invalid NT chars, instead of costly wcschr.
---
 dlls/ntdll/unix/file.c | 14 +++++++++++---
 1 file changed, 11 insertions(+), 3 deletions(-)

diff --git a/dlls/ntdll/unix/file.c b/dlls/ntdll/unix/file.c
index 097346b475d..ed949da103c 100644
--- a/dlls/ntdll/unix/file.c
+++ b/dlls/ntdll/unix/file.c
@@ -268,6 +268,16 @@ static inline BOOL is_invalid_8dot3_char( WCHAR ch )
     return ch <= 0x7f ? is_invalid[ch] : TRUE;
 }

+static inline BOOL is_invalid_nt_char( WCHAR ch )
+{
+    static const char is_invalid[0x7f] =
+    {
+        1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
+        ['*'] = 1, ['?'] = 1, ['<'] = 1, ['>'] = 1, ['|'] = 1, ['"'] = 1,
+    };
+    return ch <= 0x7f ? is_invalid[ch] : FALSE;
+}
+
 /* check if the device can be a mounted volume */
 static inline BOOL is_valid_mounted_device( const struct stat *st )
 {
@@ -3474,7 +3484,6 @@ static NTSTATUS nt_to_unix_file_name_no_root( const UNICODE_STRING *nameW, char
                                               UINT disposition )
 {
     static const WCHAR unixW[] = {'u','n','i','x'};
-    static const WCHAR invalid_charsW[] = { INVALID_NT_CHARS, 0 };

NTSTATUS status = STATUS_SUCCESS;
     const WCHAR *name;
@@ -3501,8 +3510,7 @@ static NTSTATUS nt_to_unix_file_name_no_root( const UNICODE_STRING *nameW, char
     for (pos = 0; pos < name_len && pos <= MAX_DIR_ENTRY_LEN; pos++)
     {
         if (name[pos] == '\') break;
-        if (name[pos] < 32 || wcschr( invalid_charsW, name[pos] ))
-            return STATUS_OBJECT_NAME_INVALID;
+        if (is_invalid_nt_char( name[pos] )) return STATUS_OBJECT_NAME_INVALID;
         prefix[pos] = (name[pos] >= 'A' && name[pos] <= 'Z') ? name[pos] + 'a' - 'A' : name[pos];
     }
     if (pos > MAX_DIR_ENTRY_LEN) return STATUS_OBJECT_NAME_INVALID;

[PATCH 4/5] ntdll: Use invalid char lookup table in lookup_unix_name.

[Rémi Bernon]

From: Rémi Bernon

To lookup invalid unix chars, instead of costly wcschr.
---
 dlls/ntdll/unix/file.c | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)

diff --git a/dlls/ntdll/unix/file.c b/dlls/ntdll/unix/file.c
index ed949da103c..67a4d4b6c89 100644
--- a/dlls/ntdll/unix/file.c
+++ b/dlls/ntdll/unix/file.c
@@ -169,7 +169,6 @@ typedef struct
 #define IS_SEPARATOR(ch)   ((ch) == '\' || (ch) == '/')

#define INVALID_NT_CHARS   '*','?','<','>','|','"'
-#define INVALID_DOS_CHARS  INVALID_NT_CHARS,'+','=',',',';','[',']',' ','\345'

#define MAX_DIR_ENTRY_LEN 255  /* max length of a directory entry in chars */

@@ -278,6 +277,12 @@ static inline BOOL is_invalid_nt_char( WCHAR ch )
     return ch <= 0x7f ? is_invalid[ch] : FALSE;
 }

+static inline BOOL is_invalid_unix_char( WCHAR ch )
+{
+    if (ch == '/') return TRUE;
+    return is_invalid_nt_char( ch );
+}
+
 /* check if the device can be a mounted volume */
 static inline BOOL is_valid_mounted_device( const struct stat *st )
 {
@@ -3364,7 +3369,6 @@ done:
 static NTSTATUS lookup_unix_name( const WCHAR *name, int name_len, char **buffer, int unix_len, int pos,
                                   UINT disposition, BOOL is_unix )
 {
-    static const WCHAR invalid_charsW[] = { INVALID_NT_CHARS, '/', 0 };
     NTSTATUS status;
     int ret;
     struct stat st;
@@ -3391,7 +3395,7 @@ static NTSTATUS lookup_unix_name( const WCHAR *name, int name_len, char **buffer
         {
             if (!*ptr) return STATUS_OBJECT_NAME_INVALID;
             if (is_unix) continue;
-            if (*ptr < 32 || wcschr( invalid_charsW, *ptr )) return STATUS_OBJECT_NAME_INVALID;
+            if (is_invalid_unix_char( *ptr )) return STATUS_OBJECT_NAME_INVALID;
         }
     }

[PATCH 5/5] ntdll: Lookup extension from the end in hash_short_file_name.

[Rémi Bernon]

From: Rémi Bernon

---
 dlls/ntdll/unix/file.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/dlls/ntdll/unix/file.c b/dlls/ntdll/unix/file.c
index 67a4d4b6c89..f1d7742e238 100644
--- a/dlls/ntdll/unix/file.c
+++ b/dlls/ntdll/unix/file.c
@@ -1369,7 +1369,8 @@ static ULONG hash_short_file_name( const WCHAR *name, int length, LPWSTR buffer
     }

/* Find last dot for start of the extension */
-    for (p = name + 1, ext = NULL; p < end - 1; p++) if (*p == '.') ext = p;
+    for (ext = end - 2; ext > name; ext--) if (*ext == '.') break;
+    if (ext == name) ext = NULL;

/* Copy first 4 chars, replacing invalid chars with '_' */
     for (i = 4, p = name, dst = buffer; i > 0; i--, p++)