xref: /rk3399_ARM-atf/docs/porting-guide.rst (revision cf6c30e08ba54a22a0ce8abd6eac8b1d1ccad324)
1Trusted Firmware-A Porting Guide
2================================
3
4
5.. section-numbering::
6    :suffix: .
7
8.. contents::
9
10--------------
11
12Introduction
13------------
14
15Porting Trusted Firmware-A (TF-A) to a new platform involves making some
16mandatory and optional modifications for both the cold and warm boot paths.
17Modifications consist of:
18
19-  Implementing a platform-specific function or variable,
20-  Setting up the execution context in a certain way, or
21-  Defining certain constants (for example #defines).
22
23The platform-specific functions and variables are declared in
24`include/plat/common/platform.h`_. The firmware provides a default implementation
25of variables and functions to fulfill the optional requirements. These
26implementations are all weakly defined; they are provided to ease the porting
27effort. Each platform port can override them with its own implementation if the
28default implementation is inadequate.
29
30Some modifications are common to all Boot Loader (BL) stages. Section 2
31discusses these in detail. The subsequent sections discuss the remaining
32modifications for each BL stage in detail.
33
34This document should be read in conjunction with the TF-A `User Guide`_.
35
36Please refer to the `Platform compatibility policy`_ for the policy regarding
37compatibility and deprecation of these porting interfaces.
38
39Only Arm development platforms (such as FVP and Juno) may use the
40functions/definitions in ``include/plat/arm/common/`` and the corresponding
41source files in ``plat/arm/common/``. This is done so that there are no
42dependencies between platforms maintained by different people/companies. If you
43want to use any of the functionality present in ``plat/arm`` files, please
44create a pull request that moves the code to ``plat/common`` so that it can be
45discussed.
46
47Common modifications
48--------------------
49
50This section covers the modifications that should be made by the platform for
51each BL stage to correctly port the firmware stack. They are categorized as
52either mandatory or optional.
53
54Common mandatory modifications
55------------------------------
56
57A platform port must enable the Memory Management Unit (MMU) as well as the
58instruction and data caches for each BL stage. Setting up the translation
59tables is the responsibility of the platform port because memory maps differ
60across platforms. A memory translation library (see ``lib/xlat_tables/``) is
61provided to help in this setup.
62
63Note that although this library supports non-identity mappings, this is intended
64only for re-mapping peripheral physical addresses and allows platforms with high
65I/O addresses to reduce their virtual address space. All other addresses
66corresponding to code and data must currently use an identity mapping.
67
68Also, the only translation granule size supported in TF-A is 4KB, as various
69parts of the code assume that is the case. It is not possible to switch to
7016 KB or 64 KB granule sizes at the moment.
71
72In Arm standard platforms, each BL stage configures the MMU in the
73platform-specific architecture setup function, ``blX_plat_arch_setup()``, and uses
74an identity mapping for all addresses.
75
76If the build option ``USE_COHERENT_MEM`` is enabled, each platform can allocate a
77block of identity mapped secure memory with Device-nGnRE attributes aligned to
78page boundary (4K) for each BL stage. All sections which allocate coherent
79memory are grouped under ``coherent_ram``. For ex: Bakery locks are placed in a
80section identified by name ``bakery_lock`` inside ``coherent_ram`` so that its
81possible for the firmware to place variables in it using the following C code
82directive:
83
84::
85
86    __section("bakery_lock")
87
88Or alternatively the following assembler code directive:
89
90::
91
92    .section bakery_lock
93
94The ``coherent_ram`` section is a sum of all sections like ``bakery_lock`` which are
95used to allocate any data structures that are accessed both when a CPU is
96executing with its MMU and caches enabled, and when it's running with its MMU
97and caches disabled. Examples are given below.
98
99The following variables, functions and constants must be defined by the platform
100for the firmware to work correctly.
101
102File : platform_def.h [mandatory]
103~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
104
105Each platform must ensure that a header file of this name is in the system
106include path with the following constants defined. This will require updating
107the list of ``PLAT_INCLUDES`` in the ``platform.mk`` file.
108
109Platform ports may optionally use the file `include/plat/common/common_def.h`_,
110which provides typical values for some of the constants below. These values are
111likely to be suitable for all platform ports.
112
113-  **#define : PLATFORM_LINKER_FORMAT**
114
115   Defines the linker format used by the platform, for example
116   ``elf64-littleaarch64``.
117
118-  **#define : PLATFORM_LINKER_ARCH**
119
120   Defines the processor architecture for the linker by the platform, for
121   example ``aarch64``.
122
123-  **#define : PLATFORM_STACK_SIZE**
124
125   Defines the normal stack memory available to each CPU. This constant is used
126   by `plat/common/aarch64/platform_mp_stack.S`_ and
127   `plat/common/aarch64/platform_up_stack.S`_.
128
129-  **define : CACHE_WRITEBACK_GRANULE**
130
131   Defines the size in bits of the largest cache line across all the cache
132   levels in the platform.
133
134-  **#define : FIRMWARE_WELCOME_STR**
135
136   Defines the character string printed by BL1 upon entry into the ``bl1_main()``
137   function.
138
139-  **#define : PLATFORM_CORE_COUNT**
140
141   Defines the total number of CPUs implemented by the platform across all
142   clusters in the system.
143
144-  **#define : PLAT_NUM_PWR_DOMAINS**
145
146   Defines the total number of nodes in the power domain topology
147   tree at all the power domain levels used by the platform.
148   This macro is used by the PSCI implementation to allocate
149   data structures to represent power domain topology.
150
151-  **#define : PLAT_MAX_PWR_LVL**
152
153   Defines the maximum power domain level that the power management operations
154   should apply to. More often, but not always, the power domain level
155   corresponds to affinity level. This macro allows the PSCI implementation
156   to know the highest power domain level that it should consider for power
157   management operations in the system that the platform implements. For
158   example, the Base AEM FVP implements two clusters with a configurable
159   number of CPUs and it reports the maximum power domain level as 1.
160
161-  **#define : PLAT_MAX_OFF_STATE**
162
163   Defines the local power state corresponding to the deepest power down
164   possible at every power domain level in the platform. The local power
165   states for each level may be sparsely allocated between 0 and this value
166   with 0 being reserved for the RUN state. The PSCI implementation uses this
167   value to initialize the local power states of the power domain nodes and
168   to specify the requested power state for a PSCI_CPU_OFF call.
169
170-  **#define : PLAT_MAX_RET_STATE**
171
172   Defines the local power state corresponding to the deepest retention state
173   possible at every power domain level in the platform. This macro should be
174   a value less than PLAT_MAX_OFF_STATE and greater than 0. It is used by the
175   PSCI implementation to distinguish between retention and power down local
176   power states within PSCI_CPU_SUSPEND call.
177
178-  **#define : PLAT_MAX_PWR_LVL_STATES**
179
180   Defines the maximum number of local power states per power domain level
181   that the platform supports. The default value of this macro is 2 since
182   most platforms just support a maximum of two local power states at each
183   power domain level (power-down and retention). If the platform needs to
184   account for more local power states, then it must redefine this macro.
185
186   Currently, this macro is used by the Generic PSCI implementation to size
187   the array used for PSCI_STAT_COUNT/RESIDENCY accounting.
188
189-  **#define : BL1_RO_BASE**
190
191   Defines the base address in secure ROM where BL1 originally lives. Must be
192   aligned on a page-size boundary.
193
194-  **#define : BL1_RO_LIMIT**
195
196   Defines the maximum address in secure ROM that BL1's actual content (i.e.
197   excluding any data section allocated at runtime) can occupy.
198
199-  **#define : BL1_RW_BASE**
200
201   Defines the base address in secure RAM where BL1's read-write data will live
202   at runtime. Must be aligned on a page-size boundary.
203
204-  **#define : BL1_RW_LIMIT**
205
206   Defines the maximum address in secure RAM that BL1's read-write data can
207   occupy at runtime.
208
209-  **#define : BL2_BASE**
210
211   Defines the base address in secure RAM where BL1 loads the BL2 binary image.
212   Must be aligned on a page-size boundary. This constant is not applicable
213   when BL2_IN_XIP_MEM is set to '1'.
214
215-  **#define : BL2_LIMIT**
216
217   Defines the maximum address in secure RAM that the BL2 image can occupy.
218   This constant is not applicable when BL2_IN_XIP_MEM is set to '1'.
219
220-  **#define : BL2_RO_BASE**
221
222   Defines the base address in secure XIP memory where BL2 RO section originally
223   lives. Must be aligned on a page-size boundary. This constant is only needed
224   when BL2_IN_XIP_MEM is set to '1'.
225
226-  **#define : BL2_RO_LIMIT**
227
228   Defines the maximum address in secure XIP memory that BL2's actual content
229   (i.e. excluding any data section allocated at runtime) can occupy. This
230   constant is only needed when BL2_IN_XIP_MEM is set to '1'.
231
232-  **#define : BL2_RW_BASE**
233
234   Defines the base address in secure RAM where BL2's read-write data will live
235   at runtime. Must be aligned on a page-size boundary. This constant is only
236   needed when BL2_IN_XIP_MEM is set to '1'.
237
238-  **#define : BL2_RW_LIMIT**
239
240   Defines the maximum address in secure RAM that BL2's read-write data can
241   occupy at runtime. This constant is only needed when BL2_IN_XIP_MEM is set
242   to '1'.
243
244-  **#define : BL31_BASE**
245
246   Defines the base address in secure RAM where BL2 loads the BL31 binary
247   image. Must be aligned on a page-size boundary.
248
249-  **#define : BL31_LIMIT**
250
251   Defines the maximum address in secure RAM that the BL31 image can occupy.
252
253For every image, the platform must define individual identifiers that will be
254used by BL1 or BL2 to load the corresponding image into memory from non-volatile
255storage. For the sake of performance, integer numbers will be used as
256identifiers. The platform will use those identifiers to return the relevant
257information about the image to be loaded (file handler, load address,
258authentication information, etc.). The following image identifiers are
259mandatory:
260
261-  **#define : BL2_IMAGE_ID**
262
263   BL2 image identifier, used by BL1 to load BL2.
264
265-  **#define : BL31_IMAGE_ID**
266
267   BL31 image identifier, used by BL2 to load BL31.
268
269-  **#define : BL33_IMAGE_ID**
270
271   BL33 image identifier, used by BL2 to load BL33.
272
273If Trusted Board Boot is enabled, the following certificate identifiers must
274also be defined:
275
276-  **#define : TRUSTED_BOOT_FW_CERT_ID**
277
278   BL2 content certificate identifier, used by BL1 to load the BL2 content
279   certificate.
280
281-  **#define : TRUSTED_KEY_CERT_ID**
282
283   Trusted key certificate identifier, used by BL2 to load the trusted key
284   certificate.
285
286-  **#define : SOC_FW_KEY_CERT_ID**
287
288   BL31 key certificate identifier, used by BL2 to load the BL31 key
289   certificate.
290
291-  **#define : SOC_FW_CONTENT_CERT_ID**
292
293   BL31 content certificate identifier, used by BL2 to load the BL31 content
294   certificate.
295
296-  **#define : NON_TRUSTED_FW_KEY_CERT_ID**
297
298   BL33 key certificate identifier, used by BL2 to load the BL33 key
299   certificate.
300
301-  **#define : NON_TRUSTED_FW_CONTENT_CERT_ID**
302
303   BL33 content certificate identifier, used by BL2 to load the BL33 content
304   certificate.
305
306-  **#define : FWU_CERT_ID**
307
308   Firmware Update (FWU) certificate identifier, used by NS_BL1U to load the
309   FWU content certificate.
310
311-  **#define : PLAT_CRYPTOCELL_BASE**
312
313   This defines the base address of Arm® TrustZone® CryptoCell and must be
314   defined if CryptoCell crypto driver is used for Trusted Board Boot. For
315   capable Arm platforms, this driver is used if ``ARM_CRYPTOCELL_INTEG`` is
316   set.
317
318If the AP Firmware Updater Configuration image, BL2U is used, the following
319must also be defined:
320
321-  **#define : BL2U_BASE**
322
323   Defines the base address in secure memory where BL1 copies the BL2U binary
324   image. Must be aligned on a page-size boundary.
325
326-  **#define : BL2U_LIMIT**
327
328   Defines the maximum address in secure memory that the BL2U image can occupy.
329
330-  **#define : BL2U_IMAGE_ID**
331
332   BL2U image identifier, used by BL1 to fetch an image descriptor
333   corresponding to BL2U.
334
335If the SCP Firmware Update Configuration Image, SCP_BL2U is used, the following
336must also be defined:
337
338-  **#define : SCP_BL2U_IMAGE_ID**
339
340   SCP_BL2U image identifier, used by BL1 to fetch an image descriptor
341   corresponding to SCP_BL2U.
342   NOTE: TF-A does not provide source code for this image.
343
344If the Non-Secure Firmware Updater ROM, NS_BL1U is used, the following must
345also be defined:
346
347-  **#define : NS_BL1U_BASE**
348
349   Defines the base address in non-secure ROM where NS_BL1U executes.
350   Must be aligned on a page-size boundary.
351   NOTE: TF-A does not provide source code for this image.
352
353-  **#define : NS_BL1U_IMAGE_ID**
354
355   NS_BL1U image identifier, used by BL1 to fetch an image descriptor
356   corresponding to NS_BL1U.
357
358If the Non-Secure Firmware Updater, NS_BL2U is used, the following must also
359be defined:
360
361-  **#define : NS_BL2U_BASE**
362
363   Defines the base address in non-secure memory where NS_BL2U executes.
364   Must be aligned on a page-size boundary.
365   NOTE: TF-A does not provide source code for this image.
366
367-  **#define : NS_BL2U_IMAGE_ID**
368
369   NS_BL2U image identifier, used by BL1 to fetch an image descriptor
370   corresponding to NS_BL2U.
371
372For the the Firmware update capability of TRUSTED BOARD BOOT, the following
373macros may also be defined:
374
375-  **#define : PLAT_FWU_MAX_SIMULTANEOUS_IMAGES**
376
377   Total number of images that can be loaded simultaneously. If the platform
378   doesn't specify any value, it defaults to 10.
379
380If a SCP_BL2 image is supported by the platform, the following constants must
381also be defined:
382
383-  **#define : SCP_BL2_IMAGE_ID**
384
385   SCP_BL2 image identifier, used by BL2 to load SCP_BL2 into secure memory
386   from platform storage before being transferred to the SCP.
387
388-  **#define : SCP_FW_KEY_CERT_ID**
389
390   SCP_BL2 key certificate identifier, used by BL2 to load the SCP_BL2 key
391   certificate (mandatory when Trusted Board Boot is enabled).
392
393-  **#define : SCP_FW_CONTENT_CERT_ID**
394
395   SCP_BL2 content certificate identifier, used by BL2 to load the SCP_BL2
396   content certificate (mandatory when Trusted Board Boot is enabled).
397
398If a BL32 image is supported by the platform, the following constants must
399also be defined:
400
401-  **#define : BL32_IMAGE_ID**
402
403   BL32 image identifier, used by BL2 to load BL32.
404
405-  **#define : TRUSTED_OS_FW_KEY_CERT_ID**
406
407   BL32 key certificate identifier, used by BL2 to load the BL32 key
408   certificate (mandatory when Trusted Board Boot is enabled).
409
410-  **#define : TRUSTED_OS_FW_CONTENT_CERT_ID**
411
412   BL32 content certificate identifier, used by BL2 to load the BL32 content
413   certificate (mandatory when Trusted Board Boot is enabled).
414
415-  **#define : BL32_BASE**
416
417   Defines the base address in secure memory where BL2 loads the BL32 binary
418   image. Must be aligned on a page-size boundary.
419
420-  **#define : BL32_LIMIT**
421
422   Defines the maximum address that the BL32 image can occupy.
423
424If the Test Secure-EL1 Payload (TSP) instantiation of BL32 is supported by the
425platform, the following constants must also be defined:
426
427-  **#define : TSP_SEC_MEM_BASE**
428
429   Defines the base address of the secure memory used by the TSP image on the
430   platform. This must be at the same address or below ``BL32_BASE``.
431
432-  **#define : TSP_SEC_MEM_SIZE**
433
434   Defines the size of the secure memory used by the BL32 image on the
435   platform. ``TSP_SEC_MEM_BASE`` and ``TSP_SEC_MEM_SIZE`` must fully
436   accommodate the memory required by the BL32 image, defined by ``BL32_BASE``
437   and ``BL32_LIMIT``.
438
439-  **#define : TSP_IRQ_SEC_PHY_TIMER**
440
441   Defines the ID of the secure physical generic timer interrupt used by the
442   TSP's interrupt handling code.
443
444If the platform port uses the translation table library code, the following
445constants must also be defined:
446
447-  **#define : PLAT_XLAT_TABLES_DYNAMIC**
448
449   Optional flag that can be set per-image to enable the dynamic allocation of
450   regions even when the MMU is enabled. If not defined, only static
451   functionality will be available, if defined and set to 1 it will also
452   include the dynamic functionality.
453
454-  **#define : MAX_XLAT_TABLES**
455
456   Defines the maximum number of translation tables that are allocated by the
457   translation table library code. To minimize the amount of runtime memory
458   used, choose the smallest value needed to map the required virtual addresses
459   for each BL stage. If ``PLAT_XLAT_TABLES_DYNAMIC`` flag is enabled for a BL
460   image, ``MAX_XLAT_TABLES`` must be defined to accommodate the dynamic regions
461   as well.
462
463-  **#define : MAX_MMAP_REGIONS**
464
465   Defines the maximum number of regions that are allocated by the translation
466   table library code. A region consists of physical base address, virtual base
467   address, size and attributes (Device/Memory, RO/RW, Secure/Non-Secure), as
468   defined in the ``mmap_region_t`` structure. The platform defines the regions
469   that should be mapped. Then, the translation table library will create the
470   corresponding tables and descriptors at runtime. To minimize the amount of
471   runtime memory used, choose the smallest value needed to register the
472   required regions for each BL stage. If ``PLAT_XLAT_TABLES_DYNAMIC`` flag is
473   enabled for a BL image, ``MAX_MMAP_REGIONS`` must be defined to accommodate
474   the dynamic regions as well.
475
476-  **#define : PLAT_VIRT_ADDR_SPACE_SIZE**
477
478   Defines the total size of the virtual address space in bytes. For example,
479   for a 32 bit virtual address space, this value should be ``(1ULL << 32)``.
480
481-  **#define : PLAT_PHY_ADDR_SPACE_SIZE**
482
483   Defines the total size of the physical address space in bytes. For example,
484   for a 32 bit physical address space, this value should be ``(1ULL << 32)``.
485
486If the platform port uses the IO storage framework, the following constants
487must also be defined:
488
489-  **#define : MAX_IO_DEVICES**
490
491   Defines the maximum number of registered IO devices. Attempting to register
492   more devices than this value using ``io_register_device()`` will fail with
493   -ENOMEM.
494
495-  **#define : MAX_IO_HANDLES**
496
497   Defines the maximum number of open IO handles. Attempting to open more IO
498   entities than this value using ``io_open()`` will fail with -ENOMEM.
499
500-  **#define : MAX_IO_BLOCK_DEVICES**
501
502   Defines the maximum number of registered IO block devices. Attempting to
503   register more devices this value using ``io_dev_open()`` will fail
504   with -ENOMEM. MAX_IO_BLOCK_DEVICES should be less than MAX_IO_DEVICES.
505   With this macro, multiple block devices could be supported at the same
506   time.
507
508If the platform needs to allocate data within the per-cpu data framework in
509BL31, it should define the following macro. Currently this is only required if
510the platform decides not to use the coherent memory section by undefining the
511``USE_COHERENT_MEM`` build flag. In this case, the framework allocates the
512required memory within the the per-cpu data to minimize wastage.
513
514-  **#define : PLAT_PCPU_DATA_SIZE**
515
516   Defines the memory (in bytes) to be reserved within the per-cpu data
517   structure for use by the platform layer.
518
519The following constants are optional. They should be defined when the platform
520memory layout implies some image overlaying like in Arm standard platforms.
521
522-  **#define : BL31_PROGBITS_LIMIT**
523
524   Defines the maximum address in secure RAM that the BL31's progbits sections
525   can occupy.
526
527-  **#define : TSP_PROGBITS_LIMIT**
528
529   Defines the maximum address that the TSP's progbits sections can occupy.
530
531If the platform port uses the PL061 GPIO driver, the following constant may
532optionally be defined:
533
534-  **PLAT_PL061_MAX_GPIOS**
535   Maximum number of GPIOs required by the platform. This allows control how
536   much memory is allocated for PL061 GPIO controllers. The default value is
537
538   #. $(eval $(call add_define,PLAT_PL061_MAX_GPIOS))
539
540If the platform port uses the partition driver, the following constant may
541optionally be defined:
542
543-  **PLAT_PARTITION_MAX_ENTRIES**
544   Maximum number of partition entries required by the platform. This allows
545   control how much memory is allocated for partition entries. The default
546   value is 128.
547   `For example, define the build flag in platform.mk`_:
548   PLAT_PARTITION_MAX_ENTRIES := 12
549   $(eval $(call add_define,PLAT_PARTITION_MAX_ENTRIES))
550
551The following constant is optional. It should be defined to override the default
552behaviour of the ``assert()`` function (for example, to save memory).
553
554-  **PLAT_LOG_LEVEL_ASSERT**
555   If ``PLAT_LOG_LEVEL_ASSERT`` is higher or equal than ``LOG_LEVEL_VERBOSE``,
556   ``assert()`` prints the name of the file, the line number and the asserted
557   expression. Else if it is higher than ``LOG_LEVEL_INFO``, it prints the file
558   name and the line number. Else if it is lower than ``LOG_LEVEL_INFO``, it
559   doesn't print anything to the console. If ``PLAT_LOG_LEVEL_ASSERT`` isn't
560   defined, it defaults to ``LOG_LEVEL``.
561
562If the platform port uses the Activity Monitor Unit, the following constants
563may be defined:
564
565-  **PLAT_AMU_GROUP1_COUNTERS_MASK**
566   This mask reflects the set of group counters that should be enabled.  The
567   maximum number of group 1 counters supported by AMUv1 is 16 so the mask
568   can be at most 0xffff. If the platform does not define this mask, no group 1
569   counters are enabled. If the platform defines this mask, the following
570   constant needs to also be defined.
571
572-  **PLAT_AMU_GROUP1_NR_COUNTERS**
573   This value is used to allocate an array to save and restore the counters
574   specified by ``PLAT_AMU_GROUP1_COUNTERS_MASK`` on CPU suspend.
575   This value should be equal to the highest bit position set in the
576   mask, plus 1.  The maximum number of group 1 counters in AMUv1 is 16.
577
578File : plat_macros.S [mandatory]
579~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
580
581Each platform must ensure a file of this name is in the system include path with
582the following macro defined. In the Arm development platforms, this file is
583found in ``plat/arm/board/<plat_name>/include/plat_macros.S``.
584
585-  **Macro : plat_crash_print_regs**
586
587   This macro allows the crash reporting routine to print relevant platform
588   registers in case of an unhandled exception in BL31. This aids in debugging
589   and this macro can be defined to be empty in case register reporting is not
590   desired.
591
592   For instance, GIC or interconnect registers may be helpful for
593   troubleshooting.
594
595Handling Reset
596--------------
597
598BL1 by default implements the reset vector where execution starts from a cold
599or warm boot. BL31 can be optionally set as a reset vector using the
600``RESET_TO_BL31`` make variable.
601
602For each CPU, the reset vector code is responsible for the following tasks:
603
604#. Distinguishing between a cold boot and a warm boot.
605
606#. In the case of a cold boot and the CPU being a secondary CPU, ensuring that
607   the CPU is placed in a platform-specific state until the primary CPU
608   performs the necessary steps to remove it from this state.
609
610#. In the case of a warm boot, ensuring that the CPU jumps to a platform-
611   specific address in the BL31 image in the same processor mode as it was
612   when released from reset.
613
614The following functions need to be implemented by the platform port to enable
615reset vector code to perform the above tasks.
616
617Function : plat_get_my_entrypoint() [mandatory when PROGRAMMABLE_RESET_ADDRESS == 0]
618~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
619
620::
621
622    Argument : void
623    Return   : uintptr_t
624
625This function is called with the MMU and caches disabled
626(``SCTLR_EL3.M`` = 0 and ``SCTLR_EL3.C`` = 0). The function is responsible for
627distinguishing between a warm and cold reset for the current CPU using
628platform-specific means. If it's a warm reset, then it returns the warm
629reset entrypoint point provided to ``plat_setup_psci_ops()`` during
630BL31 initialization. If it's a cold reset then this function must return zero.
631
632This function does not follow the Procedure Call Standard used by the
633Application Binary Interface for the Arm 64-bit architecture. The caller should
634not assume that callee saved registers are preserved across a call to this
635function.
636
637This function fulfills requirement 1 and 3 listed above.
638
639Note that for platforms that support programming the reset address, it is
640expected that a CPU will start executing code directly at the right address,
641both on a cold and warm reset. In this case, there is no need to identify the
642type of reset nor to query the warm reset entrypoint. Therefore, implementing
643this function is not required on such platforms.
644
645Function : plat_secondary_cold_boot_setup() [mandatory when COLD_BOOT_SINGLE_CPU == 0]
646~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
647
648::
649
650    Argument : void
651
652This function is called with the MMU and data caches disabled. It is responsible
653for placing the executing secondary CPU in a platform-specific state until the
654primary CPU performs the necessary actions to bring it out of that state and
655allow entry into the OS. This function must not return.
656
657In the Arm FVP port, when using the normal boot flow, each secondary CPU powers
658itself off. The primary CPU is responsible for powering up the secondary CPUs
659when normal world software requires them. When booting an EL3 payload instead,
660they stay powered on and are put in a holding pen until their mailbox gets
661populated.
662
663This function fulfills requirement 2 above.
664
665Note that for platforms that can't release secondary CPUs out of reset, only the
666primary CPU will execute the cold boot code. Therefore, implementing this
667function is not required on such platforms.
668
669Function : plat_is_my_cpu_primary() [mandatory when COLD_BOOT_SINGLE_CPU == 0]
670~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
671
672::
673
674    Argument : void
675    Return   : unsigned int
676
677This function identifies whether the current CPU is the primary CPU or a
678secondary CPU. A return value of zero indicates that the CPU is not the
679primary CPU, while a non-zero return value indicates that the CPU is the
680primary CPU.
681
682Note that for platforms that can't release secondary CPUs out of reset, only the
683primary CPU will execute the cold boot code. Therefore, there is no need to
684distinguish between primary and secondary CPUs and implementing this function is
685not required.
686
687Function : platform_mem_init() [mandatory]
688~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
689
690::
691
692    Argument : void
693    Return   : void
694
695This function is called before any access to data is made by the firmware, in
696order to carry out any essential memory initialization.
697
698Function: plat_get_rotpk_info()
699~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
700
701::
702
703    Argument : void *, void **, unsigned int *, unsigned int *
704    Return   : int
705
706This function is mandatory when Trusted Board Boot is enabled. It returns a
707pointer to the ROTPK stored in the platform (or a hash of it) and its length.
708The ROTPK must be encoded in DER format according to the following ASN.1
709structure:
710
711::
712
713    AlgorithmIdentifier  ::=  SEQUENCE  {
714        algorithm         OBJECT IDENTIFIER,
715        parameters        ANY DEFINED BY algorithm OPTIONAL
716    }
717
718    SubjectPublicKeyInfo  ::=  SEQUENCE  {
719        algorithm         AlgorithmIdentifier,
720        subjectPublicKey  BIT STRING
721    }
722
723In case the function returns a hash of the key:
724
725::
726
727    DigestInfo ::= SEQUENCE {
728        digestAlgorithm   AlgorithmIdentifier,
729        digest            OCTET STRING
730    }
731
732The function returns 0 on success. Any other value is treated as error by the
733Trusted Board Boot. The function also reports extra information related
734to the ROTPK in the flags parameter:
735
736::
737
738    ROTPK_IS_HASH      : Indicates that the ROTPK returned by the platform is a
739                         hash.
740    ROTPK_NOT_DEPLOYED : This allows the platform to skip certificate ROTPK
741                         verification while the platform ROTPK is not deployed.
742                         When this flag is set, the function does not need to
743                         return a platform ROTPK, and the authentication
744                         framework uses the ROTPK in the certificate without
745                         verifying it against the platform value. This flag
746                         must not be used in a deployed production environment.
747
748Function: plat_get_nv_ctr()
749~~~~~~~~~~~~~~~~~~~~~~~~~~~
750
751::
752
753    Argument : void *, unsigned int *
754    Return   : int
755
756This function is mandatory when Trusted Board Boot is enabled. It returns the
757non-volatile counter value stored in the platform in the second argument. The
758cookie in the first argument may be used to select the counter in case the
759platform provides more than one (for example, on platforms that use the default
760TBBR CoT, the cookie will correspond to the OID values defined in
761TRUSTED_FW_NVCOUNTER_OID or NON_TRUSTED_FW_NVCOUNTER_OID).
762
763The function returns 0 on success. Any other value means the counter value could
764not be retrieved from the platform.
765
766Function: plat_set_nv_ctr()
767~~~~~~~~~~~~~~~~~~~~~~~~~~~
768
769::
770
771    Argument : void *, unsigned int
772    Return   : int
773
774This function is mandatory when Trusted Board Boot is enabled. It sets a new
775counter value in the platform. The cookie in the first argument may be used to
776select the counter (as explained in plat_get_nv_ctr()). The second argument is
777the updated counter value to be written to the NV counter.
778
779The function returns 0 on success. Any other value means the counter value could
780not be updated.
781
782Function: plat_set_nv_ctr2()
783~~~~~~~~~~~~~~~~~~~~~~~~~~~~
784
785::
786
787    Argument : void *, const auth_img_desc_t *, unsigned int
788    Return   : int
789
790This function is optional when Trusted Board Boot is enabled. If this
791interface is defined, then ``plat_set_nv_ctr()`` need not be defined. The
792first argument passed is a cookie and is typically used to
793differentiate between a Non Trusted NV Counter and a Trusted NV
794Counter. The second argument is a pointer to an authentication image
795descriptor and may be used to decide if the counter is allowed to be
796updated or not. The third argument is the updated counter value to
797be written to the NV counter.
798
799The function returns 0 on success. Any other value means the counter value
800either could not be updated or the authentication image descriptor indicates
801that it is not allowed to be updated.
802
803Common mandatory function modifications
804---------------------------------------
805
806The following functions are mandatory functions which need to be implemented
807by the platform port.
808
809Function : plat_my_core_pos()
810~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
811
812::
813
814    Argument : void
815    Return   : unsigned int
816
817This function returns the index of the calling CPU which is used as a
818CPU-specific linear index into blocks of memory (for example while allocating
819per-CPU stacks). This function will be invoked very early in the
820initialization sequence which mandates that this function should be
821implemented in assembly and should not rely on the availability of a C
822runtime environment. This function can clobber x0 - x8 and must preserve
823x9 - x29.
824
825This function plays a crucial role in the power domain topology framework in
826PSCI and details of this can be found in `Power Domain Topology Design`_.
827
828Function : plat_core_pos_by_mpidr()
829~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
830
831::
832
833    Argument : u_register_t
834    Return   : int
835
836This function validates the ``MPIDR`` of a CPU and converts it to an index,
837which can be used as a CPU-specific linear index into blocks of memory. In
838case the ``MPIDR`` is invalid, this function returns -1. This function will only
839be invoked by BL31 after the power domain topology is initialized and can
840utilize the C runtime environment. For further details about how TF-A
841represents the power domain topology and how this relates to the linear CPU
842index, please refer `Power Domain Topology Design`_.
843
844Common optional modifications
845-----------------------------
846
847The following are helper functions implemented by the firmware that perform
848common platform-specific tasks. A platform may choose to override these
849definitions.
850
851Function : plat_set_my_stack()
852~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
853
854::
855
856    Argument : void
857    Return   : void
858
859This function sets the current stack pointer to the normal memory stack that
860has been allocated for the current CPU. For BL images that only require a
861stack for the primary CPU, the UP version of the function is used. The size
862of the stack allocated to each CPU is specified by the platform defined
863constant ``PLATFORM_STACK_SIZE``.
864
865Common implementations of this function for the UP and MP BL images are
866provided in `plat/common/aarch64/platform_up_stack.S`_ and
867`plat/common/aarch64/platform_mp_stack.S`_
868
869Function : plat_get_my_stack()
870~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
871
872::
873
874    Argument : void
875    Return   : uintptr_t
876
877This function returns the base address of the normal memory stack that
878has been allocated for the current CPU. For BL images that only require a
879stack for the primary CPU, the UP version of the function is used. The size
880of the stack allocated to each CPU is specified by the platform defined
881constant ``PLATFORM_STACK_SIZE``.
882
883Common implementations of this function for the UP and MP BL images are
884provided in `plat/common/aarch64/platform_up_stack.S`_ and
885`plat/common/aarch64/platform_mp_stack.S`_
886
887Function : plat_report_exception()
888~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
889
890::
891
892    Argument : unsigned int
893    Return   : void
894
895A platform may need to report various information about its status when an
896exception is taken, for example the current exception level, the CPU security
897state (secure/non-secure), the exception type, and so on. This function is
898called in the following circumstances:
899
900-  In BL1, whenever an exception is taken.
901-  In BL2, whenever an exception is taken.
902
903The default implementation doesn't do anything, to avoid making assumptions
904about the way the platform displays its status information.
905
906For AArch64, this function receives the exception type as its argument.
907Possible values for exceptions types are listed in the
908`include/common/bl_common.h`_ header file. Note that these constants are not
909related to any architectural exception code; they are just a TF-A convention.
910
911For AArch32, this function receives the exception mode as its argument.
912Possible values for exception modes are listed in the
913`include/lib/aarch32/arch.h`_ header file.
914
915Function : plat_reset_handler()
916~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
917
918::
919
920    Argument : void
921    Return   : void
922
923A platform may need to do additional initialization after reset. This function
924allows the platform to do the platform specific intializations. Platform
925specific errata workarounds could also be implemented here. The API should
926preserve the values of callee saved registers x19 to x29.
927
928The default implementation doesn't do anything. If a platform needs to override
929the default implementation, refer to the `Firmware Design`_ for general
930guidelines.
931
932Function : plat_disable_acp()
933~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
934
935::
936
937    Argument : void
938    Return   : void
939
940This API allows a platform to disable the Accelerator Coherency Port (if
941present) during a cluster power down sequence. The default weak implementation
942doesn't do anything. Since this API is called during the power down sequence,
943it has restrictions for stack usage and it can use the registers x0 - x17 as
944scratch registers. It should preserve the value in x18 register as it is used
945by the caller to store the return address.
946
947Function : plat_error_handler()
948~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
949
950::
951
952    Argument : int
953    Return   : void
954
955This API is called when the generic code encounters an error situation from
956which it cannot continue. It allows the platform to perform error reporting or
957recovery actions (for example, reset the system). This function must not return.
958
959The parameter indicates the type of error using standard codes from ``errno.h``.
960Possible errors reported by the generic code are:
961
962-  ``-EAUTH``: a certificate or image could not be authenticated (when Trusted
963   Board Boot is enabled)
964-  ``-ENOENT``: the requested image or certificate could not be found or an IO
965   error was detected
966-  ``-ENOMEM``: resources exhausted. TF-A does not use dynamic memory, so this
967   error is usually an indication of an incorrect array size
968
969The default implementation simply spins.
970
971Function : plat_panic_handler()
972~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
973
974::
975
976    Argument : void
977    Return   : void
978
979This API is called when the generic code encounters an unexpected error
980situation from which it cannot recover. This function must not return,
981and must be implemented in assembly because it may be called before the C
982environment is initialized.
983
984Note: The address from where it was called is stored in x30 (Link Register).
985The default implementation simply spins.
986
987Function : plat_get_bl_image_load_info()
988~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
989
990::
991
992    Argument : void
993    Return   : bl_load_info_t *
994
995This function returns pointer to the list of images that the platform has
996populated to load. This function is invoked in BL2 to load the
997BL3xx images.
998
999Function : plat_get_next_bl_params()
1000~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1001
1002::
1003
1004    Argument : void
1005    Return   : bl_params_t *
1006
1007This function returns a pointer to the shared memory that the platform has
1008kept aside to pass TF-A related information that next BL image needs. This
1009function is invoked in BL2 to pass this information to the next BL
1010image.
1011
1012Function : plat_get_stack_protector_canary()
1013~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1014
1015::
1016
1017    Argument : void
1018    Return   : u_register_t
1019
1020This function returns a random value that is used to initialize the canary used
1021when the stack protector is enabled with ENABLE_STACK_PROTECTOR. A predictable
1022value will weaken the protection as the attacker could easily write the right
1023value as part of the attack most of the time. Therefore, it should return a
1024true random number.
1025
1026Note: For the protection to be effective, the global data need to be placed at
1027a lower address than the stack bases. Failure to do so would allow an attacker
1028to overwrite the canary as part of the stack buffer overflow attack.
1029
1030Function : plat_flush_next_bl_params()
1031~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1032
1033::
1034
1035    Argument : void
1036    Return   : void
1037
1038This function flushes to main memory all the image params that are passed to
1039next image. This function is invoked in BL2 to flush this information
1040to the next BL image.
1041
1042Function : plat_log_get_prefix()
1043~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1044
1045::
1046
1047    Argument : unsigned int
1048    Return   : const char *
1049
1050This function defines the prefix string corresponding to the `log_level` to be
1051prepended to all the log output from TF-A. The `log_level` (argument) will
1052correspond to one of the standard log levels defined in debug.h. The platform
1053can override the common implementation to define a different prefix string for
1054the log output. The implementation should be robust to future changes that
1055increase the number of log levels.
1056
1057Function : plat_get_mbedtls_heap()
1058~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1059
1060::
1061
1062    Arguments : void **heap_addr, size_t *heap_size
1063    Return    : int
1064
1065This function is invoked during Mbed TLS library initialisation to get
1066a heap, by means of a starting address and a size. This heap will then be used
1067internally by the Mbed TLS library. The heap is requested from the current BL
1068stage, i.e. the current BL image inside which Mbed TLS is used.
1069
1070In the default implementation a heap is statically allocated inside every image
1071(i.e. every BL stage) that utilises Mbed TLS. So, in this case, the function
1072simply returns the address and size of this "pre-allocated" heap. However, by
1073overriding the default implementation, platforms have the potential to optimise
1074memory usage. For example, on some Arm platforms, the Mbed TLS heap is shared
1075between BL1 and BL2 stages and, thus, the necessary space is not reserved
1076twice.
1077
1078On success the function should return 0 and a negative error code otherwise.
1079
1080Modifications specific to a Boot Loader stage
1081---------------------------------------------
1082
1083Boot Loader Stage 1 (BL1)
1084-------------------------
1085
1086BL1 implements the reset vector where execution starts from after a cold or
1087warm boot. For each CPU, BL1 is responsible for the following tasks:
1088
1089#. Handling the reset as described in section 2.2
1090
1091#. In the case of a cold boot and the CPU being the primary CPU, ensuring that
1092   only this CPU executes the remaining BL1 code, including loading and passing
1093   control to the BL2 stage.
1094
1095#. Identifying and starting the Firmware Update process (if required).
1096
1097#. Loading the BL2 image from non-volatile storage into secure memory at the
1098   address specified by the platform defined constant ``BL2_BASE``.
1099
1100#. Populating a ``meminfo`` structure with the following information in memory,
1101   accessible by BL2 immediately upon entry.
1102
1103   ::
1104
1105       meminfo.total_base = Base address of secure RAM visible to BL2
1106       meminfo.total_size = Size of secure RAM visible to BL2
1107
1108   By default, BL1 places this ``meminfo`` structure at the end of secure
1109   memory visible to BL2.
1110
1111   It is possible for the platform to decide where it wants to place the
1112   ``meminfo`` structure for BL2 or restrict the amount of memory visible to
1113   BL2 by overriding the weak default implementation of
1114   ``bl1_plat_handle_post_image_load`` API.
1115
1116The following functions need to be implemented by the platform port to enable
1117BL1 to perform the above tasks.
1118
1119Function : bl1_early_platform_setup() [mandatory]
1120~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1121
1122::
1123
1124    Argument : void
1125    Return   : void
1126
1127This function executes with the MMU and data caches disabled. It is only called
1128by the primary CPU.
1129
1130On Arm standard platforms, this function:
1131
1132-  Enables a secure instance of SP805 to act as the Trusted Watchdog.
1133
1134-  Initializes a UART (PL011 console), which enables access to the ``printf``
1135   family of functions in BL1.
1136
1137-  Enables issuing of snoop and DVM (Distributed Virtual Memory) requests to
1138   the CCI slave interface corresponding to the cluster that includes the
1139   primary CPU.
1140
1141Function : bl1_plat_arch_setup() [mandatory]
1142~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1143
1144::
1145
1146    Argument : void
1147    Return   : void
1148
1149This function performs any platform-specific and architectural setup that the
1150platform requires. Platform-specific setup might include configuration of
1151memory controllers and the interconnect.
1152
1153In Arm standard platforms, this function enables the MMU.
1154
1155This function helps fulfill requirement 2 above.
1156
1157Function : bl1_platform_setup() [mandatory]
1158~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1159
1160::
1161
1162    Argument : void
1163    Return   : void
1164
1165This function executes with the MMU and data caches enabled. It is responsible
1166for performing any remaining platform-specific setup that can occur after the
1167MMU and data cache have been enabled.
1168
1169if support for multiple boot sources is required, it initializes the boot
1170sequence used by plat_try_next_boot_source().
1171
1172In Arm standard platforms, this function initializes the storage abstraction
1173layer used to load the next bootloader image.
1174
1175This function helps fulfill requirement 4 above.
1176
1177Function : bl1_plat_sec_mem_layout() [mandatory]
1178~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1179
1180::
1181
1182    Argument : void
1183    Return   : meminfo *
1184
1185This function should only be called on the cold boot path. It executes with the
1186MMU and data caches enabled. The pointer returned by this function must point to
1187a ``meminfo`` structure containing the extents and availability of secure RAM for
1188the BL1 stage.
1189
1190::
1191
1192    meminfo.total_base = Base address of secure RAM visible to BL1
1193    meminfo.total_size = Size of secure RAM visible to BL1
1194
1195This information is used by BL1 to load the BL2 image in secure RAM. BL1 also
1196populates a similar structure to tell BL2 the extents of memory available for
1197its own use.
1198
1199This function helps fulfill requirements 4 and 5 above.
1200
1201Function : bl1_plat_prepare_exit() [optional]
1202~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1203
1204::
1205
1206    Argument : entry_point_info_t *
1207    Return   : void
1208
1209This function is called prior to exiting BL1 in response to the
1210``BL1_SMC_RUN_IMAGE`` SMC request raised by BL2. It should be used to perform
1211platform specific clean up or bookkeeping operations before transferring
1212control to the next image. It receives the address of the ``entry_point_info_t``
1213structure passed from BL2. This function runs with MMU disabled.
1214
1215Function : bl1_plat_set_ep_info() [optional]
1216~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1217
1218::
1219
1220    Argument : unsigned int image_id, entry_point_info_t *ep_info
1221    Return   : void
1222
1223This function allows platforms to override ``ep_info`` for the given ``image_id``.
1224
1225The default implementation just returns.
1226
1227Function : bl1_plat_get_next_image_id() [optional]
1228~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1229
1230::
1231
1232    Argument : void
1233    Return   : unsigned int
1234
1235This and the following function must be overridden to enable the FWU feature.
1236
1237BL1 calls this function after platform setup to identify the next image to be
1238loaded and executed. If the platform returns ``BL2_IMAGE_ID`` then BL1 proceeds
1239with the normal boot sequence, which loads and executes BL2. If the platform
1240returns a different image id, BL1 assumes that Firmware Update is required.
1241
1242The default implementation always returns ``BL2_IMAGE_ID``. The Arm development
1243platforms override this function to detect if firmware update is required, and
1244if so, return the first image in the firmware update process.
1245
1246Function : bl1_plat_get_image_desc() [optional]
1247~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1248
1249::
1250
1251    Argument : unsigned int image_id
1252    Return   : image_desc_t *
1253
1254BL1 calls this function to get the image descriptor information ``image_desc_t``
1255for the provided ``image_id`` from the platform.
1256
1257The default implementation always returns a common BL2 image descriptor. Arm
1258standard platforms return an image descriptor corresponding to BL2 or one of
1259the firmware update images defined in the Trusted Board Boot Requirements
1260specification.
1261
1262Function : bl1_plat_handle_pre_image_load() [optional]
1263~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1264
1265::
1266
1267    Argument : unsigned int image_id
1268    Return   : int
1269
1270This function can be used by the platforms to update/use image information
1271corresponding to ``image_id``. This function is invoked in BL1, both in cold
1272boot and FWU code path, before loading the image.
1273
1274Function : bl1_plat_handle_post_image_load() [optional]
1275~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1276
1277::
1278
1279    Argument : unsigned int image_id
1280    Return   : int
1281
1282This function can be used by the platforms to update/use image information
1283corresponding to ``image_id``. This function is invoked in BL1, both in cold
1284boot and FWU code path, after loading and authenticating the image.
1285
1286The default weak implementation of this function calculates the amount of
1287Trusted SRAM that can be used by BL2 and allocates a ``meminfo_t``
1288structure at the beginning of this free memory and populates it. The address
1289of ``meminfo_t`` structure is updated in ``arg1`` of the entrypoint
1290information to BL2.
1291
1292Function : bl1_plat_fwu_done() [optional]
1293~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1294
1295::
1296
1297    Argument : unsigned int image_id, uintptr_t image_src,
1298               unsigned int image_size
1299    Return   : void
1300
1301BL1 calls this function when the FWU process is complete. It must not return.
1302The platform may override this function to take platform specific action, for
1303example to initiate the normal boot flow.
1304
1305The default implementation spins forever.
1306
1307Function : bl1_plat_mem_check() [mandatory]
1308~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1309
1310::
1311
1312    Argument : uintptr_t mem_base, unsigned int mem_size,
1313               unsigned int flags
1314    Return   : int
1315
1316BL1 calls this function while handling FWU related SMCs, more specifically when
1317copying or authenticating an image. Its responsibility is to ensure that the
1318region of memory identified by ``mem_base`` and ``mem_size`` is mapped in BL1, and
1319that this memory corresponds to either a secure or non-secure memory region as
1320indicated by the security state of the ``flags`` argument.
1321
1322This function can safely assume that the value resulting from the addition of
1323``mem_base`` and ``mem_size`` fits into a ``uintptr_t`` type variable and does not
1324overflow.
1325
1326This function must return 0 on success, a non-null error code otherwise.
1327
1328The default implementation of this function asserts therefore platforms must
1329override it when using the FWU feature.
1330
1331Boot Loader Stage 2 (BL2)
1332-------------------------
1333
1334The BL2 stage is executed only by the primary CPU, which is determined in BL1
1335using the ``platform_is_primary_cpu()`` function. BL1 passed control to BL2 at
1336``BL2_BASE``. BL2 executes in Secure EL1 and and invokes
1337``plat_get_bl_image_load_info()`` to retrieve the list of images to load from
1338non-volatile storage to secure/non-secure RAM. After all the images are loaded
1339then BL2 invokes ``plat_get_next_bl_params()`` to get the list of executable
1340images to be passed to the next BL image.
1341
1342The following functions must be implemented by the platform port to enable BL2
1343to perform the above tasks.
1344
1345Function : bl2_early_platform_setup2() [mandatory]
1346~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1347
1348::
1349
1350    Argument : u_register_t, u_register_t, u_register_t, u_register_t
1351    Return   : void
1352
1353This function executes with the MMU and data caches disabled. It is only called
1354by the primary CPU. The 4 arguments are passed by BL1 to BL2 and these arguments
1355are platform specific.
1356
1357On Arm standard platforms, the arguments received are :
1358
1359    arg0 - Points to load address of HW_CONFIG if present
1360
1361    arg1 - ``meminfo`` structure populated by BL1. The platform copies
1362    the contents of ``meminfo`` as it may be subsequently overwritten by BL2.
1363
1364On Arm standard platforms, this function also:
1365
1366-  Initializes a UART (PL011 console), which enables access to the ``printf``
1367   family of functions in BL2.
1368
1369-  Initializes the storage abstraction layer used to load further bootloader
1370   images. It is necessary to do this early on platforms with a SCP_BL2 image,
1371   since the later ``bl2_platform_setup`` must be done after SCP_BL2 is loaded.
1372
1373Function : bl2_plat_arch_setup() [mandatory]
1374~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1375
1376::
1377
1378    Argument : void
1379    Return   : void
1380
1381This function executes with the MMU and data caches disabled. It is only called
1382by the primary CPU.
1383
1384The purpose of this function is to perform any architectural initialization
1385that varies across platforms.
1386
1387On Arm standard platforms, this function enables the MMU.
1388
1389Function : bl2_platform_setup() [mandatory]
1390~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1391
1392::
1393
1394    Argument : void
1395    Return   : void
1396
1397This function may execute with the MMU and data caches enabled if the platform
1398port does the necessary initialization in ``bl2_plat_arch_setup()``. It is only
1399called by the primary CPU.
1400
1401The purpose of this function is to perform any platform initialization
1402specific to BL2.
1403
1404In Arm standard platforms, this function performs security setup, including
1405configuration of the TrustZone controller to allow non-secure masters access
1406to most of DRAM. Part of DRAM is reserved for secure world use.
1407
1408Function : bl2_plat_handle_pre_image_load() [optional]
1409~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1410
1411::
1412
1413    Argument : unsigned int
1414    Return   : int
1415
1416This function can be used by the platforms to update/use image information
1417for given ``image_id``. This function is currently invoked in BL2 before
1418loading each image.
1419
1420Function : bl2_plat_handle_post_image_load() [optional]
1421~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1422
1423::
1424
1425    Argument : unsigned int
1426    Return   : int
1427
1428This function can be used by the platforms to update/use image information
1429for given ``image_id``. This function is currently invoked in BL2 after
1430loading each image.
1431
1432Function : bl2_plat_preload_setup [optional]
1433~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1434
1435::
1436
1437    Argument : void
1438    Return   : void
1439
1440This optional function performs any BL2 platform initialization
1441required before image loading, that is not done later in
1442bl2_platform_setup(). Specifically, if support for multiple
1443boot sources is required, it initializes the boot sequence used by
1444plat_try_next_boot_source().
1445
1446Function : plat_try_next_boot_source() [optional]
1447~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1448
1449::
1450
1451    Argument : void
1452    Return   : int
1453
1454This optional function passes to the next boot source in the redundancy
1455sequence.
1456
1457This function moves the current boot redundancy source to the next
1458element in the boot sequence. If there are no more boot sources then it
1459must return 0, otherwise it must return 1. The default implementation
1460of this always returns 0.
1461
1462Boot Loader Stage 2 (BL2) at EL3
1463--------------------------------
1464
1465When the platform has a non-TF-A Boot ROM it is desirable to jump
1466directly to BL2 instead of TF-A BL1. In this case BL2 is expected to
1467execute at EL3 instead of executing at EL1. Refer to the `Firmware
1468Design`_ for more information.
1469
1470All mandatory functions of BL2 must be implemented, except the functions
1471bl2_early_platform_setup and bl2_el3_plat_arch_setup, because
1472their work is done now by bl2_el3_early_platform_setup and
1473bl2_el3_plat_arch_setup. These functions should generally implement
1474the bl1_plat_xxx() and bl2_plat_xxx() functionality combined.
1475
1476
1477Function : bl2_el3_early_platform_setup() [mandatory]
1478~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1479
1480::
1481
1482	Argument : u_register_t, u_register_t, u_register_t, u_register_t
1483	Return   : void
1484
1485This function executes with the MMU and data caches disabled. It is only called
1486by the primary CPU. This function receives four parameters which can be used
1487by the platform to pass any needed information from the Boot ROM to BL2.
1488
1489On Arm standard platforms, this function does the following:
1490
1491-  Initializes a UART (PL011 console), which enables access to the ``printf``
1492   family of functions in BL2.
1493
1494-  Initializes the storage abstraction layer used to load further bootloader
1495   images. It is necessary to do this early on platforms with a SCP_BL2 image,
1496   since the later ``bl2_platform_setup`` must be done after SCP_BL2 is loaded.
1497
1498- Initializes the private variables that define the memory layout used.
1499
1500Function : bl2_el3_plat_arch_setup() [mandatory]
1501~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1502
1503::
1504
1505	Argument : void
1506	Return   : void
1507
1508This function executes with the MMU and data caches disabled. It is only called
1509by the primary CPU.
1510
1511The purpose of this function is to perform any architectural initialization
1512that varies across platforms.
1513
1514On Arm standard platforms, this function enables the MMU.
1515
1516Function : bl2_el3_plat_prepare_exit() [optional]
1517~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1518
1519::
1520
1521	Argument : void
1522	Return   : void
1523
1524This function is called prior to exiting BL2 and run the next image.
1525It should be used to perform platform specific clean up or bookkeeping
1526operations before transferring control to the next image. This function
1527runs with MMU disabled.
1528
1529FWU Boot Loader Stage 2 (BL2U)
1530------------------------------
1531
1532The AP Firmware Updater Configuration, BL2U, is an optional part of the FWU
1533process and is executed only by the primary CPU. BL1 passes control to BL2U at
1534``BL2U_BASE``. BL2U executes in Secure-EL1 and is responsible for:
1535
1536#. (Optional) Transferring the optional SCP_BL2U binary image from AP secure
1537   memory to SCP RAM. BL2U uses the SCP_BL2U ``image_info`` passed by BL1.
1538   ``SCP_BL2U_BASE`` defines the address in AP secure memory where SCP_BL2U
1539   should be copied from. Subsequent handling of the SCP_BL2U image is
1540   implemented by the platform specific ``bl2u_plat_handle_scp_bl2u()`` function.
1541   If ``SCP_BL2U_BASE`` is not defined then this step is not performed.
1542
1543#. Any platform specific setup required to perform the FWU process. For
1544   example, Arm standard platforms initialize the TZC controller so that the
1545   normal world can access DDR memory.
1546
1547The following functions must be implemented by the platform port to enable
1548BL2U to perform the tasks mentioned above.
1549
1550Function : bl2u_early_platform_setup() [mandatory]
1551~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1552
1553::
1554
1555    Argument : meminfo *mem_info, void *plat_info
1556    Return   : void
1557
1558This function executes with the MMU and data caches disabled. It is only
1559called by the primary CPU. The arguments to this function is the address
1560of the ``meminfo`` structure and platform specific info provided by BL1.
1561
1562The platform may copy the contents of the ``mem_info`` and ``plat_info`` into
1563private storage as the original memory may be subsequently overwritten by BL2U.
1564
1565On Arm CSS platforms ``plat_info`` is interpreted as an ``image_info_t`` structure,
1566to extract SCP_BL2U image information, which is then copied into a private
1567variable.
1568
1569Function : bl2u_plat_arch_setup() [mandatory]
1570~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1571
1572::
1573
1574    Argument : void
1575    Return   : void
1576
1577This function executes with the MMU and data caches disabled. It is only
1578called by the primary CPU.
1579
1580The purpose of this function is to perform any architectural initialization
1581that varies across platforms, for example enabling the MMU (since the memory
1582map differs across platforms).
1583
1584Function : bl2u_platform_setup() [mandatory]
1585~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1586
1587::
1588
1589    Argument : void
1590    Return   : void
1591
1592This function may execute with the MMU and data caches enabled if the platform
1593port does the necessary initialization in ``bl2u_plat_arch_setup()``. It is only
1594called by the primary CPU.
1595
1596The purpose of this function is to perform any platform initialization
1597specific to BL2U.
1598
1599In Arm standard platforms, this function performs security setup, including
1600configuration of the TrustZone controller to allow non-secure masters access
1601to most of DRAM. Part of DRAM is reserved for secure world use.
1602
1603Function : bl2u_plat_handle_scp_bl2u() [optional]
1604~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1605
1606::
1607
1608    Argument : void
1609    Return   : int
1610
1611This function is used to perform any platform-specific actions required to
1612handle the SCP firmware. Typically it transfers the image into SCP memory using
1613a platform-specific protocol and waits until SCP executes it and signals to the
1614Application Processor (AP) for BL2U execution to continue.
1615
1616This function returns 0 on success, a negative error code otherwise.
1617This function is included if SCP_BL2U_BASE is defined.
1618
1619Boot Loader Stage 3-1 (BL31)
1620----------------------------
1621
1622During cold boot, the BL31 stage is executed only by the primary CPU. This is
1623determined in BL1 using the ``platform_is_primary_cpu()`` function. BL1 passes
1624control to BL31 at ``BL31_BASE``. During warm boot, BL31 is executed by all
1625CPUs. BL31 executes at EL3 and is responsible for:
1626
1627#. Re-initializing all architectural and platform state. Although BL1 performs
1628   some of this initialization, BL31 remains resident in EL3 and must ensure
1629   that EL3 architectural and platform state is completely initialized. It
1630   should make no assumptions about the system state when it receives control.
1631
1632#. Passing control to a normal world BL image, pre-loaded at a platform-
1633   specific address by BL2. On ARM platforms, BL31 uses the ``bl_params`` list
1634   populated by BL2 in memory to do this.
1635
1636#. Providing runtime firmware services. Currently, BL31 only implements a
1637   subset of the Power State Coordination Interface (PSCI) API as a runtime
1638   service. See Section 3.3 below for details of porting the PSCI
1639   implementation.
1640
1641#. Optionally passing control to the BL32 image, pre-loaded at a platform-
1642   specific address by BL2. BL31 exports a set of APIs that allow runtime
1643   services to specify the security state in which the next image should be
1644   executed and run the corresponding image. On ARM platforms, BL31 uses the
1645   ``bl_params`` list populated by BL2 in memory to do this.
1646
1647If BL31 is a reset vector, It also needs to handle the reset as specified in
1648section 2.2 before the tasks described above.
1649
1650The following functions must be implemented by the platform port to enable BL31
1651to perform the above tasks.
1652
1653Function : bl31_early_platform_setup2() [mandatory]
1654~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1655
1656::
1657
1658    Argument : u_register_t, u_register_t, u_register_t, u_register_t
1659    Return   : void
1660
1661This function executes with the MMU and data caches disabled. It is only called
1662by the primary CPU. BL2 can pass 4 arguments to BL31 and these arguments are
1663platform specific.
1664
1665In Arm standard platforms, the arguments received are :
1666
1667    arg0 - The pointer to the head of `bl_params_t` list
1668    which is list of executable images following BL31,
1669
1670    arg1 - Points to load address of SOC_FW_CONFIG if present
1671
1672    arg2 - Points to load address of HW_CONFIG if present
1673
1674    arg3 - A special value to verify platform parameters from BL2 to BL31. Not
1675    used in release builds.
1676
1677The function runs through the `bl_param_t` list and extracts the entry point
1678information for BL32 and BL33. It also performs the following:
1679
1680-  Initialize a UART (PL011 console), which enables access to the ``printf``
1681   family of functions in BL31.
1682
1683-  Enable issuing of snoop and DVM (Distributed Virtual Memory) requests to the
1684   CCI slave interface corresponding to the cluster that includes the primary
1685   CPU.
1686
1687Function : bl31_plat_arch_setup() [mandatory]
1688~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1689
1690::
1691
1692    Argument : void
1693    Return   : void
1694
1695This function executes with the MMU and data caches disabled. It is only called
1696by the primary CPU.
1697
1698The purpose of this function is to perform any architectural initialization
1699that varies across platforms.
1700
1701On Arm standard platforms, this function enables the MMU.
1702
1703Function : bl31_platform_setup() [mandatory]
1704~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1705
1706::
1707
1708    Argument : void
1709    Return   : void
1710
1711This function may execute with the MMU and data caches enabled if the platform
1712port does the necessary initialization in ``bl31_plat_arch_setup()``. It is only
1713called by the primary CPU.
1714
1715The purpose of this function is to complete platform initialization so that both
1716BL31 runtime services and normal world software can function correctly.
1717
1718On Arm standard platforms, this function does the following:
1719
1720-  Initialize the generic interrupt controller.
1721
1722   Depending on the GIC driver selected by the platform, the appropriate GICv2
1723   or GICv3 initialization will be done, which mainly consists of:
1724
1725   -  Enable secure interrupts in the GIC CPU interface.
1726   -  Disable the legacy interrupt bypass mechanism.
1727   -  Configure the priority mask register to allow interrupts of all priorities
1728      to be signaled to the CPU interface.
1729   -  Mark SGIs 8-15 and the other secure interrupts on the platform as secure.
1730   -  Target all secure SPIs to CPU0.
1731   -  Enable these secure interrupts in the GIC distributor.
1732   -  Configure all other interrupts as non-secure.
1733   -  Enable signaling of secure interrupts in the GIC distributor.
1734
1735-  Enable system-level implementation of the generic timer counter through the
1736   memory mapped interface.
1737
1738-  Grant access to the system counter timer module
1739
1740-  Initialize the power controller device.
1741
1742   In particular, initialise the locks that prevent concurrent accesses to the
1743   power controller device.
1744
1745Function : bl31_plat_runtime_setup() [optional]
1746~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1747
1748::
1749
1750    Argument : void
1751    Return   : void
1752
1753The purpose of this function is allow the platform to perform any BL31 runtime
1754setup just prior to BL31 exit during cold boot. The default weak
1755implementation of this function will invoke ``console_switch_state()`` to switch
1756console output to consoles marked for use in the ``runtime`` state.
1757
1758Function : bl31_plat_get_next_image_ep_info() [mandatory]
1759~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1760
1761::
1762
1763    Argument : uint32_t
1764    Return   : entry_point_info *
1765
1766This function may execute with the MMU and data caches enabled if the platform
1767port does the necessary initializations in ``bl31_plat_arch_setup()``.
1768
1769This function is called by ``bl31_main()`` to retrieve information provided by
1770BL2 for the next image in the security state specified by the argument. BL31
1771uses this information to pass control to that image in the specified security
1772state. This function must return a pointer to the ``entry_point_info`` structure
1773(that was copied during ``bl31_early_platform_setup()``) if the image exists. It
1774should return NULL otherwise.
1775
1776Function : bl31_plat_enable_mmu [optional]
1777~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1778
1779::
1780
1781    Argument : uint32_t
1782    Return   : void
1783
1784This function enables the MMU. The boot code calls this function with MMU and
1785caches disabled. This function should program necessary registers to enable
1786translation, and upon return, the MMU on the calling PE must be enabled.
1787
1788The function must honor flags passed in the first argument. These flags are
1789defined by the translation library, and can be found in the file
1790``include/lib/xlat_tables/xlat_mmu_helpers.h``.
1791
1792On DynamIQ systems, this function must not use stack while enabling MMU, which
1793is how the function in xlat table library version 2 is implemented.
1794
1795Function : plat_init_apiakey [optional]
1796~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1797
1798::
1799
1800    Argument : void
1801    Return   : uint64_t *
1802
1803This function populates the ``plat_apiakey`` array that contains the values used
1804to set the ``APIAKey{Hi,Lo}_EL1`` registers. It returns a pointer to this array.
1805
1806The value should be obtained from a reliable source of randomness.
1807
1808This function is only needed if ARMv8.3 pointer authentication is used in the
1809Trusted Firmware by building with ``ENABLE_PAUTH=1``.
1810
1811Function : plat_get_syscnt_freq2() [mandatory]
1812~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1813
1814::
1815
1816    Argument : void
1817    Return   : unsigned int
1818
1819This function is used by the architecture setup code to retrieve the counter
1820frequency for the CPU's generic timer. This value will be programmed into the
1821``CNTFRQ_EL0`` register. In Arm standard platforms, it returns the base frequency
1822of the system counter, which is retrieved from the first entry in the frequency
1823modes table.
1824
1825#define : PLAT_PERCPU_BAKERY_LOCK_SIZE [optional]
1826~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1827
1828When ``USE_COHERENT_MEM = 0``, this constant defines the total memory (in
1829bytes) aligned to the cache line boundary that should be allocated per-cpu to
1830accommodate all the bakery locks.
1831
1832If this constant is not defined when ``USE_COHERENT_MEM = 0``, the linker
1833calculates the size of the ``bakery_lock`` input section, aligns it to the
1834nearest ``CACHE_WRITEBACK_GRANULE``, multiplies it with ``PLATFORM_CORE_COUNT``
1835and stores the result in a linker symbol. This constant prevents a platform
1836from relying on the linker and provide a more efficient mechanism for
1837accessing per-cpu bakery lock information.
1838
1839If this constant is defined and its value is not equal to the value
1840calculated by the linker then a link time assertion is raised. A compile time
1841assertion is raised if the value of the constant is not aligned to the cache
1842line boundary.
1843
1844SDEI porting requirements
1845~~~~~~~~~~~~~~~~~~~~~~~~~
1846
1847The SDEI dispatcher requires the platform to provide the following macros
1848and functions, of which some are optional, and some others mandatory.
1849
1850Macros
1851......
1852
1853Macro: PLAT_SDEI_NORMAL_PRI [mandatory]
1854^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1855
1856This macro must be defined to the EL3 exception priority level associated with
1857Normal SDEI events on the platform. This must have a higher value (therefore of
1858lower priority) than ``PLAT_SDEI_CRITICAL_PRI``.
1859
1860Macro: PLAT_SDEI_CRITICAL_PRI [mandatory]
1861^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1862
1863This macro must be defined to the EL3 exception priority level associated with
1864Critical SDEI events on the platform. This must have a lower value (therefore of
1865higher priority) than ``PLAT_SDEI_NORMAL_PRI``.
1866
1867**Note**: SDEI exception priorities must be the lowest among Secure priorities.
1868Among the SDEI exceptions, Critical SDEI priority must be higher than Normal
1869SDEI priority.
1870
1871Functions
1872.........
1873
1874Function: int plat_sdei_validate_entry_point(uintptr_t ep) [optional]
1875^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1876
1877::
1878
1879  Argument: uintptr_t
1880  Return: int
1881
1882This function validates the address of client entry points provided for both
1883event registration and *Complete and Resume* SDEI calls. The function takes one
1884argument, which is the address of the handler the SDEI client requested to
1885register. The function must return ``0`` for successful validation, or ``-1``
1886upon failure.
1887
1888The default implementation always returns ``0``. On Arm platforms, this function
1889is implemented to translate the entry point to physical address, and further to
1890ensure that the address is located in Non-secure DRAM.
1891
1892Function: void plat_sdei_handle_masked_trigger(uint64_t mpidr, unsigned int intr) [optional]
1893^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1894
1895::
1896
1897  Argument: uint64_t
1898  Argument: unsigned int
1899  Return: void
1900
1901SDEI specification requires that a PE comes out of reset with the events masked.
1902The client therefore is expected to call ``PE_UNMASK`` to unmask SDEI events on
1903the PE. No SDEI events can be dispatched until such time.
1904
1905Should a PE receive an interrupt that was bound to an SDEI event while the
1906events are masked on the PE, the dispatcher implementation invokes the function
1907``plat_sdei_handle_masked_trigger``. The MPIDR of the PE that received the
1908interrupt and the interrupt ID are passed as parameters.
1909
1910The default implementation only prints out a warning message.
1911
1912Power State Coordination Interface (in BL31)
1913--------------------------------------------
1914
1915The TF-A implementation of the PSCI API is based around the concept of a
1916*power domain*. A *power domain* is a CPU or a logical group of CPUs which
1917share some state on which power management operations can be performed as
1918specified by `PSCI`_. Each CPU in the system is assigned a cpu index which is
1919a unique number between ``0`` and ``PLATFORM_CORE_COUNT - 1``. The
1920*power domains* are arranged in a hierarchical tree structure and each
1921*power domain* can be identified in a system by the cpu index of any CPU that
1922is part of that domain and a *power domain level*. A processing element (for
1923example, a CPU) is at level 0. If the *power domain* node above a CPU is a
1924logical grouping of CPUs that share some state, then level 1 is that group of
1925CPUs (for example, a cluster), and level 2 is a group of clusters (for
1926example, the system). More details on the power domain topology and its
1927organization can be found in `Power Domain Topology Design`_.
1928
1929BL31's platform initialization code exports a pointer to the platform-specific
1930power management operations required for the PSCI implementation to function
1931correctly. This information is populated in the ``plat_psci_ops`` structure. The
1932PSCI implementation calls members of the ``plat_psci_ops`` structure for performing
1933power management operations on the power domains. For example, the target
1934CPU is specified by its ``MPIDR`` in a PSCI ``CPU_ON`` call. The ``pwr_domain_on()``
1935handler (if present) is called for the CPU power domain.
1936
1937The ``power-state`` parameter of a PSCI ``CPU_SUSPEND`` call can be used to
1938describe composite power states specific to a platform. The PSCI implementation
1939defines a generic representation of the power-state parameter, which is an
1940array of local power states where each index corresponds to a power domain
1941level. Each entry contains the local power state the power domain at that power
1942level could enter. It depends on the ``validate_power_state()`` handler to
1943convert the power-state parameter (possibly encoding a composite power state)
1944passed in a PSCI ``CPU_SUSPEND`` call to this representation.
1945
1946The following functions form part of platform port of PSCI functionality.
1947
1948Function : plat_psci_stat_accounting_start() [optional]
1949~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1950
1951::
1952
1953    Argument : const psci_power_state_t *
1954    Return   : void
1955
1956This is an optional hook that platforms can implement for residency statistics
1957accounting before entering a low power state. The ``pwr_domain_state`` field of
1958``state_info`` (first argument) can be inspected if stat accounting is done
1959differently at CPU level versus higher levels. As an example, if the element at
1960index 0 (CPU power level) in the ``pwr_domain_state`` array indicates a power down
1961state, special hardware logic may be programmed in order to keep track of the
1962residency statistics. For higher levels (array indices > 0), the residency
1963statistics could be tracked in software using PMF. If ``ENABLE_PMF`` is set, the
1964default implementation will use PMF to capture timestamps.
1965
1966Function : plat_psci_stat_accounting_stop() [optional]
1967~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1968
1969::
1970
1971    Argument : const psci_power_state_t *
1972    Return   : void
1973
1974This is an optional hook that platforms can implement for residency statistics
1975accounting after exiting from a low power state. The ``pwr_domain_state`` field
1976of ``state_info`` (first argument) can be inspected if stat accounting is done
1977differently at CPU level versus higher levels. As an example, if the element at
1978index 0 (CPU power level) in the ``pwr_domain_state`` array indicates a power down
1979state, special hardware logic may be programmed in order to keep track of the
1980residency statistics. For higher levels (array indices > 0), the residency
1981statistics could be tracked in software using PMF. If ``ENABLE_PMF`` is set, the
1982default implementation will use PMF to capture timestamps.
1983
1984Function : plat_psci_stat_get_residency() [optional]
1985~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1986
1987::
1988
1989    Argument : unsigned int, const psci_power_state_t *, int
1990    Return   : u_register_t
1991
1992This is an optional interface that is is invoked after resuming from a low power
1993state and provides the time spent resident in that low power state by the power
1994domain at a particular power domain level. When a CPU wakes up from suspend,
1995all its parent power domain levels are also woken up. The generic PSCI code
1996invokes this function for each parent power domain that is resumed and it
1997identified by the ``lvl`` (first argument) parameter. The ``state_info`` (second
1998argument) describes the low power state that the power domain has resumed from.
1999The current CPU is the first CPU in the power domain to resume from the low
2000power state and the ``last_cpu_idx`` (third parameter) is the index of the last
2001CPU in the power domain to suspend and may be needed to calculate the residency
2002for that power domain.
2003
2004Function : plat_get_target_pwr_state() [optional]
2005~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2006
2007::
2008
2009    Argument : unsigned int, const plat_local_state_t *, unsigned int
2010    Return   : plat_local_state_t
2011
2012The PSCI generic code uses this function to let the platform participate in
2013state coordination during a power management operation. The function is passed
2014a pointer to an array of platform specific local power state ``states`` (second
2015argument) which contains the requested power state for each CPU at a particular
2016power domain level ``lvl`` (first argument) within the power domain. The function
2017is expected to traverse this array of upto ``ncpus`` (third argument) and return
2018a coordinated target power state by the comparing all the requested power
2019states. The target power state should not be deeper than any of the requested
2020power states.
2021
2022A weak definition of this API is provided by default wherein it assumes
2023that the platform assigns a local state value in order of increasing depth
2024of the power state i.e. for two power states X & Y, if X < Y
2025then X represents a shallower power state than Y. As a result, the
2026coordinated target local power state for a power domain will be the minimum
2027of the requested local power state values.
2028
2029Function : plat_get_power_domain_tree_desc() [mandatory]
2030~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2031
2032::
2033
2034    Argument : void
2035    Return   : const unsigned char *
2036
2037This function returns a pointer to the byte array containing the power domain
2038topology tree description. The format and method to construct this array are
2039described in `Power Domain Topology Design`_. The BL31 PSCI initialization code
2040requires this array to be described by the platform, either statically or
2041dynamically, to initialize the power domain topology tree. In case the array
2042is populated dynamically, then plat_core_pos_by_mpidr() and
2043plat_my_core_pos() should also be implemented suitably so that the topology
2044tree description matches the CPU indices returned by these APIs. These APIs
2045together form the platform interface for the PSCI topology framework.
2046
2047Function : plat_setup_psci_ops() [mandatory]
2048~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2049
2050::
2051
2052    Argument : uintptr_t, const plat_psci_ops **
2053    Return   : int
2054
2055This function may execute with the MMU and data caches enabled if the platform
2056port does the necessary initializations in ``bl31_plat_arch_setup()``. It is only
2057called by the primary CPU.
2058
2059This function is called by PSCI initialization code. Its purpose is to let
2060the platform layer know about the warm boot entrypoint through the
2061``sec_entrypoint`` (first argument) and to export handler routines for
2062platform-specific psci power management actions by populating the passed
2063pointer with a pointer to BL31's private ``plat_psci_ops`` structure.
2064
2065A description of each member of this structure is given below. Please refer to
2066the Arm FVP specific implementation of these handlers in
2067`plat/arm/board/fvp/fvp_pm.c`_ as an example. For each PSCI function that the
2068platform wants to support, the associated operation or operations in this
2069structure must be provided and implemented (Refer section 4 of
2070`Firmware Design`_ for the PSCI API supported in TF-A). To disable a PSCI
2071function in a platform port, the operation should be removed from this
2072structure instead of providing an empty implementation.
2073
2074plat_psci_ops.cpu_standby()
2075...........................
2076
2077Perform the platform-specific actions to enter the standby state for a cpu
2078indicated by the passed argument. This provides a fast path for CPU standby
2079wherein overheads of PSCI state management and lock acquisition is avoided.
2080For this handler to be invoked by the PSCI ``CPU_SUSPEND`` API implementation,
2081the suspend state type specified in the ``power-state`` parameter should be
2082STANDBY and the target power domain level specified should be the CPU. The
2083handler should put the CPU into a low power retention state (usually by
2084issuing a wfi instruction) and ensure that it can be woken up from that
2085state by a normal interrupt. The generic code expects the handler to succeed.
2086
2087plat_psci_ops.pwr_domain_on()
2088.............................
2089
2090Perform the platform specific actions to power on a CPU, specified
2091by the ``MPIDR`` (first argument). The generic code expects the platform to
2092return PSCI_E_SUCCESS on success or PSCI_E_INTERN_FAIL for any failure.
2093
2094plat_psci_ops.pwr_domain_off()
2095..............................
2096
2097Perform the platform specific actions to prepare to power off the calling CPU
2098and its higher parent power domain levels as indicated by the ``target_state``
2099(first argument). It is called by the PSCI ``CPU_OFF`` API implementation.
2100
2101The ``target_state`` encodes the platform coordinated target local power states
2102for the CPU power domain and its parent power domain levels. The handler
2103needs to perform power management operation corresponding to the local state
2104at each power level.
2105
2106For this handler, the local power state for the CPU power domain will be a
2107power down state where as it could be either power down, retention or run state
2108for the higher power domain levels depending on the result of state
2109coordination. The generic code expects the handler to succeed.
2110
2111plat_psci_ops.pwr_domain_suspend_pwrdown_early() [optional]
2112...........................................................
2113
2114This optional function may be used as a performance optimization to replace
2115or complement pwr_domain_suspend() on some platforms. Its calling semantics
2116are identical to pwr_domain_suspend(), except the PSCI implementation only
2117calls this function when suspending to a power down state, and it guarantees
2118that data caches are enabled.
2119
2120When HW_ASSISTED_COHERENCY = 0, the PSCI implementation disables data caches
2121before calling pwr_domain_suspend(). If the target_state corresponds to a
2122power down state and it is safe to perform some or all of the platform
2123specific actions in that function with data caches enabled, it may be more
2124efficient to move those actions to this function. When HW_ASSISTED_COHERENCY
2125= 1, data caches remain enabled throughout, and so there is no advantage to
2126moving platform specific actions to this function.
2127
2128plat_psci_ops.pwr_domain_suspend()
2129..................................
2130
2131Perform the platform specific actions to prepare to suspend the calling
2132CPU and its higher parent power domain levels as indicated by the
2133``target_state`` (first argument). It is called by the PSCI ``CPU_SUSPEND``
2134API implementation.
2135
2136The ``target_state`` has a similar meaning as described in
2137the ``pwr_domain_off()`` operation. It encodes the platform coordinated
2138target local power states for the CPU power domain and its parent
2139power domain levels. The handler needs to perform power management operation
2140corresponding to the local state at each power level. The generic code
2141expects the handler to succeed.
2142
2143The difference between turning a power domain off versus suspending it is that
2144in the former case, the power domain is expected to re-initialize its state
2145when it is next powered on (see ``pwr_domain_on_finish()``). In the latter
2146case, the power domain is expected to save enough state so that it can resume
2147execution by restoring this state when its powered on (see
2148``pwr_domain_suspend_finish()``).
2149
2150When suspending a core, the platform can also choose to power off the GICv3
2151Redistributor and ITS through an implementation-defined sequence. To achieve
2152this safely, the ITS context must be saved first. The architectural part is
2153implemented by the ``gicv3_its_save_disable()`` helper, but most of the needed
2154sequence is implementation defined and it is therefore the responsibility of
2155the platform code to implement the necessary sequence. Then the GIC
2156Redistributor context can be saved using the ``gicv3_rdistif_save()`` helper.
2157Powering off the Redistributor requires the implementation to support it and it
2158is the responsibility of the platform code to execute the right implementation
2159defined sequence.
2160
2161When a system suspend is requested, the platform can also make use of the
2162``gicv3_distif_save()`` helper to save the context of the GIC Distributor after
2163it has saved the context of the Redistributors and ITS of all the cores in the
2164system. The context of the Distributor can be large and may require it to be
2165allocated in a special area if it cannot fit in the platform's global static
2166data, for example in DRAM. The Distributor can then be powered down using an
2167implementation-defined sequence.
2168
2169plat_psci_ops.pwr_domain_pwr_down_wfi()
2170.......................................
2171
2172This is an optional function and, if implemented, is expected to perform
2173platform specific actions including the ``wfi`` invocation which allows the
2174CPU to powerdown. Since this function is invoked outside the PSCI locks,
2175the actions performed in this hook must be local to the CPU or the platform
2176must ensure that races between multiple CPUs cannot occur.
2177
2178The ``target_state`` has a similar meaning as described in the ``pwr_domain_off()``
2179operation and it encodes the platform coordinated target local power states for
2180the CPU power domain and its parent power domain levels. This function must
2181not return back to the caller.
2182
2183If this function is not implemented by the platform, PSCI generic
2184implementation invokes ``psci_power_down_wfi()`` for power down.
2185
2186plat_psci_ops.pwr_domain_on_finish()
2187....................................
2188
2189This function is called by the PSCI implementation after the calling CPU is
2190powered on and released from reset in response to an earlier PSCI ``CPU_ON`` call.
2191It performs the platform-specific setup required to initialize enough state for
2192this CPU to enter the normal world and also provide secure runtime firmware
2193services.
2194
2195The ``target_state`` (first argument) is the prior state of the power domains
2196immediately before the CPU was turned on. It indicates which power domains
2197above the CPU might require initialization due to having previously been in
2198low power states. The generic code expects the handler to succeed.
2199
2200plat_psci_ops.pwr_domain_suspend_finish()
2201.........................................
2202
2203This function is called by the PSCI implementation after the calling CPU is
2204powered on and released from reset in response to an asynchronous wakeup
2205event, for example a timer interrupt that was programmed by the CPU during the
2206``CPU_SUSPEND`` call or ``SYSTEM_SUSPEND`` call. It performs the platform-specific
2207setup required to restore the saved state for this CPU to resume execution
2208in the normal world and also provide secure runtime firmware services.
2209
2210The ``target_state`` (first argument) has a similar meaning as described in
2211the ``pwr_domain_on_finish()`` operation. The generic code expects the platform
2212to succeed.
2213
2214If the Distributor, Redistributors or ITS have been powered off as part of a
2215suspend, their context must be restored in this function in the reverse order
2216to how they were saved during suspend sequence.
2217
2218plat_psci_ops.system_off()
2219..........................
2220
2221This function is called by PSCI implementation in response to a ``SYSTEM_OFF``
2222call. It performs the platform-specific system poweroff sequence after
2223notifying the Secure Payload Dispatcher.
2224
2225plat_psci_ops.system_reset()
2226............................
2227
2228This function is called by PSCI implementation in response to a ``SYSTEM_RESET``
2229call. It performs the platform-specific system reset sequence after
2230notifying the Secure Payload Dispatcher.
2231
2232plat_psci_ops.validate_power_state()
2233....................................
2234
2235This function is called by the PSCI implementation during the ``CPU_SUSPEND``
2236call to validate the ``power_state`` parameter of the PSCI API and if valid,
2237populate it in ``req_state`` (second argument) array as power domain level
2238specific local states. If the ``power_state`` is invalid, the platform must
2239return PSCI_E_INVALID_PARAMS as error, which is propagated back to the
2240normal world PSCI client.
2241
2242plat_psci_ops.validate_ns_entrypoint()
2243......................................
2244
2245This function is called by the PSCI implementation during the ``CPU_SUSPEND``,
2246``SYSTEM_SUSPEND`` and ``CPU_ON`` calls to validate the non-secure ``entry_point``
2247parameter passed by the normal world. If the ``entry_point`` is invalid,
2248the platform must return PSCI_E_INVALID_ADDRESS as error, which is
2249propagated back to the normal world PSCI client.
2250
2251plat_psci_ops.get_sys_suspend_power_state()
2252...........................................
2253
2254This function is called by the PSCI implementation during the ``SYSTEM_SUSPEND``
2255call to get the ``req_state`` parameter from platform which encodes the power
2256domain level specific local states to suspend to system affinity level. The
2257``req_state`` will be utilized to do the PSCI state coordination and
2258``pwr_domain_suspend()`` will be invoked with the coordinated target state to
2259enter system suspend.
2260
2261plat_psci_ops.get_pwr_lvl_state_idx()
2262.....................................
2263
2264This is an optional function and, if implemented, is invoked by the PSCI
2265implementation to convert the ``local_state`` (first argument) at a specified
2266``pwr_lvl`` (second argument) to an index between 0 and
2267``PLAT_MAX_PWR_LVL_STATES`` - 1. This function is only needed if the platform
2268supports more than two local power states at each power domain level, that is
2269``PLAT_MAX_PWR_LVL_STATES`` is greater than 2, and needs to account for these
2270local power states.
2271
2272plat_psci_ops.translate_power_state_by_mpidr()
2273..............................................
2274
2275This is an optional function and, if implemented, verifies the ``power_state``
2276(second argument) parameter of the PSCI API corresponding to a target power
2277domain. The target power domain is identified by using both ``MPIDR`` (first
2278argument) and the power domain level encoded in ``power_state``. The power domain
2279level specific local states are to be extracted from ``power_state`` and be
2280populated in the ``output_state`` (third argument) array. The functionality
2281is similar to the ``validate_power_state`` function described above and is
2282envisaged to be used in case the validity of ``power_state`` depend on the
2283targeted power domain. If the ``power_state`` is invalid for the targeted power
2284domain, the platform must return PSCI_E_INVALID_PARAMS as error. If this
2285function is not implemented, then the generic implementation relies on
2286``validate_power_state`` function to translate the ``power_state``.
2287
2288This function can also be used in case the platform wants to support local
2289power state encoding for ``power_state`` parameter of PSCI_STAT_COUNT/RESIDENCY
2290APIs as described in Section 5.18 of `PSCI`_.
2291
2292plat_psci_ops.get_node_hw_state()
2293.................................
2294
2295This is an optional function. If implemented this function is intended to return
2296the power state of a node (identified by the first parameter, the ``MPIDR``) in
2297the power domain topology (identified by the second parameter, ``power_level``),
2298as retrieved from a power controller or equivalent component on the platform.
2299Upon successful completion, the implementation must map and return the final
2300status among ``HW_ON``, ``HW_OFF`` or ``HW_STANDBY``. Upon encountering failures, it
2301must return either ``PSCI_E_INVALID_PARAMS`` or ``PSCI_E_NOT_SUPPORTED`` as
2302appropriate.
2303
2304Implementations are not expected to handle ``power_levels`` greater than
2305``PLAT_MAX_PWR_LVL``.
2306
2307plat_psci_ops.system_reset2()
2308.............................
2309
2310This is an optional function. If implemented this function is
2311called during the ``SYSTEM_RESET2`` call to perform a reset
2312based on the first parameter ``reset_type`` as specified in
2313`PSCI`_. The parameter ``cookie`` can be used to pass additional
2314reset information. If the ``reset_type`` is not supported, the
2315function must return ``PSCI_E_NOT_SUPPORTED``. For architectural
2316resets, all failures must return ``PSCI_E_INVALID_PARAMETERS``
2317and vendor reset can return other PSCI error codes as defined
2318in `PSCI`_. On success this function will not return.
2319
2320plat_psci_ops.write_mem_protect()
2321.................................
2322
2323This is an optional function. If implemented it enables or disables the
2324``MEM_PROTECT`` functionality based on the value of ``val``.
2325A non-zero value enables ``MEM_PROTECT`` and a value of zero
2326disables it. Upon encountering failures it must return a negative value
2327and on success it must return 0.
2328
2329plat_psci_ops.read_mem_protect()
2330................................
2331
2332This is an optional function. If implemented it returns the current
2333state of ``MEM_PROTECT`` via the ``val`` parameter.  Upon encountering
2334failures it must return a negative value and on success it must
2335return 0.
2336
2337plat_psci_ops.mem_protect_chk()
2338...............................
2339
2340This is an optional function. If implemented it checks if a memory
2341region defined by a base address ``base`` and with a size of ``length``
2342bytes is protected by ``MEM_PROTECT``.  If the region is protected
2343then it must return 0, otherwise it must return a negative number.
2344
2345Interrupt Management framework (in BL31)
2346----------------------------------------
2347
2348BL31 implements an Interrupt Management Framework (IMF) to manage interrupts
2349generated in either security state and targeted to EL1 or EL2 in the non-secure
2350state or EL3/S-EL1 in the secure state. The design of this framework is
2351described in the `IMF Design Guide`_
2352
2353A platform should export the following APIs to support the IMF. The following
2354text briefly describes each API and its implementation in Arm standard
2355platforms. The API implementation depends upon the type of interrupt controller
2356present in the platform. Arm standard platform layer supports both
2357`Arm Generic Interrupt Controller version 2.0 (GICv2)`_
2358and `3.0 (GICv3)`_. Juno builds the Arm platform layer to use GICv2 and the
2359FVP can be configured to use either GICv2 or GICv3 depending on the build flag
2360``FVP_USE_GIC_DRIVER`` (See FVP platform specific build options in
2361`User Guide`_ for more details).
2362
2363See also: `Interrupt Controller Abstraction APIs`__.
2364
2365.. __: platform-interrupt-controller-API.rst
2366
2367Function : plat_interrupt_type_to_line() [mandatory]
2368~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2369
2370::
2371
2372    Argument : uint32_t, uint32_t
2373    Return   : uint32_t
2374
2375The Arm processor signals an interrupt exception either through the IRQ or FIQ
2376interrupt line. The specific line that is signaled depends on how the interrupt
2377controller (IC) reports different interrupt types from an execution context in
2378either security state. The IMF uses this API to determine which interrupt line
2379the platform IC uses to signal each type of interrupt supported by the framework
2380from a given security state. This API must be invoked at EL3.
2381
2382The first parameter will be one of the ``INTR_TYPE_*`` values (see
2383`IMF Design Guide`_) indicating the target type of the interrupt, the second parameter is the
2384security state of the originating execution context. The return result is the
2385bit position in the ``SCR_EL3`` register of the respective interrupt trap: IRQ=1,
2386FIQ=2.
2387
2388In the case of Arm standard platforms using GICv2, S-EL1 interrupts are
2389configured as FIQs and Non-secure interrupts as IRQs from either security
2390state.
2391
2392In the case of Arm standard platforms using GICv3, the interrupt line to be
2393configured depends on the security state of the execution context when the
2394interrupt is signalled and are as follows:
2395
2396-  The S-EL1 interrupts are signaled as IRQ in S-EL0/1 context and as FIQ in
2397   NS-EL0/1/2 context.
2398-  The Non secure interrupts are signaled as FIQ in S-EL0/1 context and as IRQ
2399   in the NS-EL0/1/2 context.
2400-  The EL3 interrupts are signaled as FIQ in both S-EL0/1 and NS-EL0/1/2
2401   context.
2402
2403Function : plat_ic_get_pending_interrupt_type() [mandatory]
2404~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2405
2406::
2407
2408    Argument : void
2409    Return   : uint32_t
2410
2411This API returns the type of the highest priority pending interrupt at the
2412platform IC. The IMF uses the interrupt type to retrieve the corresponding
2413handler function. ``INTR_TYPE_INVAL`` is returned when there is no interrupt
2414pending. The valid interrupt types that can be returned are ``INTR_TYPE_EL3``,
2415``INTR_TYPE_S_EL1`` and ``INTR_TYPE_NS``. This API must be invoked at EL3.
2416
2417In the case of Arm standard platforms using GICv2, the *Highest Priority
2418Pending Interrupt Register* (``GICC_HPPIR``) is read to determine the id of
2419the pending interrupt. The type of interrupt depends upon the id value as
2420follows.
2421
2422#. id < 1022 is reported as a S-EL1 interrupt
2423#. id = 1022 is reported as a Non-secure interrupt.
2424#. id = 1023 is reported as an invalid interrupt type.
2425
2426In the case of Arm standard platforms using GICv3, the system register
2427``ICC_HPPIR0_EL1``, *Highest Priority Pending group 0 Interrupt Register*,
2428is read to determine the id of the pending interrupt. The type of interrupt
2429depends upon the id value as follows.
2430
2431#. id = ``PENDING_G1S_INTID`` (1020) is reported as a S-EL1 interrupt
2432#. id = ``PENDING_G1NS_INTID`` (1021) is reported as a Non-secure interrupt.
2433#. id = ``GIC_SPURIOUS_INTERRUPT`` (1023) is reported as an invalid interrupt type.
2434#. All other interrupt id's are reported as EL3 interrupt.
2435
2436Function : plat_ic_get_pending_interrupt_id() [mandatory]
2437~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2438
2439::
2440
2441    Argument : void
2442    Return   : uint32_t
2443
2444This API returns the id of the highest priority pending interrupt at the
2445platform IC. ``INTR_ID_UNAVAILABLE`` is returned when there is no interrupt
2446pending.
2447
2448In the case of Arm standard platforms using GICv2, the *Highest Priority
2449Pending Interrupt Register* (``GICC_HPPIR``) is read to determine the id of the
2450pending interrupt. The id that is returned by API depends upon the value of
2451the id read from the interrupt controller as follows.
2452
2453#. id < 1022. id is returned as is.
2454#. id = 1022. The *Aliased Highest Priority Pending Interrupt Register*
2455   (``GICC_AHPPIR``) is read to determine the id of the non-secure interrupt.
2456   This id is returned by the API.
2457#. id = 1023. ``INTR_ID_UNAVAILABLE`` is returned.
2458
2459In the case of Arm standard platforms using GICv3, if the API is invoked from
2460EL3, the system register ``ICC_HPPIR0_EL1``, *Highest Priority Pending Interrupt
2461group 0 Register*, is read to determine the id of the pending interrupt. The id
2462that is returned by API depends upon the value of the id read from the
2463interrupt controller as follows.
2464
2465#. id < ``PENDING_G1S_INTID`` (1020). id is returned as is.
2466#. id = ``PENDING_G1S_INTID`` (1020) or ``PENDING_G1NS_INTID`` (1021). The system
2467   register ``ICC_HPPIR1_EL1``, *Highest Priority Pending Interrupt group 1
2468   Register* is read to determine the id of the group 1 interrupt. This id
2469   is returned by the API as long as it is a valid interrupt id
2470#. If the id is any of the special interrupt identifiers,
2471   ``INTR_ID_UNAVAILABLE`` is returned.
2472
2473When the API invoked from S-EL1 for GICv3 systems, the id read from system
2474register ``ICC_HPPIR1_EL1``, *Highest Priority Pending group 1 Interrupt
2475Register*, is returned if is not equal to GIC_SPURIOUS_INTERRUPT (1023) else
2476``INTR_ID_UNAVAILABLE`` is returned.
2477
2478Function : plat_ic_acknowledge_interrupt() [mandatory]
2479~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2480
2481::
2482
2483    Argument : void
2484    Return   : uint32_t
2485
2486This API is used by the CPU to indicate to the platform IC that processing of
2487the highest pending interrupt has begun. It should return the raw, unmodified
2488value obtained from the interrupt controller when acknowledging an interrupt.
2489The actual interrupt number shall be extracted from this raw value using the API
2490`plat_ic_get_interrupt_id()`__.
2491
2492.. __: platform-interrupt-controller-API.rst#function-unsigned-int-plat-ic-get-interrupt-id-unsigned-int-raw-optional
2493
2494This function in Arm standard platforms using GICv2, reads the *Interrupt
2495Acknowledge Register* (``GICC_IAR``). This changes the state of the highest
2496priority pending interrupt from pending to active in the interrupt controller.
2497It returns the value read from the ``GICC_IAR``, unmodified.
2498
2499In the case of Arm standard platforms using GICv3, if the API is invoked
2500from EL3, the function reads the system register ``ICC_IAR0_EL1``, *Interrupt
2501Acknowledge Register group 0*. If the API is invoked from S-EL1, the function
2502reads the system register ``ICC_IAR1_EL1``, *Interrupt Acknowledge Register
2503group 1*. The read changes the state of the highest pending interrupt from
2504pending to active in the interrupt controller. The value read is returned
2505unmodified.
2506
2507The TSP uses this API to start processing of the secure physical timer
2508interrupt.
2509
2510Function : plat_ic_end_of_interrupt() [mandatory]
2511~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2512
2513::
2514
2515    Argument : uint32_t
2516    Return   : void
2517
2518This API is used by the CPU to indicate to the platform IC that processing of
2519the interrupt corresponding to the id (passed as the parameter) has
2520finished. The id should be the same as the id returned by the
2521``plat_ic_acknowledge_interrupt()`` API.
2522
2523Arm standard platforms write the id to the *End of Interrupt Register*
2524(``GICC_EOIR``) in case of GICv2, and to ``ICC_EOIR0_EL1`` or ``ICC_EOIR1_EL1``
2525system register in case of GICv3 depending on where the API is invoked from,
2526EL3 or S-EL1. This deactivates the corresponding interrupt in the interrupt
2527controller.
2528
2529The TSP uses this API to finish processing of the secure physical timer
2530interrupt.
2531
2532Function : plat_ic_get_interrupt_type() [mandatory]
2533~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2534
2535::
2536
2537    Argument : uint32_t
2538    Return   : uint32_t
2539
2540This API returns the type of the interrupt id passed as the parameter.
2541``INTR_TYPE_INVAL`` is returned if the id is invalid. If the id is valid, a valid
2542interrupt type (one of ``INTR_TYPE_EL3``, ``INTR_TYPE_S_EL1`` and ``INTR_TYPE_NS``) is
2543returned depending upon how the interrupt has been configured by the platform
2544IC. This API must be invoked at EL3.
2545
2546Arm standard platforms using GICv2 configures S-EL1 interrupts as Group0 interrupts
2547and Non-secure interrupts as Group1 interrupts. It reads the group value
2548corresponding to the interrupt id from the relevant *Interrupt Group Register*
2549(``GICD_IGROUPRn``). It uses the group value to determine the type of interrupt.
2550
2551In the case of Arm standard platforms using GICv3, both the *Interrupt Group
2552Register* (``GICD_IGROUPRn``) and *Interrupt Group Modifier Register*
2553(``GICD_IGRPMODRn``) is read to figure out whether the interrupt is configured
2554as Group 0 secure interrupt, Group 1 secure interrupt or Group 1 NS interrupt.
2555
2556Crash Reporting mechanism (in BL31)
2557-----------------------------------
2558
2559BL31 implements a crash reporting mechanism which prints the various registers
2560of the CPU to enable quick crash analysis and debugging. This mechanism relies
2561on the platform implementing ``plat_crash_console_init``,
2562``plat_crash_console_putc`` and ``plat_crash_console_flush``.
2563
2564The file ``plat/common/aarch64/crash_console_helpers.S`` contains sample
2565implementation of all of them. Platforms may include this file to their
2566makefiles in order to benefit from them. By default, they will cause the crash
2567output to be routed over the normal console infrastructure and get printed on
2568consoles configured to output in crash state. ``console_set_scope()`` can be
2569used to control whether a console is used for crash output.
2570NOTE: Platforms are responsible for making sure that they only mark consoles for
2571use in the crash scope that are able to support this, i.e. that are written in
2572assembly and conform with the register clobber rules for putc() (x0-x2, x16-x17)
2573and flush() (x0-x3, x16-x17) crash callbacks.
2574
2575In some cases (such as debugging very early crashes that happen before the
2576normal boot console can be set up), platforms may want to control crash output
2577more explicitly. These platforms may instead provide custom implementations for
2578these. They are executed outside of a C environment and without a stack. Many
2579console drivers provide functions named ``console_xxx_core_init/putc/flush``
2580that are designed to be used by these functions. See Arm platforms (like juno)
2581for an example of this.
2582
2583Function : plat_crash_console_init [mandatory]
2584~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2585
2586::
2587
2588    Argument : void
2589    Return   : int
2590
2591This API is used by the crash reporting mechanism to initialize the crash
2592console. It must only use the general purpose registers x0 through x7 to do the
2593initialization and returns 1 on success.
2594
2595Function : plat_crash_console_putc [mandatory]
2596~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2597
2598::
2599
2600    Argument : int
2601    Return   : int
2602
2603This API is used by the crash reporting mechanism to print a character on the
2604designated crash console. It must only use general purpose registers x1 and
2605x2 to do its work. The parameter and the return value are in general purpose
2606register x0.
2607
2608Function : plat_crash_console_flush [mandatory]
2609~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2610
2611::
2612
2613    Argument : void
2614    Return   : int
2615
2616This API is used by the crash reporting mechanism to force write of all buffered
2617data on the designated crash console. It should only use general purpose
2618registers x0 through x5 to do its work. The return value is 0 on successful
2619completion; otherwise the return value is -1.
2620
2621External Abort handling and RAS Support
2622---------------------------------------
2623
2624Function : plat_ea_handler
2625~~~~~~~~~~~~~~~~~~~~~~~~~~
2626
2627::
2628
2629    Argument : int
2630    Argument : uint64_t
2631    Argument : void *
2632    Argument : void *
2633    Argument : uint64_t
2634    Return   : void
2635
2636This function is invoked by the RAS framework for the platform to handle an
2637External Abort received at EL3. The intention of the function is to attempt to
2638resolve the cause of External Abort and return; if that's not possible, to
2639initiate orderly shutdown of the system.
2640
2641The first parameter (``int ea_reason``) indicates the reason for External Abort.
2642Its value is one of ``ERROR_EA_*`` constants defined in ``ea_handle.h``.
2643
2644The second parameter (``uint64_t syndrome``) is the respective syndrome
2645presented to EL3 after having received the External Abort. Depending on the
2646nature of the abort (as can be inferred from the ``ea_reason`` parameter), this
2647can be the content of either ``ESR_EL3`` or ``DISR_EL1``.
2648
2649The third parameter (``void *cookie``) is unused for now. The fourth parameter
2650(``void *handle``) is a pointer to the preempted context. The fifth parameter
2651(``uint64_t flags``) indicates the preempted security state. These parameters
2652are received from the top-level exception handler.
2653
2654If ``RAS_EXTENSION`` is set to ``1``, the default implementation of this
2655function iterates through RAS handlers registered by the platform. If any of the
2656RAS handlers resolve the External Abort, no further action is taken.
2657
2658If ``RAS_EXTENSION`` is set to ``0``, or if none of the platform RAS handlers
2659could resolve the External Abort, the default implementation prints an error
2660message, and panics.
2661
2662Function : plat_handle_uncontainable_ea
2663~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2664
2665::
2666
2667    Argument : int
2668    Argument : uint64_t
2669    Return   : void
2670
2671This function is invoked by the RAS framework when an External Abort of
2672Uncontainable type is received at EL3. Due to the critical nature of
2673Uncontainable errors, the intention of this function is to initiate orderly
2674shutdown of the system, and is not expected to return.
2675
2676This function must be implemented in assembly.
2677
2678The first and second parameters are the same as that of ``plat_ea_handler``.
2679
2680The default implementation of this function calls
2681``report_unhandled_exception``.
2682
2683Function : plat_handle_double_fault
2684~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2685
2686::
2687
2688    Argument : int
2689    Argument : uint64_t
2690    Return   : void
2691
2692This function is invoked by the RAS framework when another External Abort is
2693received at EL3 while one is already being handled. I.e., a call to
2694``plat_ea_handler`` is outstanding. Due to its critical nature, the intention of
2695this function is to initiate orderly shutdown of the system, and is not expected
2696recover or return.
2697
2698This function must be implemented in assembly.
2699
2700The first and second parameters are the same as that of ``plat_ea_handler``.
2701
2702The default implementation of this function calls
2703``report_unhandled_exception``.
2704
2705Function : plat_handle_el3_ea
2706~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2707
2708::
2709
2710    Return   : void
2711
2712This function is invoked when an External Abort is received while executing in
2713EL3. Due to its critical nature, the intention of this function is to initiate
2714orderly shutdown of the system, and is not expected recover or return.
2715
2716This function must be implemented in assembly.
2717
2718The default implementation of this function calls
2719``report_unhandled_exception``.
2720
2721Build flags
2722-----------
2723
2724There are some build flags which can be defined by the platform to control
2725inclusion or exclusion of certain BL stages from the FIP image. These flags
2726need to be defined in the platform makefile which will get included by the
2727build system.
2728
2729-  **NEED_BL33**
2730   By default, this flag is defined ``yes`` by the build system and ``BL33``
2731   build option should be supplied as a build option. The platform has the
2732   option of excluding the BL33 image in the ``fip`` image by defining this flag
2733   to ``no``. If any of the options ``EL3_PAYLOAD_BASE`` or ``PRELOADED_BL33_BASE``
2734   are used, this flag will be set to ``no`` automatically.
2735
2736C Library
2737---------
2738
2739To avoid subtle toolchain behavioral dependencies, the header files provided
2740by the compiler are not used. The software is built with the ``-nostdinc`` flag
2741to ensure no headers are included from the toolchain inadvertently. Instead the
2742required headers are included in the TF-A source tree. The library only
2743contains those C library definitions required by the local implementation. If
2744more functionality is required, the needed library functions will need to be
2745added to the local implementation.
2746
2747Some C headers have been obtained from `FreeBSD`_ and `SCC`_, while others have
2748been written specifically for TF-A. Fome implementation files have been obtained
2749from `FreeBSD`_, others have been written specifically for TF-A as well. The
2750files can be found in ``include/lib/libc`` and ``lib/libc``.
2751
2752SCC can be found in http://www.simple-cc.org/. A copy of the `FreeBSD`_ sources
2753can be obtained from http://github.com/freebsd/freebsd.
2754
2755Storage abstraction layer
2756-------------------------
2757
2758In order to improve platform independence and portability an storage abstraction
2759layer is used to load data from non-volatile platform storage.
2760
2761Each platform should register devices and their drivers via the Storage layer.
2762These drivers then need to be initialized by bootloader phases as
2763required in their respective ``blx_platform_setup()`` functions. Currently
2764storage access is only required by BL1 and BL2 phases. The ``load_image()``
2765function uses the storage layer to access non-volatile platform storage.
2766
2767It is mandatory to implement at least one storage driver. For the Arm
2768development platforms the Firmware Image Package (FIP) driver is provided as
2769the default means to load data from storage (see the "Firmware Image Package"
2770section in the `User Guide`_). The storage layer is described in the header file
2771``include/drivers/io/io_storage.h``. The implementation of the common library
2772is in ``drivers/io/io_storage.c`` and the driver files are located in
2773``drivers/io/``.
2774
2775Each IO driver must provide ``io_dev_*`` structures, as described in
2776``drivers/io/io_driver.h``. These are returned via a mandatory registration
2777function that is called on platform initialization. The semi-hosting driver
2778implementation in ``io_semihosting.c`` can be used as an example.
2779
2780The Storage layer provides mechanisms to initialize storage devices before
2781IO operations are called. The basic operations supported by the layer
2782include ``open()``, ``close()``, ``read()``, ``write()``, ``size()`` and ``seek()``.
2783Drivers do not have to implement all operations, but each platform must
2784provide at least one driver for a device capable of supporting generic
2785operations such as loading a bootloader image.
2786
2787The current implementation only allows for known images to be loaded by the
2788firmware. These images are specified by using their identifiers, as defined in
2789``include/plat/common/common_def.h`` (or a separate header file included from
2790there). The platform layer (``plat_get_image_source()``) then returns a reference
2791to a device and a driver-specific ``spec`` which will be understood by the driver
2792to allow access to the image data.
2793
2794The layer is designed in such a way that is it possible to chain drivers with
2795other drivers. For example, file-system drivers may be implemented on top of
2796physical block devices, both represented by IO devices with corresponding
2797drivers. In such a case, the file-system "binding" with the block device may
2798be deferred until the file-system device is initialised.
2799
2800The abstraction currently depends on structures being statically allocated
2801by the drivers and callers, as the system does not yet provide a means of
2802dynamically allocating memory. This may also have the affect of limiting the
2803amount of open resources per driver.
2804
2805--------------
2806
2807*Copyright (c) 2013-2019, Arm Limited and Contributors. All rights reserved.*
2808
2809.. _include/plat/common/platform.h: ../include/plat/common/platform.h
2810.. _include/plat/arm/common/plat_arm.h: ../include/plat/arm/common/plat_arm.h%5D
2811.. _User Guide: user-guide.rst
2812.. _include/plat/common/common_def.h: ../include/plat/common/common_def.h
2813.. _include/plat/arm/common/arm_def.h: ../include/plat/arm/common/arm_def.h
2814.. _plat/common/aarch64/platform_mp_stack.S: ../plat/common/aarch64/platform_mp_stack.S
2815.. _plat/common/aarch64/platform_up_stack.S: ../plat/common/aarch64/platform_up_stack.S
2816.. _For example, define the build flag in platform.mk: PLAT_PL061_MAX_GPIOS%20:=%20160
2817.. _Power Domain Topology Design: psci-pd-tree.rst
2818.. _include/common/bl_common.h: ../include/common/bl_common.h
2819.. _include/lib/aarch32/arch.h: ../include/lib/aarch32/arch.h
2820.. _Firmware Design: firmware-design.rst
2821.. _PSCI: http://infocenter.arm.com/help/topic/com.arm.doc.den0022c/DEN0022C_Power_State_Coordination_Interface.pdf
2822.. _plat/arm/board/fvp/fvp_pm.c: ../plat/arm/board/fvp/fvp_pm.c
2823.. _Platform compatibility policy: ./platform-compatibility-policy.rst
2824.. _IMF Design Guide: interrupt-framework-design.rst
2825.. _Arm Generic Interrupt Controller version 2.0 (GICv2): http://infocenter.arm.com/help/topic/com.arm.doc.ihi0048b/index.html
2826.. _3.0 (GICv3): http://infocenter.arm.com/help/topic/com.arm.doc.ihi0069b/index.html
2827.. _FreeBSD: https://www.freebsd.org
2828.. _SCC: http://www.simple-cc.org/
2829