qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

The question about migration/checkpointing


From: Sergei Lyubski
Subject: The question about migration/checkpointing
Date: Fri, 26 Jul 2024 12:41:47 -0700

  Hi guys,

My name is Sergey. My company uses QEMU for  CPU performance simulation/evaluaton.
Sorry, I found  your emails in QEMU  relatively recent commits  into ram.c .  Why ram.c . Because I see some weirdness in the  behavior.
  
Our goal is :
     a)  To boot  Linux , to launch some  TEST/SPEC, to drop a few checkpoints.
     b)  To resume  execution from  those checkpoints, to attach a performance simulator and
           to evaluate performance at the "hot" code  areas.

  How we tried to do it .

     a) run QEMU with -monitor telnet options
     b) connect  to QEMU with telnet 
     c)  after  LINUX  booted  in  QEMU monitor (telnet)  executed
            migrate -d file://<FULL  CKPT PATH> 
            quit
      d)  run QEMU  with option -incoming file://<FULL CKPT PATH>

  Now !!

  If  I  drop the checkpoint  when running  on linux a long ( endless) test which  prints on      the console,  everything works (!)  and  the console is properly restored.
   
  When I drop  the checkpoint  when the linux console is in shell  prompt, the restoration does  not  work.
   I see either error message  from  ram_load_precopy()
       error_report("Unknown combination of migration flags: 0x%x", flags);

       or

   the console is locked, QEMU restarts and  runs  (ps, top).   The only  solution is to kill 
   QEMU .  After that,  QEMU console is distorted and requires to type RESET.

    Could you please comment on the above issues. 
    If you need more info I can send you my screen shots.
    Please note that  that the similar behavior is observed on  both
     qemu 9.0  ( stable-9.0)  and
     qemu 8.x

Thanks in advance.
Sergey Lyubskiy


reply via email to

[Prev in Thread] Current Thread [Next in Thread]