This patch adds support to return SUCCESS if a pending interrupt is
detected during a CPU_SUSPEND call to a power down state. The check
is performed as late as possible without losing the ability to return
to the caller. This reduces the overhead incurred by a CPU in
undergoing a complete power cycle when a wakeup interrupt is already
pending.
FixesARM-Software/tf-issues#102
Change-Id: I1aff04a74b704a2f529734428030d1d10750fd4b
This patch allows the platform to validate the power_state and
entrypoint information from the normal world early on in PSCI
calls so that we can return the error safely. New optional
pm_ops hooks `validate_power_state` and `validate_ns_entrypoint`
are introduced to do this.
As a result of these changes, all the other pm_ops handlers except
the PSCI_ON handler are expected to be successful. Also, the PSCI
implementation will now assert if a PSCI API is invoked without the
corresponding pm_ops handler being registered by the platform.
NOTE : PLATFORM PORTS WILL BREAK ON MERGE OF THIS COMMIT. The
pm hooks have 2 additional optional callbacks and the return type
of the other hooks have changed.
FixesARM-Software/tf-issues#229
Change-Id: I036bc0cff2349187c7b8b687b9ee0620aa7e24dc
This patch adds support to save the "power state" parameter before the
affinity level specific handlers are called in a CPU_SUSPEND call.
This avoids the need to pass the power_state as a parameter to the
handlers and Secure Payload Dispatcher (SPD) suspend spd_pm_ops.
The power_state arguments in the spd_pm_ops operations are now reserved
and must not be used. The SPD can query the relevant power_state fields
by using the psci_get_suspend_afflvl() & psci_get_suspend_stateid() APIs.
NOTE: THIS PATCH WILL BREAK THE SPD_PM_OPS INTERFACE. HENCE THE SECURE
PAYLOAD DISPATCHERS WILL NEED TO BE REWORKED TO USE THE NEW INTERFACE.
Change-Id: I1293d7dc8cf29cfa6a086a009eee41bcbf2f238e
This patch replaces the internal psci_save_ns_entry() API with a
psci_get_ns_ep_info() API. The new function splits the work done by the
previous one such that it populates and returns an 'entry_point_info_t'
structure with the information to enter the normal world upon completion
of the CPU_SUSPEND or CPU_ON call. This information is used to populate
the non-secure context structure separately.
This allows the new internal API `psci_get_ns_ep_info` to return error
and enable the code to return safely.
Change-Id: Ifd87430a4a3168eac0ebac712f59c93cbad1b231
This patch moves the check for valid CPU state during
PSCI_CPU_ON to before the non secure entry point is programmed
so as to enable it to return early on error.
Change-Id: I1b1a21be421e2b2a6e33db236e91dee8688efffa
This patch removes the non-secure entry point information being passed
to the platform pm_ops which is not needed. Also, it removes the `mpidr`
parameter for platform pm hooks which are meant to do power management
operations only on the current cpu.
NOTE: PLATFORM PORTS MUST BE UPDATED AFTER MERGING THIS COMMIT.
Change-Id: If632376a990b7f3b355f910e78771884bf6b12e7
This patch extends the build option `USE_COHERENT_MEMORY` to
conditionally remove coherent memory from the memory maps of
all boot loader stages. The patch also adds necessary
documentation for coherent memory removal in firmware-design,
porting and user guides.
FixesARM-Software/tf-issues#106
Change-Id: I260e8768c6a5c2efc402f5804a80657d8ce38773
This patch moves the bakery locks out of coherent memory to normal memory.
This implies that the lock information needs to be placed on a separate cache
line for each cpu. Hence the bakery_lock_info_t structure is allocated in the
per-cpu data so as to minimize memory wastage. A similar platform per-cpu
data is introduced for the platform locks.
As a result of the above changes, the bakery lock api is completely changed.
Earlier, a reference to the lock structure was passed to the lock implementation.
Now a unique-id (essentially an index into the per-cpu data array) and an offset
into the per-cpu data for bakery_info_t needs to be passed to the lock
implementation.
Change-Id: I1e76216277448713c6c98b4c2de4fb54198b39e0
This patch adds helper macros for barrier operations that specify
the type of barrier (dmb, dsb) and the shareability domain (system,
inner-shareable) it affects.
Change-Id: I4bf95103e79da212c4fbdbc13d91ad8ac385d9f5
This patch is an optimization in the bakery_lock_get() function
which removes the wfe() when waiting for other contenders to choose
their ticket i.e when their `entering` flag is set. Since the time
taken to execute bakery_get_ticket() by other contenders is bounded,
this wait is a bounded time wait. Hence the removal of wfe() and the
corresponding sev() and dsb() in bakery_get_ticket() may result
in better time performance during lock acquisition.
Change-Id: I141bb21294226b54cb6e89e7cac0175c553afd8d
This patch maps the non-secure region of DRAM1 in the MMU. The
non-secure region comprises the whole DRAM1 (0x80000000 -
0xffffffff) excluding the top 16 MB (0xff000000 - 0xffffffff).
The TrustZone controller configures this 16 MB region as secure
memory, so it can not be accessed in non-secure mode.
The number of MMU tables in BL2 has been increased to 3 because
the new size of the non-secure region in DRAM requires an extra L2
table.
Change-Id: I5ad080c6e181f6b6060e15cebb1d18b7fa128cf5
This patch fixes a crash due to corruption of cpu_ops
data structure. During the secondary CPU boot, after the
cpu_ops has been initialized in the per cpu-data, the
dcache lines need to invalidated so that the update in
memory can be seen later on when the dcaches are turned ON.
Also, after initializing the psci per cpu data, the dcache
lines are flushed so that they are written back to memory
and dirty dcache lines are avoided.
FixesARM-Software/tf-issues#271
Change-Id: Ia90f55e9882690ead61226eea5a5a9146d35f313
Patch 20d51ca moved the shared data page from the top of the
Trusted SRAM to the bottom, changing the load addresses of BL3-1
and BL3-2.
This patch updates BL3-1 and BL3-2 addresses in the instructions
to run the Trusted Firmware on FVP using BL3-1 as reset vector.
This patch is similar to but distinct from bfb1dd5 and 7ea4c43.
Change-Id: I6b467f9a82360a5e2181db99fea881487de52704
This patch allows the BL3-1 NOBITS section to overlap the BL1 R/W
section since the former will always be used after the latter.
Similarly, the BL3-2 NOBITS section can overlay the BL2 image
when BL3-2 is loaded in Trusted SRAM.
Due to the current size of the images, there is no actual overlap.
Nevertheless, this reorganization may help to optimise the Trusted
SRAM usage when the images size grows.
Note that because BL3-1 NOBITS section is allowed to overlap the
BL1 R/W section, BL1 global variables will remain valid only until
execution reaches the BL3-1 entry point during a cold boot.
Documentation updated accordingly.
FixesARM-software/tf-issues#254
Change-Id: Id538f4d1c7f1f7858108280fd7b97e138572b879
Previously, the User Guide recommended launching the Foundation
FVP with the parameter --no-secure-memory, which disabled security
control of the address map. This was due to missing support for
secure memory regions in v1 of the Foundation FVP. This is no longer
needed as secure memory is now supported on the Foundation FVP.
This patch updates the User Guide to recommend enabling secure
memory instead.
Change-Id: Ifae53c10ff6e1c7c6724af20e05a3d3a88f6a5ad
This patch initializes the version field in the bl30_image_info
structure when loading BL30. This initialization must be done before
calling load_image().
FixesARM-software/tf-issues#274
Change-Id: I74a05167d66fff51d257ad611abc7b5436e5d912
Calls to system register read accessors functions may be optimised
out by the compiler if called twice in a row for the same register.
This is because the compiler is not aware that the result from
the instruction may be modified by external agents. Therefore, if
nothing modifies the register between the 2 reads as far as the
compiler knows then it might consider that it is useless to read
it twice and emit only 1 call.
This behaviour is faulty for registers that may not have the same
value if read twice in succession. E.g.: counters, timer
control/countdown registers, GICv3 interrupt status registers and
so on.
The same problem happens for calls to system register write
accessors functions. The compiler might optimise out some calls
if it considers that it will produce the same result. Again, this
behaviour is faulty for cases where intermediate writes to these
registers make a difference in the system.
This patch fixes the problem by making these assembly register
accesses volatile.
FixesARM-software/tf-issues#273
Change-Id: I33903bc4cc4eea8a8d87bc2c757909fbb0138925
This patch allows to define the name of the FIP at build time by
defining the FIP_NAME variable. If FIP_NAME is not defined, default
name 'fip.bin' is used.
Documentation updated accordingly.
Change-Id: Ic41f42aac379b0c958b3dfd02863ba8ba7108710
This patch fixes the assertion failure when CPU_SUSPEND is invoked with
an affinity level higher than supported by the platform by adding suitable
checks for affinity level within `psci_cpu_suspend`. Also added suitable
bound checks within `psci_aff_map_get_idx` to prevent indexing beyond array
limits.
FixesARM-software/tf-issues#260
Change-Id: I04b75c49729e6c6d1983add590f60146c8fc3630
If one of the files passed to fip_create on the command line doesn't
exist, it will print an error message but produce an incomplete
fip.bin file and report success. This behaviour could potentially
hide errors made in the command line arguments.
This patch addresses the issue by having the tool bail out if one of
the supplied files can't be processed.
Signed-off-by: Kévin Petit <kevin.petit@arm.com>
FixesARM-software/tf-issues#279
Change-Id: I1c7d87d09eb4c063005b7969bdaad1d043c29dec
In both FVP and Juno ports, IRQ #16, which is a PPI, is incorrectly
identified as secure SGI #8 through the constant IRQ_SEC_SGI_8.
This patch removes it.
FixesARM-software/tf-issues#282
Change-Id: I9e52d849611ffcd2b1f28e56dd156c5b217ed63e
This patch fixes the array size of mpidr_aff_map_nodes_t which
was less by one element.
FixesARM-software/tf-issues#264
Change-Id: I48264f6f9e7046a3d0f4cbcd63b9ba49657e8818
The 'fiptool' target doesn't depend on fip_create's source files,
neither directly nor indirectly. As a result, the FIP tool is not
rebuilt whenever its source files change.
This patch makes the ${FIPTOOL} target into a phony target so that the
FIP tool's sub-makefile is always called. The sub-makefile correctly
handles the dependencies. It also moves the completion message into
the sub-makefile so that it is only displayed when the tool is
actually recompiled.
FixesARM-software/tf-issues#278
Change-Id: Ia027519fe51d3c42be30665d1ad20a7b89fa350f
This patch fixes a bug in the bakery lock implementation where a data
synchronisation barrier instruction is not issued before sending an event as
mandated by the ARMv8 ARM. This can cause a event to be signalled before the
related memory accesses have completed resulting in erroneous execution.
FixesARM-software/tf-issues#272
Change-Id: I5ce02bf70afb001d967b9fa4c3f77442931d5349
This patch reassigns the crash console on Juno and FVP to use the runtime
BL3-1 console. The crash console is changed to SoC UART0 (UART2) from the
previous FPGA UART0 (UART0) on Juno. In FVP, it is changed from UART0 to
UART1.
FixesARM-software/tf-issues#256
Change-Id: I7df54f86ca00ec2652c27261dd66a94c12610816
This patch replaces the usage of the GIC private driver in Juno with
the generic ARM GIC driver. The private driver is no longer necessary
and has been removed from the Juno port.
FixesARM-software/tf-issues#253
Change-Id: I6aaabc252e5e6fb5fcf44ab6d0febd9b38791056
This patch introduces several improvements to the ARM GIC driver:
* In function gicd_set_itargetsr(), target CPU is specified using
the same bit mask detailed in the GICD_ITARGETSRn register instead
of the CPU linear ID, removing the dependency between bit position
and linear ID in the platform porting. The current CPU bit mask
may be obtained by reading GICD_ITARGETSR0.
* PPIs and SGIs are initialized in arm_gic_pcpu_distif_setup().
SPIs are initialized in arm_gic_distif_setup().
* By default, non secure interrupts are assigned the maximum
priority allowed to a non secure interrupt (defined by
GIC_HIGHEST_NS_PRIORITY).
* GICR base address is allowed to be NULL for GICv1 and GICv2.
Change-Id: Ie2837fe860d43b2282e582dfdb13c39c6186f232
This patch optimizes the Cortex-A57 cluster power down sequence by not
flushing the Level1 data cache. The L1 data cache and the L2 unified
cache are inclusive. A flush of the L2 by set/way flushes any dirty
lines from the L1 as well. This is a known safe deviation from the
Cortex-A57 TRM defined power down sequence. This optimization can be
enabled by the platform through the 'SKIP_A57_L1_FLUSH_PWR_DWN' build
flag. Each Cortex-A57 based platform must make its own decision on
whether to use the optimization.
This patch also renames the cpu-errata-workarounds.md to
cpu-specific-build-macros.md as this facilitates documentation
of both CPU Specific errata and CPU Specific Optimization
build macros.
Change-Id: I299b9fe79e9a7e08e8a0dffb7d345f9a00a71480
This the patch replaces the DSB SY with DSB ISH
after disabling L2 prefetches during the Cortex-A57
power down sequence.
Change-Id: I048d12d830c1b974b161224eff079fb9f8ecf52d
Prior to this patch, the errata workarounds were applied for any version
of the CPU in the release build and in the debug build an assert
failure resulted when the revision did not match. This patch applies
errata workarounds in the Cortex-A57 reset handler only if the 'variant'
and 'revision' fields read from the MIDR_EL1 match. In the debug build,
a warning message is printed for each errata workaround which is not
applied.
The patch modifies the register usage in 'reset_handler` so
as to adhere to ARM procedure calling standards.
FixesARM-software/tf-issues#242
Change-Id: I51b1f876474599db885afa03346e38a476f84c29
This patch adds level specific cache maintenance functions
to cache_helpers.S. The new functions 'dcsw_op_levelx',
where '1 <= x <= 3', allow to perform cache maintenance by
set/way for that particular level of cache. With this patch,
functions to support cache maintenance upto level 3 have
been implemented since it is the highest cache level for
most ARM SoCs.
These functions are now utilized in CPU specific power down
sequences to implement them as mandated by processor specific
technical reference manual.
Change-Id: Icd90ce6b51cff5a12863bcda01b93601417fd45c
Soby Mathew
danh-arm
Juan Castillo
Sandrine Bailleux
Joakim Bech
Kévin Petit
Achin Gupta