On Thu, 2016-10-13 at 22:42 +0900, Sergey Senozhatsky wrote:
quoted
quoted
https://git.kernel.org/cgit/linux/kernel/git/luto/linux.git/commi
t/?h=x86/vmap_stack&id=0a39cfa6fbb5d5635c85253cc7d6b44b54822afd
https://git.kernel.org/cgit/linux/kernel/git/luto/linux.git/commi
t/?h=x86/vmap_stack&id=bf8cfa200b5a01383ea39fc8ce2f32909767baa8
That truly sounds like something we'd rather avoid in the TX/RX
paths though, which should perform well.
didn't fix.
It couldn't, since the new helpers weren't used in mac80211 in those
patches yet.
so I finally had some time to do a better bug-reporter job.
I added a bunch of printk-s and several virt_addr_valid()-s
to ieee80211_aes_ccm_encrypt().
and right befoe the Oops I see the following report from
virt_addr_valid()
FAIL: 00004100002cba02 > ffffc900802cba02 || 1 -> (00004100002cba02
quoted
quoted
39) == 130
Yeah, we already know that in this function the aad variable is on the
stack, it explicitly is.
The question, though, is why precisely that fails in the crypto code.
Can you send the Oops report itself?
johannes
On (10/13/16 15:45), Johannes Berg wrote:
On Thu, 2016-10-13 at 22:42 +0900, Sergey Senozhatsky wrote:
quoted
quoted
quoted
https://git.kernel.org/cgit/linux/kernel/git/luto/linux.git/commi
t/?h=x86/vmap_stack&id=0a39cfa6fbb5d5635c85253cc7d6b44b54822afd
https://git.kernel.org/cgit/linux/kernel/git/luto/linux.git/commi
t/?h=x86/vmap_stack&id=bf8cfa200b5a01383ea39fc8ce2f32909767baa8
That truly sounds like something we'd rather avoid in the TX/RX
paths though, which should perform well.
didn't fix.
It couldn't, since the new helpers weren't used in mac80211 in those
patches yet.
indeed. I thought they were.
quoted
FAIL: 00004100002cba02 > ffffc900802cba02 || 1 -> (00004100002cba02
quoted
quoted
39) == 130
The question, though, is why precisely that fails in the crypto code.
Can you send the Oops report itself?
kernel: BUG: unable to handle kernel NULL pointer dereference at (null)
kernel: IP: [<ffffffff8146d2f4>] ieee80211_aes_ccm_decrypt+0x107/0x27f
kernel: PGD 0
kernel:
kernel: Oops: 0000 [#1] PREEMPT SMP
kernel: Modules linked in: nls_iso8859_1 nls_cp437 vfat fat mousedev psmouse serio_raw atkbd libps2 i915 coretemp i2c_algo_bit hwmon crc32c_intel mxm_wmi drm_kms_helper cfbfillrect syscopyarea cfbimgblt sysfillrect iwlmvm sysimgblt fb_sys_fops i2c_i801 cfbcopyarea ie31200_edac drm iwlwifi i2c
kernel: CPU: 3 PID: 245 Comm: irq/28-iwlwifi Not tainted 4.8.0-next-20161013-dbg-00002-ge789862-dirty #112
kernel: task: ffff88041bf01800 task.stack: ffffc900002d0000
kernel: RIP: 0010:[<ffffffff8146d2f4>] [<ffffffff8146d2f4>] ieee80211_aes_ccm_decrypt+0x107/0x27f
kernel: RSP: 0018:ffffc900002d3770 EFLAGS: 00010246
kernel: RAX: ffffc900002d3930 RBX: ffff8804133cf606 RCX: 0000000000082000
kernel: RDX: 0000000000000000 RSI: 0000000000000018 RDI: 0000000000000a02
kernel: RBP: ffffc900002d39b8 R08: 00000000000005e4 R09: 00000004100002d3
kernel: R10: 000000000000001c R11: ffff8803e66d2d20 R12: ffff8804191c2780
kernel: R13: ffffc900002d39f0 R14: ffff8804133cf022 R15: 1ffff9200005a6ee
kernel: FS: 0000000000000000(0000) GS:ffff88041ea00000(0000) knlGS:0000000000000000
kernel: CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
kernel: CR2: 0000000000000000 CR3: 0000000001805000 CR4: 00000000001406e0
kernel: Stack:
kernel: 0000000000000000 0000000000000000 0000000000000000 0000000000000000
kernel: 0000000000000000 0000000000000000 0000000000000000 0000000000000000
kernel: 0000000000000000 0000000000000000 0000000000000000 0000000000000000
kernel: Call Trace:
kernel: [<ffffffff8145c405>] ieee80211_crypto_ccmp_decrypt+0x204/0x298
kernel: [<ffffffff81476cd8>] ieee80211_rx_handlers+0x7df/0x1c1d
kernel: [<ffffffff814790c8>] ieee80211_prepare_and_rx_handle+0xdc2/0xe79
kernel: [<ffffffff814792e7>] ? ieee80211_rx_napi+0x168/0x7b6
kernel: [<ffffffff8147960a>] ieee80211_rx_napi+0x48b/0x7b6
kernel: [<ffffffff8123729e>] ? debug_smp_processor_id+0x17/0x19
kernel: [<ffffffffa01cfe3b>] iwl_mvm_rx_rx_mpdu+0x6e6/0x751 [iwlmvm]
kernel: [<ffffffffa01c9c49>] iwl_mvm_rx+0x7e/0x98 [iwlmvm]
kernel: [<ffffffffa0131bca>] iwl_pcie_rx_handle+0x523/0x698 [iwlwifi]
kernel: [<ffffffffa0133027>] iwl_pcie_irq_handler+0x46f/0x65f [iwlwifi]
kernel: [<ffffffff810893d0>] ? irq_finalize_oneshot+0xd4/0xd4
kernel: [<ffffffff810893ed>] irq_thread_fn+0x1d/0x34
kernel: [<ffffffff81089661>] irq_thread+0xe6/0x1bb
kernel: [<ffffffff810894e6>] ? wake_threads_waitq+0x2c/0x2c
kernel: [<ffffffff8108957b>] ? irq_thread_dtor+0x95/0x95
kernel: [<ffffffff8105d762>] kthread+0xfc/0x104
kernel: [<ffffffff8107d36c>] ? put_lock_stats.isra.9+0xe/0x20
kernel: [<ffffffff8105d666>] ? kthread_create_on_node+0x3f/0x3f
kernel: [<ffffffff814b2852>] ret_from_fork+0x22/0x30
kernel: Code: 01 ca 49 89 d1 48 89 d1 48 c1 ea 23 48 8b 14 d5 80 23 63 82 49 c1 e9 0c 48 c1 e9 1b 48 85 d2 74 0a 0f b6 c9 48 c1 e1 04 48 01 ca <48> 8b 12 49 c1 e1 06 b9 00 00 00 80 89 7d 80 89 75 84 48 8b 3d
kernel: RIP [<ffffffff8146d2f4>] ieee80211_aes_ccm_decrypt+0x107/0x27f
kernel: RSP <ffffc900002d3770>
kernel: CR2: 0000000000000000
kernel: ---[ end trace 3cd1fcd496516f72 ]---
-ss
On (10/14/16 00:00), Sergey Senozhatsky wrote:
kernel: [<ffffffff8145c405>] ieee80211_crypto_ccmp_decrypt+0x204/0x298
kernel: [<ffffffff81476cd8>] ieee80211_rx_handlers+0x7df/0x1c1d
kernel: [<ffffffff814790c8>] ieee80211_prepare_and_rx_handle+0xdc2/0xe79
kernel: [<ffffffff814792e7>] ? ieee80211_rx_napi+0x168/0x7b6
kernel: [<ffffffff8147960a>] ieee80211_rx_napi+0x48b/0x7b6
kernel: [<ffffffff8123729e>] ? debug_smp_processor_id+0x17/0x19
kernel: [<ffffffffa01cfe3b>] iwl_mvm_rx_rx_mpdu+0x6e6/0x751 [iwlmvm]
kernel: [<ffffffffa01c9c49>] iwl_mvm_rx+0x7e/0x98 [iwlmvm]
kernel: [<ffffffffa0131bca>] iwl_pcie_rx_handle+0x523/0x698 [iwlwifi]
kernel: [<ffffffffa0133027>] iwl_pcie_irq_handler+0x46f/0x65f [iwlwifi]
kernel: [<ffffffff810893d0>] ? irq_finalize_oneshot+0xd4/0xd4
kernel: [<ffffffff810893ed>] irq_thread_fn+0x1d/0x34
kernel: [<ffffffff81089661>] irq_thread+0xe6/0x1bb
kernel: [<ffffffff810894e6>] ? wake_threads_waitq+0x2c/0x2c
kernel: [<ffffffff8108957b>] ? irq_thread_dtor+0x95/0x95
kernel: [<ffffffff8105d762>] kthread+0xfc/0x104
kernel: [<ffffffff8107d36c>] ? put_lock_stats.isra.9+0xe/0x20
kernel: [<ffffffff8105d666>] ? kthread_create_on_node+0x3f/0x3f
kernel: [<ffffffff814b2852>] ret_from_fork+0x22/0x30
kernel: Code: 01 ca 49 89 d1 48 89 d1 48 c1 ea 23 48 8b 14 d5 80 23 63 82 49 c1 e9 0c 48 c1 e9 1b 48 85 d2 74 0a 0f b6 c9 48 c1 e1 04 48 01 ca <48> 8b 12 49 c1 e1 06 b9 00 00 00 80 89 7d 80 89 75 84 48 8b 3d
kernel: RIP [<ffffffff8146d2f4>] ieee80211_aes_ccm_decrypt+0x107/0x27f
ffffffff8146d1ed <ieee80211_aes_ccm_decrypt>:
ffffffff8146d1ed: e8 9e 67 04 00 callq ffffffff814b3990 <__fentry__>
ffffffff8146d1f2: 55 push %rbp
ffffffff8146d1f3: 48 89 e5 mov %rsp,%rbp
ffffffff8146d1f6: 41 57 push %r15
ffffffff8146d1f8: 41 56 push %r14
ffffffff8146d1fa: 49 89 ce mov %rcx,%r14
ffffffff8146d1fd: 41 55 push %r13
ffffffff8146d1ff: 41 54 push %r12
ffffffff8146d201: 53 push %rbx
ffffffff8146d202: 48 83 c4 80 add $0xffffffffffffff80,%rsp
ffffffff8146d206: 8b 47 04 mov 0x4(%rdi),%eax
ffffffff8146d209: 48 8d 48 50 lea 0x50(%rax),%rcx
ffffffff8146d20d: 48 83 c0 5e add $0x5e,%rax
ffffffff8146d211: 48 c1 e8 03 shr $0x3,%rax
ffffffff8146d215: 48 c1 e0 03 shl $0x3,%rax
ffffffff8146d219: 48 29 c4 sub %rax,%rsp
ffffffff8146d21c: 4c 8d 7c 24 07 lea 0x7(%rsp),%r15
ffffffff8146d221: 49 c1 ef 03 shr $0x3,%r15
ffffffff8146d225: 4d 85 c0 test %r8,%r8
ffffffff8146d228: 4a 8d 04 fd 00 00 00 lea 0x0(,%r15,8),%rax
ffffffff8146d22f: 00
ffffffff8146d230: 48 89 85 70 ff ff ff mov %rax,-0x90(%rbp)
ffffffff8146d237: 75 0a jne ffffffff8146d243 <ieee80211_aes_ccm_decrypt+0x56>
ffffffff8146d239: b8 ea ff ff ff mov $0xffffffea,%eax
ffffffff8146d23e: e9 1a 02 00 00 jmpq ffffffff8146d45d <ieee80211_aes_ccm_decrypt+0x270>
ffffffff8146d243: 31 c0 xor %eax,%eax
ffffffff8146d245: 49 89 fc mov %rdi,%r12
ffffffff8146d248: 49 89 f5 mov %rsi,%r13
ffffffff8146d24b: 4c 89 85 58 ff ff ff mov %r8,-0xa8(%rbp)
ffffffff8146d252: 4a 8d 3c fd 00 00 00 lea 0x0(,%r15,8),%rdi
ffffffff8146d259: 00
ffffffff8146d25a: be 03 00 00 00 mov $0x3,%esi
ffffffff8146d25f: 4c 89 cb mov %r9,%rbx
ffffffff8146d262: 48 89 95 60 ff ff ff mov %rdx,-0xa0(%rbp)
ffffffff8146d269: f3 aa rep stos %al,%es:(%rdi)
ffffffff8146d26b: 48 8d 85 78 ff ff ff lea -0x88(%rbp),%rax
ffffffff8146d272: 48 89 c7 mov %rax,%rdi
ffffffff8146d275: 48 89 85 68 ff ff ff mov %rax,-0x98(%rbp)
ffffffff8146d27c: e8 46 06 dc ff callq ffffffff8122d8c7 <sg_init_table>
ffffffff8146d281: 48 8b 95 60 ff ff ff mov -0xa0(%rbp),%rdx
ffffffff8146d288: 41 b9 00 00 00 80 mov $0x80000000,%r9d
ffffffff8146d28e: 48 8b 0d 7b cd 39 00 mov 0x39cd7b(%rip),%rcx # ffffffff8180a010 <phys_base>
ffffffff8146d295: 48 8b 85 68 ff ff ff mov -0x98(%rbp),%rax
ffffffff8146d29c: 4c 8b 85 58 ff ff ff mov -0xa8(%rbp),%r8
ffffffff8146d2a3: 0f b7 32 movzwl (%rdx),%esi
ffffffff8146d2a6: 48 83 c2 02 add $0x2,%rdx
ffffffff8146d2aa: 89 d7 mov %edx,%edi
ffffffff8146d2ac: 81 e7 ff 0f 00 00 and $0xfff,%edi
ffffffff8146d2b2: 66 c1 c6 08 rol $0x8,%si
ffffffff8146d2b6: 4c 01 ca add %r9,%rdx
ffffffff8146d2b9: 0f b7 f6 movzwl %si,%esi
ffffffff8146d2bc: 72 0a jb ffffffff8146d2c8 <ieee80211_aes_ccm_decrypt+0xdb>
ffffffff8146d2be: 48 b9 00 00 00 80 ff movabs $0x77ff80000000,%rcx
ffffffff8146d2c5: 77 00 00
ffffffff8146d2c8: 48 01 ca add %rcx,%rdx
ffffffff8146d2cb: 49 89 d1 mov %rdx,%r9
ffffffff8146d2ce: 48 89 d1 mov %rdx,%rcx
ffffffff8146d2d1: 48 c1 ea 23 shr $0x23,%rdx
ffffffff8146d2d5: 48 8b 14 d5 80 23 63 mov -0x7d9cdc80(,%rdx,8),%rdx
ffffffff8146d2dc: 82
ffffffff8146d2dd: 49 c1 e9 0c shr $0xc,%r9
ffffffff8146d2e1: 48 c1 e9 1b shr $0x1b,%rcx
ffffffff8146d2e5: 48 85 d2 test %rdx,%rdx
ffffffff8146d2e8: 74 0a je ffffffff8146d2f4 <ieee80211_aes_ccm_decrypt+0x107>
ffffffff8146d2ea: 0f b6 c9 movzbl %cl,%ecx
ffffffff8146d2ed: 48 c1 e1 04 shl $0x4,%rcx
ffffffff8146d2f1: 48 01 ca add %rcx,%rdx
ffffffff8146d2f4: 48 8b 12 mov (%rdx),%rdx
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ffffffff8146d2f7: 49 c1 e1 06 shl $0x6,%r9
ffffffff8146d2fb: b9 00 00 00 80 mov $0x80000000,%ecx
ffffffff8146d300: 89 7d 80 mov %edi,-0x80(%rbp)
ffffffff8146d303: 89 75 84 mov %esi,-0x7c(%rbp)
ffffffff8146d306: 48 8b 3d 03 cd 39 00 mov 0x39cd03(%rip),%rdi # ffffffff8180a010 <phys_base>
ffffffff8146d30d: 48 83 e2 fc and $0xfffffffffffffffc,%rdx
ffffffff8146d311: 49 01 d1 add %rdx,%r9
ffffffff8146d314: 48 8b 95 78 ff ff ff mov -0x88(%rbp),%rdx
-ss
On Oct 13, 2016 6:46 AM, "Johannes Berg" [off-list ref] wrote:
On Thu, 2016-10-13 at 22:42 +0900, Sergey Senozhatsky wrote:
quoted
quoted
quoted
https://git.kernel.org/cgit/linux/kernel/git/luto/linux.git/commi
t/?h=x86/vmap_stack&id=0a39cfa6fbb5d5635c85253cc7d6b44b54822afd
https://git.kernel.org/cgit/linux/kernel/git/luto/linux.git/commi
t/?h=x86/vmap_stack&id=bf8cfa200b5a01383ea39fc8ce2f32909767baa8
That truly sounds like something we'd rather avoid in the TX/RX
paths though, which should perform well.
didn't fix.
It couldn't, since the new helpers weren't used in mac80211 in those
patches yet.
quoted
so I finally had some time to do a better bug-reporter job.
I added a bunch of printk-s and several virt_addr_valid()-s
to ieee80211_aes_ccm_encrypt().
and right befoe the Oops I see the following report from
virt_addr_valid()
FAIL: 00004100002cba02 > ffffc900802cba02 || 1 -> (00004100002cba02
quoted
quoted
39) == 130
Yeah, we already know that in this function the aad variable is on the
stack, it explicitly is.
The question, though, is why precisely that fails in the crypto code.
Can you send the Oops report itself?
It's failing before that. With CONFIG_VMAP_STACK=y, the stack may not
be physically contiguous and can't be used for DMA, so putting it in a
scatterlist is bogus in general, and the crypto code mostly wants a
scatterlist.
There are a couple (faster!) APIs for crypto that don't use
scatterlists, but I don't think AEAD works with them.
--Andy
On (10/13/16 14:49), Andy Lutomirski wrote:
[..]
quoted
quoted
FAIL: 00004100002cba02 > ffffc900802cba02 || 1 -> (00004100002cba02
quoted
quoted
39) == 130
Yeah, we already know that in this function the aad variable is on the
stack, it explicitly is.
The question, though, is why precisely that fails in the crypto code.
Can you send the Oops report itself?
It's failing before that. With CONFIG_VMAP_STACK=y, the stack may not
be physically contiguous and can't be used for DMA, so putting it in a
scatterlist is bogus in general, and the crypto code mostly wants a
scatterlist.
There are a couple (faster!) APIs for crypto that don't use
scatterlists, but I don't think AEAD works with them.
given that we have a known issue shouldn't VMAP_STACK be
disabled for now, or would you rather prefer to mark MAC80211
as incompatible: "depends on CFG80211 && !VMAP_STACK"?
-ss