G-stage translation should be considered to be user-level access in
riscv_cpu_get_phys_page_debug(), as already done in riscv_cpu_tlb_fill().
This fixes a bug that prevents gdb from reading memory while the VM is
running in VS-mode.
Signed-off-by: Hiroaki Yamamoto <hrak1529@gmail.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-ID: <20240228081028.35081-1-hrak1529@gmail.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Gate hardware A/D PTE bit updating on {m,h}envcfg.ADUE and only
enable menvcfg.ADUE on reset if svade has not been selected. Now
that we also consider svade, we have four possible configurations:
1) !svade && !svadu
use hardware updating and there's no way to disable it
(the default, which maintains past behavior. Maintaining
the default, even with !svadu is a change that fixes [1])
2) !svade && svadu
use hardware updating, but also provide {m,h}envcfg.ADUE,
allowing software to switch to exception mode
(being able to switch is a change which fixes [1])
3) svade && !svadu
use exception mode and there's no way to switch to hardware
updating
(this behavior change fixes [2])
4) svade && svadu
use exception mode, but also provide {m,h}envcfg.ADUE,
allowing software to switch to hardware updating
(this behavior change fixes [2])
Fixes: 0af3f115e6 ("target/riscv: Add *envcfg.HADE related check in address translation") [1]
Fixes: 48531f5adb ("target/riscv: implement svade") [2]
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Signed-off-by: Andrew Jones <ajones@ventanamicro.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-ID: <20240215223955.969568-6-dbarboza@ventanamicro.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
We'll re-use the logic froim vext_get_vlmax() in 2 other occurrences in
the next patch, but first we need to make it independent of both 'cpu'
and 'vtype'. To do that, add 'vlenb', 'vsew' and 'lmul' as parameters
instead.
Adapt the two existing callers. In cpu_get_tb_cpu_state(), rename 'sew'
to 'vsew' to be less ambiguous about what we're encoding into *pflags.
In HELPER(vsetvl) the following changes were made:
- add a 'vsew' var to store vsew. Use it in the shift to get 'sew';
- the existing 'lmul' var was renamed to 'vlmul';
- add a new 'lmul' var to store 'lmul' encoded like DisasContext:lmul.
Signed-off-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-ID: <20240122161107.26737-12-dbarboza@ventanamicro.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
According to RISCV Specification sect 9.5 on two stage translation when
V=1 the vsstatus(mstatus in QEMU's terms) field MXR, which makes
execute-only pages readable, only overrides VS-stage page protection.
Setting MXR at HS-level(mstatus_hs), however, overrides both VS-stage
and G-stage execute-only permissions.
The hypervisor extension changes the behavior of MXR\MPV\MPRV bits.
Due to RISCV Specification sect. 9.4.1 when MPRV=1, explicit memory
accesses are translated and protected, and endianness is applied, as
though the current virtualization mode were set to MPV and the current
nominal privilege mode were set to MPP. vsstatus.MXR makes readable
those pages marked executable at the VS translation stage.
Fixes: 36a18664ba ("target/riscv: Implement second stage MMU")
Signed-off-by: Ivan Klokov <ivan.klokov@syntacore.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Message-ID: <20231121071757.7178-3-ivan.klokov@syntacore.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
According to RISCV privileged spec sect. 5.3.2 Virtual Address Translation Process
access-fault exceptions may raise only after PMA/PMP check. Current implementation
generates an access-fault for mbare mode even if there were no PMA/PMP errors.
This patch removes the erroneous MMU mode check and generates an access-fault
exception based on the pmp_violation flag only.
Fixes: 1448689c7b ("target/riscv: Allow specifying MMU stage")
Signed-off-by: Ivan Klokov <ivan.klokov@syntacore.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Message-ID: <20231121071757.7178-2-ivan.klokov@syntacore.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
This change adds support for inserting virtual interrupts from HS-mode
into VS-mode using hvien and hvip csrs. This also allows for IRQ filtering
from HS-mode.
Also, the spec doesn't mandate the interrupt to be actually supported
in hardware. Which allows HS-mode to assert virtual interrupts to VS-mode
that have no connection to any real interrupt events.
This is defined as part of the AIA specification [0], "6.3.2 Virtual
interrupts for VS level".
[0]: https://github.com/riscv/riscv-aia/releases/download/1.0/riscv-interrupts-1.0.pdf
Signed-off-by: Rajnesh Kanwal <rkanwal@rivosinc.com>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Message-ID: <20231016111736.28721-7-rkanwal@rivosinc.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
This change adds support for inserting virtual interrupts from M-mode
into S-mode using mvien and mvip csrs. IRQ filtering is a use case of
this change, i-e M-mode can stop delegating an interrupt to S-mode and
instead enable it in MIE and receive those interrupts in M-mode and then
selectively inject the interrupt using mvien and mvip.
Also, the spec doesn't mandate the interrupt to be actually supported
in hardware. Which allows M-mode to assert virtual interrupts to S-mode
that have no connection to any real interrupt events.
This is defined as part of the AIA specification [0], "5.3 Interrupt
filtering and virtual interrupts for supervisor level".
[0]: https://github.com/riscv/riscv-aia/releases/download/1.0/riscv-interrupts-1.0.pdf
Signed-off-by: Rajnesh Kanwal <rkanwal@rivosinc.com>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Message-ID: <20231016111736.28721-6-rkanwal@rivosinc.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
This is to allow virtual interrupts to be inserted into S and VS
modes. Given virtual interrupts will be maintained in separate
mvip and hvip CSRs, riscv_cpu_update_mip will no longer be in the
path and interrupts need to be triggered for these cases from
rmw_hvip64 and rmw_mvip64 functions.
Signed-off-by: Rajnesh Kanwal <rkanwal@rivosinc.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-ID: <20231016111736.28721-5-rkanwal@rivosinc.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
RISCV_CPU(cs) uses a checked cast. When QOM cast debugging is enabled
this adds about 5% total overhead when emulating RV64 on x86-64 host.
Using a RISC-V guest with 16 vCPUs, 16 GB of guest RAM, virtio-blk
disk. The guest has a copy of the qemu source tree. The test
involves compiling the qemu source tree with 'make clean; time make -j16'.
Before making this change the compile step took 449 & 447 seconds over
two consecutive runs.
After making this change: 428 & 421 seconds.
The saving is over 5%.
Thanks: Paolo Bonzini
Thanks: Philippe Mathieu-Daudé
Signed-off-by: Richard W.M. Jones <rjones@redhat.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20231009124859.3373696-2-rjones@redhat.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
The privileged spec states:
For a memory access made to support VS-stage address translation (such as
to read/write a VS-level page table), permissions are checked as though
for a load or store, not for the original access type. However, any
exception is always reported for the original access type (instruction,
load, or store/AMO).
The current implementation converts the access type to LOAD if implicit
G-stage translation fails which results in only reporting "Load guest-page
fault". This commit removes the convertion of access type, so the reported
exception conforms to the spec.
Signed-off-by: Jason Chien <jason.chien@sifive.com>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230627074915.7686-1-jason.chien@sifive.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Pointer mask is also affected by MPRV which means cur_pmbase/pmmask
should also take MPRV into consideration. As pointer mask for instruction
is not supported currently, so we can directly update cur_pmbase/pmmask
based on address related mode and xlen affected by MPRV now.
Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Message-Id: <20230614032547.35895-3-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
We initialize cur_pmmask as -1(UINT32_MAX/UINT64_MAX) and regard it
as if pointer mask is disabled in current implementation. However,
the addresses for vector load/store will be adjusted to zero in this
case and -1(UINT32_MAX/UINT64_MAX) is valid value for pmmask when
pointer mask is enabled.
Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: LIU Zhiwei <zhiwei_liu@linux.alibaba.com>
Message-Id: <20230610094651.43786-1-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
PMP entries before (including) the matched PMP entry may only cover partial
of the TLB page, and this may split the page into regions with different
permissions. Such as for PMP0 (0x80000008~0x8000000F, R) and PMP1 (0x80000000~
0x80000FFF, RWX), write access to 0x80000000 will match PMP1. However we cannot
cache the translation result in the TLB since this will make the write access
to 0x80000008 bypass the check of PMP0. So we should check all of them instead
of the matched PMP entry in pmp_get_tlb_size() and set the tlb_size to 1 in
this case.
Set tlb_size to TARGET_PAGE_SIZE if PMP is not support or there is no PMP rules.
Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Reviewed-by: LIU Zhiwei <zhiwei_liu@linux.alibaba.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230517091519.34439-2-liweiwei@iscas.ac.cn>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Move a use of TARGET_LONG_BITS out of tcg/tcg.h.
Include the new file only where required.
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
As per the specification, in 64-bit, if any of the pte reserved bits
60-54 is set an exception should be triggered (see 4.4.1, "Addressing and
Memory Protection"). In addition, we must check the napot/pbmt bits are
not set if those extensions are not active.
Reported-by: Andrea Parri <andrea@rivosinc.com>
Signed-off-by: Alexandre Ghiti <alexghiti@rivosinc.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230420150220.60919-1-alexghiti@rivosinc.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Before changing the flow check for sv39/48/57.
According to specification (for Supervisor mode):
Sv39 implementations support a 39-bit virtual address space, divided into 4 KiB
pages.
Instruction fetch addresses and load and store effective addresses, which are
64 bits,
must have bits 63–39 all equal to bit 38, or else a page-fault exception will
occur.
Likewise for Sv48 and Sv57.
So the high bits are equal to bit 38 for sv39.
According to specification (for Hypervisor mode):
For Sv39x4, address bits of the guest physical address 63:41 must all be zeros,
or else a
guest-page-fault exception occurs.
Likewise for Sv48x4 and Sv57x4.
For Sv48x4 address bits 63:50 must all be zeros, or else a guest-page-fault
exception occurs.
For Sv57x4 address bits 63:59 must all be zeros, or else a guest-page-fault
exception occurs.
For example we are trying to access address 0xffff_ffff_ff01_0000 with only
G-translation enabled.
So expected behavior is to generate exception. But qemu doesn't generate such
exception.
For the old check, we get
va_bits == 41, mask == (1 << 24) - 1, masked_msbs == (0xffff_ffff_ff01_0000 >>
40) & mask == mask.
Accordingly, the condition masked_msbs != 0 && masked_msbs != mask is not
fulfilled
and the check passes.
Signed-off-by: Irina Ryapolova <irina.ryapolova@syntacore.com>
Reviewed-by: Weiwei Li <liweiwei@iscas.ac.cn>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20230418075423.26217-1-irina.ryapolova@syntacore.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
