xref: /linux/Documentation/ABI/testing/sysfs-firmware-acpi (revision be239684b18e1cdcafcf8c7face4a2f562c745ad)
1What:		/sys/firmware/acpi/fpdt/
2Date:		Jan 2021
3Contact:	Zhang Rui <rui.zhang@intel.com>
4Description:
5		ACPI Firmware Performance Data Table (FPDT) provides
6		information for firmware performance data for system boot,
7		S3 suspend and S3 resume. This sysfs entry contains the
8		performance data retrieved from the FPDT.
9
10		boot:
11			firmware_start_ns: Timer value logged at the beginning
12				of firmware image execution. In nanoseconds.
13			bootloader_load_ns: Timer value logged just prior to
14				loading the OS boot loader into memory.
15				In nanoseconds.
16			bootloader_launch_ns: Timer value logged just prior to
17				launching the currently loaded OS boot loader
18				image. In nanoseconds.
19			exitbootservice_start_ns: Timer value logged at the
20				point when the OS loader calls the
21				ExitBootServices function for UEFI compatible
22				firmware. In nanoseconds.
23			exitbootservice_end_ns: Timer value logged at the point
24				just prior to the OS loader gaining control
25				back from the ExitBootServices function for
26				UEFI compatible firmware. In nanoseconds.
27		suspend:
28			suspend_start_ns: Timer value recorded at the previous
29				OS write to SLP_TYP upon entry to S3. In
30				nanoseconds.
31			suspend_end_ns: Timer value recorded at the previous
32				firmware write to SLP_TYP used to trigger
33				hardware entry to S3. In nanoseconds.
34		resume:
35			resume_count: A count of the number of S3 resume cycles
36				since the last full boot sequence.
37			resume_avg_ns: Average timer value of all resume cycles
38				logged since the last full boot sequence,
39				including the most recent resume. In nanoseconds.
40			resume_prev_ns: Timer recorded at the end of the previous
41				platform runtime firmware S3 resume, just prior to
42				handoff to the OS waking vector. In nanoseconds.
43
44What:		/sys/firmware/acpi/bgrt/
45Date:		January 2012
46Contact:	Matthew Garrett <mjg@redhat.com>
47Description:
48		The BGRT is an ACPI 5.0 feature that allows the OS
49		to obtain a copy of the firmware boot splash and
50		some associated metadata. This is intended to be used
51		by boot splash applications in order to interact with
52		the firmware boot splash in order to avoid jarring
53		transitions.
54
55		image: The image bitmap. Currently a 32-bit BMP.
56		status: 1 if the image is valid, 0 if firmware invalidated it.
57		type: 0 indicates image is in BMP format.
58
59		======== ===================================================
60		version: The version of the BGRT. Currently 1.
61		xoffset: The number of pixels between the left of the screen
62			 and the left edge of the image.
63		yoffset: The number of pixels between the top of the screen
64			 and the top edge of the image.
65		======== ===================================================
66
67What:		/sys/firmware/acpi/hotplug/
68Date:		February 2013
69Contact:	Rafael J. Wysocki <rafael.j.wysocki@intel.com>
70Description:
71		There are separate hotplug profiles for different classes of
72		devices supported by ACPI, such as containers, memory modules,
73		processors, PCI root bridges etc.  A hotplug profile for a given
74		class of devices is a collection of settings defining the way
75		that class of devices will be handled by the ACPI core hotplug
76		code.  Those profiles are represented in sysfs as subdirectories
77		of /sys/firmware/acpi/hotplug/.
78
79		The following setting is available to user space for each
80		hotplug profile:
81
82		======== =======================================================
83		enabled: If set, the ACPI core will handle notifications of
84			 hotplug events associated with the given class of
85			 devices and will allow those devices to be ejected with
86			 the help of the _EJ0 control method.  Unsetting it
87			 effectively disables hotplug for the corresponding
88			 class of devices.
89		======== =======================================================
90
91		The value of the above attribute is an integer number: 1 (set)
92		or 0 (unset).  Attempts to write any other values to it will
93		cause -EINVAL to be returned.
94
95What:		/sys/firmware/acpi/interrupts/
96Date:		February 2008
97Contact:	Len Brown <lenb@kernel.org>
98Description:
99		All ACPI interrupts are handled via a single IRQ,
100		the System Control Interrupt (SCI), which appears
101		as "acpi" in /proc/interrupts.
102
103		However, one of the main functions of ACPI is to make
104		the platform understand random hardware without
105		special driver support.  So while the SCI handles a few
106		well known (fixed feature) interrupts sources, such
107		as the power button, it can also handle a variable
108		number of a "General Purpose Events" (GPE).
109
110		A GPE vectors to a specified handler in AML, which
111		can do a anything the BIOS writer wants from
112		OS context.  GPE 0x12, for example, would vector
113		to a level or edge handler called _L12 or _E12.
114		The handler may do its business and return.
115		Or the handler may send send a Notify event
116		to a Linux device driver registered on an ACPI device,
117		such as a battery, or a processor.
118
119		To figure out where all the SCI's are coming from,
120		/sys/firmware/acpi/interrupts contains a file listing
121		every possible source, and the count of how many
122		times it has triggered::
123
124		  $ cd /sys/firmware/acpi/interrupts
125		  $ grep . *
126		  error:	     0
127		  ff_gbl_lock:	     0   enable
128		  ff_pmtimer:	     0  invalid
129		  ff_pwr_btn:	     0   enable
130		  ff_rt_clk:	     2  disable
131		  ff_slp_btn:	     0  invalid
132		  gpe00:	     0	invalid
133		  gpe01:	     0	 enable
134		  gpe02:	   108	 enable
135		  gpe03:	     0	invalid
136		  gpe04:	     0	invalid
137		  gpe05:	     0	invalid
138		  gpe06:	     0	 enable
139		  gpe07:	     0	 enable
140		  gpe08:	     0	invalid
141		  gpe09:	     0	invalid
142		  gpe0A:	     0	invalid
143		  gpe0B:	     0	invalid
144		  gpe0C:	     0	invalid
145		  gpe0D:	     0	invalid
146		  gpe0E:	     0	invalid
147		  gpe0F:	     0	invalid
148		  gpe10:	     0	invalid
149		  gpe11:	     0	invalid
150		  gpe12:	     0	invalid
151		  gpe13:	     0	invalid
152		  gpe14:	     0	invalid
153		  gpe15:	     0	invalid
154		  gpe16:	     0	invalid
155		  gpe17:	  1084	 enable
156		  gpe18:	     0	 enable
157		  gpe19:	     0	invalid
158		  gpe1A:	     0	invalid
159		  gpe1B:	     0	invalid
160		  gpe1C:	     0	invalid
161		  gpe1D:	     0	invalid
162		  gpe1E:	     0	invalid
163		  gpe1F:	     0	invalid
164		  gpe_all:	  1192
165		  sci:		  1194
166		  sci_not:	     0
167
168		===========  ==================================================
169		sci	     The number of times the ACPI SCI
170			     has been called and claimed an interrupt.
171
172		sci_not	     The number of times the ACPI SCI
173			     has been called and NOT claimed an interrupt.
174
175		gpe_all	     count of SCI caused by GPEs.
176
177		gpeXX	     count for individual GPE source
178
179		ff_gbl_lock  Global Lock
180
181		ff_pmtimer   PM Timer
182
183		ff_pwr_btn   Power Button
184
185		ff_rt_clk    Real Time Clock
186
187		ff_slp_btn   Sleep Button
188
189		error	     an interrupt that can't be accounted for above.
190
191		invalid      it's either a GPE or a Fixed Event that
192			     doesn't have an event handler.
193
194		disable	     the GPE/Fixed Event is valid but disabled.
195
196		enable       the GPE/Fixed Event is valid and enabled.
197		===========  ==================================================
198
199		Root has permission to clear any of these counters.  Eg.::
200
201		  # echo 0 > gpe11
202
203		All counters can be cleared by clearing the total "sci"::
204
205		  # echo 0 > sci
206
207		None of these counters has an effect on the function
208		of the system, they are simply statistics.
209
210		Besides this, user can also write specific strings to these files
211		to enable/disable/clear ACPI interrupts in user space, which can be
212		used to debug some ACPI interrupt storm issues.
213
214		Note that only writing to VALID GPE/Fixed Event is allowed,
215		i.e. user can only change the status of runtime GPE and
216		Fixed Event with event handler installed.
217
218		Let's take power button fixed event for example, please kill acpid
219		and other user space applications so that the machine won't shutdown
220		when pressing the power button::
221
222		  # cat ff_pwr_btn
223		  0	enabled
224		  # press the power button for 3 times;
225		  # cat ff_pwr_btn
226		  3	enabled
227		  # echo disable > ff_pwr_btn
228		  # cat ff_pwr_btn
229		  3	disabled
230		  # press the power button for 3 times;
231		  # cat ff_pwr_btn
232		  3	disabled
233		  # echo enable > ff_pwr_btn
234		  # cat ff_pwr_btn
235		  4	enabled
236		  /*
237		   * this is because the status bit is set even if the enable
238		   * bit is cleared, and it triggers an ACPI fixed event when
239		   * the enable bit is set again
240		   */
241		  # press the power button for 3 times;
242		  # cat ff_pwr_btn
243		  7	enabled
244		  # echo disable > ff_pwr_btn
245		  # press the power button for 3 times;
246		  # echo clear > ff_pwr_btn	/* clear the status bit */
247		  # echo disable > ff_pwr_btn
248		  # cat ff_pwr_btn
249		  7	enabled
250
251