From 562787a5c32ccdf182de27793a83a9f2ee86cd77 Mon Sep 17 00:00:00 2001 From: Davide Libenzi Date: Tue, 22 Sep 2009 16:43:57 -0700 Subject: anonfd: split interface into file creation and install Split the anonfd interface into a bare file pointer creation one, and a file pointer creation plus install one. There are cases, like the usage of eventfds inside other kernel interfaces, where the file pointer created by anonfd needs to be used inside the initialization of other structures. As it is right now, as soon as anon_inode_getfd() returns, the kenrle can race with userspace closing the newly installed file descriptor. This patch, while keeping the old anon_inode_getfd(), introduces a new anon_inode_getfile() (whose services are reused in anon_inode_getfd()) that allows to split the file creation phase and the fd install one. Once all the kernel structures are initialized, the code can call the proper fd_install(). Gregory manifested the need for something like this inside KVM. Signed-off-by: Davide Libenzi Cc: Alexander Viro Cc: James Morris Cc: Peter Zijlstra Cc: Gregory Haskins Acked-by: Serge Hallyn Acked-by: Roland Dreier Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- fs/eventfd.c | 67 ++++++++++++++++++++++++++++++++++++++++++++++++------------ 1 file changed, 54 insertions(+), 13 deletions(-) (limited to 'fs/eventfd.c') diff --git a/fs/eventfd.c b/fs/eventfd.c index 31d12de83a2a..8b47e4200e65 100644 --- a/fs/eventfd.c +++ b/fs/eventfd.c @@ -68,11 +68,16 @@ int eventfd_signal(struct eventfd_ctx *ctx, int n) } EXPORT_SYMBOL_GPL(eventfd_signal); +static void eventfd_free_ctx(struct eventfd_ctx *ctx) +{ + kfree(ctx); +} + static void eventfd_free(struct kref *kref) { struct eventfd_ctx *ctx = container_of(kref, struct eventfd_ctx, kref); - kfree(ctx); + eventfd_free_ctx(ctx); } /** @@ -298,9 +303,23 @@ struct eventfd_ctx *eventfd_ctx_fileget(struct file *file) } EXPORT_SYMBOL_GPL(eventfd_ctx_fileget); -SYSCALL_DEFINE2(eventfd2, unsigned int, count, int, flags) +/** + * eventfd_file_create - Creates an eventfd file pointer. + * @count: Initial eventfd counter value. + * @flags: Flags for the eventfd file. + * + * This function creates an eventfd file pointer, w/out installing it into + * the fd table. This is useful when the eventfd file is used during the + * initialization of data structures that require extra setup after the eventfd + * creation. So the eventfd creation is split into the file pointer creation + * phase, and the file descriptor installation phase. + * In this way races with userspace closing the newly installed file descriptor + * can be avoided. + * Returns an eventfd file pointer, or a proper error pointer. + */ +struct file *eventfd_file_create(unsigned int count, int flags) { - int fd; + struct file *file; struct eventfd_ctx *ctx; /* Check the EFD_* constants for consistency. */ @@ -308,26 +327,48 @@ SYSCALL_DEFINE2(eventfd2, unsigned int, count, int, flags) BUILD_BUG_ON(EFD_NONBLOCK != O_NONBLOCK); if (flags & ~EFD_FLAGS_SET) - return -EINVAL; + return ERR_PTR(-EINVAL); ctx = kmalloc(sizeof(*ctx), GFP_KERNEL); if (!ctx) - return -ENOMEM; + return ERR_PTR(-ENOMEM); kref_init(&ctx->kref); init_waitqueue_head(&ctx->wqh); ctx->count = count; ctx->flags = flags; - /* - * 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, - flags & EFD_SHARED_FCNTL_FLAGS); - if (fd < 0) - kfree(ctx); + file = anon_inode_getfile("[eventfd]", &eventfd_fops, ctx, + flags & EFD_SHARED_FCNTL_FLAGS); + if (IS_ERR(file)) + eventfd_free_ctx(ctx); + + return file; +} + +SYSCALL_DEFINE2(eventfd2, unsigned int, count, int, flags) +{ + int fd, error; + struct file *file; + + error = get_unused_fd_flags(flags & EFD_SHARED_FCNTL_FLAGS); + if (error < 0) + return error; + fd = error; + + file = eventfd_file_create(count, flags); + if (IS_ERR(file)) { + error = PTR_ERR(file); + goto err_put_unused_fd; + } + fd_install(fd, file); + return fd; + +err_put_unused_fd: + put_unused_fd(fd); + + return error; } SYSCALL_DEFINE1(eventfd, unsigned int, count) -- cgit v1.2.3