flag parameters: eventfd

This patch adds the new eventfd2 syscall.  It extends the old eventfd
syscall by one parameter which is meant to hold a flag value.  In this
patch the only flag support is EFD_CLOEXEC which causes the close-on-exec
flag for the returned file descriptor to be set.

A new name EFD_CLOEXEC is introduced which in this implementation must
have the same value as O_CLOEXEC.

The following test must be adjusted for architectures other than x86 and
x86-64 and in case the syscall numbers changed.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
#include <fcntl.h>
#include <stdio.h>
#include <unistd.h>
#include <sys/syscall.h>

#ifndef __NR_eventfd2
# ifdef __x86_64__
#  define __NR_eventfd2 290
# elif defined __i386__
#  define __NR_eventfd2 328
# else
#  error "need __NR_eventfd2"
# endif
#endif

#define EFD_CLOEXEC O_CLOEXEC

int
main (void)
{
  int fd = syscall (__NR_eventfd2, 1, 0);
  if (fd == -1)
    {
      puts ("eventfd2(0) failed");
      return 1;
    }
  int coe = fcntl (fd, F_GETFD);
  if (coe == -1)
    {
      puts ("fcntl failed");
      return 1;
    }
  if (coe & FD_CLOEXEC)
    {
      puts ("eventfd2(0) sets close-on-exec flag");
      return 1;
    }
  close (fd);

  fd = syscall (__NR_eventfd2, 1, EFD_CLOEXEC);
  if (fd == -1)
    {
      puts ("eventfd2(EFD_CLOEXEC) failed");
      return 1;
    }
  coe = fcntl (fd, F_GETFD);
  if (coe == -1)
    {
      puts ("fcntl failed");
      return 1;
    }
  if ((coe & FD_CLOEXEC) == 0)
    {
      puts ("eventfd2(EFD_CLOEXEC) does not set close-on-exec flag");
      return 1;
    }
  close (fd);

  puts ("OK");

  return 0;
}
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

[akpm@linux-foundation.org: add sys_ni stub]
Signed-off-by: Ulrich Drepper <drepper@redhat.com>
Acked-by: Davide Libenzi <davidel@xmailserver.org>
Cc: Michael Kerrisk <mtk.manpages@googlemail.com>
Cc: <linux-arch@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

8 files changed, 24 insertions, 2 deletions

diff --git a/arch/x86/ia32/ia32entry.S b/arch/x86/ia32/ia32entry.S
index c308128b9251..cf0eb31745ca 100644
--- a/arch/x86/ia32/ia32entry.S
+++ b/arch/x86/ia32/ia32entry.S
@@ -827,4 +827,5 @@ ia32_sys_call_table:
         .quad compat_sys_timerfd_settime        /* 325 */
         .quad compat_sys_timerfd_gettime
         .quad compat_sys_signalfd4
+        .quad sys_eventfd2
 ia32_syscall_end:
diff --git a/arch/x86/kernel/syscall_table_32.S b/arch/x86/kernel/syscall_table_32.S
index c12a36c9fd51..cf112cb11c37 100644
--- a/arch/x86/kernel/syscall_table_32.S
+++ b/arch/x86/kernel/syscall_table_32.S
@@ -327,3 +327,4 @@ ENTRY(sys_call_table)
         .long sys_timerfd_settime       /* 325 */
         .long sys_timerfd_gettime
         .long sys_signalfd4
+        .long sys_eventfd2
diff --git a/fs/eventfd.c b/fs/eventfd.c
index 6094265ca409..bd420e6478ad 100644
--- a/fs/eventfd.c
+++ b/fs/eventfd.c
@@ -198,11 +198,14 @@ struct file *eventfd_fget(int fd)
         return file;
 }
 
-asmlinkage long sys_eventfd(unsigned int count)
+asmlinkage long sys_eventfd2(unsigned int count, int flags)
 {
         int fd;
         struct eventfd_ctx *ctx;
 
+        if (flags & ~EFD_CLOEXEC)
+                return -EINVAL;
+
         ctx = kmalloc(sizeof(*ctx), GFP_KERNEL);
         if (!ctx)
                 return -ENOMEM;
@@ -214,9 +217,15 @@ asmlinkage long sys_eventfd(unsigned int count)
          * When we call this, the initialization must be complete, since
          * anon_inode_getfd() will install the fd.
          */
-        fd = anon_inode_getfd("[eventfd]", &eventfd_fops, ctx, 0);
+        fd = anon_inode_getfd("[eventfd]", &eventfd_fops, ctx,
+                              flags & O_CLOEXEC);
         if (fd < 0)
                 kfree(ctx);
         return fd;
 }
 
+asmlinkage long sys_eventfd(unsigned int count)
+{
+        return sys_eventfd2(count, 0);
+}
+
diff --git a/include/asm-x86/unistd_32.h b/include/asm-x86/unistd_32.h
index c310371f5613..edbd8723c939 100644
--- a/include/asm-x86/unistd_32.h
+++ b/include/asm-x86/unistd_32.h
@@ -333,6 +333,7 @@
 #define __NR_timerfd_settime    325
 #define __NR_timerfd_gettime    326
 #define __NR_signalfd4          327
+#define __NR_eventfd2           328
 
 #ifdef __KERNEL__
 
diff --git a/include/asm-x86/unistd_64.h b/include/asm-x86/unistd_64.h
index e0a9b45b2346..fb059a6feeb1 100644
--- a/include/asm-x86/unistd_64.h
+++ b/include/asm-x86/unistd_64.h
@@ -643,6 +643,8 @@ __SYSCALL(__NR_timerfd_gettime, sys_timerfd_gettime)
 __SYSCALL(__NR_paccept, sys_paccept)
 #define __NR_signalfd4                          289
 __SYSCALL(__NR_signalfd4, sys_signalfd4)
+#define __NR_eventfd2                           290
+__SYSCALL(__NR_eventfd2, sys_eventfd2)
 
 
 #ifndef __NO_STUBS
diff --git a/include/linux/eventfd.h b/include/linux/eventfd.h
index a701399b7fed..a6c0eaedb1b0 100644
--- a/include/linux/eventfd.h
+++ b/include/linux/eventfd.h
@@ -10,6 +10,12 @@
 
 #ifdef CONFIG_EVENTFD
 
+/* For O_CLOEXEC */
+#include <linux/fcntl.h>
+
+/* Flags for eventfd2.  */
+#define EFD_CLOEXEC O_CLOEXEC
+
 struct file *eventfd_fget(int fd);
 int eventfd_signal(struct file *file, int n);
 
diff --git a/include/linux/syscalls.h b/include/linux/syscalls.h
index 1c2707797845..9ab09926a7f2 100644
--- a/include/linux/syscalls.h
+++ b/include/linux/syscalls.h
@@ -617,6 +617,7 @@ asmlinkage long sys_timerfd_settime(int ufd, int flags,
                                     struct itimerspec __user *otmr);
 asmlinkage long sys_timerfd_gettime(int ufd, struct itimerspec __user *otmr);
 asmlinkage long sys_eventfd(unsigned int count);
+asmlinkage long sys_eventfd2(unsigned int count, int flags);
 asmlinkage long sys_fallocate(int fd, int mode, loff_t offset, loff_t len);
 
 int kernel_execve(const char *filename, char *const argv[], char *const envp[]);
diff --git a/kernel/sys_ni.c b/kernel/sys_ni.c
index 8627c89ae9e8..2a361ccdc7ca 100644
--- a/kernel/sys_ni.c
+++ b/kernel/sys_ni.c
@@ -164,3 +164,4 @@ cond_syscall(sys_timerfd_gettime);
 cond_syscall(compat_sys_timerfd_settime);
 cond_syscall(compat_sys_timerfd_gettime);
 cond_syscall(sys_eventfd);
+cond_syscall(sys_eventfd2);