[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[PULL 06/23] vl.c: do not execute trace_init_backends() before daemonizi
From: |
Paolo Bonzini |
Subject: |
[PULL 06/23] vl.c: do not execute trace_init_backends() before daemonizing |
Date: |
Sat, 6 Mar 2021 11:54:02 +0100 |
From: Daniel Henrique Barboza <danielhb413@gmail.com>
Commit v5.2.0-190-g0546c0609c ("vl: split various early command line
options to a separate function") moved the trace backend init code to
the qemu_process_early_options(). Which is now being called before
os_daemonize() via qemu_maybe_daemonize().
Turns out that this change of order causes a problem when executing
QEMU in daemon mode and with CONFIG_TRACE_SIMPLE. The trace thread
is now being created by the parent, and the parent is left waiting for
a trace file flush that was registered via st_init(). The result is
that the parent process never exits.
To reproduce, fire up a QEMU process with -daemonize and with
CONFIG_TRACE_SIMPLE enabled. Two QEMU process will be left in the
host:
$ sudo ./x86_64-softmmu/qemu-system-x86_64 -S -no-user-config -nodefaults \
-nographic -machine none,accel=kvm:tcg -daemonize
$ ps axf | grep qemu
529710 pts/3 S+ 0:00 | \_ grep --color=auto qemu
529697 ? Ssl 0:00 \_ ./x86_64-softmmu/qemu-system-x86_64 -S
-no-user-config -nodefaults -nographic -machine none,accel=kvm:tcg -daemonize
529699 ? Sl 0:00 \_ ./x86_64-softmmu/qemu-system-x86_64 -S
-no-user-config -nodefaults -nographic -machine none,accel=kvm:tcg -daemonize
The parent thread is hang in flush_trace_file:
$ sudo gdb ./x86_64-softmmu/qemu-system-x86_64 529697
(..)
(gdb) bt
#0 0x00007f9dac6a137d in syscall () at /lib64/libc.so.6
#1 0x00007f9dacc3c4f3 in g_cond_wait () at /lib64/libglib-2.0.so.0
#2 0x0000555d12f952da in flush_trace_file (wait=true) at ../trace/simple.c:140
#3 0x0000555d12f95b4c in st_flush_trace_buffer () at ../trace/simple.c:383
#4 0x00007f9dac5e43a7 in __run_exit_handlers () at /lib64/libc.so.6
#5 0x00007f9dac5e4550 in on_exit () at /lib64/libc.so.6
#6 0x0000555d12d454de in os_daemonize () at ../os-posix.c:255
#7 0x0000555d12d0bd5c in qemu_maybe_daemonize (pid_file=0x0) at
../softmmu/vl.c:2408
#8 0x0000555d12d0e566 in qemu_init (argc=8, argv=0x7fffc594d9b8,
envp=0x7fffc594da00) at ../softmmu/vl.c:3459
#9 0x0000555d128edac1 in main (argc=8, argv=0x7fffc594d9b8,
envp=0x7fffc594da00) at ../softmmu/main.c:49
(gdb)
Aside from the 'zombie' process in the host, this is directly impacting
Libvirt. Libvirt waits for the parent process to exit to be sure that the
QMP monitor is available in the daemonized process to fetch QEMU
capabilities, and as is now Libvirt hangs at daemon start waiting
for the parent thread to exit.
The fix is simple: just move the trace backend related code back to
be executed after daemonizing.
Fixes: 0546c0609cb5a8d90c1cbac8e0d64b5a048bbb19
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20210105181437.538366-2-danielhb413@gmail.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
softmmu/vl.c | 18 +++++++++++++-----
1 file changed, 13 insertions(+), 5 deletions(-)
diff --git a/softmmu/vl.c b/softmmu/vl.c
index 10bd8a10a3..7e8110bd6e 100644
--- a/softmmu/vl.c
+++ b/softmmu/vl.c
@@ -2361,11 +2361,6 @@ static void qemu_process_early_options(void)
cleanup_add_fd, NULL, &error_fatal);
#endif
- if (!trace_init_backends()) {
- exit(1);
- }
- trace_init_file();
-
/* Open the logfile at this point and set the log mask if necessary. */
qemu_set_log_filename(log_file, &error_fatal);
if (log_mask) {
@@ -3475,6 +3470,19 @@ void qemu_init(int argc, char **argv, char **envp)
qemu_process_help_options();
qemu_maybe_daemonize(pid_file);
+ /*
+ * The trace backend must be initialized after daemonizing.
+ * trace_init_backends() will call st_init(), which will create the
+ * trace thread in the parent, and also register st_flush_trace_buffer()
+ * in atexit(). This function will force the parent to wait for the
+ * writeout thread to finish, which will not occur, and the parent
+ * process will be left in the host.
+ */
+ if (!trace_init_backends()) {
+ exit(1);
+ }
+ trace_init_file();
+
qemu_init_main_loop(&error_fatal);
cpu_timers_init();
--
2.29.2
- [PULL 00/23] Misc patches for 2021-03-06, Paolo Bonzini, 2021/03/06
- [PULL 01/23] KVM: x86: do not fail if software breakpoint has already been removed, Paolo Bonzini, 2021/03/06
- [PULL 02/23] KVM: x86: deprecate -M kernel-irqchip=off except for -M isapc, Paolo Bonzini, 2021/03/06
- [PULL 03/23] qga-vss: Use dynamic linking for GLib, Paolo Bonzini, 2021/03/06
- [PULL 04/23] chardev: add nodelay option, Paolo Bonzini, 2021/03/06
- [PULL 05/23] qom: Check for wellformed id in user_creatable_add_type(), Paolo Bonzini, 2021/03/06
- [PULL 06/23] vl.c: do not execute trace_init_backends() before daemonizing,
Paolo Bonzini <=
- [PULL 07/23] accel: kvm: Fix memory waste under mismatch page size, Paolo Bonzini, 2021/03/06
- [PULL 10/23] elf_ops: correct loading of 32 bit PVH kernel, Paolo Bonzini, 2021/03/06
- [PULL 11/23] x86/pvh: extract only 4 bytes of start address for 32 bit kernels, Paolo Bonzini, 2021/03/06
- [PULL 13/23] scsi: Rename linux-specific SG_ERR codes to generic SCSI_HOST error codes, Paolo Bonzini, 2021/03/06
- [PULL 12/23] qemu-config: add error propagation to qemu_config_parse, Paolo Bonzini, 2021/03/06
- [PULL 15/23] scsi-generic: do not snoop the output of failed commands, Paolo Bonzini, 2021/03/06
- [PULL 16/23] scsi: inline sg_io_sense_from_errno() into the callers., Paolo Bonzini, 2021/03/06
- [PULL 20/23] meson: adjust timeouts for some slower tests, Paolo Bonzini, 2021/03/06
- [PULL 21/23] trace: fix "-trace file=...", Paolo Bonzini, 2021/03/06
- [PULL 22/23] trace: skip qemu_set_log_filename if no "-D" option was passed, Paolo Bonzini, 2021/03/06