Re: [PATCH v4] x86/sgx: Resolves SECS reclaim vs. page fault for EAUG race

[Reinette Chatre]

Hi Haitao,

On 7/26/2023 1:50 PM, Haitao Huang wrote:
> Under heavy load, the SGX EPC reclaimer (ksgxd) may reclaim the SECS EPC
> page for an enclave and set encl->secs.epc_page to NULL. But the SECS
> EPC page is used for EAUG in the SGX page fault handler without checking
> for NULL and reloading.
> 
> Fix this by checking if SECS is loaded before EAUG and loading it if it
> was reclaimed.
> 
> The SECS page holds global enclave metadata. It can only be reclaimed
> when there are no other enclave pages remaining. At that point,
> virtually nothing can be done with the enclave until the SECS page is
> paged back in.
> 
> An enclave can not run nor generate page faults without a resident SECS
> page. But it is still possible for a #PF for a non-SECS page to race
> with paging out the SECS page: when the last resident non-SECS page A
> triggers a #PF in a non-resident page B, and then page A and the SECS
> both are paged out before the #PF on B is handled.
> 
> Hitting this bug requires that race triggered with a #PF for EAUG.
> Following is a trace when it happens.
> 

Thank you very much for finding this issue as well as providing
a fix.

> [ 1682.914263] BUG: kernel NULL pointer dereference, address: 0000000000000000
> [ 1682.922966] #PF: supervisor read access in kernel mode
> [ 1682.929115] #PF: error_code(0x0000) - not-present page
> [ 1682.935264] PGD 0 P4D 0
> [ 1682.938383] Oops: 0000 [#1] PREEMPT SMP NOPTI
> [ 1682.943620] CPU: 43 PID: 2681 Comm: test_sgx Not tainted 6.3.0-rc4sgxcet #12
> [ 1682.951989] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.16.2-0-gea1b7a073390-prebuilt.qemu.org 04/01/2014
> [ 1682.965504] RIP: 0010:sgx_encl_eaug_page+0xc7/0x210
> [ 1682.971359] Code: 25 49 8b 96 98 04 00 00 48 8d 40 48 48 89 42 08 48 89 56 48 49 8d 96 98 04 00 00 48 89 56 50 49 89 86 98 04 00 00 49 8b 46 60 <8b> 10 48 c1 e2 05 488
> [ 1682.993330] RSP: 0000:ffffb2e64725bc00 EFLAGS: 00010246
> [ 1682.999585] RAX: 0000000000000000 RBX: ffff987e5abac428 RCX: 0000000000000000
> [ 1683.008059] RDX: 0000000000000001 RSI: 0000000000000000 RDI: ffff987e61aee000
> [ 1683.016533] RBP: ffffb2e64725bcf0 R08: 0000000000000000 R09: ffffb2e64725bb58
> [ 1683.025008] R10: 0000000000000000 R11: 00007f3f5c418fff R12: ffff987e61aee020
> [ 1683.033479] R13: ffff987e505bc080 R14: ffff987e61aee000 R15: ffffb2e6420fcb20
> [ 1683.041949] FS:  00007f3f5cb48740(0000) GS:ffff989cfe8c0000(0000) knlGS:0000000000000000
> [ 1683.051540] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> [ 1683.058478] CR2: 0000000000000000 CR3: 0000000115896002 CR4: 0000000000770ee0
> [ 1683.067018] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
> [ 1683.075539] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
> [ 1683.084085] PKRU: 55555554
> [ 1683.087465] Call Trace:
> [ 1683.090547]  <TASK>
> [ 1683.093220]  ? __kmem_cache_alloc_node+0x16a/0x440
> [ 1683.099034]  ? xa_load+0x6e/0xa0
> [ 1683.103038]  sgx_vma_fault+0x119/0x230
> [ 1683.107630]  __do_fault+0x36/0x140
> [ 1683.111828]  do_fault+0x12f/0x400
> [ 1683.115928]  __handle_mm_fault+0x728/0x1110
> [ 1683.121050]  handle_mm_fault+0x105/0x310
> [ 1683.125850]  do_user_addr_fault+0x1ee/0x750
> [ 1683.130957]  ? __this_cpu_preempt_check+0x13/0x20
> [ 1683.136667]  exc_page_fault+0x76/0x180
> [ 1683.141265]  asm_exc_page_fault+0x27/0x30
> [ 1683.146160] RIP: 0033:0x7ffc6496beea
> [ 1683.150563] Code: 43 48 8b 4d 10 48 c7 c3 28 00 00 00 48 83 3c 19 00 75 31 48 83 c3 08 48 81 fb 00 01 00 00 75 ec 48 8b 19 48 8d 0d 00 00 00 00 <0f> 01 d7 48 8b 5d 101
> [ 1683.172773] RSP: 002b:00007ffc64935b68 EFLAGS: 00000202
> [ 1683.179138] RAX: 0000000000000003 RBX: 00007f3800000000 RCX: 00007ffc6496beea
> [ 1683.187675] RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000000
> [ 1683.196200] RBP: 00007ffc64935b70 R08: 0000000000000000 R09: 0000000000000000
> [ 1683.204724] R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000
> [ 1683.213310] R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000000
> [ 1683.221850]  </TASK>
> [ 1683.224636] Modules linked in: isofs intel_rapl_msr intel_rapl_common binfmt_misc kvm_intel nls_iso8859_1 kvm ppdev irqbypass input_leds parport_pc joydev parport rapi
> [ 1683.291173] CR2: 0000000000000000
> [ 1683.295271] ---[ end trace 0000000000000000 ]---

Could you please trim this trace? There is more detail in
Documentation/process/submitting-patches.rst (search for
"Backtraces in commit messages"), but the ideal trace should
have just the information needed to describe the issue (no
timestamps, register dumps, etc.).

With that addressed, feel free to add:

Acked-by: Reinette Chatre <reinette.chatre@xxxxxxxxx>

Reinette