correct octets in c-string decoding errors
[sbcl.git] / src / code / fd-stream.lisp
index d98bf4a..311a4b9 100644 (file)
   #!+sb-doc
   "List of available buffers.")
 
-(defvar *available-buffers-spinlock* (sb!thread::make-spinlock
-                                      :name "lock for *AVAILABLE-BUFFERS*")
+(defvar *available-buffers-lock* (sb!thread:make-mutex
+                                  :name "lock for *AVAILABLE-BUFFERS*")
   #!+sb-doc
   "Mutex for access to *AVAILABLE-BUFFERS*.")
 
 (defmacro with-available-buffers-lock ((&optional) &body body)
-  ;; CALL-WITH-SYSTEM-SPINLOCK because
-  ;;
-  ;; 1. streams are low-level enough to be async signal safe, and in
-  ;;    particular a C-c that brings up the debugger while holding the
-  ;;    mutex would lose badly
-  ;;
-  ;; 2. this can potentially be a fairly busy (but also probably
-  ;;    uncontended) lock, so we don't want to pay the syscall per
-  ;;    release -- hence a spinlock.
-  ;;
-  ;; ...again, once we have smarted locks the spinlock here can become
-  ;; a mutex.
-  `(sb!thread::with-system-spinlock (*available-buffers-spinlock*)
+  ;; CALL-WITH-SYSTEM-MUTEX because streams are low-level enough to be
+  ;; async signal safe, and in particular a C-c that brings up the
+  ;; debugger while holding the mutex would lose badly.
+  `(sb!thread::with-system-mutex (*available-buffers-lock*)
      ,@body))
 
 (defconstant +bytes-per-buffer+ (* 4 1024)
   (element-type 'base-char)
   ;; the Unix file descriptor
   (fd -1 :type fixnum)
+  ;; What do we know about the FD?
+  (fd-type :unknown :type keyword)
   ;; controls when the output buffer is flushed
   (buffering :full :type (member :full :line :none))
   ;; controls whether the input buffer must be cleared before output
   (char-pos nil :type (or unsigned-byte null))
   ;; T if input is waiting on FD. :EOF if we hit EOF.
   (listen nil :type (member nil t :eof))
+  ;; T if serve-event is allowed when this stream blocks
+  (serve-events nil :type boolean)
 
   ;; the input buffer
   (instead (make-array 0 :element-type 'character :adjustable t :fill-pointer t) :type (array character (*)))
   (timeout nil :type (or single-float null))
   ;; pathname of the file this stream is opened to (returned by PATHNAME)
   (pathname nil :type (or pathname null))
-  (external-format :default)
+  ;; Not :DEFAULT, because we want to match CHAR-SIZE!
+  (external-format :latin-1)
   ;; fixed width, or function to call with a character
   (char-size 1 :type (or fixnum function))
   (output-bytes #'ill-out :type function)
                (aver (< head tail))
                (%queue-and-replace-output-buffer stream))
               (t
-               ;; Try a non-blocking write, queue whatever is left over.
+               ;; Try a non-blocking write, if SERVE-EVENT is allowed, queue
+               ;; whatever is left over. Otherwise wait until we can write.
                (aver (< head tail))
                (synchronize-stream-output stream)
-               (let ((length (- tail head)))
-                 (multiple-value-bind (count errno)
-                     (sb!unix:unix-write (fd-stream-fd stream) (buffer-sap obuf)
-                                         head length)
-                   (cond ((eql count length)
-                          ;; Complete write -- we can use the same buffer.
-                          (reset-buffer obuf))
-                         (count
-                          ;; Partial write -- update buffer status and queue.
-                          ;; Do not use INCF! Another thread might have moved
-                          ;; head...
-                          (setf (buffer-head obuf) (+ count head))
-                          (%queue-and-replace-output-buffer stream))
-                         #!-win32
-                         ((eql errno sb!unix:ewouldblock)
-                          ;; Blocking, queue.
-                          (%queue-and-replace-output-buffer stream))
-                         (t
-                          (simple-stream-perror "Couldn't write to ~s"
-                                                stream errno)))))))))))
+               (loop
+                 (let ((length (- tail head)))
+                   (multiple-value-bind (count errno)
+                       (sb!unix:unix-write (fd-stream-fd stream) (buffer-sap obuf)
+                                           head length)
+                     (flet ((queue-or-wait ()
+                              (if (fd-stream-serve-events stream)
+                                  (return (%queue-and-replace-output-buffer stream))
+                                  (or (wait-until-fd-usable (fd-stream-fd stream) :output
+                                                            (fd-stream-timeout stream)
+                                                            nil)
+                                      (signal-timeout 'io-timeout
+                                                      :stream stream
+                                                      :direction :output
+                                                      :seconds (fd-stream-timeout stream))))))
+                        (cond ((eql count length)
+                               ;; Complete write -- we can use the same buffer.
+                               (return (reset-buffer obuf)))
+                              (count
+                               ;; Partial write -- update buffer status and
+                               ;; queue or wait.
+                               (incf head count)
+                               (setf (buffer-head obuf) head)
+                               (queue-or-wait))
+                              #!-win32
+                              ((eql errno sb!unix:ewouldblock)
+                               ;; Blocking, queue or wair.
+                               (queue-or-wait))
+                              (t
+                               (simple-stream-perror "Couldn't write to ~s"
+                                                     stream errno)))))))))))))
 
 ;;; Helper for FLUSH-OUTPUT-BUFFER -- returns the new buffer.
 (defun %queue-and-replace-output-buffer (stream)
+  (aver (fd-stream-serve-events stream))
   (let ((queue (fd-stream-output-queue stream))
         (later (list (or (fd-stream-obuf stream) (bug "Missing obuf."))))
         (new (get-buffer)))
 ;;; This is called by the FD-HANDLER for the stream when output is
 ;;; possible.
 (defun write-output-from-queue (stream)
+  (aver (fd-stream-serve-events stream))
   (synchronize-stream-output stream)
   (let (not-first-p)
     (tagbody
 ;;; this is not something we want to export. Nikodemus thinks the
 ;;; right thing is to support a low-level non-stream like IO layer,
 ;;; akin to java.nio.
-(defun output-raw-bytes (stream thing &optional start end)
+(declaim (inline output-raw-bytes))
+(define-deprecated-function :late "1.0.8.16" output-raw-bytes write-sequence
+    (stream thing &optional start end)
   (write-or-buffer-output stream thing (or start 0) (or end (length thing))))
-
-(define-compiler-macro output-raw-bytes (stream thing &optional start end)
-  (deprecation-warning 'output-raw-bytes)
-  (let ((x (gensym "THING")))
-    `(let ((,x ,thing))
-       (write-or-buffer-output ,stream ,x (or ,start 0) (or ,end (length ,x))))))
 \f
 ;;;; output routines and related noise
 
   (error 'c-string-encoding-error
          :external-format external-format
          :code code))
-
-(defun c-string-decoding-error (external-format octets)
+(defun c-string-decoding-error (external-format sap offset count)
   (error 'c-string-decoding-error
          :external-format external-format
-         :octets octets))
+         :octets (sap-ref-octets sap offset count)))
 
 ;;; Returning true goes into end of file handling, false will enter another
 ;;; round of input buffer filling followed by re-entering character decode.
   (octets-to-string-fun (missing-arg) :type function)
   (string-to-octets-fun (missing-arg) :type function))
 
+(defun ef-char-size (ef-entry)
+  (if (variable-width-external-format-p ef-entry)
+      (bytes-for-char-fun ef-entry)
+      (funcall (bytes-for-char-fun ef-entry) #\x)))
+
 (defun wrap-external-format-functions (external-format fun)
   (let ((result (%copy-external-format external-format)))
     (macrolet ((frob (accessor)
 
 (defun get-external-format-or-lose (external-format)
   (or (get-external-format external-format)
-      (error "Undefined external-format ~A" external-format)))
+      (error "Undefined external-format: ~S" external-format)))
 
 (defun external-format-keyword (external-format)
   (typecase external-format
 ;;; number of bytes per element.
 (defun pick-output-routine (type buffering &optional external-format)
   (when (subtypep type 'character)
-    (let ((entry (get-external-format external-format)))
-      (when entry
-        (return-from pick-output-routine
-          (values (ecase buffering
-                    (:none (ef-write-char-none-buffered-fun entry))
-                    (:line (ef-write-char-line-buffered-fun entry))
-                    (:full (ef-write-char-full-buffered-fun entry)))
-                  'character
-                  1
-                  (ef-write-n-bytes-fun entry)
-                  (canonize-external-format external-format entry))))))
+    (let ((entry (get-external-format-or-lose external-format)))
+      (return-from pick-output-routine
+        (values (ecase buffering
+                  (:none (ef-write-char-none-buffered-fun entry))
+                  (:line (ef-write-char-line-buffered-fun entry))
+                  (:full (ef-write-char-full-buffered-fun entry)))
+                'character
+                1
+                (ef-write-n-bytes-fun entry)
+                (ef-char-size entry)
+                (canonize-external-format external-format entry)))))
   (dolist (entry *output-routines*)
     (when (and (subtypep type (first entry))
                (eq buffering (second entry))
   ;; This answers T at EOF on win32, I think.
   (not (sb!win32:fd-listen (fd-stream-fd stream)))
   #!-win32
-  (sb!unix:with-restarted-syscall (count errno)
-    (sb!alien:with-alien ((read-fds (sb!alien:struct sb!unix:fd-set)))
-      (sb!unix:fd-zero read-fds)
-      (sb!unix:fd-set (fd-stream-fd stream) read-fds)
-      (sb!unix:unix-fast-select (1+ (fd-stream-fd stream))
-                                (sb!alien:addr read-fds)
-                                nil nil 0 0))
-    (case count
-      ((1) nil)
-      ((0) t)
-      (otherwise
-       (simple-stream-perror "couldn't check whether ~S is readable"
-                             stream
-                             errno)))))
+  (not (sb!unix:unix-simple-poll (fd-stream-fd stream) :input 0)))
 
 ;;; If the read would block wait (using SERVE-EVENT) till input is available,
 ;;; then fill the input buffer, and return the number of bytes read. Throws
            (errno 0)
            (count 0))
     (tagbody
-       ;; Check for blocking input before touching the stream, as if
-       ;; we happen to wait we are liable to be interrupted, and the
-       ;; interrupt handler may use the same stream.
-       (if (sysread-may-block-p stream)
+       ;; Check for blocking input before touching the stream if we are to
+       ;; serve events: if the FD is blocking, we don't want to try an uninterruptible
+       ;; read(). Regular files should never block, so we can elide the check.
+       (if (and (neq :regular (fd-stream-fd-type stream))
+                (sysread-may-block-p stream))
            (go :wait-for-input)
            (go :main))
        ;; These (:CLOSED-FLAME and :READ-ERROR) tags are here so what
      :wait-for-input
        ;; This tag is here so we can unwind outside the WITHOUT-INTERRUPTS
        ;; to wait for input if read tells us EWOULDBLOCK.
-       (unless (wait-until-fd-usable fd :input (fd-stream-timeout stream))
-         (signal-timeout 'io-timeout :stream stream :direction :read
+       (unless (wait-until-fd-usable fd :input (fd-stream-timeout stream)
+                                     (fd-stream-serve-events stream))
+         (signal-timeout 'io-timeout
+                         :stream stream
+                         :direction :input
                          :seconds (fd-stream-timeout stream)))
      :main
        ;; Since the read should not block, we'll disable the
        ;; resulting thunk is stack-allocatable.
        ((lambda (return-reason)
           (ecase return-reason
-            ((nil))             ; fast path normal cases
+            ((nil))                     ; fast path normal cases
             ((:wait-for-input) (go :wait-for-input))
             ((:closed-flame)   (go :closed-flame))
             ((:read-error)     (go :read-error))))
 ;;; bytes per element (and for character types string input routine).
 (defun pick-input-routine (type &optional external-format)
   (when (subtypep type 'character)
-    (let ((entry (get-external-format external-format)))
-      (when entry
-        (return-from pick-input-routine
-          (values (ef-read-char-fun entry)
-                  'character
-                  1
-                  (ef-read-n-chars-fun entry)
-                  (canonize-external-format external-format entry))))))
+    (let ((entry (get-external-format-or-lose external-format)))
+      (return-from pick-input-routine
+        (values (ef-read-char-fun entry)
+                'character
+                1
+                (ef-read-n-chars-fun entry)
+                (ef-char-size entry)
+                (canonize-external-format external-format entry)))))
   (dolist (entry *input-routines*)
     (when (and (subtypep type (first entry))
                (or (not (fourth entry))
                 (return-from ,in-function total-copied)))
             (setf (buffer-head ibuf) head)
             ;; Maybe we need to refill the stream buffer.
-            (cond ( ;; If there were enough data in the stream buffer, we're done.
-                   (= total-copied requested)
+            (cond ( ;; If was data in the stream buffer, we're done.
+                   (plusp total-copied)
                    (return total-copied))
                   ( ;; If EOF, we're done in another way.
                    (or (eq decode-break-reason 'eof)
                                    (incf head size)
                                    nil))
                            (when decode-break-reason
-                             (c-string-decoding-error ,name decode-break-reason))
+                             (c-string-decoding-error
+                              ,name sap head decode-break-reason))
                            (when (zerop (char-code char))
                              (return count))))
                  (string (make-string length :element-type element-type)))
                       (incf head size)
                       nil))
               (when decode-break-reason
-                (c-string-decoding-error ,name decode-break-reason))
+                (c-string-decoding-error
+                 ,name sap head decode-break-reason))
               (setf (aref string index) char)))))
 
       (defun ,output-c-string-function (string)
          (character-stream-p (subtypep target-type 'character))
          (bivalent-stream-p (eq element-type :default))
          normalized-external-format
+         char-size
          (bin-routine #'ill-bin)
          (bin-type nil)
          (bin-size nil)
     (when output-p
       (setf (fd-stream-char-pos fd-stream) 0))
 
-    (when (and character-stream-p
-               (eq external-format :default))
+    (when (and character-stream-p (eq external-format :default))
       (/show0 "/getting default external format")
       (setf external-format (default-external-format)))
 
     (when input-p
       (when (or (not character-stream-p) bivalent-stream-p)
-        (multiple-value-setq (bin-routine bin-type bin-size read-n-characters
-                                          normalized-external-format)
-          (pick-input-routine (if bivalent-stream-p '(unsigned-byte 8)
-                                  target-type)
-                              external-format))
+        (setf (values bin-routine bin-type bin-size read-n-characters
+                      char-size normalized-external-format)
+              (pick-input-routine (if bivalent-stream-p '(unsigned-byte 8)
+                                      target-type)
+                                  external-format))
         (unless bin-routine
           (error "could not find any input routine for ~S" target-type)))
       (when character-stream-p
-        (multiple-value-setq (cin-routine cin-type cin-size read-n-characters
-                                          normalized-external-format)
-          (pick-input-routine target-type external-format))
+        (setf (values cin-routine cin-type cin-size read-n-characters
+                      char-size normalized-external-format)
+              (pick-input-routine target-type external-format))
         (unless cin-routine
           (error "could not find any input routine for ~S" target-type)))
       (setf (fd-stream-in fd-stream) cin-routine
       (setf input-size (or cin-size bin-size))
       (setf input-type (or cin-type bin-type))
       (when normalized-external-format
-        (setf (fd-stream-external-format fd-stream)
-              normalized-external-format))
+        (setf (fd-stream-external-format fd-stream) normalized-external-format
+              (fd-stream-char-size fd-stream) char-size))
       (when (= (or cin-size 1) (or bin-size 1) 1)
         (setf (fd-stream-n-bin fd-stream) ;XXX
               (if (and character-stream-p (not bivalent-stream-p))
 
     (when output-p
       (when (or (not character-stream-p) bivalent-stream-p)
-        (multiple-value-setq (bout-routine bout-type bout-size output-bytes
-                                           normalized-external-format)
-          (let ((buffering (fd-stream-buffering fd-stream)))
-            (if bivalent-stream-p
-                (pick-output-routine '(unsigned-byte 8)
-                                     (if (eq :line buffering)
-                                         :full
-                                         buffering)
-                                     external-format)
-                (pick-output-routine target-type buffering external-format))))
+        (setf (values bout-routine bout-type bout-size output-bytes
+                      char-size normalized-external-format)
+              (let ((buffering (fd-stream-buffering fd-stream)))
+                (if bivalent-stream-p
+                    (pick-output-routine '(unsigned-byte 8)
+                                         (if (eq :line buffering)
+                                             :full
+                                             buffering)
+                                         external-format)
+                    (pick-output-routine target-type buffering external-format))))
         (unless bout-routine
           (error "could not find any output routine for ~S buffered ~S"
                  (fd-stream-buffering fd-stream)
                  target-type)))
       (when character-stream-p
-        (multiple-value-setq (cout-routine cout-type cout-size output-bytes
-                                           normalized-external-format)
-          (pick-output-routine target-type
-                               (fd-stream-buffering fd-stream)
-                               external-format))
+        (setf (values cout-routine cout-type cout-size output-bytes
+                      char-size normalized-external-format)
+              (pick-output-routine target-type
+                                   (fd-stream-buffering fd-stream)
+                                   external-format))
         (unless cout-routine
           (error "could not find any output routine for ~S buffered ~S"
                  (fd-stream-buffering fd-stream)
                  target-type)))
       (when normalized-external-format
-        (setf (fd-stream-external-format fd-stream)
-              normalized-external-format))
+        (setf (fd-stream-external-format fd-stream) normalized-external-format
+              (fd-stream-char-size fd-stream) char-size))
       (when character-stream-p
         (setf (fd-stream-output-bytes fd-stream) output-bytes))
       (setf (fd-stream-out fd-stream) cout-routine
   (flush-output-buffer stream)
   (do ()
       ((null (fd-stream-output-queue stream)))
+    (aver (fd-stream-serve-events stream))
     (serve-all-events)))
 
 (defun fd-stream-get-file-position (stream)
 ;;; FILE is the name of the file (will be returned by PATHNAME).
 ;;;
 ;;; NAME is used to identify the stream when printed.
+;;;
+;;; If SERVE-EVENTS is true, SERVE-EVENT machinery is used to
+;;; handle blocking IO on the stream.
 (defun make-fd-stream (fd
                        &key
                        (input nil input-p)
                        (element-type 'base-char)
                        (buffering :full)
                        (external-format :default)
+                       serve-events
                        timeout
                        file
                        original
         ((not (or input output))
          (error "File descriptor must be opened either for input or output.")))
   (let ((stream (%make-fd-stream :fd fd
+                                 :fd-type (progn
+                                            #!-win32 (sb!unix:fd-type fd)
+                                            ;; KLUDGE.
+                                            #!+win32 (if serve-events
+                                                         :unknown
+                                                         :regular))
                                  :name name
                                  :file file
                                  :original original
                                  :pathname pathname
                                  :buffering buffering
                                  :dual-channel-p dual-channel-p
-                                 :external-format external-format
                                  :bivalent-p (eq element-type :default)
-                                 :char-size (external-format-char-size external-format)
+                                 :serve-events serve-events
                                  :timeout
                                  (if timeout
                                      (coerce timeout 'single-float)
             (cond ((numberp fd)
                    (case direction
                      ((:input :output :io)
+                      ;; For O_APPEND opened files, lseek returns 0 until first write.
+                      ;; So we jump ahead here.
+                      (when (eq if-exists :append)
+                        (sb!unix:unix-lseek fd 0 sb!unix:l_xtnd))
                       (make-fd-stream fd
                                       :input input
                                       :output output
                                       :delete-original delete-original
                                       :pathname pathname
                                       :dual-channel-p nil
+                                      :serve-events nil
                                       :input-buffer-p t
                                       :auto-close t))
                      (:probe
     (setf *stdin*
           (make-fd-stream 0 :name "standard input" :input t :buffering :line
                           :element-type :default
+                          :serve-events t
                           :external-format (stdstream-external-format nil)))
     (setf *stdout*
           (make-fd-stream 1 :name "standard output" :output t :buffering :line
                 (make-fd-stream tty :name "the terminal"
                                 :input t :output t :buffering :line
                                 :external-format (stdstream-external-format t)
+                                :serve-events t
                                 :auto-close t))
           (setf *tty* (make-two-way-stream *stdin* *stdout*))))
     (princ (get-output-stream-string *error-output*) *stderr*))