Add PMU support for generic platform. Generic platform solely relies on
the device tree to parse all pmu related information. If any event is
not described in device tree, generic platform will not support it.
Reviewed-by: Anup Patel <anup.patel@wdc.com>
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Follow the standard conventon for static function names:
All global functions should be start with sbi_<module name>_
All static functions should be start with <module name>_
Reviewed-by: Anup Patel <anup.patel@wdc.com>
Signed-off-by: Atish Patra <atish.patra@wdc.com>
RISC-V SBI v0.3 specification defines a set of firmware events that can
provide additional information about the current firmware context. All
of the firmware event monitoring are enabled now. The firmware
events must be defined as raw perf event with MSB set as specified in the
specification.
Reviewed-by: Anup Patel <anup.patel@wdc.com>
Signed-off-by: Atish Patra <atish.patra@wdc.com>
RISC-V SBI specfication 0.3 defines a PMU extension that allows supervisor
mode to start/stop/configure pmu related events. This patch implements
all of the functionality defined in the specification.
Reviewed-by: Anup Patel <anup.patel@wdc.com>
Signed-off-by: Atish Patra <atish.patra@wdc.com>
The PMU DT node bindings are defined in docs/pmu_support.md
Add few fdt helper functions to parse the DT node and update the
event-counter mapping tables.
Reviewed-by: Anup Patel <anup.patel@wdc.com>
Signed-off-by: Atish Patra <atish.patra@wdc.com>
RISC-V SBI v0.3 specification defined a PMU extension to configure/start/stop
the hardware/firmware pmu events.
Implement PMU support in OpenSBI library. The implementation is agnostic of
event to counter mapping & mhpmevent value configuration. That means, it
expects platform hooks will be used to set up the mapping and provide
the mhpmevent value at runtime.
Reviewed-by: Anup Patel <anup.patel@wdc.com>
Signed-off-by: Atish Patra <atish.patra@wdc.com>
A platform hook to initialize PMU allows platform vendors to provide
their own mechanism to define pmu event-counter mappings in addition
to the DT based approach.
Another platform hook that allows platform vendors customize the
final mhpmevent value configuration.
Reviewed-by: Anup Patel <anup.patel@wdc.com>
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Currently, csr_read/write_num functions are used to read/write PMP related
CSRs where CSR value is decided at runtime. Expand this function to include
PMU related CSRs as well.
Reviewed-by: Anup Patel <anup.patel@wdc.com>
Reviewed-by: Xiang W <wxjstz@126.com>
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Currently, all bits in mcountern are enabled unconditionally at boot time.
With SBI PMU extension, all the programmable counters should enabled only
during performance monitoring for a particular event. However, this is done
only if mcountinhibit is implemented because the supervisor mode can not
start/stop any event without mcountinhibit.
Similarly, supervisor should take care enabling scounteren which allows
U-mode to access programmable pmu counters. All the non-programmable ones
(CY, TM, IR) should be enabled in M-mode because some userspace may rely on
builtins such as __builtin_readcyclecounter. Supervisor OS can still disable
them during initial configuration.
Reviewed-by: Anup Patel <anup.patel@wdc.com>
Signed-off-by: Atish Patra <atish.patra@wdc.com>
RISC-V privilege specification allows the implementation to have less
than 64 bits.
Add a function to detect the number of implemented bits in mhpmcounter
dynamically at runtime.
Reviewed-by: Xiang W <wxjstz@126.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
Signed-off-by: Atish Patra <atish.patra@wdc.com>
RISC-V ISA specification v1.11 defined mcountinhibit CSR that allows
software to stop any counter from incrementing. The SBI PMU extension
depends on this CSR support in hardware.
Define mcountinhibit as a hart specific feature and detect it at runtime.
Reviewed-by: Anup Patel <anup.patel@wdc.com>
Signed-off-by: Atish Patra <atish.patra@wdc.com>
SBI PMU extension requires a firmware to be aware of the event to
counter/mhpmevent mappings supported by the hardware. One approach
is to encode that information in the device tree.
Define a device tree binding that allows a hardware to describe
all the PMU mappings required in concise format.
Reviewed-by: Anup Patel <anup.patel@wdc.com>
Signed-off-by: Atish Patra <atish.patra@wdc.com>
We know about pmp granularity and number of bits supported by PMP.
Show those information in the boot time info print
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
As per RISC-V privilege specification, a platform may choose to implement
a coarser granularity scheme for PMP addresses. In that case, we shouldn't
allow any pmp region size smaller than the platform supports. A platform
may not also implement all the bits for a PMP address specified in the priv
specification.
The pmp range granularity and address bits should be detected dynamically
before detecing PMP regions. Any pmp modification request beyond these detected
value must not succeed.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
The "reg" property in a device node may not be the correct address always.
If a parent node defines a "ranges" property, the child address need to be
translated with respect to parents address. If the ranges property is not
present, it will just use 1:1 translation.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
U-Boot now supports loading Linux kernel image via network and storage
media. Thus, we don't need to use a combined payload containing both
U-Boot & Linux kernel image to boot Linux from U-Boot prompt.
Remove the old documentation.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
The doxygen config option "USE_MDFILE_AS_MAINPAGE" can be used to set
the main page in doxygen generated pdf. This allows us to remove the
"#mainpage" from the README file which markdown doesn't parse.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
Currently, unprivileged access functions uses few additional instructions
which are redundant. It also uses specific registers which are not necessary.
Remove those additional instructions and let the compiler choose the
registers.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
We have now clear distinction between platform and hart features.
Modify the boot print messages to print hart specific features in
a string format.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Tested-by: Jonathan Balkind <jbalkind@cs.princeton.edu>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
We have now clear distinction between platform and hart features.
Modify the boot print messages to print platform features in a string
format. In the process, this patch moved relatively larger functions
to its own file from platform.h header file.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Tested-by: Jonathan Balkind <jbalkind@cs.princeton.edu>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
As per the RISC-V specification, time value can be obtained from a time
CSR implemented in hardware or a MMIO based IP block. Qemu virt machine
already supports timer csr while CLINT provides the timer for other
platforms.
Implement a hart specific timer feature that can be detected at runtime.
As CSR based timer implementation are faster than MMIO address based, it
is always preferred over MMIO based one.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Tested-by: Jonathan Balkind <jbalkind@cs.princeton.edu>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
Any platform feature that belongs to a hart, have already been moved to
hart features and are detected at run time. The remaining platform features
are identical to platform default features.
Use the platform default features instead of defining a separate copy of it.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Tested-by: Jonathan Balkind <jbalkind@cs.princeton.edu>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
PMP & performance counters belong to a hart rather than a platform.
In addition to that, these features enable reading/writing from a
particular csr. Thus, they can be detected and set at runtime rather
than compile time.
Move PMP/MCOUNTEREN/SCOUNTEREN features to hart and detect them at runtime.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Tested-by: Jonathan Balkind <jbalkind@cs.princeton.edu>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
There may be some features which are hart specific rather than a platform
specific feature. Add a framework to support that.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Tested-by: Jonathan Balkind <jbalkind@cs.princeton.edu>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
As per RISC-V ISA, pmp is not mandatory. Currently, we only add reserved
memory node in DT only if PMP is present. That allows supervisor to access
the memory where OpenSBI continue to exist without realizing it for non-pmp
based platforms. It may result in corrupting OpenSBI. That's why OpenSBI
should at least let the supervisor know where it continue to exist.
This a best effort service provided by OpenSBI expecting that supervisor
software is not buggy and properly sets up its memory after parsing the
reserved-memory device tree node.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
There may be few common mathematics helper functions which can be used
anywhere in OpenSBI project.
Add a separate math helper function file to add these functions.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
As RISC-V ISA allows many CSRs such as pmp, s/mcounteren to be optional
in hardware, OpenSBI should provide an option to dynamically detect
these csr access capability at run time.
Implement a csr read/write access check helper macros.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Tested-by: Jonathan Balkind <jbalkind@cs.princeton.edu>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
Unprivileged trap handler can be reused for any cases where the executing
code expects a trap.
Rename it to "expected" trap handler as it will be used in other cases in
future.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Tested-by: Jonathan Balkind <jbalkind@cs.princeton.edu>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
Enable vector context in mstatus by updating the corresponding bits
in mstatus if vector extension is supported by the hart.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
As per the PLIC specification, maximum priority threshold value is 0x7.
Even though, writing a higher value doesn't cause any error in qemu
hifive unleashed, there may be some implementation which checks the upper
and may result in an illegal access.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
OpenPiton is designed to run on different FPGAs with different
configurations. Use the fdt parser to retrieve the required values
from device tree instead of using fixed values.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Tested-by: Jonathan Balkind <jbalkind@cs.princeton.edu>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
FDT helper file contain both fdt fixup and parsing functions.
Split the fixup related functions to a separate file for a better code
organization.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
Different DT based platforms from the sam family may reuse IP blocks with
different configurations. These different configurations can be obtained
by parsing the device tree.
Add a FDT parser framework that can parse various device configurations from
device tree. Currently, the parsing algorithms doesn't cover all the use cases
or possible combination of DT configurations. It will be improved over time.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
OpenPiton is a research platform from Princeton University [1].
"OpenPiton is the world's first open source, general purpose,
multithreaded manycore processor. It is a tiled manycore
framework scalable from one to 1/2 billion cores."
Add OpenSBI support for OpenPiton. As it is based on ariane core,
it reuses the platform code from arine project.
[1]. https://github.com/PrincetonUniversity/openpiton
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
All the common PLIC specific macros are already defined in plic.c.
Remove it from platform code. While at it, Fix the other coding style
issues.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
Commit 71d2b837c4 (lib: Move all coldboot wait APIs to sbi_init.c)
caused a regression while moving the code from sbi_hart.c to sbi_init.c.
As per original commit text, WFI can be implemented as a NOP according
to the RISC-V privilege specification. Software should ensure that
relevant interrupt pending bits are set. Otherwise, loop back to WFI.
Fix the regression by applying the original patch to sbi_init.c.
Fixes: 71d2b837c4 ("lib: Move all coldboot wait APIs to sbi_init.c")
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
Upstream U-Boot now have SMP support and doesn't require any additional
patches for HiFive Unleashed.
Update the documentation.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
As per the RISC-V privilege specification, WFI can be implemented as
a NOP. Software should ensure that relevant interrupt pending bits
are set. Otherwise, loop back to WFI.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
SBI HSM extension defines possible hart status values in the
specification.
Define all possible status values. Add a helper function to
convert hart state to status because hart states are internal
to OpenSBI only and may not match the status values defined in
the specification.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
This patch adds support for hart hotplug in OpenSBI using a generic WFI
based approach. Hart hotplug can be achieved via SBI HSM extension which
allows supervisor mode software to start or stop any harts anytime.
Any platform wishes to implement platform specific hart hotplug must
implement both hart_start and hart_stop in addition to enable platform
feature SBI_PLATFORM_HAS_HART_HOTPLUG.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
A platform may have a specific method to bring-up secondary harts for the
first time but may rely on generic WFI based approach for hart hotplug in
absence of a platform specific hart hotplug method.
Define a platform feature flag for such platforms. The platform needs to
implement platform specific bring-up method in hart_start and not define
hart_stop method in this case. They must only define
SBI_PLATFORM_HAS_HART_SECONDARY_BOOT.
SBI_PLATFORM_HAS_HART_HOTPLUG should only be defined when the platform
intend to support both hart_start and hart_stop and do not intend to rely
on generic WFI based approach.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
As per the latest SBI specification, all online harts should receive
IPI if hbase is set to -1.
Set the target mask to all available hart mask if hbase is -1.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
There is a typo in atomic operations code which prevents the
usage of riscv atomic instructions even if it is supported.
Fix the typo.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
To initialize PLIC in sane state, we should:
1. set maximum threshold value of M-mode PLIC contexts
2. set maximum threshold value of S-mode PLIC contexts
3. set irq priorities to miniumum
Fix the comment and initialize the threshold/priorities correctly.
Signed-off-by: Anup Patel <anup.patel@wdc.com>
Signed-off-by: Atish Patra <atish.patra@wdc.com>
As per the SBI specification, the return value in sbiret is undefined
if not explicitly described in the function. However, supervisor may
check this value by mistake and get a garbage value.
Initialize it to zero to avoid nasty supervisor bugs.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
It was reported that tlb range flush is not working on fu540.
Only tlb full flush seems to work on fu540 probably due to some
hardware errata.
Set the tlb flush limit to zero so that all tlb flush requests
are converted to full flush.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
If a platform requires to perform a tlb full flush, they should set
the tlb_range_flush_limit value to zero. However, corresponding platform
API ignore the value and continue to return the default value.
Add a platform ops to retrieve platform specific tlb range flush limit.
The platform variable becomes redundant in presence of the platform ops.
Take this opportunity to remove the variable as well.
The default is still set to smallest page size in RISC-V (4KB), as there
is no way to figure out a best value for all platforms. Individual platform
should set it to the optimal value for their platform.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
The hypervisor specification support hfence calls which can be used
issue tlb flush requests at both level of address translation. Currently,
these requests are issued only via SBI which are defined in v0.2.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
Currently, the toolchains do not have support for hfence instruction.
Hence, the instruction are hardcode until we have toolchain support.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
This patch adds new IPI extension which replaces ipi related
v0.1 extensions. This also adds a new API for ipi sending as trap
handling is not necessary in v0.2 SBI IPI related extensions.
It also modifies the IPI sending code which now accepts hart mask as a value
instead of S-mode virtual address. Thus, the caller should set it to exact hart
mask value everytime.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
This patch adds support for TIME extension which replaces v0.1
timer extension.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
We just need to distinguish only between FENCE and non FENCE related
IPIs as all of the fence related requests are handled via fifo now.
Remove the unnecessary IPI types related to individual fence types.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
Take this opportunity to move the enums to macros as enums make
sbi_ecall_interface.h unusable in assembly files.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
SBI v0.2 specification allows vendor extensions and it should be
implemented in a independent of the core sbi library.
Introduce a single platform callback that will let platforms handle
all vendor extensions in platform specific code if they want.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
SBI v0.2 introduces a base specification which is mandatory to
implement for any SBI implementations that is not legacy.
Add support for the base extension.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
An ecall handler should only return error if valid SBI function
fails. Otherwise, it should succeed with appropriate error in a0.
Get rid of unnecessary setting of the temporary return variable to
zero for the cases where errors are not expected.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
Current SBI implementation is now considered as version 0.1 and will be
removed/replaced with newer extension/functions in future.
Rename the existing implementations accordingly to be in sync with the
specification.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
Reviewed-by: Zong Li <zong.li@sifive.com>
Test payload uses an SBI call and uses the macros defined in interface
header which is not the correct place to have these definitions.
The interface header file should be used to keep SBI specification
related macros.
Keep all the test payload related code in test itself.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
Use platform defined flush range limit value only if it is non-zero.
Otherwise, use the default value.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Currently, the tlb range flush threshold is fixed and set to 4k for
all platforms. However, it should be platform specific as it completely
depends upon how platform actually implements sfence instruction.
Define a platform feature that allows every individual platform to set
different values. If a platform doesn't define it, just use a page size as
the threshold.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
In RISC-V, tlb flush happens at a page granularity. That's why OpenSBI
also have a tlb range flush limit which decides the which tlb flush
requests should be upgraded to full flush to avoid long delays.
Currently, this is set to 1G which would result in a many sfence.vma
execution in a tight loop for a large range.
Change the threshold to 4k to speed things up.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
U-Boot readme for fu540 platform suggest that fdt_addr_r should be used
as DT address after DT is copied via tftpboot.
Update the OpenSBI docs to reflect that. Remove other stale informations
as well.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
Currently, we upgrade to a full tlb flush only If a tlb flush request
size is greater than the threshold. This is done as sfence in RISC-V
can only flush 4KB at a time. Doing a large number of flushes page by
page impacts the performance. It is better to do a full tlbflush if the
request size is at least equal to the threshold size.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
If compiler supports riscv atomic instructions, we should
use them instead of legacy gcc built-in macros
__sync_lock_test_and_set in atomic exchange functions.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
To read 64bit time in 32 bit we have to read lower & upper half
separately and 'or' them together. However, upper half time may
have changed by the time we read lower half. Thus, the resultant
64 bit time may not be accurate.
Consider lower half time value only if upper half time value has
not changed.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
OpenSBI manages outstanding TLB flush requests by queueing
them in a fifo synchronously. An ipi sync which uses an
atomic operation on MMIO address is no longer required.
Remove the ipi sync method from platform header and all usage.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Anup Patel <anup.patel@wdc.com>
Linux kernel expects tlb flush SBI call to be completely synchronous i.e.
the SBI call should only return once corresponding *fence* instruction is
executed.
OpenSBI manages the outstanding TLB flush requests by keeping them in a
per hart based fifo. However, there are few corner cases that may lead to
race conditions while updating the fifo.
Currently, the caller hart waits for IPI acknowledgement via clint
address which is not a very good method as synchronization on MMIO may not
be supported in every platform. Moreover, the waiter doesn't have any way of
identifying if the IPI is received for specific tlb flush request or any
other IPI. This may lead to unpredictable behavior in supervisor/user space.
Fix this by waiting on individual fifo entries rather than MMIO address.
Currently, a relaxed loop is being used because wfi again involves MMIO write
which would be slower compared to relaxed loop. To avoid deadlock, fifo
is processed every time a hart loops for fifo enqueue or fifo sync to consume
the tlb flush requests sent by other harts.
Signed-off-by: Anup Patel <anup.patel@wdc.com>
Signed-off-by: Atish Patra <atish.patra@wdc.com>
The Linux kernel will maintain the device tree (DT) for HiFive Unleashed
from release v5.2-rc6. This DT is incompatible with previous DT
present for Microsemi expansion board in OpenSBI.
Since, OpenSBI will directly load the pre-built DTB from kernel now onwards,
no need to keep the out-of-date DT in openSBI. Remove this DT and
it's related documentation.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Acked-by: Anup Patel <anup.patel@wdc.com>
With 5.2-rc6 release, Linux kernel hosts the DT for Unleashed board
which is incompatible with the default DT from FSBL.
Update the document to use the DT from kernel.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Acked-by: Anup Patel <anup.patel@wdc.com>
tinyfdt.c was originally added to provide a minimal implementation of
fdt parsing. However, we have already included libfdt in OpenSBI for
more complicated operations.
Remove tinfdt and replace its functiolity using libfdt.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Acked-by: Anup Patel <anup.patel@wdc.com>
The drivers and libfdt are built as libsbiutils.a
instead of libplatsbi.a.
libsbiutils.a are not built per platform specific. Thus,
enable all drivers by default.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Acked-by: Anup Patel <anup.patel@wdc.com>
All string functions are part of libsbi. It makes more sense
to rename them to sbi_string.x as the libsbi can be linked
with external libraries that can have similar implementation.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Acked-by: Anup Patel <anup.patel@wdc.com>
Currently, platform/common contains platform/non-platform specific
common minimal drivers and libraries. This is helpful is all platforms
are built within opensbi framework.
Move them to lib/utils so that any external platform code also can
reuse the minimalistic drivers or other common libraries.
This patch doesn't introduce any functional changes.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Acked-by: Anup Patel <anup.patel@wdc.com>
libsbi needs some of the custom libc functions. It should be directly
included in libsbi instead of platform specific libraries.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Acked-by: Anup Patel <anup.patel@wdc.com>
All dependant patches are merged in U-Boot and Linux kernel now.
Update the unleashed guide to reflect that and fix the documentation
about tftp loading path as well.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Simulatenous requests for tlbflush IPIs can have overlapping address
ranges.
Ignore if address range is same or within already existing fifo entries.
Update the tlb flush info in fifo directly if the one of the existing
entry lies within the new flush request.
Delete all entries if flush all request is recieved for the vma.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Implement a lock enabled iteration for fifo so that
caller can determine if next entry can be skipped or
any existing entries in fifo can be updated before enqueue.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Currently, there is no provision for tracking multiple IPIs sent
to a single hart at the same time by different harts.
Use a fifo manage the outstanding requests. While dequeueing, read all
the entries once, because we have only 1 bit to track the type of IPI.
Once the queue is full, busy wait until the there is space available in
queue. This is not the most elegant approach. It should be changed in
favor of a wakeup event once available in opensbi.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
OpenSBI should show error trace only if any valid SBI function
does not perform as expected.
However, OpenSBI should show notify the caller with a negative
error if given SBI function ID is not valid.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Scratch space setup needs to be done once for reboot for each
hart. _start_warm may be called several times if hart hotplug
is implemented.
Move scratch space setup to the beginning so that it is done
only once.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Currently, global page mappings are not flushed if start and size
arguments are zero.
Flush entire TLB if both size and start argument is passed as zero.
Fixes : 90cb491 (lib: Implement sfence.vma correctly)
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Currently, OpenSBI doesn't distinguish between sfence.vma
and sfence.vm asid calls. Moreover, it ignores the page
ranges and just flush entire TLB everytime.
Fix the sfence implementation by keeping all the tlb flush
info in scratch area.
The relevant Linux kernel code was added by
https://patchwork.kernel.org/project/linux-riscv/list/?series=89695
However, this patch is backward compatible with older version kernel
that doesn't have the above patches as well.
Fixes#87
Signed-off-by: Atish Patra <atish.patra@wdc.com>
A warm reset using reset button may put icache and registers
in non-coherent state.
Flush the icache and reset all registers for every hart.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
All the non-boot harts run in a tight loop which may cause a heavy load
on the memory bus. This may delay the boot hart to complete the cold boot
process.
Introduce few nop that will ease up the traffic.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
