about summary refs log tree commit diff
path: root/wcsmbs
diff options
context:
space:
mode:
authorZack Weinberg <zackw@panix.com>2018-02-21 19:12:51 -0500
committerZack Weinberg <zackw@panix.com>2018-03-13 08:31:56 -0400
commit2cc7bad0ae0a412e75270be5ed41d45c03e7a931 (patch)
treea726dff1dc98e5fabf47685d10f9c681265db95b /wcsmbs
parent778f1974863d63e858b6d0105e41d6f0c30732d3 (diff)
downloadglibc-2cc7bad0ae0a412e75270be5ed41d45c03e7a931.tar.gz
glibc-2cc7bad0ae0a412e75270be5ed41d45c03e7a931.tar.xz
glibc-2cc7bad0ae0a412e75270be5ed41d45c03e7a931.zip
[BZ 1190] Make EOF sticky in stdio.
C99 specifies that the EOF condition on a file is "sticky": once EOF
has been encountered, all subsequent reads should continue to return
EOF until the file is closed or something clears the "end-of-file
indicator" (e.g. fseek, clearerr).  This is arguably a change from
C89, where the wording was ambiguous; the BSDs always had sticky EOF,
but the System V lineage would attempt to read from the underlying fd
again.  GNU libc has followed System V for as long as we've been
using libio, but nowadays C99 conformance and BSD compatibility are
more important than System V compatibility.

You might wonder if changing the _underflow impls is sufficient to
apply the C99 semantics to all of the many stdio functions that
perform input.  It should be enough to cover all paths to _IO_SYSREAD,
and the only other functions that call _IO_SYSREAD are the _seekoff
impls, which is OK because seeking clears EOF, and the _xsgetn impls,
which, as far as I can tell, are unused within glibc.

The test programs in this patch use a pseudoterminal to set up the
necessary conditions.  To facilitate this I added a new test-support
function that sets up a pair of pty file descriptors for you; it's
almost the same as BSD openpty, the only differences are that it
allocates the optionally-returned tty pathname with malloc, and that
it crashes if anything goes wrong.

	[BZ #1190]
        [BZ #19476]
	* libio/fileops.c (_IO_new_file_underflow): Return EOF immediately
	if the _IO_EOF_SEEN bit is already set; update commentary.
	* libio/oldfileops.c (_IO_old_file_underflow): Likewise.
	* libio/wfileops.c (_IO_wfile_underflow): Likewise.

	* support/support_openpty.c, support/tty.h: New files.
	* support/Makefile (libsupport-routines): Add support_openpty.

	* libio/tst-fgetc-after-eof.c, wcsmbs/test-fgetwc-after-eof.c:
	New test cases.
	* libio/Makefile (tests): Add tst-fgetc-after-eof.
	* wcsmbs/Makefile (tests): Add tst-fgetwc-after-eof.
Diffstat (limited to 'wcsmbs')
-rw-r--r--wcsmbs/Makefile2
-rw-r--r--wcsmbs/tst-fgetwc-after-eof.c114
2 files changed, 115 insertions, 1 deletions
diff --git a/wcsmbs/Makefile b/wcsmbs/Makefile
index 3ee91d2e1a..63a6fbab58 100644
--- a/wcsmbs/Makefile
+++ b/wcsmbs/Makefile
@@ -50,7 +50,7 @@ strop-tests :=  wcscmp wcsncmp wmemcmp wcslen wcschr wcsrchr wcscpy wcsnlen \
 tests := tst-wcstof wcsmbs-tst1 tst-wcsnlen tst-btowc tst-mbrtowc \
 	 tst-wcrtomb tst-wcpncpy tst-mbsrtowcs tst-wchar-h tst-mbrtowc2 \
 	 tst-c16c32-1 wcsatcliff tst-wcstol-locale tst-wcstod-nan-locale \
-	 tst-wcstod-round test-char-types \
+	 tst-wcstod-round test-char-types tst-fgetwc-after-eof \
 	 $(addprefix test-,$(strop-tests))
 
 include ../Rules
diff --git a/wcsmbs/tst-fgetwc-after-eof.c b/wcsmbs/tst-fgetwc-after-eof.c
new file mode 100644
index 0000000000..9103529246
--- /dev/null
+++ b/wcsmbs/tst-fgetwc-after-eof.c
@@ -0,0 +1,114 @@
+/* Bug 1190: EOF conditions are supposed to be sticky.
+   Copyright (C) 2018 Free Software Foundation.
+   Copying and distribution of this file, with or without modification,
+   are permitted in any medium without royalty provided the copyright
+   notice and this notice are preserved. This file is offered as-is,
+   without any warranty.  */
+
+/* ISO C1999 specification of fgetwc:
+
+       #include <stdio.h>
+       #include <wchar.h>
+       wint_t fgetwc (FILE *stream);
+
+   Description
+
+     If the end-of-file indicator for the input stream pointed to by
+     stream is not set and a next wide character is present, the
+     fgetwc function obtains that wide character as a wchar_t
+     converted to a wint_t and advances the associated file position
+     indicator for the stream (if defined).
+
+  Returns
+
+     If the end-of-file indicator for the stream is set, or if the
+     stream is at end-of-file, the end- of-file indicator for the
+     stream is set and the fgetwc function returns WEOF. Otherwise,
+     the fgetwc function returns the next wide character from the
+     input stream pointed to by stream. If a read error occurs, the
+     error indicator for the stream is set and the fgetwc function
+     returns WEOF. If an encoding error occurs (including too few
+     bytes), the value of the macro EILSEQ is stored in errno and the
+     fgetwc function returns WEOF.
+
+   The requirement to return WEOF "if the end-of-file indicator for the
+   stream is set" was new in C99; the language in the 1995 edition of
+   the standard was ambiguous.  Historically, BSD-derived Unix always
+   had the C99 behavior, whereas in System V fgetwc would attempt to
+   call read() again before returning EOF again.  Prior to version 2.28,
+   glibc followed the System V behavior even though this does not
+   comply with C99.
+
+   See
+   <https://sourceware.org/bugzilla/show_bug.cgi?id=1190>,
+   <https://sourceware.org/bugzilla/show_bug.cgi?id=19476>,
+   and the thread at
+   <https://sourceware.org/ml/libc-alpha/2012-09/msg00343.html>
+   for more detail.  */
+
+#include <support/tty.h>
+#include <support/check.h>
+
+#include <fcntl.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+#include <wchar.h>
+
+#define XWRITE(fd, s, msg) do {                         \
+    if (write (fd, s, sizeof s - 1) != sizeof s - 1)    \
+      {                                                 \
+        perror ("write " msg);                          \
+        return 1;                                       \
+      }                                                 \
+  } while (0)
+
+int
+do_test (void)
+{
+  /* The easiest way to set up the conditions under which you can
+     notice whether the end-of-file indicator is sticky, is with a
+     pseudo-tty.  This is also the case which applications are most
+     likely to care about.  And it avoids any question of whether and
+     how it is legitimate to access the same physical file with two
+     independent FILE objects.  */
+  int outer_fd, inner_fd;
+  FILE *fp;
+
+  support_openpty (&outer_fd, &inner_fd, 0, 0, 0);
+  fp = fdopen (inner_fd, "r+");
+  if (!fp)
+    {
+      perror ("fdopen");
+      return 1;
+    }
+
+  XWRITE (outer_fd, "abc\n\004", "first line + EOF");
+  TEST_COMPARE (fgetwc (fp), L'a');
+  TEST_COMPARE (fgetwc (fp), L'b');
+  TEST_COMPARE (fgetwc (fp), L'c');
+  TEST_COMPARE (fgetwc (fp), L'\n');
+  TEST_COMPARE (fgetwc (fp), WEOF);
+
+  TEST_VERIFY_EXIT (feof (fp));
+  TEST_VERIFY_EXIT (!ferror (fp));
+
+  XWRITE (outer_fd, "d\n", "second line");
+
+  /* At this point, there is a new full line of input waiting in the
+     kernelside input buffer, but we should still observe EOF from
+     stdio, because the end-of-file indicator has not been cleared.  */
+  TEST_COMPARE (fgetwc (fp), WEOF);
+
+  /* Clearing EOF should reveal the next line of input.  */
+  clearerr (fp);
+  TEST_COMPARE (fgetwc (fp), L'd');
+  TEST_COMPARE (fgetwc (fp), L'\n');
+
+  fclose (fp);
+  close (outer_fd);
+  return 0;
+}
+
+#include <support/test-driver.c>