On 10/13/21 9:46 AM, Frédéric Pétrot wrote:
Hello,
Le 07/10/2021 à 19:47, Richard Henderson a écrit :
Shortly, the set of supported XL will not be just 32 and 64,
and representing that properly using the enumeration will be
imperative.
Two places, booting and gdb, intentionally use misa_mxl_max
to emphasize the use of the reset value of misa.mxl, and not
the current cpu state.
Signed-off-by: Richard Henderson <richard.hender...@linaro.org>
---
target/riscv/cpu.h | 9 ++++++++-
hw/riscv/boot.c | 2 +-
semihosting/arm-compat-semi.c | 2 +-
target/riscv/cpu.c | 24 ++++++++++++++----------
target/riscv/cpu_helper.c | 12 ++++++------
target/riscv/csr.c | 24 ++++++++++++------------
target/riscv/gdbstub.c | 2 +-
target/riscv/monitor.c | 4 ++--
8 files changed, 45 insertions(+), 34 deletions(-)
diff --git a/target/riscv/cpu.h b/target/riscv/cpu.h
@@ -522,7 +522,7 @@ static RISCVException write_mstatus(CPURISCVState *env, int
csrno,
dirty = ((mstatus & MSTATUS_FS) == MSTATUS_FS) |
((mstatus & MSTATUS_XS) == MSTATUS_XS);
- if (riscv_cpu_is_32bit(env)) {
+ if (riscv_cpu_mxl(env) == MXL_RV32) {
mstatus = set_field(mstatus, MSTATUS32_SD, dirty);
} else {
mstatus = set_field(mstatus, MSTATUS64_SD, dirty);
I believe we miss the settings of the SXL and UXL fields that are needed
by cpu_get_xl
diff --git a/target/riscv/csr.c b/target/riscv/csr.c
index 9c0753bc8b..c4a479ddd2 100644
--- a/target/riscv/csr.c
+++ b/target/riscv/csr.c
@@ -526,6 +526,9 @@ static RISCVException write_mstatus(CPURISCVState *env, int
csrno,
mstatus = set_field(mstatus, MSTATUS32_SD, dirty);
} else {
mstatus = set_field(mstatus, MSTATUS64_SD, dirty);
+ /* SXL and UXL fields are for now read only */
+ mstatus = set_field(mstatus, MSTATUS64_SXL, MXL_RV64);
+ mstatus = set_field(mstatus, MSTATUS64_UXL, MXL_RV64);
}
env->mstatus = mstatus;
Yes, I missed that. I was relying too much on the UXL patch set which I did not
incorporate here. This is a good minimal addition to keep things bisectable.
r~