diff options
| author | Alex Bennée <alex.bennee@linaro.org> | 2021-11-29 14:09:26 +0000 |
|---|---|---|
| committer | Alex Bennée <alex.bennee@linaro.org> | 2021-11-29 15:12:37 +0000 |
| commit | aff0e204cb1f1c036a496c94c15f5dfafcd9b4b4 (patch) | |
| tree | 3b7ca5c6f32f9635df139d4961c2c2aaa87d9007 /softmmu/physmem.c | |
| parent | 48e14066ac10581db4e69f75eda107cfdafa6022 (diff) | |
accel/tcg: suppress IRQ check for special TBs
When we set cpu->cflags_next_tb it is because we want to carefully
control the execution of the next TB. Currently there is a race that
causes the second stage of watchpoint handling to get ignored if an
IRQ is processed before we finish executing the instruction that
triggers the watchpoint. Use the new CF_NOIRQ facility to avoid the
race.
We also suppress IRQs when handling precise self modifying code to
avoid unnecessary bouncing.
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Cc: Pavel Dovgalyuk <pavel.dovgalyuk@ispras.ru>
Fixes: https://gitlab.com/qemu-project/qemu/-/issues/245
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20211129140932.4115115-3-alex.bennee@linaro.org>
Diffstat (limited to 'softmmu/physmem.c')
| -rw-r--r-- | softmmu/physmem.c | 4 |
1 files changed, 2 insertions, 2 deletions
diff --git a/softmmu/physmem.c b/softmmu/physmem.c index 314f8b439c..3524c04c2a 100644 --- a/softmmu/physmem.c +++ b/softmmu/physmem.c @@ -912,7 +912,7 @@ void cpu_check_watchpoint(CPUState *cpu, vaddr addr, vaddr len, */ if (!cpu->can_do_io) { /* Force execution of one insn next time. */ - cpu->cflags_next_tb = 1 | CF_LAST_IO | curr_cflags(cpu); + cpu->cflags_next_tb = 1 | CF_LAST_IO | CF_NOIRQ | curr_cflags(cpu); cpu_loop_exit_restore(cpu, ra); } /* @@ -946,7 +946,7 @@ void cpu_check_watchpoint(CPUState *cpu, vaddr addr, vaddr len, cpu_loop_exit(cpu); } else { /* Force execution of one insn next time. */ - cpu->cflags_next_tb = 1 | CF_LAST_IO | curr_cflags(cpu); + cpu->cflags_next_tb = 1 | CF_LAST_IO | CF_NOIRQ | curr_cflags(cpu); mmap_unlock(); cpu_loop_exit_noexc(cpu); } |