about summary refs log tree commit diff
path: root/sysdeps/posix/pwritev_common.c
diff options
context:
space:
mode:
authorAdhemerval Zanella <adhemerval.zanella@linaro.org>2017-05-02 14:39:58 -0300
committerAdhemerval Zanella <adhemerval.zanella@linaro.org>2017-05-15 16:33:45 -0300
commitc79a72aa5cb8357c216a71015c7448a9259c8531 (patch)
treecaa8eba1f2827c9aa553c3543e85ada1ea4e37e7 /sysdeps/posix/pwritev_common.c
parentcdd45522b6e87369afc341402bda7a95bdf83380 (diff)
downloadglibc-c79a72aa5cb8357c216a71015c7448a9259c8531.tar.gz
glibc-c79a72aa5cb8357c216a71015c7448a9259c8531.tar.xz
glibc-c79a72aa5cb8357c216a71015c7448a9259c8531.zip
posix: Fix and simplify default p{read,write}v implementation
Currently all architectures but microblaze use wire-up syscall for
p{readv,write}v.  Microblaze still uses the syscall emulation using
sysdep/posix/p{readv,writev}.c and it was reported in some ocasions
[1] [2] that it might have some issues with some linux specific
usage (mainly with O_DIRECT and the alignment requirement).

Although it is not an issue for virtually all the system, this
patch refactors the sysdeps/posix p{read,write}v syscall to avoid
such issue (by using posix_memalign on the buffer used on
p{read,write} call) and by refactoring it common files to avoid
the need check on defines to correct set the alias and internal
symbols.

Checked on microblaze-linux-gnu check with run-built-tests=no and
by using the sysdeps/posix implementation on x86_64-linux-gnu (just
for sanity test where it shown no regression).

	* sysdeps/posix/preadv.c: Use sysdeps/posix/preadv_common.c.
	* sysdeps/posix/preadv64.c: Likewise.
	* sysdeps/unix/sysv/linux/preadv.c: Likewise.
	* sysdeps/unix/sysv/linux/preadv64.c: Likewise.
	* sysdeps/posix/pwritev.c: Use sysdeps/posix/pwritev_common.c.
	* sysdeps/posix/pwritev64.c: Likewise.
	* sysdeps/unix/sysv/linux/pwritev.c: Likewise.
	* sysdeps/unix/sysv/linux/pwritev64.c: Likewise.
	* sysdeps/posix/preadv_common.c: New file.
	* sysdeps/posix/pwritev_common.c: Likewise.

[1] http://www.mail-archive.com/qemu-devel@nongnu.org/msg25282.html
[2] https://bugzilla.redhat.com/show_bug.cgi?id=563103#c8
Diffstat (limited to 'sysdeps/posix/pwritev_common.c')
-rw-r--r--sysdeps/posix/pwritev_common.c72
1 files changed, 72 insertions, 0 deletions
diff --git a/sysdeps/posix/pwritev_common.c b/sysdeps/posix/pwritev_common.c
new file mode 100644
index 0000000000..03830650e4
--- /dev/null
+++ b/sysdeps/posix/pwritev_common.c
@@ -0,0 +1,72 @@
+/* Write data into multiple buffers.  Base implementation for pwritev
+   and pwritev64.
+   Copyright (C) 2017 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, see
+   <http://www.gnu.org/licenses/>.  */
+
+#include <unistd.h>
+#include <sys/uio.h>
+#include <sys/param.h>
+#include <errno.h>
+#include <malloc.h>
+
+#include <ldsodefs.h>
+
+/* Write data pointed by the buffers described by IOVEC, which is a
+   vector of COUNT 'struct iovec's, to file descriptor FD at the given
+   position OFFSET without change the file pointer.  The data is
+   written in the order specified.  Operates just like 'write' (see
+   <unistd.h>) except that the data are taken from IOVEC instead of a
+   contiguous buffer.  */
+ssize_t
+PWRITEV (int fd, const struct iovec *vector, int count, OFF_T offset)
+{
+  /* Find the total number of bytes to be read.  */
+  size_t bytes = 0;
+  for (int i = 0; i < count; ++i)
+    {
+      /* Check for ssize_t overflow.  */
+      if (SSIZE_MAX - bytes < vector[i].iov_len)
+	{
+	  __set_errno (EINVAL);
+	  return -1;
+	}
+      bytes += vector[i].iov_len;
+    }
+
+  /* Allocate a temporary buffer to hold the data.  It could be done with a
+     stack allocation, but due limitations on some system (Linux with
+     O_DIRECT) it aligns the buffer to pagesize.  A possible optimization
+     would be querying if the syscall would impose any alignment constraint,
+     but 1. it is system specific (not meant in generic implementation), and
+     2. it would make the implementation more complex, and 3. it will require
+     another syscall (fcntl).  */
+  void *buffer = NULL;
+  if (__posix_memalign (&buffer, GLRO(dl_pagesize), bytes) != 0)
+    return -1;
+
+  /* Copy the data from BUFFER into the memory specified by VECTOR.  */
+  char *ptr = buffer;
+  for (int i = 0; i < count; ++i)
+    ptr = __mempcpy ((void *) ptr, (void *) vector[i].iov_base,
+		     vector[i].iov_len);
+
+  ssize_t ret = PWRITE (fd, buffer, bytes, offset);
+
+  free (buffer);
+
+  return ret;
+}