sysml/uniprof

A stack tracer/profiler for Xen domains

uniprof's Issues

Could not access information for domid, Does domid exist?

I'm using xen which comes with ubuntu 16.04.

root@or-xen-d0:/var/log/xen# xl list
Name                                        ID   Mem VCPUs	State	Time(s)
Domain-0                                     0 23962     2     r-----    3004.6
Mini-OS                                     51    32     1     r-----    2024.6

root@or-xen-d0:/var/log/xen# uniprof -F 1 -T 1 - 51
Could not access information for domid 51. (Does domid 51 exist?)

strace:

root@or-xen-d0:/var/log/xen# strace uniprof -F 1 -T 1 - 51
...
open("/proc/xen/privcmd", O_RDWR)       = 3
fcntl(3, F_GETFD)                       = 0
fcntl(3, F_SETFD, FD_CLOEXEC)           = 0
mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS|MAP_LOCKED, -1, 0) = 0x7f9b995c9000
madvise(0x7f9b995c9000, 4096, MADV_DONTFORK) = 0
ioctl(3, _IOC(0, 0x50, 0x00, 0x30), 0x7ffca3114730) = -1 EACCES (Permission denied)
write(2, "Could not access information for"..., 66Could not access information for domid 51. (Does domid 51 exist?)
) = 66
exit_group(-5)                          = ?

What should I do now?

get_domain_state() signals an error

I'm trying to run uniprof with Unikraft, but unfortunately get_domain_state () throws an error.
I tried with libxc, but also with libxencall and both fail at this point.
For configuration I used this version of Xen.

int get_domain_state(int domid, unsigned int *state) {
	int retval;
#if defined(HYPERCALL_XENCALL)
	struct xen_domctl domctl;
	domctl.domain = (domid_t)domid;
	domctl.interface_version = XEN_DOMCTL_INTERFACE_VERSION;
	domctl.cmd = XEN_DOMCTL_getdomaininfo;
	retval = xencall1(callh, __HYPERVISOR_domctl, (unsigned long)(&domctl));
	*state = domctl.u.getdomaininfo.flags;
	return retval;
#elif defined(HYPERCALL_LIBXC)
	xc_dominfo_t info;
	retval = xc_domain_getinfo(xc_handle, domid, 1, &info);
	*state |= (info.shutdown_reason << XEN_DOMINF_shutdownshift);
	if (info.dying)
		*state |= XEN_DOMINF_dying;
	if (info.hvm)
		*state |= XEN_DOMINF_hvm_guest;
	if (info.shutdown || info.crashed)
		*state |= XEN_DOMINF_shutdown;
	if (info.paused)
		*state |= XEN_DOMINF_paused;
	if (info.blocked)
		*state |= XEN_DOMINF_blocked;
	if (info.running)
		*state |= XEN_DOMINF_running;
	if (info.debugged)
		*state |= XEN_DOMINF_debugged;
	if (retval == 1)
		return 0;
	return retval;
#endif
}

In this function xencall1 and xc_domain_getinfo always return -1 so what I think is that hypercalls don't work as they should. But I don't know what the cause is.

bus error

Just mucking around with an existing Xen server and Linux guests:

# ./uniprof out.prof 27
Bus error (core dumped)
# gdb ./uniprof cone
[...]
Program terminated with signal SIGBUS, Bus error.
#0  __memmove_sse2_unaligned_erms () at ../sysdeps/x86_64/multiarch/../multiarch/memmove-vec-unaligned-erms.S:292
292	../sysdeps/x86_64/multiarch/../multiarch/memmove-vec-unaligned-erms.S: No such file or directory.
(gdb) bt
#0  __memmove_sse2_unaligned_erms () at ../sysdeps/x86_64/multiarch/../multiarch/memmove-vec-unaligned-erms.S:292
#1  0x000055df7f056024 in walk_stack_fp ()
#2  0x000055df7f056175 in do_stack_trace_fp ()
#3  0x000055df7f0554f0 in main ()

Of course, I also had to unpause the domain to let it continue:

# xl list
Name                                        ID   Mem VCPUs	State	Time(s)
Domain-0                                     0  2438     4     r-----    9222.0
vm0                                          1   512     1     r-----     167.5
vm2hvm                                      27   512     3     --p---     157.2
vm1hvm                                      28   512     1     r-----     277.9
# xl unpause vm2hvm

If it's crashing when trying to walk frame pointers, then sure, Linux is likely running things that aren't using them...

It doesn't crash all the time. It does sometimes work.