From: Sean Christopherson <sean.j.christopherson@intel.com>
To: Jarkko Sakkinen <jarkko.sakkinen@linux.intel.com>,
intel-sgx-kernel-dev@lists.01.org
Cc: linux-kernel@vger.kernel.org, platform-driver-x86@vger.kernel.org
Subject: Re: [intel-sgx-kernel-dev] [PATCH v5 10/11] intel_sgx: glue code for in-kernel LE
Date: Tue, 14 Nov 2017 10:16:43 -0800 [thread overview]
Message-ID: <1510683403.11044.2.camel@intel.com> (raw)
In-Reply-To: <20171113194528.28557-11-jarkko.sakkinen@linux.intel.com>
On Mon, 2017-11-13 at 21:45 +0200, Jarkko Sakkinen wrote:
> --- a/drivers/platform/x86/intel_sgx/sgx_main.c
> +++ b/drivers/platform/x86/intel_sgx/sgx_main.c
> @@ -88,6 +88,37 @@ u64 sgx_encl_size_max_64;
> u64 sgx_xfrm_mask = 0x3;
> u32 sgx_misc_reserved;
> u32 sgx_xsave_size_tbl[64];
> +bool sgx_unlocked_msrs;
> +u64 sgx_le_pubkeyhash[4];
> +
> +static DECLARE_RWSEM(sgx_file_sem);
> +
> +static int sgx_open(struct inode *inode, struct file *file)
> +{
> + int ret;
> +
> + down_read(&sgx_file_sem);
> +
> + ret = sgx_le_start(&sgx_le_ctx);
> + if (ret) {
> + up_read(&sgx_file_sem);
> + return ret;
> + }
> +
> + return 0;
> +}
> +
> +static int sgx_release(struct inode *inode, struct file *file)
> +{
> + up_read(&sgx_file_sem);
> +
> + if (down_write_trylock(&sgx_file_sem)) {
> + sgx_le_stop(&sgx_le_ctx);
> + up_write(&sgx_file_sem);
> + }
> +
> + return 0;
> +}
This semaphore approach is broken due to the LE process using an anon inode for
/dev/sgx, which results in sgx_release being called without an accompanying call
to sgx_open. This causes deadlocks due to a semaphore underrun.
https://lists.01.org/pipermail/intel-sgx-kernel-dev/2017-November/000901.html
[ 242.659272] INFO: task lsdt:9425 blocked for more than 120 seconds.
[ 242.659783] Not tainted 4.14.0-rc4+ #18
[ 242.660063] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this
[ 242.660558] lsdt D 0 9425 1 0x00000004
[ 242.660559] Call Trace:
[ 242.660564] __schedule+0x3c2/0x8b0
[ 242.660567] schedule+0x36/0x80
[ 242.660568] rwsem_down_read_failed+0x10a/0x170
[ 242.660569] call_rwsem_down_read_failed+0x18/0x30
[ 242.660570] ? call_rwsem_down_read_failed+0x18/0x30
[ 242.660571] down_read+0x20/0x40
[ 242.660572] sgx_open+0x19/0x40 [intel_sgx]
[ 242.660574] chrdev_open+0xbf/0x1b0
[ 242.660576] do_dentry_open+0x1f8/0x300
[ 242.660577] ? cdev_put+0x30/0x30
[ 242.660578] vfs_open+0x4f/0x70
[ 242.660579] path_openat+0x2ae/0x13a0
[ 242.660581] ? mem_cgroup_uncharge_swap+0x60/0x90
[ 242.660582] do_filp_open+0x99/0x110
[ 242.660583] ? __check_object_size+0xfc/0x1a0
[ 242.660585] ? __alloc_fd+0xb0/0x170
[ 242.660586] do_sys_open+0x124/0x210
[ 242.660587] ? do_sys_open+0x124/0x210
[ 242.660588] SyS_open+0x1e/0x20
[ 242.660589] entry_SYSCALL_64_fastpath+0x1e/0xa9
[ 242.660590] RIP: 0033:0x7f426cf9ec7d
[ 242.660591] RSP: 002b:00007f426b31ea60 EFLAGS: 00000293 ORIG_RAX:
[ 242.660592] RAX: ffffffffffffffda RBX: 000000c4200ba000 RCX: 00007f426cf9ec7d
[ 242.660592] RDX: 0000000000000000 RSI: 0000000000000002 RDI: 000000000068cca7
[ 242.660593] RBP: 00007f426b31ec10 R08: 0000000000f6bc30 R09: 0000000000000000
[ 242.660593] R10: 00007f4264000078 R11: 0000000000000293 R12: 0000000000000001
[ 242.660594] R13: 0000000000000000 R14: 00007f426d31b13d R15: 00007f42640008c0
> #ifdef CONFIG_COMPAT
> long sgx_compat_ioctl(struct file *filep, unsigned int cmd, unsigned long
> arg)
> @@ -141,8 +172,10 @@ static unsigned long sgx_get_unmapped_area(struct file
> *file,
> return addr;
> }
>
> -static const struct file_operations sgx_fops = {
> +const struct file_operations sgx_fops = {
> .owner = THIS_MODULE,
> + .open = sgx_open,
> + .release = sgx_release,
> .unlocked_ioctl = sgx_ioctl,
> #ifdef CONFIG_COMPAT
> .compat_ioctl = sgx_compat_ioctl,
next prev parent reply other threads:[~2017-11-14 18:20 UTC|newest]
Thread overview: 78+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-11-13 19:45 [PATCH v5 00/11] Intel SGX Driver Jarkko Sakkinen
2017-11-13 19:45 ` [PATCH v5 01/11] intel_sgx: updated MAINTAINERS Jarkko Sakkinen
2017-11-17 21:54 ` Darren Hart
2017-11-24 19:18 ` Jarkko Sakkinen
2017-11-13 19:45 ` [PATCH v5 02/11] x86: add SGX definition to cpufeature Jarkko Sakkinen
2017-11-13 19:45 ` [PATCH v5 03/11] x86: define the feature control MSR's SGX enable bit Jarkko Sakkinen
2017-11-17 21:48 ` Darren Hart
2017-11-13 19:45 ` [PATCH v5 04/11] x86: define the feature control MSR's SGX launch control bit Jarkko Sakkinen
2017-11-13 19:45 ` [PATCH v5 05/11] x86: add SGX MSRs to msr-index.h Jarkko Sakkinen
2017-11-13 19:45 ` [PATCH v5 06/11] intel_sgx: driver for Intel Software Guard Extensions Jarkko Sakkinen
2017-11-13 23:41 ` James Morris
2017-11-14 20:12 ` Jarkko Sakkinen
2017-11-15 10:04 ` Jarkko Sakkinen
2017-11-14 17:55 ` [intel-sgx-kernel-dev] " Sean Christopherson
2017-11-14 20:28 ` Jarkko Sakkinen
2017-11-15 18:20 ` Sean Christopherson
2017-12-13 23:18 ` Christopherson, Sean J
2017-12-15 15:00 ` Jarkko Sakkinen
2017-12-19 18:52 ` Christopherson, Sean J
2017-12-19 23:11 ` Jarkko Sakkinen
2017-12-19 23:24 ` Christopherson, Sean J
2017-12-20 10:13 ` Jarkko Sakkinen
2017-11-13 19:45 ` [PATCH v5 07/11] intel_sgx: ptrace() support Jarkko Sakkinen
2017-11-16 9:28 ` Thomas Gleixner
2017-11-23 10:25 ` Jarkko Sakkinen
2017-11-13 19:45 ` [PATCH v5 08/11] intel_sgx: in-kernel launch enclave Jarkko Sakkinen
2017-11-14 17:05 ` [intel-sgx-kernel-dev] " Sean Christopherson
2017-11-14 20:05 ` Jarkko Sakkinen
2017-11-20 22:21 ` Jarkko Sakkinen
2017-11-15 11:50 ` Peter Zijlstra
2017-11-20 22:25 ` Jarkko Sakkinen
2017-11-20 22:43 ` Thomas Gleixner
2017-11-20 23:43 ` Jarkko Sakkinen
2017-11-20 23:48 ` Thomas Gleixner
2017-11-21 12:23 ` Jarkko Sakkinen
2017-11-21 23:36 ` Thomas Gleixner
2017-11-13 19:45 ` [PATCH v5 09/11] fs/pipe.c: export create_pipe_files() and replace_fd() Jarkko Sakkinen
2017-11-16 9:15 ` Thomas Gleixner
2017-11-20 22:30 ` Jarkko Sakkinen
2017-11-13 19:45 ` [PATCH v5 10/11] intel_sgx: glue code for in-kernel LE Jarkko Sakkinen
2017-11-14 18:16 ` Sean Christopherson [this message]
2017-11-14 20:31 ` [intel-sgx-kernel-dev] " Jarkko Sakkinen
2017-11-15 10:10 ` Jarkko Sakkinen
2017-11-17 23:07 ` Darren Hart
2017-11-25 12:52 ` Jarkko Sakkinen
2017-11-25 18:01 ` Jarkko Sakkinen
2017-11-13 19:45 ` [PATCH v5 11/11] intel_sgx: driver documentation Jarkko Sakkinen
2017-11-14 3:01 ` [intel-sgx-kernel-dev] " Kai Huang
2017-11-14 19:47 ` Jarkko Sakkinen
2017-11-14 21:12 ` Kai Huang
2017-11-14 8:36 ` Borislav Petkov
2017-11-14 20:49 ` Jarkko Sakkinen
2017-11-14 21:53 ` Borislav Petkov
2017-11-20 22:37 ` Jarkko Sakkinen
2017-11-20 22:42 ` Borislav Petkov
2017-11-20 23:41 ` Jarkko Sakkinen
2017-11-21 11:10 ` Borislav Petkov
2017-11-15 11:54 ` Peter Zijlstra
2017-11-20 22:46 ` Jarkko Sakkinen
2017-11-21 12:38 ` Jarkko Sakkinen
2017-11-21 12:47 ` Borislav Petkov
2017-11-21 23:45 ` Jethro Beekman
2017-11-22 0:10 ` Borislav Petkov
2017-11-22 0:27 ` Jethro Beekman
2017-11-22 11:00 ` Borislav Petkov
2017-11-22 16:07 ` Jethro Beekman
2017-11-17 21:43 ` Darren Hart
2017-11-17 23:34 ` Thomas Gleixner
2017-11-17 23:46 ` Darren Hart
2017-11-20 23:12 ` Jarkko Sakkinen
2017-11-20 23:08 ` Jarkko Sakkinen
2017-11-27 17:03 ` Sean Christopherson
2017-11-27 19:41 ` Sean Christopherson
2017-11-28 20:37 ` Jarkko Sakkinen
2017-11-28 20:46 ` Jarkko Sakkinen
2017-11-24 17:26 ` Jarkko Sakkinen
2017-11-15 10:35 ` [PATCH v5 00/11] Intel SGX Driver Thomas Gleixner
2017-11-20 22:20 ` Jarkko Sakkinen
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1510683403.11044.2.camel@intel.com \
--to=sean.j.christopherson@intel.com \
--cc=intel-sgx-kernel-dev@lists.01.org \
--cc=jarkko.sakkinen@linux.intel.com \
--cc=linux-kernel@vger.kernel.org \
--cc=platform-driver-x86@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).