This is a note to let you know that I've just added the patch titled nitro_enclaves: Fix stale file descriptors on failed usercopy to the 5.12-stable tree which can be found at: http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary The filename of the patch is: nitro_enclaves-fix-stale-file-descriptors-on-failed-usercopy.patch and it can be found in the queue-5.12 subdirectory. If you, or anyone else, feels it should not be added to the stable tree, please let <stable@xxxxxxxxxxxxxxx> know about it. >From f1ce3986baa62cffc3c5be156994de87524bab99 Mon Sep 17 00:00:00 2001 From: Mathias Krause <minipli@xxxxxxxxxxxxxx> Date: Thu, 29 Apr 2021 19:59:41 +0300 Subject: nitro_enclaves: Fix stale file descriptors on failed usercopy From: Mathias Krause <minipli@xxxxxxxxxxxxxx> commit f1ce3986baa62cffc3c5be156994de87524bab99 upstream. A failing usercopy of the slot uid will lead to a stale entry in the file descriptor table as put_unused_fd() won't release it. This enables userland to refer to a dangling 'file' object through that still valid file descriptor, leading to all kinds of use-after-free exploitation scenarios. Exchanging put_unused_fd() for close_fd(), ksys_close() or alike won't solve the underlying issue, as the file descriptor might have been replaced in the meantime, e.g. via userland calling close() on it (leading to a NULL pointer dereference in the error handling code as 'fget(enclave_fd)' will return a NULL pointer) or by dup2()'ing a completely different file object to that very file descriptor, leading to the same situation: a dangling file descriptor pointing to a freed object -- just in this case to a file object of user's choosing. Generally speaking, after the call to fd_install() the file descriptor is live and userland is free to do whatever with it. We cannot rely on it to still refer to our enclave object afterwards. In fact, by abusing userfaultfd() userland can hit the condition without any racing and abuse the error handling in the nitro code as it pleases. To fix the above issues, defer the call to fd_install() until all possible errors are handled. In this case it's just the usercopy, so do it directly in ne_create_vm_ioctl() itself. Signed-off-by: Mathias Krause <minipli@xxxxxxxxxxxxxx> Signed-off-by: Andra Paraschiv <andraprs@xxxxxxxxxx> Cc: stable <stable@xxxxxxxxxxxxxxx> Link: https://lore.kernel.org/r/20210429165941.27020-2-andraprs@xxxxxxxxxx Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx> --- drivers/virt/nitro_enclaves/ne_misc_dev.c | 43 +++++++++++------------------- 1 file changed, 17 insertions(+), 26 deletions(-) --- a/drivers/virt/nitro_enclaves/ne_misc_dev.c +++ b/drivers/virt/nitro_enclaves/ne_misc_dev.c @@ -1524,7 +1524,8 @@ static const struct file_operations ne_e * enclave file descriptor to be further used for enclave * resources handling e.g. memory regions and CPUs. * @ne_pci_dev : Private data associated with the PCI device. - * @slot_uid: Generated unique slot id associated with an enclave. + * @slot_uid: User pointer to store the generated unique slot id + * associated with an enclave to. * * Context: Process context. This function is called with the ne_pci_dev enclave * mutex held. @@ -1532,7 +1533,7 @@ static const struct file_operations ne_e * * Enclave fd on success. * * Negative return value on failure. */ -static int ne_create_vm_ioctl(struct ne_pci_dev *ne_pci_dev, u64 *slot_uid) +static int ne_create_vm_ioctl(struct ne_pci_dev *ne_pci_dev, u64 __user *slot_uid) { struct ne_pci_dev_cmd_reply cmd_reply = {}; int enclave_fd = -1; @@ -1634,7 +1635,18 @@ static int ne_create_vm_ioctl(struct ne_ list_add(&ne_enclave->enclave_list_entry, &ne_pci_dev->enclaves_list); - *slot_uid = ne_enclave->slot_uid; + if (copy_to_user(slot_uid, &ne_enclave->slot_uid, sizeof(ne_enclave->slot_uid))) { + /* + * As we're holding the only reference to 'enclave_file', fput() + * will call ne_enclave_release() which will do a proper cleanup + * of all so far allocated resources, leaving only the unused fd + * for us to free. + */ + fput(enclave_file); + put_unused_fd(enclave_fd); + + return -EFAULT; + } fd_install(enclave_fd, enclave_file); @@ -1671,34 +1683,13 @@ static long ne_ioctl(struct file *file, switch (cmd) { case NE_CREATE_VM: { int enclave_fd = -1; - struct file *enclave_file = NULL; struct ne_pci_dev *ne_pci_dev = ne_devs.ne_pci_dev; - int rc = -EINVAL; - u64 slot_uid = 0; + u64 __user *slot_uid = (void __user *)arg; mutex_lock(&ne_pci_dev->enclaves_list_mutex); - - enclave_fd = ne_create_vm_ioctl(ne_pci_dev, &slot_uid); - if (enclave_fd < 0) { - rc = enclave_fd; - - mutex_unlock(&ne_pci_dev->enclaves_list_mutex); - - return rc; - } - + enclave_fd = ne_create_vm_ioctl(ne_pci_dev, slot_uid); mutex_unlock(&ne_pci_dev->enclaves_list_mutex); - if (copy_to_user((void __user *)arg, &slot_uid, sizeof(slot_uid))) { - enclave_file = fget(enclave_fd); - /* Decrement file refs to have release() called. */ - fput(enclave_file); - fput(enclave_file); - put_unused_fd(enclave_fd); - - return -EFAULT; - } - return enclave_fd; } Patches currently in stable-queue which might be from minipli@xxxxxxxxxxxxxx are queue-5.12/nitro_enclaves-fix-stale-file-descriptors-on-failed-usercopy.patch