Separate the implement of T-HEAD c9xx errata to allow any platform
with bug related to c9xx cores can use it.

Signed-off-by: Inochi Amaoto <inochiama@outlook.com>
Reviewed-by: Anup Patel <anup@brainfault.org>

Separate the implement of T-HEAD c9xx pmu to allow any platform with
c9xx cores can use it.

Signed-off-by: Inochi Amaoto <inochiama@outlook.com>
Reviewed-by: Anup Patel <anup@brainfault.org>

The CSR encoding for t-head c9xx cores is shared across all the
platforms with these cores. So move header thead_c9xx.h to the
thead subdir.

Signed-off-by: Inochi Amaoto <inochiama@outlook.com>
Acked-by: Guo Ren <guoren@kernel.org>
Reviewed-by: Anup Patel <anup@brainfault.org>

Since commit 112daa2e ("lib: sbi: Maximize the use of HART index in
sbi_domain") the platform parameter is unused.

Fixes: 112daa2e

 ("lib: sbi: Maximize the use of HART index in sbi_domain")
Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
Reviewed-by: Anup Patel <anup@brainfault.org>

Now that SBI v2.0 specification is frozen, we change spec version
implemented by OpenSBI to v2.0.

Signed-off-by: Anup Patel <apatel@ventanamicro.com>
Reviewed-by: Andrew Jones <ajones@ventanamicro.com>

Unlike C.LWSP/C.LDSP, these encodings can be used with the zero
register, so checking that the rs2 field is non-zero is unnecessary.

Additionally, the previous check was incorrect since it was checking
the immediate field of the instruction instead of the rs2 field.

Signed-off-by: Amanieu d'Antras <amanieu@gmail.com>
Reviewed-by: Anup Patel <anup@brainfault.org>

RISC-V privileged specification v1.12 introduced the mconfigptr CSR
which will hold the physical address of a configuration data
structure.

Signed-off-by: Yangjie Zhang <jay1273062855@outlook.com>
Reviewed-by: Anup Patel <anup@brainfault.org>

"Zicsr" isa extension has been separated from "I" extension.
This patch add the isa requirement of "Zicsr" extension in
platform requirements documentation.

Signed-off-by: Yangjie Zhang <jay1273062855@outlook.com>
Reviewed-by: Anup Patel <anup@brainfault.org>

In starfive_gpio_direction_output() readl() is called twice to read the
gpio direction register. The result of the first read is discarded.

Remove the redundant read.

Fixes: 908be1b8

 ("gpio/starfive: add gpio driver and support gpio reset")
Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
Reviewed-by: Anup Patel <anup@brainfault.org>

sbi_pmu_ctr_cfg_match() receives data from a lower privilege level mode.
We must catch maliciously wrong values.

We already check against total_ctrs. But we do not check that total_ctrs is
less than SBI_PMU_HW_CTR_MAX + SBI_PMU_FW_CTR_MAX.

Check that the number of hardware counters is in the valid range.

Addresses-Coverity-ID: 1566114 Out-of-bounds write
Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
Reviewed-by: Atish Patra <atishp@rivosinc.com>

The define is hard to read. The continue statement does not do what was
intended.

* Remove do {} while (false);
* Change the name to set_multi_letter_ext
  - Other local macros are lower case too.
  - Refer to the fact that this is only used for multi-letter extensions.

Addresses-Coverity-ID: 1568359 Unexpected control flow
Fixes: d72f5f17

 ("lib: utils: Add detection of Smepmp from ISA string in FDT")
Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
Reviewed-by: Xiang W <wxjstz@126.com>
Reviewed-by: Anup Patel <anup@brainfault.org>

hart_exts == NULL can only occur if offset and node address lead to an
overflow resulting in exactly NULL. As we don't catch other values of
overflow it does not make sense to treat this one as special.

Addresses-Coverity-ID: 1568355 Logically dead code
Addresses-Coverity-ID: 1568358 Logically dead code
Fixes: 6259b2ec

 ("lib: utils/fdt: Fix fdt_parse_isa_extensions() implementation")
Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
Reviewed-by: Xiang W <wxjstz@126.com>
Reviewed-by: Anup Patel <anup@brainfault.org>

'1' is a 32 bit integer. When shifting it by more than 31 bits it becomes
zero and we get an incorrect return value.

Addresses-Coverity-ID: 1568356 Bad bit shift operation
Fixes: 296e70d6

 ("lib: sbi: Extend sbi_hartmask to support both hartid and hartindex")
Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
Reviewed-by: Xiang W <wxjstz@126.com>
Reviewed-by: Anup Patel <anup@brainfault.org>

This patch adds unspecified permission flags for the PLICSW region
and updates the permission of the PLMT region.

With this update, both regions will become M-mode only read/write
regions in the root domain.

  Domain0 Region00: 0x00000000f0300000-0x00000000f0300fff M: (I,R,W) S/U: (R,W)
  Domain0 Region01: 0x0000000000040000-0x000000000005ffff M: (R,W) S/U: ()
  Domain0 Region02: 0x0000000000000000-0x000000000003ffff M: (R,X) S/U: ()
> Domain0 Region03: 0x00000000e6000000-0x00000000e60fffff M: (I,R,W) S/U: ()
> Domain0 Region04: 0x00000000e6400000-0x00000000e67fffff M: (I,R,W) S/U: ()
  Domain0 Region05: 0x0000000000000000-0xffffffffffffffff M: () S/U: (R,W,X)

The PMP rules of AE350-AX65 (single-core) w/ Smepmp:

  p/x $pmpcfg0
  $1 = {0x1f9b9b9d9b1e00,
  pmp0cfg = {0x0},
                    L--AAXWR
  pmp1cfg = {0x1e} (00011110), pmpaddr1: 0xf0300000 ~   0xf0300fff  (UART1)
  pmp2cfg = {0x9b} (10011011), pmpaddr2:    0x40000 ~      0x5ffff
  pmp3cfg = {0x9d} (10011101), pmpaddr3:        0x0 ~      0x3ffff
  pmp4cfg = {0x9b} (10011011), pmpaddr4: 0xe6000000 ~   0xe60fffff  (PLMT)
  pmp5cfg = {0x9b} (10011011), pmpaddr5: 0xe6400000 ~   0xe67fffff  (PLICSW)
  pmp6cfg = {0x1f} (00011111), pmpaddr6:        0x0 ~ 0xffffffffff
  pmp7cfg = {0x0 }}

The PMP rules of AE350-AX45MP (qual-core) w/o Smepmp:

  p/x $pmpcfg0
  $1 = {0x1f181818181b,
                     L--AAXWR
  pmp0cfg = {0x1b}, (00011011), pmpaddr0: 0xf0300000 ~  0xf0300fff  (UART1)
  pmp1cfg = {0x18}, (00011000), pmpaddr1:    0x40000 ~     0x5ffff
  pmp2cfg = {0x18}, (00011000), pmpaddr2:        0x0 ~     0x3ffff
  pmp3cfg = {0x18}, (00011000), pmpaddr3: 0xe6000000 ~  0xe60fffff  (PLMT)
  pmp4cfg = {0x18}, (00011000), pmpaddr4: 0xe6400000 ~  0xe67fffff  (PLICSW)
  pmp5cfg = {0x1f}, (00011111), pmpaddr5:        0x0 ~ 0x1ffffffff
  pmp6cfg = {0x0 }}

Note that starting from this patch, we restrict the S/U-mode read
permission to the PLMT region, since we should read the TIME CSR
in a lower privilege mode.

Signed-off-by: Yu Chien Peter Lin <peterlin@andestech.com>
Reviewed-by: Anup Patel <anup@brainfault.org>

The added memory region should start from the base address.
Otherwise, the range will be shifted by reg_offset and not
able to merge consecutive NAPOT regions in the root domain,
resulting in wasted PMP entries.

Fixes: e8bc1621

 ("lib: utils/serial: Add shared regions for
serial drivers")

Signed-off-by: Yu Chien Peter Lin <peterlin@andestech.com>
Reviewed-by: Anup Patel <anup@brainfault.org>

The base of .text is defined by `FW_TEXT_START`, not `FW_TEXT_BASE`.

Signed-off-by: Hoa Nguyen <hn@hnpl.org>
Reviewed-by: Xiang W <wxjstz@126.com>
Reviewed-by: Anup Patel <anup@brainfault.org>

Since everything is statically linked, we don't need to expose symbols
for dynamic linking.

For a default build this saves about 2 KiB of useless read only data in
.dynsym, .dynstr, .hash, .gnu.hash sections.

Signed-off-by: Vivian Wang <dramforever@live.com>
Reviewed-by: Anup Patel <anup@brainfault.org>

Since everything is statically linked, we won't actually have
R_RISCV_{32,64} relocations. No need to handle these.

Fixes: 0f20e8ad

 ("firmware: Support position independent execution")
Signed-off-by: Vivian Wang <dramforever@live.com>
Reviewed-by: Anup Patel <anup@brainfault.org>

The .rela.dyn section should be exactly the size of the relocations,
without padding. On RV64, .rela* sections are already aligned and
there's no need for padding. On RV32, this adds padding up to 4 bytes,
which, if present, confuses the relocation loop into processing an extra
entry past the end of .rela*, and it crashes with an invalid memory
access.

Fixes: 0f20e8ad

 ("firmware: Support position independent execution")
Signed-off-by: Vivian Wang <dramforever@live.com>
Reviewed-by: Anup Patel <anup@brainfault.org>

The TLB entries remain functional all the time once added in T-HEAD th1520
and Sophgo sg2042 (even if the MMU is then disabled afterwards). If there
are some stale TLB entries that contains the address of SBI, it will cause
unexpected memory access and issue a illegal instruction error. To avoid
this, a TLB flush is needed to drop these TLB entries before any memory
access in the trap handler.

To handle this workaroud, add a custom trap handler with executing TLB flush
first in the T-HEAD platform to fix affected socs.

Signed-off-by: Inochi Amaoto <inochiama@outlook.com>
Reviewed-by: Anup Patel <anup@brainfault.org>

The syscon region used by OpenSBI should be marked as a shared
read-write region between M-mode and SU-mode.

Signed-off-by: Anup Patel <apatel@ventanamicro.com>
Signed-off-by: Mayuresh Chitale <mchitale@ventanamicro.com>

The serial driver regions used by OpenSBI should be marked as a shared
read-write regions between M-mode and SU-mode as those are accessed
by earlycon and the corresponding tty serial drivers running in 'S' mode.
When the smepmp extension is enabled, PMP entries for these shared regions
will get programmed.

Signed-off-by: Mayuresh Chitale <mchitale@ventanamicro.com>
Signed-off-by: Anup Patel <apatel@ventanamicro.com>

Based on sections 4.c and 4.d in Ch.2 of the Smepmp spec the PMP entries
must be programmed as below:
1. Program M-only entries
2. Enable mseccfg.MML
3. Program shared-region entries
4. Program SU-only entries

Co-developed-by: Mayuresh Chitale <mchitale@ventanamicro.com>
Signed-off-by: Mayuresh Chitale <mchitale@ventanamicro.com>
Signed-off-by: Anup Patel <apatel@ventanamicro.com>

The Smepmp read-only shared region must have pmpcfg.L, pmpcfg.R,
pmpcfg.W, and pmpcfg.X bits set so sbi_hart_get_smepmp_flags()
must return pmp_flags accordingly.

Signed-off-by: Anup Patel <apatel@ventanamicro.com>
Signed-off-by: Mayuresh Chitale <mchitale@ventanamicro.com>

The mseccfg.MML bit is a sticky bit which remains unchanged once set
so no need to clear it in sbi_hart_smepmp_configure().

Signed-off-by: Anup Patel <apatel@ventanamicro.com>
Signed-off-by: Mayuresh Chitale <mchitale@ventanamicro.com>

Let us factor-out Smepmp configuaration as separate function so
that code is more readable.

Signed-off-by: Anup Patel <apatel@ventanamicro.com>
Signed-off-by: Mayuresh Chitale <mchitale@ventanamicro.com>

The macros to test permissions must perform an exact match of all the
bits in the input with the desired permission bits. Otherwise, the check
returns false positives in those cases where only some of the desired
permissions are set in the input.

Fixes: 6c202c5e

 ("include: sbi: Add Smepmp specific access flags for PMP
entries")
Signed-off-by: Mayuresh Chitale <mchitale@ventanamicro.com>
Reviewed-by: Anup Patel <anup@brainfault.org>

The sbi_hartmask_for_each_hart() macro is slow and has only one user
so let us completely remove the sbi_hartmask_for_each_hart() macro.

Signed-off-by: Anup Patel <apatel@ventanamicro.com>

Let us maximize the use of HART index in sbi_domain because hartindex
based hartmask access and sbi_scratch lookup is faster.

Signed-off-by: Anup Patel <apatel@ventanamicro.com>

The sbi_scratch_last_hartid() macro is not of much use on platforms
with really sparse hartids so let us replace use of this macro with
other approaches.

Signed-off-by: Anup Patel <apatel@ventanamicro.com>

Let us prefer hartindex over hartid in IPI framework which in-turn
forces IPI users to also prefer hartindex.

Signed-off-by: Anup Patel <apatel@ventanamicro.com>

The sbi_hartid_to_scratch() involves translating hartid to hartindex
which is expensive so let's use sbi_hartindex_to_scratch() instead.

Signed-off-by: Anup Patel <apatel@ventanamicro.com>

Currently, the sbi_hartmask is indexed by hartid which puts a
limit on hartid to be less than SBI_HARTMASK_MAX_BITS.

We extend the sbi_hartmask implementation to use hartindex and
support updating sbi_hartmask using hartid. This removes the
limit on hartid and existing code works largely unmodified.

Signed-off-by: Xiang W <wxjstz@126.com>
Signed-off-by: Anup Patel <apatel@ventanamicro.com>

The hartid to hartindex mapping is now tracked in sbi_scratch so we
don't need sbi_platform_hart_index() and sbi_platform_hart_invalid()
functions hence let us remove them.

Signed-off-by: Anup Patel <apatel@ventanamicro.com>

We introduce HART index and related helper functions in sbi_scratch
where HART index is contiguous and each HART index maps to a physical
HART id such that 0 <= HART index and HART index < SBI_HARTMASK_MAX_BITS.

The HART index to HART id mapping follows the index2id mapping provided
by the platform. If the platform does not provide index2id mapping then
identity mapping is assumed.

Signed-off-by: Anup Patel <apatel@ventanamicro.com>

We found the mstatus.FS status is not set correctly after the SET_F64_REG()
and SET_F32_REG(). We should set mstatus.FS dirty after we emulate the FPU
instructions.

Co-developed-by: Roy Lin <roy.lin@sifive.com>
Signed-off-by: Roy Lin <roy.lin@sifive.com>
Signed-off-by: Greentime Hu <greentime.hu@sifive.com>
Reviewed-by: Anup Patel <anup@brainfault.org>

If the system is defined from tlb_fifo_num_entries, the scratch may be
too small to hold the fifo, so it is alloc through the heap.

Signed-off-by: Xiang W <wxjstz@126.com>
Signed-off-by: Xing Xiaoguang <xiaoguang.xing@sophgo.com>
Reviewed-by: Anup Patel <anup@brainfault.org>

For some platforms with a particularly high number of harts, if the
tlb fifo is too small, it case harts to wait. Platforms should be
allowed to specify the size of the tlb fifo.

Signed-off-by: Xiang W <wxjstz@126.com>
Signed-off-by: Xing Xiaoguang <xiaoguang.xing@sophgo.com>
Reviewed-by: Anup Patel <anup@brainfault.org>

After the hardware hpm counter is stopped, it should not raise any new
interrupt as it is already stopped. So add the hw_counter_disable_irq
callback to allow the custom pmu device to control this behavior.

Signed-off-by: Inochi Amaoto <inochiama@outlook.com>
Reviewed-by: Guo Ren <guoren@kernel.org>
Reviewed-by: Anup Patel <anup@brainfault.org>
Tested-by: Samuel Holland <samuel@sholland.org>

T-HEAD c9xx pmu needs to clear OV bits of MCOUNTEROF in any condition
to avoid unnecessary OF interrupts.

In addition, the S-mode SCOUNTEROF only have OF bit set when the related
bits of MCOUNTERWEN is set, so also configure MCOUNTERWEN to allow kernel
to access valid SCOUNTEROF.

Signed-off-by: Haijiao Liu <haijiao.liu@sophgo.com>
Co-authored-by: Inochi Amaoto <inochiama@outlook.com>
Signed-off-by: Inochi Amaoto <inochiama@outlook.com>
Reviewed-by: Anup Patel <anup@brainfault.org>
Tested-by: Samuel Holland <samuel@sholland.org>