xref: /linux/Documentation/arch/x86/buslock.rst (revision 79d2e1919a2728ef49d938eb20ebd5903c14dfb0)
1.. SPDX-License-Identifier: GPL-2.0
2
3.. include:: <isonum.txt>
4
5===============================
6Bus lock detection and handling
7===============================
8
9:Copyright: |copy| 2021 Intel Corporation
10:Authors: - Fenghua Yu <fenghua.yu@intel.com>
11          - Tony Luck <tony.luck@intel.com>
12
13Problem
14=======
15
16A split lock is any atomic operation whose operand crosses two cache lines.
17Since the operand spans two cache lines and the operation must be atomic,
18the system locks the bus while the CPU accesses the two cache lines.
19
20A bus lock is acquired through either split locked access to writeback (WB)
21memory or any locked access to non-WB memory. This is typically thousands of
22cycles slower than an atomic operation within a cache line. It also disrupts
23performance on other cores and brings the whole system to its knees.
24
25Detection
26=========
27
28Intel processors may support either or both of the following hardware
29mechanisms to detect split locks and bus locks. Some AMD processors also
30support bus lock detect.
31
32#AC exception for split lock detection
33--------------------------------------
34
35Beginning with the Tremont Atom CPU split lock operations may raise an
36Alignment Check (#AC) exception when a split lock operation is attempted.
37
38#DB exception for bus lock detection
39------------------------------------
40
41Some CPUs have the ability to notify the kernel by an #DB trap after a user
42instruction acquires a bus lock and is executed. This allows the kernel to
43terminate the application or to enforce throttling.
44
45Software handling
46=================
47
48The kernel #AC and #DB handlers handle bus lock based on the kernel
49parameter "split_lock_detect". Here is a summary of different options:
50
51+------------------+----------------------------+-----------------------+
52|split_lock_detect=|#AC for split lock		|#DB for bus lock	|
53+------------------+----------------------------+-----------------------+
54|off	  	   |Do nothing			|Do nothing		|
55+------------------+----------------------------+-----------------------+
56|warn		   |Kernel OOPs			|Warn once per task and |
57|(default)	   |Warn once per task, add a	|and continues to run.  |
58|		   |delay, add synchronization	|			|
59|		   |to prevent more than one	|			|
60|		   |core from executing a	|			|
61|		   |split lock in parallel.	|			|
62|		   |sysctl split_lock_mitigate	|			|
63|		   |can be used to avoid the	|			|
64|		   |delay and synchronization	|			|
65|		   |When both features are	|			|
66|		   |supported, warn in #AC	|			|
67+------------------+----------------------------+-----------------------+
68|fatal		   |Kernel OOPs			|Send SIGBUS to user.	|
69|		   |Send SIGBUS to user		|			|
70|		   |When both features are	|			|
71|		   |supported, fatal in #AC	|			|
72+------------------+----------------------------+-----------------------+
73|ratelimit:N	   |Do nothing			|Limit bus lock rate to	|
74|(0 < N <= 1000)   |				|N bus locks per second	|
75|		   |				|system wide and warn on|
76|		   |				|bus locks.		|
77+------------------+----------------------------+-----------------------+
78
79Usages
80======
81
82Detecting and handling bus lock may find usages in various areas:
83
84It is critical for real time system designers who build consolidated real
85time systems. These systems run hard real time code on some cores and run
86"untrusted" user processes on other cores. The hard real time cannot afford
87to have any bus lock from the untrusted processes to hurt real time
88performance. To date the designers have been unable to deploy these
89solutions as they have no way to prevent the "untrusted" user code from
90generating split lock and bus lock to block the hard real time code to
91access memory during bus locking.
92
93It's also useful for general computing to prevent guests or user
94applications from slowing down the overall system by executing instructions
95with bus lock.
96
97
98Guidance
99========
100off
101---
102
103Disable checking for split lock and bus lock. This option can be useful if
104there are legacy applications that trigger these events at a low rate so
105that mitigation is not needed.
106
107warn
108----
109
110A warning is emitted when a bus lock is detected which allows to identify
111the offending application. This is the default behavior.
112
113fatal
114-----
115
116In this case, the bus lock is not tolerated and the process is killed.
117
118ratelimit
119---------
120
121A system wide bus lock rate limit N is specified where 0 < N <= 1000. This
122allows a bus lock rate up to N bus locks per second. When the bus lock rate
123is exceeded then any task which is caught via the buslock #DB exception is
124throttled by enforced sleeps until the rate goes under the limit again.
125
126This is an effective mitigation in cases where a minimal impact can be
127tolerated, but an eventual Denial of Service attack has to be prevented. It
128allows to identify the offending processes and analyze whether they are
129malicious or just badly written.
130
131Selecting a rate limit of 1000 allows the bus to be locked for up to about
132seven million cycles each second (assuming 7000 cycles for each bus
133lock). On a 2 GHz processor that would be about 0.35% system slowdown.
134