[PATCH 06/18] flag parameters: eventfd

From: Ulrich Drepper
Date: Sun May 04 2008 - 23:45:59 EST

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 flag EFD_CLOEXEC is used instead of reusing O_CLOEXEC to prevent
exhaustion of the int bit field in case we need more flags.

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 (1 << 0)

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;
}
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

arch/x86/ia32/ia32entry.S | 1 +
arch/x86/kernel/syscall_table_32.S | 1 +
fs/eventfd.c | 20 ++++++++++++++++++--
include/asm-x86/unistd_32.h | 1 +
include/asm-x86/unistd_64.h | 2 ++
include/linux/eventfd.h | 3 +++
include/linux/syscalls.h | 1 +
7 files changed, 27 insertions(+), 2 deletions(-)


Signed-off-by: Ulrich Drepper <drepper@xxxxxxxxxx>

diff --git a/arch/x86/ia32/ia32entry.S b/arch/x86/ia32/ia32entry.S
index b5e329d..37e4992 100644
--- a/arch/x86/ia32/ia32entry.S
+++ b/arch/x86/ia32/ia32entry.S
@@ -732,4 +732,5 @@ 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 adff556..f59aba5 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 343942d..ddb543d 100644
--- a/fs/eventfd.c
+++ b/fs/eventfd.c
@@ -16,6 +16,7 @@
#include <linux/anon_inodes.h>
#include <linux/eventfd.h>
#include <linux/syscalls.h>
+#include <linux/flagsremap.h>

struct eventfd_ctx {
wait_queue_head_t wqh;
@@ -198,10 +199,20 @@ struct file *eventfd_fget(int fd)
return file;
}

-asmlinkage long sys_eventfd(unsigned int count)
+static const struct flags_rmap eventfd_file_flags_remap[] = {
+ { EFD_CLOEXEC, O_CLOEXEC },
+};
+
+asmlinkage long sys_eventfd2(unsigned int count, int flags)
{
int fd;
struct eventfd_ctx *ctx;
+ int fflags;
+
+ if (flags_remap(eventfd_file_flags_remap,
+ ARRAY_SIZE(eventfd_file_flags_remap),
+ flags, &fflags) != 0)
+ return -EINVAL;

ctx = kmalloc(sizeof(*ctx), GFP_KERNEL);
if (!ctx)
@@ -214,9 +226,14 @@ 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, fflags);
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 8317d94..b3daf50 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 fe26e36..5ef6778 100644
--- a/include/asm-x86/unistd_64.h
+++ b/include/asm-x86/unistd_64.h
@@ -643,6 +643,8 @@ __SYSCALL(__NR_fallocate, sys_fallocate)
__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 a701399..0c2f68b 100644
--- a/include/linux/eventfd.h
+++ b/include/linux/eventfd.h
@@ -10,6 +10,9 @@

#ifdef CONFIG_EVENTFD

+/* Flags for eventfd2. */
+#define EFD_CLOEXEC (1 << 0)
+
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 0522f36..aeb2df8 100644
--- a/include/linux/syscalls.h
+++ b/include/linux/syscalls.h
@@ -617,6 +617,7 @@ asmlinkage long sys_set_robust_list(struct robust_list_head __user *head,
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[]);
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/