]>
Commit | Line | Data |
---|---|---|
63c89da2 LE |
1 | ;------------------------------------------------------------------------------\r |
2 | ; @file\r | |
3 | ; Pen any hot-added CPU in a 16-bit, real mode HLT loop, after it leaves SMM by\r | |
4 | ; executing the RSM instruction.\r | |
5 | ;\r | |
6 | ; Copyright (c) 2020, Red Hat, Inc.\r | |
7 | ;\r | |
8 | ; SPDX-License-Identifier: BSD-2-Clause-Patent\r | |
9 | ;\r | |
10 | ; The routine implemented here is stored into normal RAM, under 1MB, at the\r | |
11 | ; beginning of a page that is allocated as EfiReservedMemoryType. On any\r | |
12 | ; hot-added CPU, it is executed after *at least* the first RSM (i.e., after\r | |
13 | ; SMBASE relocation).\r | |
14 | ;\r | |
15 | ; The first execution of this code occurs as follows:\r | |
16 | ;\r | |
17 | ; - The hot-added CPU is in RESET state.\r | |
18 | ;\r | |
19 | ; - The ACPI CPU hotplug event handler triggers a broadcast SMI, from the OS.\r | |
20 | ;\r | |
21 | ; - Existent CPUs (BSP and APs) enter SMM.\r | |
22 | ;\r | |
23 | ; - The hot-added CPU remains in RESET state, but an SMI is pending for it now.\r | |
24 | ; (See "SYSTEM MANAGEMENT INTERRUPT (SMI)" in the Intel SDM.)\r | |
25 | ;\r | |
26 | ; - In SMM, pre-existent CPUs that are not elected SMM Monarch, keep themselves\r | |
27 | ; busy with their wait loops.\r | |
28 | ;\r | |
29 | ; - From the root MMI handler, the SMM Monarch:\r | |
30 | ;\r | |
31 | ; - places this routine in the reserved page,\r | |
32 | ;\r | |
33 | ; - clears the "about to leave SMM" byte in SMRAM,\r | |
34 | ;\r | |
35 | ; - clears the last byte of the reserved page,\r | |
36 | ;\r | |
37 | ; - sends an INIT-SIPI-SIPI sequence to the hot-added CPU,\r | |
38 | ;\r | |
39 | ; - un-gates the default SMI handler by APIC ID.\r | |
40 | ;\r | |
41 | ; - The startup vector in the SIPI that is sent by the SMM Monarch points to\r | |
42 | ; this code; i.e., to the reserved page. (Example: 0x9_F000.)\r | |
43 | ;\r | |
44 | ; - The SMM Monarch starts polling the "about to leave SMM" byte in SMRAM.\r | |
45 | ;\r | |
46 | ; - The hot-added CPU boots, and immediately enters SMM due to the pending SMI.\r | |
47 | ; It starts executing the default SMI handler.\r | |
48 | ;\r | |
49 | ; - Importantly, the SMRAM Save State Map captures the following information,\r | |
50 | ; when the hot-added CPU enters SMM:\r | |
51 | ;\r | |
52 | ; - CS selector: assumes the 16 most significant bits of the 20-bit (i.e.,\r | |
53 | ; below 1MB) startup vector from the SIPI. (Example: 0x9F00.)\r | |
54 | ;\r | |
55 | ; - CS attributes: Accessed, Readable, User (S=1), CodeSegment (bit#11),\r | |
56 | ; Present.\r | |
57 | ;\r | |
58 | ; - CS limit: 0xFFFF.\r | |
59 | ;\r | |
60 | ; - CS base: the CS selector value shifted left by 4 bits. That is, the CS\r | |
61 | ; base equals the SIPI startup vector. (Example: 0x9_F000.)\r | |
62 | ;\r | |
63 | ; - IP: the least significant 4 bits from the SIPI startup vector. Because\r | |
64 | ; the routine is page-aligned, these bits are zero (hence IP is zero).\r | |
65 | ;\r | |
66 | ; - ES, SS, DS, FS, GS selectors: 0.\r | |
67 | ;\r | |
68 | ; - ES, SS, DS, FS, GS attributes: same as the CS attributes, minus\r | |
69 | ; CodeSegment (bit#11).\r | |
70 | ;\r | |
71 | ; - ES, SS, DS, FS, GS limits: 0xFFFF.\r | |
72 | ;\r | |
73 | ; - ES, SS, DS, FS, GS bases: 0.\r | |
74 | ;\r | |
75 | ; - The hot-added CPU sets its new SMBASE value in the SMRAM Save State Map.\r | |
76 | ;\r | |
77 | ; - The hot-added CPU sets the "about to leave SMM" byte in SMRAM, then\r | |
78 | ; executes the RSM instruction immediately after, leaving SMM.\r | |
79 | ;\r | |
80 | ; - The SMM Monarch notices that the "about to leave SMM" byte in SMRAM has\r | |
81 | ; been set, and starts polling the last byte in the reserved page.\r | |
82 | ;\r | |
83 | ; - The hot-added CPU jumps ("returns") to the code below (in the reserved\r | |
84 | ; page), according to the register state listed in the SMRAM Save State Map.\r | |
85 | ;\r | |
86 | ; - The hot-added CPU sets the last byte of the reserved page, then halts\r | |
87 | ; itself.\r | |
88 | ;\r | |
89 | ; - The SMM Monarch notices that the hot-added CPU is done with SMBASE\r | |
90 | ; relocation.\r | |
91 | ;\r | |
92 | ; Note that, if the OS is malicious and sends INIT-SIPI-SIPI to the hot-added\r | |
93 | ; CPU before allowing the ACPI CPU hotplug event handler to trigger a broadcast\r | |
94 | ; SMI, then said broadcast SMI will yank the hot-added CPU directly into SMM,\r | |
95 | ; without becoming pending for it (as the hot-added CPU is no longer in RESET\r | |
96 | ; state). This is OK, because:\r | |
97 | ;\r | |
98 | ; - The default SMI handler copes with this, as it is gated by APIC ID. The\r | |
99 | ; hot-added CPU won't start the actual SMBASE relocation until the SMM\r | |
100 | ; Monarch lets it.\r | |
101 | ;\r | |
102 | ; - The INIT-SIPI-SIPI sequence that the SMM Monarch sends to the hot-added CPU\r | |
103 | ; will be ignored in this sate (it won't even be latched). See "SMI HANDLER\r | |
104 | ; EXECUTION ENVIRONMENT" in the Intel SDM: "INIT operations are inhibited\r | |
105 | ; when the processor enters SMM".\r | |
106 | ;\r | |
107 | ; - When the hot-added CPU (e.g., CPU#1) executes the RSM (having relocated\r | |
108 | ; SMBASE), it returns to the OS. The OS can use CPU#1 to attack the last byte\r | |
109 | ; of the reserved page, while another CPU (e.g., CPU#2) is relocating SMBASE,\r | |
110 | ; in order to trick the SMM Monarch (e.g., CPU#0) to open the APIC ID gate\r | |
111 | ; for yet another CPU (e.g., CPU#3). However, the SMM Monarch won't look at\r | |
112 | ; the last byte of the reserved page, until CPU#2 sets the "about to leave\r | |
113 | ; SMM" byte in SMRAM. This leaves a very small window (just one instruction's\r | |
114 | ; worth before the RSM) for CPU#3 to "catch up" with CPU#2, and overwrite\r | |
115 | ; CPU#2's SMBASE with its own.\r | |
116 | ;\r | |
117 | ; In other words, we do not / need not prevent a malicious OS from booting the\r | |
118 | ; hot-added CPU early; instead we provide benign OSes with a pen for hot-added\r | |
119 | ; CPUs.\r | |
120 | ;------------------------------------------------------------------------------\r | |
121 | \r | |
122 | SECTION .data\r | |
123 | BITS 16\r | |
124 | \r | |
125 | GLOBAL ASM_PFX (mPostSmmPen) ; UINT8[]\r | |
126 | GLOBAL ASM_PFX (mPostSmmPenSize) ; UINT16\r | |
127 | \r | |
128 | ASM_PFX (mPostSmmPen):\r | |
129 | ;\r | |
130 | ; Point DS at the same reserved page.\r | |
131 | ;\r | |
132 | mov ax, cs\r | |
133 | mov ds, ax\r | |
134 | \r | |
135 | ;\r | |
136 | ; Inform the SMM Monarch that we're done with SMBASE relocation, by setting\r | |
137 | ; the last byte in the reserved page.\r | |
138 | ;\r | |
139 | mov byte [ds : word 0xFFF], 1\r | |
140 | \r | |
141 | ;\r | |
142 | ; Halt now, until we get woken by another SMI, or (more likely) the OS\r | |
143 | ; reboots us with another INIT-SIPI-SIPI.\r | |
144 | ;\r | |
145 | HltLoop:\r | |
146 | cli\r | |
147 | hlt\r | |
148 | jmp HltLoop\r | |
149 | \r | |
150 | ASM_PFX (mPostSmmPenSize):\r | |
151 | dw $ - ASM_PFX (mPostSmmPen)\r |