firecracker: Cannot obtain Guest OS call graph

To profile the whole processing pipeline from the host to the guest and back for my toy TCP server, I am trying to obtain the call graph of the guest and host from the host to generate a per-syscall/kernel-function-call/guest-kernel-function-call breakdown of the latency per packet. I would like to get the following info (copied from here):

            38770.00 20.4% __ticket_spin_lock        [guest.kernel.kallsyms]
            22560.00 11.9% ftrace_likely_update      [kernel.kallsyms]
             9208.00  4.8% __lock_acquire            [kernel.kallsyms]

However, perf produces no samples in perf.data for the guest OS.

sudo perf kvm --guest --guestmount=/tmp/guestmount record -a sleep 10
sudo perf kvm --guest --guestmount=/tmp/guestmount record -a -e cpu-clock:GH -g -o perf.data
^C[ perf record: Woken up 10 times to write data ]
[ perf record: Captured and wrote 3.720 MB perf.data (25457 samples) ]

sudo perf kvm --guest --guestmount=/tmp/guestmount report -i perf.data --stdio
# To display the perf.data header info, please use --header/--header-only options.
#
#
# Total Lost Samples: 0
#
# Samples: 25K of event 'cpu-clock:GH'
# Event count (approx.): 6364250000
# Children      Self  Command  Shared Object  Symbol
# ........  ........  .......  .............  ......
#

I recompiled the guest kernel with all the appropriate parameters. I am able to get the stack for the host (no guest kallsyms can be found in the output):

sudo perf kvm --host --guest --guestmount=/tmp/guestmount record -a -F 99 -g -o perf.data
sudo perf kvm --host --guest --guestmount=/tmp/guestmount report -i perf.data --stdio

# Samples: 40K of event 'cycles:ppp'
# Event count (approx.): 20557903667
#
# Overhead  Command          Shared Object            Symbol                                                                                                    
# ........  ...............  .......................  ..........................................................................................................
#
    38.37%  fc_vcpu0         [kernel.kallsyms]        [k] vmx_vcpu_run
     7.03%  fc_vcpu0         [kernel.kallsyms]        [k] kvm_lapic_find_highest_irr
     5.21%  fc_vcpu0         [kernel.kallsyms]        [k] kvm_arch_vcpu_runnable
     4.19%  fc_vcpu0         [kernel.kallsyms]        [k] __apic_update_ppr
     4.06%  fc_vcpu0         [kernel.kallsyms]        [k] ktime_get
     2.20%  fc_vcpu0         [kernel.kallsyms]        [k] kvm_vcpu_check_block

and from inside the guest:

perf record -F 99 -p 2936 -g sleep 10

# Samples: 218  of event 'cpu-clock'
# Event count (approx.): 2202020180
#
# Children      Self  Command          Shared Object        Symbol                            
# ........  ........  ...............  ...................  ..................................
#
    43.58%     5.96%  tcp_server  ld-musl-x86_64.so.1  [.] __clock_gettime
            |          
            |--40.37%--__clock_gettime
            |          |          
            |          |--36.70%--__vdso_clock_gettime
            |          |          |          
            |          |          |--15.60%--0x7ffc6e0d9991

But unfortunately not both host and guest call graph combined… I would really appreciate any advice or suggestions on how I could fix or troubleshoot this problem.