From patchwork Fri Aug 30 00:17:02 2019 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Sean Christopherson X-Patchwork-Id: 11122729 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 597271908 for ; Fri, 30 Aug 2019 00:17:09 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id 35B992189D for ; Fri, 30 Aug 2019 00:17:09 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1726988AbfH3ARJ (ORCPT ); Thu, 29 Aug 2019 20:17:09 -0400 Received: from mga02.intel.com ([134.134.136.20]:30403 "EHLO mga02.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1726739AbfH3ARI (ORCPT ); Thu, 29 Aug 2019 20:17:08 -0400 X-Amp-Result: SKIPPED(no attachment in message) X-Amp-File-Uploaded: False Received: from fmsmga004.fm.intel.com ([10.253.24.48]) by orsmga101.jf.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 29 Aug 2019 17:17:08 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.64,445,1559545200"; d="scan'208";a="205917816" Received: from sjchrist-coffee.jf.intel.com ([10.54.74.41]) by fmsmga004.fm.intel.com with ESMTP; 29 Aug 2019 17:17:07 -0700 From: Sean Christopherson To: Jarkko Sakkinen Cc: linux-sgx@vger.kernel.org Subject: [PATCH v3 1/5] x86/sgx: Convert encl->flags from an unsigned int to an atomic Date: Thu, 29 Aug 2019 17:17:02 -0700 Message-Id: <20190830001706.29309-2-sean.j.christopherson@intel.com> X-Mailer: git-send-email 2.22.0 In-Reply-To: <20190830001706.29309-1-sean.j.christopherson@intel.com> References: <20190830001706.29309-1-sean.j.christopherson@intel.com> MIME-Version: 1.0 Sender: linux-sgx-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-sgx@vger.kernel.org Make encl->flags an atomic to ensure that all enclave state is visible in memory prior to SGX_ENCL_CREATED being set. Without a compiler memory barrier, adding pages and or initializing the enclave could theoretically consume stale data. Alternatively, a smp_wmb() + smp_rmb() could be added specifically for SGX_ENCL_CREATED, but making flags an atomic allows future patches to introduce similar checks, e.g. to force serialize ioctls, and the downsides to using an atomic are negligible, e.g. locked transactions are only introduced in slow paths since most flows can use atomic_read() to query flags. Suggested-by: Jarkko Sakkinen Signed-off-by: Sean Christopherson --- arch/x86/kernel/cpu/sgx/driver.c | 3 ++- arch/x86/kernel/cpu/sgx/encl.c | 18 +++++++++++------- arch/x86/kernel/cpu/sgx/encl.h | 2 +- arch/x86/kernel/cpu/sgx/ioctl.c | 16 ++++++++-------- arch/x86/kernel/cpu/sgx/reclaim.c | 10 +++++----- 5 files changed, 27 insertions(+), 22 deletions(-) diff --git a/arch/x86/kernel/cpu/sgx/driver.c b/arch/x86/kernel/cpu/sgx/driver.c index e740d71e2311..a1da479ffd3a 100644 --- a/arch/x86/kernel/cpu/sgx/driver.c +++ b/arch/x86/kernel/cpu/sgx/driver.c @@ -31,6 +31,7 @@ static int sgx_open(struct inode *inode, struct file *file) if (!encl) return -ENOMEM; + atomic_set(&encl->flags, 0); kref_init(&encl->refcount); INIT_LIST_HEAD(&encl->va_pages); INIT_RADIX_TREE(&encl->page_tree, GFP_KERNEL); @@ -77,7 +78,7 @@ static int sgx_release(struct inode *inode, struct file *file) }; mutex_lock(&encl->lock); - encl->flags |= SGX_ENCL_DEAD; + atomic_or(SGX_ENCL_DEAD, &encl->flags); mutex_unlock(&encl->lock); kref_put(&encl->refcount, sgx_encl_release); diff --git a/arch/x86/kernel/cpu/sgx/encl.c b/arch/x86/kernel/cpu/sgx/encl.c index 08ca0cfb1742..b6cd2a800eb2 100644 --- a/arch/x86/kernel/cpu/sgx/encl.c +++ b/arch/x86/kernel/cpu/sgx/encl.c @@ -107,6 +107,7 @@ static struct sgx_encl_page *sgx_encl_load_page(struct sgx_encl *encl, { struct sgx_epc_page *epc_page; struct sgx_encl_page *entry; + unsigned int encl_flags; /* If process was forked, VMA is still there but vm_private_data is set * to NULL. @@ -114,8 +115,9 @@ static struct sgx_encl_page *sgx_encl_load_page(struct sgx_encl *encl, if (!encl) return ERR_PTR(-EFAULT); - if ((encl->flags & SGX_ENCL_DEAD) || - !(encl->flags & SGX_ENCL_INITIALIZED)) + encl_flags = atomic_read(&encl->flags); + if ((encl_flags & SGX_ENCL_DEAD) || + !(encl_flags & SGX_ENCL_INITIALIZED)) return ERR_PTR(-EFAULT); entry = radix_tree_lookup(&encl->page_tree, addr >> PAGE_SHIFT); @@ -217,7 +219,7 @@ int sgx_encl_mm_add(struct sgx_encl *encl, struct mm_struct *mm) struct sgx_encl_mm *encl_mm; int ret; - if (encl->flags & SGX_ENCL_DEAD) + if (atomic_read(&encl->flags) & SGX_ENCL_DEAD) return -EINVAL; /* @@ -395,6 +397,7 @@ static int sgx_vma_access(struct vm_area_struct *vma, unsigned long addr, struct sgx_encl_page *entry = NULL; unsigned long align; char data[sizeof(unsigned long)]; + unsigned int encl_flags; int offset; int cnt; int ret = 0; @@ -406,9 +409,10 @@ static int sgx_vma_access(struct vm_area_struct *vma, unsigned long addr, if (!encl) return -EFAULT; - if (!(encl->flags & SGX_ENCL_DEBUG) || - !(encl->flags & SGX_ENCL_INITIALIZED) || - (encl->flags & SGX_ENCL_DEAD)) + encl_flags = atomic_read(&encl->flags); + if (!(encl_flags & SGX_ENCL_DEBUG) || + !(encl_flags & SGX_ENCL_INITIALIZED) || + (encl_flags & SGX_ENCL_DEAD)) return -EFAULT; for (i = 0; i < len; i += cnt) { @@ -495,7 +499,7 @@ void sgx_encl_destroy(struct sgx_encl *encl) struct radix_tree_iter iter; void **slot; - encl->flags |= SGX_ENCL_DEAD; + atomic_or(SGX_ENCL_DEAD, &encl->flags); radix_tree_for_each_slot(slot, &encl->page_tree, &iter, 0) { entry = *slot; diff --git a/arch/x86/kernel/cpu/sgx/encl.h b/arch/x86/kernel/cpu/sgx/encl.h index 37b5c4bcda7a..ffc09e1a0fc6 100644 --- a/arch/x86/kernel/cpu/sgx/encl.h +++ b/arch/x86/kernel/cpu/sgx/encl.h @@ -67,7 +67,7 @@ struct sgx_encl_mm { }; struct sgx_encl { - unsigned int flags; + atomic_t flags; u64 secs_attributes; u64 allowed_attributes; unsigned int page_cnt; diff --git a/arch/x86/kernel/cpu/sgx/ioctl.c b/arch/x86/kernel/cpu/sgx/ioctl.c index 1740eca4f163..d05cf051ca45 100644 --- a/arch/x86/kernel/cpu/sgx/ioctl.c +++ b/arch/x86/kernel/cpu/sgx/ioctl.c @@ -23,7 +23,7 @@ static struct sgx_va_page *sgx_encl_grow(struct sgx_encl *encl, BUILD_BUG_ON(SGX_VA_SLOT_COUNT != (SGX_ENCL_PAGE_VA_OFFSET_MASK >> 3) + 1); - if (encl->flags & disallowed_flags) + if (atomic_read(&encl->flags) & disallowed_flags) return ERR_PTR(-EFAULT); if (!(encl->page_cnt % SGX_VA_SLOT_COUNT)) { @@ -42,7 +42,7 @@ static struct sgx_va_page *sgx_encl_grow(struct sgx_encl *encl, err = ERR_CAST(va_page->epc_page); kfree(va_page); return err; - } else if (encl->flags & disallowed_flags) { + } else if (atomic_read(&encl->flags) & disallowed_flags) { sgx_free_page(va_page->epc_page); kfree(va_page); return ERR_PTR(-EFAULT); @@ -219,7 +219,7 @@ static int sgx_encl_create(struct sgx_encl *encl, struct sgx_secs *secs) } if (secs->attributes & SGX_ATTR_DEBUG) - encl->flags |= SGX_ENCL_DEBUG; + atomic_or(SGX_ENCL_DEBUG, &encl->flags); encl->secs.encl = encl; encl->secs_attributes = secs->attributes; @@ -233,7 +233,7 @@ static int sgx_encl_create(struct sgx_encl *encl, struct sgx_secs *secs) * allows other flows to check if the enclave has been created without * taking encl->lock. */ - encl->flags |= SGX_ENCL_CREATED; + atomic_or(SGX_ENCL_CREATED, &encl->flags); mutex_unlock(&encl->lock); return 0; @@ -478,7 +478,7 @@ static long sgx_ioc_enclave_add_page(struct file *filep, void __user *arg) struct sgx_enclave_add_page addp; struct sgx_secinfo secinfo; - if (!(encl->flags & SGX_ENCL_CREATED)) + if (!(atomic_read(&encl->flags) & SGX_ENCL_CREATED)) return -EINVAL; if (copy_from_user(&addp, arg, sizeof(addp))) @@ -544,7 +544,7 @@ static int sgx_encl_init(struct sgx_encl *encl, struct sgx_sigstruct *sigstruct, mutex_lock(&encl->lock); - if (encl->flags & (SGX_ENCL_INITIALIZED | SGX_ENCL_DEAD)) { + if (atomic_read(&encl->flags) & (SGX_ENCL_INITIALIZED | SGX_ENCL_DEAD)) { ret = -EFAULT; goto err_out; } @@ -579,7 +579,7 @@ static int sgx_encl_init(struct sgx_encl *encl, struct sgx_sigstruct *sigstruct, } else if (encls_returned_code(ret)) { pr_debug("EINIT returned %d\n", ret); } else { - encl->flags |= SGX_ENCL_INITIALIZED; + atomic_or(SGX_ENCL_INITIALIZED, &encl->flags); } err_out: @@ -611,7 +611,7 @@ static long sgx_ioc_enclave_init(struct file *filep, void __user *arg) struct page *initp_page; int ret; - if (!(encl->flags & SGX_ENCL_CREATED)) + if (!(atomic_read(&encl->flags) & SGX_ENCL_CREATED)) return -EINVAL; if (copy_from_user(&einit, arg, sizeof(einit))) diff --git a/arch/x86/kernel/cpu/sgx/reclaim.c b/arch/x86/kernel/cpu/sgx/reclaim.c index 3f10a8ff00b7..e7a559c41f86 100644 --- a/arch/x86/kernel/cpu/sgx/reclaim.c +++ b/arch/x86/kernel/cpu/sgx/reclaim.c @@ -156,7 +156,7 @@ static bool sgx_reclaimer_evict(struct sgx_epc_page *epc_page) mmput_async(encl_mm->mm); - if (!ret || (encl->flags & SGX_ENCL_DEAD)) + if (!ret || (atomic_read(&encl->flags) & SGX_ENCL_DEAD)) break; } @@ -170,7 +170,7 @@ static bool sgx_reclaimer_evict(struct sgx_epc_page *epc_page) * accessed page is more common and avoiding lock contention in that * case is a boon to performance. */ - if (!ret && !(encl->flags & SGX_ENCL_DEAD)) + if (!ret && !(atomic_read(&encl->flags) & SGX_ENCL_DEAD)) return false; mutex_lock(&encl->lock); @@ -210,7 +210,7 @@ static void sgx_reclaimer_block(struct sgx_epc_page *epc_page) mutex_lock(&encl->lock); - if (!(encl->flags & SGX_ENCL_DEAD)) { + if (!(atomic_read(&encl->flags) & SGX_ENCL_DEAD)) { ret = __eblock(sgx_epc_addr(epc_page)); if (encls_failed(ret)) ENCLS_WARN(ret, "EBLOCK"); @@ -314,7 +314,7 @@ static void sgx_encl_ewb(struct sgx_epc_page *epc_page, unsigned int pt) encl_page->desc &= ~SGX_ENCL_PAGE_RECLAIMED; - if (!(encl->flags & SGX_ENCL_DEAD)) { + if (!(atomic_read(&encl->flags) & SGX_ENCL_DEAD)) { va_page = list_first_entry(&encl->va_pages, struct sgx_va_page, list); va_offset = sgx_alloc_va_slot(va_page); @@ -383,7 +383,7 @@ static void sgx_reclaimer_write(struct sgx_epc_page *epc_page) encl->secs_child_cnt--; if (!encl->secs_child_cnt && - (encl->flags & (SGX_ENCL_DEAD | SGX_ENCL_INITIALIZED))) + (atomic_read(&encl->flags) & (SGX_ENCL_DEAD | SGX_ENCL_INITIALIZED))) sgx_encl_ewb(encl->secs.epc_page, SGX_SECINFO_SECS); mutex_unlock(&encl->lock); From patchwork Fri Aug 30 00:17:03 2019 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Sean Christopherson X-Patchwork-Id: 11122733 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id CAF1929B1 for ; Fri, 30 Aug 2019 00:17:09 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id B18A4215EA for ; Fri, 30 Aug 2019 00:17:09 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1727007AbfH3ARJ (ORCPT ); Thu, 29 Aug 2019 20:17:09 -0400 Received: from mga02.intel.com ([134.134.136.20]:30403 "EHLO mga02.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1726825AbfH3ARJ (ORCPT ); Thu, 29 Aug 2019 20:17:09 -0400 X-Amp-Result: SKIPPED(no attachment in message) X-Amp-File-Uploaded: False Received: from fmsmga004.fm.intel.com ([10.253.24.48]) by orsmga101.jf.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 29 Aug 2019 17:17:08 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.64,445,1559545200"; d="scan'208";a="205917820" Received: from sjchrist-coffee.jf.intel.com ([10.54.74.41]) by fmsmga004.fm.intel.com with ESMTP; 29 Aug 2019 17:17:07 -0700 From: Sean Christopherson To: Jarkko Sakkinen Cc: linux-sgx@vger.kernel.org Subject: [PATCH v3 2/5] x86/sgx: Reject concurrent ioctls on single enclave Date: Thu, 29 Aug 2019 17:17:03 -0700 Message-Id: <20190830001706.29309-3-sean.j.christopherson@intel.com> X-Mailer: git-send-email 2.22.0 In-Reply-To: <20190830001706.29309-1-sean.j.christopherson@intel.com> References: <20190830001706.29309-1-sean.j.christopherson@intel.com> MIME-Version: 1.0 Sender: linux-sgx-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-sgx@vger.kernel.org Except for ENCLAVE_SET_ATTRIBUTE, all SGX ioctls() must be executed serially to successfully initialize an enclave, e.g. the kernel already strictly requires ECREATE->EADD->EINIT, and concurrent EADDs will result in an unstable MRENCLAVE. Explicitly enforce serialization by returning EINVAL if userspace attempts an ioctl while a different ioctl for the same enclave is in progress. The primary beneficiary of explicit serialization is sgx_encl_grow() as it no longer has to deal with dropping and reacquiring encl->lock when a new VA page needs to be allocated. Eliminating the lock juggling in sgx_encl_grow() paves the way for fixing a lock ordering bug in ENCLAVE_ADD_PAGE without having to also juggle mm->mmap_sem. Serializing ioctls also fixes a race between ENCLAVE_CREATE and ENCLAVE_SET_ATTRIBUTE, as the latter does not take encl->lock, e.g. concurrent updates to allowed_attributes could result in a stale value. The race could also be fixed by taking encl->lock or making allowed_attributes atomic, but both add unnecessary overhead with this change applied. Note, manipulating the list of VA pages (encl->va_pages) during EADD must be done while holding encl->lock, the list of VA pages is visible to the reclaimer after the first EADD completes. Signed-off-by: Sean Christopherson --- arch/x86/kernel/cpu/sgx/encl.h | 1 + arch/x86/kernel/cpu/sgx/ioctl.c | 99 ++++++++++++++++----------------- 2 files changed, 50 insertions(+), 50 deletions(-) diff --git a/arch/x86/kernel/cpu/sgx/encl.h b/arch/x86/kernel/cpu/sgx/encl.h index ffc09e1a0fc6..c7608964d69e 100644 --- a/arch/x86/kernel/cpu/sgx/encl.h +++ b/arch/x86/kernel/cpu/sgx/encl.h @@ -56,6 +56,7 @@ enum sgx_encl_flags { SGX_ENCL_INITIALIZED = BIT(1), SGX_ENCL_DEBUG = BIT(2), SGX_ENCL_DEAD = BIT(3), + SGX_ENCL_IOCTL = BIT(4), }; struct sgx_encl_mm { diff --git a/arch/x86/kernel/cpu/sgx/ioctl.c b/arch/x86/kernel/cpu/sgx/ioctl.c index d05cf051ca45..1aa881e69ac1 100644 --- a/arch/x86/kernel/cpu/sgx/ioctl.c +++ b/arch/x86/kernel/cpu/sgx/ioctl.c @@ -14,8 +14,7 @@ #include #include "driver.h" -static struct sgx_va_page *sgx_encl_grow(struct sgx_encl *encl, - unsigned int disallowed_flags) +static struct sgx_va_page *sgx_encl_grow(struct sgx_encl *encl) { struct sgx_va_page *va_page = NULL; void *err; @@ -23,36 +22,19 @@ static struct sgx_va_page *sgx_encl_grow(struct sgx_encl *encl, BUILD_BUG_ON(SGX_VA_SLOT_COUNT != (SGX_ENCL_PAGE_VA_OFFSET_MASK >> 3) + 1); - if (atomic_read(&encl->flags) & disallowed_flags) - return ERR_PTR(-EFAULT); - if (!(encl->page_cnt % SGX_VA_SLOT_COUNT)) { - mutex_unlock(&encl->lock); - va_page = kzalloc(sizeof(*va_page), GFP_KERNEL); - if (!va_page) { - mutex_lock(&encl->lock); + if (!va_page) return ERR_PTR(-ENOMEM); - } va_page->epc_page = sgx_alloc_va_page(); - mutex_lock(&encl->lock); - if (IS_ERR(va_page->epc_page)) { err = ERR_CAST(va_page->epc_page); kfree(va_page); return err; - } else if (atomic_read(&encl->flags) & disallowed_flags) { - sgx_free_page(va_page->epc_page); - kfree(va_page); - return ERR_PTR(-EFAULT); - } else if (encl->page_cnt % SGX_VA_SLOT_COUNT) { - sgx_free_page(va_page->epc_page); - kfree(va_page); - va_page = NULL; - } else { - list_add(&va_page->list, &encl->va_pages); } + + WARN_ON_ONCE(encl->page_cnt % SGX_VA_SLOT_COUNT); } encl->page_cnt++; return va_page; @@ -174,13 +156,14 @@ static int sgx_encl_create(struct sgx_encl *encl, struct sgx_secs *secs) struct file *backing; long ret; - mutex_lock(&encl->lock); + if (atomic_read(&encl->flags) & SGX_ENCL_CREATED) + return -EINVAL; - va_page = sgx_encl_grow(encl, SGX_ENCL_CREATED | SGX_ENCL_DEAD); - if (IS_ERR(va_page)) { - ret = PTR_ERR(va_page); - goto err_out_unlock; - } + va_page = sgx_encl_grow(encl); + if (IS_ERR(va_page)) + return PTR_ERR(va_page); + else if (va_page) + list_add(&va_page->list, &encl->va_pages); ssaframesize = sgx_calc_ssaframesize(secs->miscselect, secs->xfrm); if (sgx_validate_secs(secs, ssaframesize)) { @@ -230,12 +213,11 @@ static int sgx_encl_create(struct sgx_encl *encl, struct sgx_secs *secs) /* * Set SGX_ENCL_CREATED only after the enclave is fully prepped. This - * allows other flows to check if the enclave has been created without - * taking encl->lock. + * allows setting and checking enclave creation without having to take + * encl->lock. */ atomic_or(SGX_ENCL_CREATED, &encl->flags); - mutex_unlock(&encl->lock); return 0; err_out: @@ -249,8 +231,6 @@ static int sgx_encl_create(struct sgx_encl *encl, struct sgx_secs *secs) err_out_shrink: sgx_encl_shrink(encl, va_page); -err_out_unlock: - mutex_unlock(&encl->lock); return ret; } @@ -270,9 +250,8 @@ static int sgx_encl_create(struct sgx_encl *encl, struct sgx_secs *secs) * 0 on success, * -errno otherwise */ -static long sgx_ioc_enclave_create(struct file *filep, void __user *arg) +static long sgx_ioc_enclave_create(struct sgx_encl *encl, void __user *arg) { - struct sgx_encl *encl = filep->private_data; struct sgx_enclave_create ecreate; struct page *secs_page; struct sgx_secs *secs; @@ -404,14 +383,21 @@ static int sgx_encl_add_page(struct sgx_encl *encl, return PTR_ERR(epc_page); } - mutex_lock(&encl->lock); - - va_page = sgx_encl_grow(encl, SGX_ENCL_INITIALIZED | SGX_ENCL_DEAD); + va_page = sgx_encl_grow(encl); if (IS_ERR(va_page)) { ret = PTR_ERR(va_page); goto err_out_free; } + mutex_lock(&encl->lock); + + /* + * Adding to encl->va_pages must be done under encl->lock. Ditto for + * deleting (via sgx_encl_shrink()) in the error path. + */ + if (va_page) + list_add(&va_page->list, &encl->va_pages); + ret = radix_tree_insert(&encl->page_tree, PFN_DOWN(encl_page->desc), encl_page); if (ret) @@ -431,12 +417,12 @@ static int sgx_encl_add_page(struct sgx_encl *encl, err_out_shrink: sgx_encl_shrink(encl, va_page); + mutex_unlock(&encl->lock); err_out_free: sgx_free_page(epc_page); kfree(encl_page); - mutex_unlock(&encl->lock); return ret; } @@ -472,9 +458,8 @@ static int sgx_encl_add_page(struct sgx_encl *encl, * -ENOMEM if any memory allocation, including EPC, fails, * -errno otherwise */ -static long sgx_ioc_enclave_add_page(struct file *filep, void __user *arg) +static long sgx_ioc_enclave_add_page(struct sgx_encl *encl, void __user *arg) { - struct sgx_encl *encl = filep->private_data; struct sgx_enclave_add_page addp; struct sgx_secinfo secinfo; @@ -602,9 +587,8 @@ static int sgx_encl_init(struct sgx_encl *encl, struct sgx_sigstruct *sigstruct, * SGX error code on EINIT failure, * -errno otherwise */ -static long sgx_ioc_enclave_init(struct file *filep, void __user *arg) +static long sgx_ioc_enclave_init(struct sgx_encl *encl, void __user *arg) { - struct sgx_encl *encl = filep->private_data; struct sgx_einittoken *einittoken; struct sgx_sigstruct *sigstruct; struct sgx_enclave_init einit; @@ -658,9 +642,9 @@ static long sgx_ioc_enclave_init(struct file *filep, void __user *arg) * * Return: 0 on success, -errno otherwise */ -static long sgx_ioc_enclave_set_attribute(struct file *filep, void __user *arg) +static long sgx_ioc_enclave_set_attribute(struct sgx_encl *encl, + void __user *arg) { - struct sgx_encl *encl = filep->private_data; struct sgx_enclave_set_attribute params; struct file *attribute_file; int ret; @@ -687,16 +671,31 @@ static long sgx_ioc_enclave_set_attribute(struct file *filep, void __user *arg) long sgx_ioctl(struct file *filep, unsigned int cmd, unsigned long arg) { + struct sgx_encl *encl = filep->private_data; + int ret; + + if (atomic_fetch_or(SGX_ENCL_IOCTL, &encl->flags) & SGX_ENCL_IOCTL) + return -EBUSY; + switch (cmd) { case SGX_IOC_ENCLAVE_CREATE: - return sgx_ioc_enclave_create(filep, (void __user *)arg); + ret = sgx_ioc_enclave_create(encl, (void __user *)arg); + break; case SGX_IOC_ENCLAVE_ADD_PAGE: - return sgx_ioc_enclave_add_page(filep, (void __user *)arg); + ret = sgx_ioc_enclave_add_page(encl, (void __user *)arg); + break; case SGX_IOC_ENCLAVE_INIT: - return sgx_ioc_enclave_init(filep, (void __user *)arg); + ret = sgx_ioc_enclave_init(encl, (void __user *)arg); + break; case SGX_IOC_ENCLAVE_SET_ATTRIBUTE: - return sgx_ioc_enclave_set_attribute(filep, (void __user *)arg); + ret = sgx_ioc_enclave_set_attribute(encl, (void __user *)arg); + break; default: - return -ENOIOCTLCMD; + ret = -ENOIOCTLCMD; + break; } + + atomic_andnot(SGX_ENCL_IOCTL, &encl->flags); + + return ret; } From patchwork Fri Aug 30 00:17:04 2019 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Sean Christopherson X-Patchwork-Id: 11122735 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 25D4E174A for ; Fri, 30 Aug 2019 00:17:10 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id 043A62173E for ; Fri, 30 Aug 2019 00:17:10 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1726825AbfH3ARJ (ORCPT ); Thu, 29 Aug 2019 20:17:09 -0400 Received: from mga02.intel.com ([134.134.136.20]:30403 "EHLO mga02.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1726014AbfH3ARJ (ORCPT ); Thu, 29 Aug 2019 20:17:09 -0400 X-Amp-Result: SKIPPED(no attachment in message) X-Amp-File-Uploaded: False Received: from fmsmga004.fm.intel.com ([10.253.24.48]) by orsmga101.jf.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 29 Aug 2019 17:17:08 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.64,445,1559545200"; d="scan'208";a="205917824" Received: from sjchrist-coffee.jf.intel.com ([10.54.74.41]) by fmsmga004.fm.intel.com with ESMTP; 29 Aug 2019 17:17:07 -0700 From: Sean Christopherson To: Jarkko Sakkinen Cc: linux-sgx@vger.kernel.org Subject: [PATCH v3 3/5] x86/sgx: Take encl->lock inside of mm->mmap_sem for EADD Date: Thu, 29 Aug 2019 17:17:04 -0700 Message-Id: <20190830001706.29309-4-sean.j.christopherson@intel.com> X-Mailer: git-send-email 2.22.0 In-Reply-To: <20190830001706.29309-1-sean.j.christopherson@intel.com> References: <20190830001706.29309-1-sean.j.christopherson@intel.com> MIME-Version: 1.0 Sender: linux-sgx-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-sgx@vger.kernel.org Reverse the order in which encl->lock and mm->mmap_sem are taken during ENCLAVE_ADD_PAGE so as to adhere to SGX's lock ordering requirements. Refactor EEXTEND and the final bookeeping out of __sgx_encl_add_page() so that mm->mmap_sem can be dropped after EADD without spreading the lock/unlock across multiple functions. Signed-off-by: Sean Christopherson --- arch/x86/kernel/cpu/sgx/ioctl.c | 55 ++++++++++++++++++++------------- 1 file changed, 33 insertions(+), 22 deletions(-) diff --git a/arch/x86/kernel/cpu/sgx/ioctl.c b/arch/x86/kernel/cpu/sgx/ioctl.c index 1aa881e69ac1..d34a13d98148 100644 --- a/arch/x86/kernel/cpu/sgx/ioctl.c +++ b/arch/x86/kernel/cpu/sgx/ioctl.c @@ -308,43 +308,40 @@ static int sgx_validate_secinfo(struct sgx_secinfo *secinfo) static int __sgx_encl_add_page(struct sgx_encl *encl, struct sgx_encl_page *encl_page, struct sgx_epc_page *epc_page, - struct sgx_secinfo *secinfo, unsigned long src, - unsigned long mrmask) + struct sgx_secinfo *secinfo, unsigned long src) { struct sgx_pageinfo pginfo; struct vm_area_struct *vma; int ret; - int i; pginfo.secs = (unsigned long)sgx_epc_addr(encl->secs.epc_page); pginfo.addr = SGX_ENCL_PAGE_ADDR(encl_page); pginfo.metadata = (unsigned long)secinfo; pginfo.contents = src; - down_read(¤t->mm->mmap_sem); - /* Query vma's VM_MAYEXEC as an indirect path_noexec() check. */ if (encl_page->vm_max_prot_bits & VM_EXEC) { vma = find_vma(current->mm, src); - if (!vma) { - up_read(¤t->mm->mmap_sem); + if (!vma) return -EFAULT; - } - if (!(vma->vm_flags & VM_MAYEXEC)) { - up_read(¤t->mm->mmap_sem); + if (!(vma->vm_flags & VM_MAYEXEC)) return -EACCES; - } } __uaccess_begin(); ret = __eadd(&pginfo, sgx_epc_addr(epc_page)); __uaccess_end(); - up_read(¤t->mm->mmap_sem); + return ret ? -EFAULT : 0; +} - if (ret) - return -EFAULT; +static int __sgx_encl_extend(struct sgx_encl *encl, + struct sgx_epc_page *epc_page, + unsigned long mrmask) +{ + int ret; + int i; for_each_set_bit(i, &mrmask, 16) { ret = __eextend(sgx_epc_addr(encl->secs.epc_page), @@ -355,12 +352,6 @@ static int __sgx_encl_add_page(struct sgx_encl *encl, return -EFAULT; } } - - encl_page->encl = encl; - encl_page->epc_page = epc_page; - encl->secs_child_cnt++; - sgx_mark_page_reclaimable(encl_page->epc_page); - return 0; } @@ -389,6 +380,8 @@ static int sgx_encl_add_page(struct sgx_encl *encl, goto err_out_free; } + down_read(¤t->mm->mmap_sem); + mutex_lock(&encl->lock); /* @@ -398,17 +391,35 @@ static int sgx_encl_add_page(struct sgx_encl *encl, if (va_page) list_add(&va_page->list, &encl->va_pages); + /* + * Insert prior to EADD in case of OOM. EADD modifies MRENCLAVE, i.e. + * can't be gracefully unwound, while failure on EADD/EXTEND is limited + * to userspace errors (or kernel/hardware bugs). + */ ret = radix_tree_insert(&encl->page_tree, PFN_DOWN(encl_page->desc), encl_page); - if (ret) + if (ret) { + up_read(¤t->mm->mmap_sem); goto err_out_shrink; + } ret = __sgx_encl_add_page(encl, encl_page, epc_page, secinfo, - addp->src, addp->mrmask); + addp->src); + up_read(¤t->mm->mmap_sem); + + if (ret) + goto err_out; + + ret = __sgx_encl_extend(encl, epc_page, addp->mrmask); if (ret) goto err_out; + encl_page->encl = encl; + encl_page->epc_page = epc_page; + encl->secs_child_cnt++; + sgx_mark_page_reclaimable(encl_page->epc_page); mutex_unlock(&encl->lock); + return 0; err_out: From patchwork Fri Aug 30 00:17:05 2019 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Sean Christopherson X-Patchwork-Id: 11122731 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 81BC518EC for ; Fri, 30 Aug 2019 00:17:09 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id 600FE215EA for ; Fri, 30 Aug 2019 00:17:09 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1726739AbfH3ARJ (ORCPT ); Thu, 29 Aug 2019 20:17:09 -0400 Received: from mga04.intel.com ([192.55.52.120]:31183 "EHLO mga04.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1726014AbfH3ARI (ORCPT ); Thu, 29 Aug 2019 20:17:08 -0400 X-Amp-Result: SKIPPED(no attachment in message) X-Amp-File-Uploaded: False Received: from fmsmga004.fm.intel.com ([10.253.24.48]) by fmsmga104.fm.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 29 Aug 2019 17:17:08 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.64,445,1559545200"; d="scan'208";a="205917827" Received: from sjchrist-coffee.jf.intel.com ([10.54.74.41]) by fmsmga004.fm.intel.com with ESMTP; 29 Aug 2019 17:17:07 -0700 From: Sean Christopherson To: Jarkko Sakkinen Cc: linux-sgx@vger.kernel.org Subject: [PATCH v3 4/5] x86/sgx: Reject all ioctls on dead enclaves Date: Thu, 29 Aug 2019 17:17:05 -0700 Message-Id: <20190830001706.29309-5-sean.j.christopherson@intel.com> X-Mailer: git-send-email 2.22.0 In-Reply-To: <20190830001706.29309-1-sean.j.christopherson@intel.com> References: <20190830001706.29309-1-sean.j.christopherson@intel.com> MIME-Version: 1.0 Sender: linux-sgx-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-sgx@vger.kernel.org Do not allow userspace to operate on a dead enclave. Note, moving the SGX_ENCL_DEAD for EINIT outside of encl->lock is safe now that sgx_ioctl() prevents concurrent calls. SGX_ENCL_DEAD is only set when the fd is released, i.e. EINIT can no longer be reached, or within an ioctl call. Signed-off-by: Sean Christopherson --- arch/x86/kernel/cpu/sgx/ioctl.c | 10 +++++++--- 1 file changed, 7 insertions(+), 3 deletions(-) diff --git a/arch/x86/kernel/cpu/sgx/ioctl.c b/arch/x86/kernel/cpu/sgx/ioctl.c index d34a13d98148..3800281c5945 100644 --- a/arch/x86/kernel/cpu/sgx/ioctl.c +++ b/arch/x86/kernel/cpu/sgx/ioctl.c @@ -540,7 +540,7 @@ static int sgx_encl_init(struct sgx_encl *encl, struct sgx_sigstruct *sigstruct, mutex_lock(&encl->lock); - if (atomic_read(&encl->flags) & (SGX_ENCL_INITIALIZED | SGX_ENCL_DEAD)) { + if (atomic_read(&encl->flags) & SGX_ENCL_INITIALIZED) { ret = -EFAULT; goto err_out; } @@ -683,11 +683,15 @@ static long sgx_ioc_enclave_set_attribute(struct sgx_encl *encl, long sgx_ioctl(struct file *filep, unsigned int cmd, unsigned long arg) { struct sgx_encl *encl = filep->private_data; - int ret; + int ret, encl_flags; - if (atomic_fetch_or(SGX_ENCL_IOCTL, &encl->flags) & SGX_ENCL_IOCTL) + encl_flags = atomic_fetch_or(SGX_ENCL_IOCTL, &encl->flags); + if (encl_flags & SGX_ENCL_IOCTL) return -EBUSY; + if (encl_flags & SGX_ENCL_DEAD) + return -EFAULT; + switch (cmd) { case SGX_IOC_ENCLAVE_CREATE: ret = sgx_ioc_enclave_create(encl, (void __user *)arg); From patchwork Fri Aug 30 00:17:06 2019 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Sean Christopherson X-Patchwork-Id: 11122727 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 2D398174A for ; Fri, 30 Aug 2019 00:17:09 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id 150A22189D for ; Fri, 30 Aug 2019 00:17:09 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1726983AbfH3ARI (ORCPT ); Thu, 29 Aug 2019 20:17:08 -0400 Received: from mga04.intel.com ([192.55.52.120]:31183 "EHLO mga04.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1726825AbfH3ARI (ORCPT ); Thu, 29 Aug 2019 20:17:08 -0400 X-Amp-Result: SKIPPED(no attachment in message) X-Amp-File-Uploaded: False Received: from fmsmga004.fm.intel.com ([10.253.24.48]) by fmsmga104.fm.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 29 Aug 2019 17:17:08 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.64,445,1559545200"; d="scan'208";a="205917830" Received: from sjchrist-coffee.jf.intel.com ([10.54.74.41]) by fmsmga004.fm.intel.com with ESMTP; 29 Aug 2019 17:17:07 -0700 From: Sean Christopherson To: Jarkko Sakkinen Cc: linux-sgx@vger.kernel.org Subject: [PATCH v3 5/5] x86/sgx: Destroy the enclave if EEXTEND fails Date: Thu, 29 Aug 2019 17:17:06 -0700 Message-Id: <20190830001706.29309-6-sean.j.christopherson@intel.com> X-Mailer: git-send-email 2.22.0 In-Reply-To: <20190830001706.29309-1-sean.j.christopherson@intel.com> References: <20190830001706.29309-1-sean.j.christopherson@intel.com> MIME-Version: 1.0 Sender: linux-sgx-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-sgx@vger.kernel.org Mark an enclave as dead and release its resources if EEXTEND fails, as the driver cannot gracefully unwind from EEXTEND failure and does not provide userspace enough information to restart the ioctl. Allowing EEXTEND to be restarted is not a requirement as EEXTEND can only fail in the event of a kernel or hardware bug. Signed-off-by: Sean Christopherson --- arch/x86/kernel/cpu/sgx/ioctl.c | 1 + 1 file changed, 1 insertion(+) diff --git a/arch/x86/kernel/cpu/sgx/ioctl.c b/arch/x86/kernel/cpu/sgx/ioctl.c index 3800281c5945..fffdf25ee3f9 100644 --- a/arch/x86/kernel/cpu/sgx/ioctl.c +++ b/arch/x86/kernel/cpu/sgx/ioctl.c @@ -349,6 +349,7 @@ static int __sgx_encl_extend(struct sgx_encl *encl, if (ret) { if (encls_failed(ret)) ENCLS_WARN(ret, "EEXTEND"); + sgx_encl_destroy(encl); return -EFAULT; } }