[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [PATCH 00/18] Stop all qemu-cpu threads on a breakpoint
From: |
Alex Bennée |
Subject: |
Re: [PATCH 00/18] Stop all qemu-cpu threads on a breakpoint |
Date: |
Wed, 08 Jan 2025 15:56:49 +0000 |
User-agent: |
mu4e 1.12.8; emacs 29.4 |
Ilya Leoshkevich <iii@linux.ibm.com> writes:
> Hi,
>
> On reporting a breakpoint in a non-non-stop mode, GDB remotes must stop
> all threads. Currently qemu-user doesn't do that, breaking the
> debugging session for at least two reasons: concurrent access to the
> GDB socket, and an assertion within GDB [1].
>
> This series fixes this by importing pause_all_vcpus() from qemu-system.
> This in turn requires introducing BQL and a few stubs to qemu-user.
Is there a conclusion to this design choice? I'd like to avoid bringing
in a bunch of system-mode infrastructure if the existing exclusive code
would work. For that I'll defer to the linux-user maintainer or Richard
who knows the code better than I do.
I could certainly harvest the early clean-up patches to keep the delta
low while the details are worked out. Is there going to be a v2?
>
> Best regards,
> Ilya
>
> [1] https://gitlab.com/qemu-project/qemu/-/issues/2465
>
> Ilya Leoshkevich (18):
> gdbstub: Make gdb_get_char() static
> gdbstub: Move phy_memory_mode to GDBSystemState
> gdbstub: Move gdb_syscall_mode to GDBSyscallState
> gdbstub: Factor out gdb_try_stop()
> accel/tcg: Factor out cpu_exec_user()
> qemu-thread: Introduce QEMU_MUTEX_INITIALIZER
> qemu-thread: Introduce QEMU_COND_INITIALIZER
> replay: Add replay_mutex_{lock,unlock}() stubs for qemu-user
> qemu-timer: Provide qemu_clock_enable() stub for qemu-user
> cpu: Use BQL in qemu-user
> accel/tcg: Unify user implementations of qemu_cpu_kick()
> cpu: Track CPUs executing syscalls
> cpu: Implement cpu_thread_is_idle() for qemu-user
> cpu: Introduce cpu_is_paused()
> cpu: Set current_cpu early in qemu-user
> cpu: Allow pausing and resuming CPUs in qemu-user
> gdbstub: Pause all CPUs before sending stop replies
> tests/tcg: Stress test thread breakpoints
>
> accel/tcg/user-exec-stub.c | 4 -
> accel/tcg/user-exec.c | 55 ++++++
> bsd-user/aarch64/target_arch_cpu.h | 6 +-
> bsd-user/arm/target_arch_cpu.h | 5 +-
> bsd-user/freebsd/os-syscall.c | 10 +
> bsd-user/i386/target_arch_cpu.h | 5 +-
> bsd-user/main.c | 8 +-
> bsd-user/x86_64/target_arch_cpu.h | 5 +-
> cpu-common.c | 179 ++++++++++++++++++
> gdbstub/gdbstub.c | 17 +-
> gdbstub/internals.h | 4 +-
> gdbstub/syscalls.c | 20 +-
> gdbstub/system.c | 18 +-
> gdbstub/user.c | 28 ++-
> include/exec/cpu-common.h | 15 ++
> include/exec/replay-core.h | 13 ++
> include/hw/core/cpu.h | 1 +
> include/qemu/thread-posix.h | 8 +
> include/qemu/thread-win32.h | 8 +
> include/sysemu/cpus.h | 6 -
> include/sysemu/replay.h | 13 --
> linux-user/aarch64/cpu_loop.c | 5 +-
> linux-user/alpha/cpu_loop.c | 5 +-
> linux-user/arm/cpu_loop.c | 5 +-
> linux-user/hexagon/cpu_loop.c | 5 +-
> linux-user/hppa/cpu_loop.c | 5 +-
> linux-user/i386/cpu_loop.c | 5 +-
> linux-user/loongarch64/cpu_loop.c | 5 +-
> linux-user/m68k/cpu_loop.c | 5 +-
> linux-user/main.c | 9 +-
> linux-user/microblaze/cpu_loop.c | 5 +-
> linux-user/mips/cpu_loop.c | 5 +-
> linux-user/openrisc/cpu_loop.c | 5 +-
> linux-user/ppc/cpu_loop.c | 5 +-
> linux-user/riscv/cpu_loop.c | 5 +-
> linux-user/s390x/cpu_loop.c | 5 +-
> linux-user/sh4/cpu_loop.c | 5 +-
> linux-user/sparc/cpu_loop.c | 5 +-
> linux-user/syscall.c | 12 ++
> linux-user/xtensa/cpu_loop.c | 5 +-
> replay/stubs-system.c | 8 +
> stubs/meson.build | 8 +
> stubs/qemu-timer.c | 6 +
> stubs/replay-mutex.c | 10 +
> stubs/replay-tools.c | 8 -
> system/cpus.c | 172 +----------------
> tests/tcg/multiarch/Makefile.target | 13 +-
> .../gdbstub/test-thread-breakpoint-stress.py | 28 +++
> .../tcg/multiarch/thread-breakpoint-stress.c | 92 +++++++++
> 49 files changed, 552 insertions(+), 327 deletions(-)
> create mode 100644 stubs/qemu-timer.c
> create mode 100644 stubs/replay-mutex.c
> create mode 100644
> tests/tcg/multiarch/gdbstub/test-thread-breakpoint-stress.py
> create mode 100644 tests/tcg/multiarch/thread-breakpoint-stress.c
--
Alex Bennée
Virtualisation Tech Lead @ Linaro
- Re: [PATCH 00/18] Stop all qemu-cpu threads on a breakpoint,
Alex Bennée <=