(cl:in-package :cl-user)
-;; Actually there's no real side-effect here. The impurity we're
-;; avoiding is the sigchld handler that RUN-PROGRAM sets up, which
-;; interfers with the manual unix process control done by the test
-;; framework (sometimes the handler will manage to WAIT3 a process
-;; before run-tests WAITPIDs it).
-
-(let* ((process (sb-ext:run-program "/bin/cat" '() :wait nil
- :output :stream :input :stream))
- (out (process-input process))
- (in (process-output process)))
+;; In addition to definitions lower down the impurity we're avoiding
+;; is the sigchld handler that RUN-PROGRAM sets up, which interfers
+;; with the manual unix process control done by the test framework
+;; (sometimes the handler will manage to WAIT3 a process before
+;; run-tests WAITPIDs it).
+
+(with-test (:name :run-program-cat-1)
+ (let* ((process (sb-ext:run-program "/bin/cat" '() :wait nil
+ :output :stream :input :stream))
+ (out (process-input process))
+ (in (process-output process)))
+ (unwind-protect
+ (loop for i from 0 to 255 do
+ (write-byte i out)
+ (force-output out)
+ (assert (= (read-byte in) i)))
+ (process-close process))))
+
+;;; Test driving an external program (cat) through pipes wrapped in
+;;; composite streams.
+
+(require :sb-posix)
+
+(defun make-pipe ()
+ (multiple-value-bind (in out) (sb-posix:pipe)
+ (let ((input (sb-sys:make-fd-stream in
+ :input t
+ :external-format :ascii
+ :buffering :none :name "in"))
+ (output (sb-sys:make-fd-stream out
+ :output t
+ :external-format :ascii
+ :buffering :none :name "out")))
+ (make-two-way-stream input output))))
+
+(defparameter *cat-in-pipe* (make-pipe))
+(defparameter *cat-in* (make-synonym-stream '*cat-in-pipe*))
+(defparameter *cat-out-pipe* (make-pipe))
+(defparameter *cat-out* (make-synonym-stream '*cat-out-pipe*))
+
+(with-test (:name :run-program-cat-2)
+ (let ((cat (run-program "/bin/cat" nil :input *cat-in* :output *cat-out*
+ :wait nil)))
+ (dolist (test '("This is a test!"
+ "This is another test!"
+ "This is the last test...."))
+ (write-line test *cat-in*)
+ (assert (equal test (read-line *cat-out*))))
+ (process-close cat)))
+
+;;; The above test used to use ed, but there were buffering issues: on some platforms
+;;; buffering of stdin and stdout depends on their TTYness, and ed isn't sufficiently
+;;; agressive about flushing them. So, here's another test using :PTY.
+
+(defparameter *tmpfile* "run-program-ed-test.tmp")
+
+(with-open-file (f *tmpfile*
+ :direction :output
+ :if-exists :supersede)
+ (write-line "bar" f))
+
+(defparameter *ed*
+ (run-program "/bin/ed" (list *tmpfile*) :wait nil :pty t))
+
+(defparameter *ed-pipe* (make-two-way-stream (process-pty *ed*) (process-pty *ed*)))
+(defparameter *ed-in* (make-synonym-stream '*ed-pipe*))
+(defparameter *ed-out* (make-synonym-stream '*ed-pipe*))
+
+(defun read-linish (stream)
+ (with-output-to-string (s)
+ (loop for c = (read-char stream)
+ while (and c (not (eq #\newline c)) (not (eq #\return c)))
+ do (write-char c s))))
+
+(defun assert-ed (command response)
+ (when command
+ (write-line command *ed-in*)
+ (force-output *ed-in*))
+ (let ((got (read-linish *ed-out*)))
+ (unless (equal response got)
+ (error "wanted ~S from ed, got ~S" response got)))
+ *ed*)
+
+(unwind-protect
+ (with-test (:name :run-program-ed)
+ (assert-ed nil "4")
+ (assert-ed ".s/bar/baz/g" "")
+ (assert-ed "w" "4")
+ (assert-ed "q" "")
+ (process-wait *ed*)
+ (with-open-file (f *tmpfile*)
+ (assert (equal "baz" (read-line f)))))
+ (delete-file *tmpfile*))
+
+;; Around 1.0.12 there was a regression when :INPUT or :OUTPUT was a
+;; pathname designator. Since these use the same code, it should
+;; suffice to test just :INPUT.
+(let ((file))
(unwind-protect
- (loop for i from 0 to 255 do
- (write-byte i out)
- (force-output out)
- (assert (= (read-byte in) i)))
- (process-close process)))
+ (progn (with-open-file (f "run-program-test.tmp" :direction :output)
+ (setf file (truename f))
+ (write-line "Foo" f))
+ (assert (run-program "cat" ()
+ :input file :output t
+ :search t :wait t)))
+ (when file
+ (delete-file file))))
+
+;;; This used to crash on Darwin and trigger recursive lock errors on
+;;; every platform.
+(with-test (:name (:run-program :stress))
+ ;; Do it a hundred times in batches of 10 so that with a low limit
+ ;; of the number of processes the test can have a chance to pass.
+ (loop
+ repeat 10 do
+ (map nil
+ #'sb-ext:process-wait
+ (loop repeat 10
+ collect
+ (sb-ext:run-program "/bin/echo" '
+ ("It would be nice if this didn't crash.")
+ :wait nil :output nil)))))
+
+(with-test (:name (:run-program :pty-stream))
+ (assert (equal "OK"
+ (subseq
+ (with-output-to-string (s)
+ (assert (= 42 (process-exit-code
+ (run-program "/bin/sh" '("-c" "echo OK; exit 42") :wait t
+ :pty s))))
+ s)
+ 0
+ 2))))
+
+;; Check whether RUN-PROGRAM puts its child process into the foreground
+;; when stdin is inherited. If it fails to do so we will receive a SIGTTIN.
+;;
+;; We can't check for the signal itself since run-program.c resets the
+;; forked process' signal mask to defaults. But the default is `stop'
+;; of which we can be notified asynchronously by providing a status hook.
+(with-test (:name (:run-program :inherit-stdin)
+ :fails-on :sbcl)
+ (error "Hangs at least on threaded Darwin and threaded x86-64/Linux.")
+ (let (stopped)
+ (flet ((status-hook (proc)
+ (case (sb-ext:process-status proc)
+ (:stopped (setf stopped t)))))
+ (let ((proc (sb-ext:run-program "/bin/ed" nil :search nil :wait nil
+ :input t :output t
+ :status-hook #'status-hook)))
+ ;; Give the program a generous time to generate the SIGTTIN.
+ ;; If it hasn't done so after that time we can consider it
+ ;; to be working (i.e. waiting for input without generating SIGTTIN).
+ (sleep 0.5)
+ ;; either way we have to signal it to terminate
+ (process-kill proc sb-posix:sigterm)
+ (process-close proc)
+ (assert (not stopped))))))
+
projects / sbcl.git / blobdiff