QEMU Version 1.4.0 - SLIRP hangs VM

(Note: problem is not present in version 1.3.0)

Stacktrace: please see attached gdb log file.

Steps to reproduce:

1. gdb -x debug-qemu.gdb testing/qemu-1.4.0/ppc64-softmmu/qemu-system-ppc64

Contents of debug-qemu.gdb:

run -L ./testing/qemu-1.4.0/pc-bios  -name "[DEBUG] Software und System-Entwicklung IBM POWER7" -cpu POWER7_v2.3 -M pseries -m 1024 -rtc base=utc -nodefaults -vga std -monitor vc -serial vc -netdev type=user,id=mynet0,hostfwd=tcp:127.0.0.1:9011-10.0.2.11:22 -device virtio-net-pci,netdev=mynet0 -drive file=images/suse-ppc.img,if=virtio,index=0,media=disk,cache=unsafe -kernel images/iso/suseboot/vmlinux -append "root=/dev/mapper/system-root ro audit=0 selinux=0 apparmor=0" -initrd images/iso/suseboot/initrd.img -drive if=scsi,bus=0,unit=0,media=cdrom


2. build information
    QEMU: ppc64 full emulation, version 1.4.0
    Host OS: Windows XP
    Guest OS: openSUSE 12.2 kernel 3.4.6-2.10-ppc64

    PATH=/usr/i686-pc-mingw32/sys-root/mingw/bin:$PATH
    PKG_CONFIG_LIBDIR=/usr/i686-pc-mingw32/sys-root/mingw/lib/pkgconfig
    THREADS=4
    export PKG_CONFIG_LIBDIR
    
    sed -i 's/--static-libs/--static --libs/' configure
    CC=i686-pc-mingw32-gcc ./configure \
          --target-list=ppc64-softmmu \
          --enable-debug \
          --enable-sdl \
          --static \
          --enable-fdt && \
    sed -i 's/ -Wl,--dynamicbase//g; s/-Wl,--nxcompat //g;'  config-host.mak && \
    make -j$THREADS && {
          echo "renaming binw.exe to bin.exe..."
          for i in `echo $TARGET_LIST | tr ',' ' '`; do
                 BINARCH=`echo $i | sed 's/-softmmu//'`
                 mv $i/qemu-system-${BINARCH}w.exe \
                        $i/qemu-system-$BINARCH.exe
          done
    }

   
3. From VM: 
    Command to hang VM: zypper dup
    Last message before VM hang:  Retrieving repository 'openSUSE-12.2-12.2-0' metadata -----------------------[|]



On Fri, Mar 22, 2013 at 05:51:14PM -0000, Kenneth Salerno wrote:
> Public bug reported:
> 
> (Note: problem is not present in version 1.3.0)

Please run git-bisect(1) to identify the commit that introduced the bug.

Information about git-bisect(1):
http://git-scm.com/book/en/Git-Tools-Debugging-with-Git#Binary-Search
http://code-worrier.com/blog/git-bisect-basics/
https://www.kernel.org/pub/software/scm/git/docs/git-bisect.html

Stefan


Result of git bisect:

commit 837d1f978224f7e7b020c71ffb10b291952cc596
Merge: a6fc23e 2b35e93
Author: Blue Swirl <email address hidden>
Date:   Sat Jan 12 12:46:57 2013 +0000

    Merge branch 's390-reorg' of git://repo.or.cz/qemu/rth
    
    * 's390-reorg' of git://repo.or.cz/qemu/rth: (149 commits)
      target-s390: Claim maintainership
      target-s390: Use noreturn for exception and load_psw
      target-s390: Use TCG_CALL_NO_WG for misc helpers
      target-s390: Use TCG_CALL_NO_WG for integer helpers
      target-s390: Use TCG_CALL_NO_WG for floating-point helpers
      target-s390: Use TCG_CALL_NO_WG for memory helpers
      target-s390: Perform COMPARE AND SWAP inline
      target-s390: Optimize get_address
      target-s390: Optimize ADDC/SUBB
      target-s390: Optimize ADDU/SUBU CC testing
      target-s390: Tidy comparisons
      target-s390: Optmize emitting discards
      target-s390: Optimize XC
      target-s390: Fix cpu_clone_regs
      target-s390: Implement LOAD/SET FP AND SIGNAL
      target-s390: Implement SET ROUNDING MODE
      target-s390: Use uint64_to_float128
      target-s390: Implement LCDFR
      target-s390: Check insn operand specifications
      target-s390: Implement CPSDR
      ...


There are 1,484 files affected by this commit - Is there a way I can narrow this down even further within this commit?

Also attaching "strace -f" output from guest during execution of "zypper dup" leading to hang, as well as corresponding qemu gdb log following hang.

The last lines of the strace are:

time(NULL)                              = 1364222018
stat("/etc/localtime", {st_mode=S_IFREG|0644, st_size=3519, ...}) = 0
uname({sys="Linux", node="suse-ppc", ...}) = 0
write(3, "2013-03-25 10:33:38 <1> suse-ppc"..., 127) = 127
time(NULL)                              = 1364222018
stat("/etc/localtime", {st_mode=S_IFREG|0644, st_size=3519, ...}) = 0
uname({sys="Linux", node="suse-ppc", ...}) = 0
write(3, "2013-03-25 10:33:38 <1> suse-ppc"..., 137) = 137
close(9)                                = 0
time(NULL)                              = 1364222018
stat("/etc/localtime", {st_mode=S_IFREG|0644, st_size=3519, ...}) = 0
uname({sys="Linux", node="suse-ppc", ...}) = 0
write(3, "2013-03-25 10:33:38 <1> suse-ppc"..., 127) = 127
time(NULL)                              = 1364222018
stat("/etc/localtime", {st_mode=S_IFREG|0644, st_size=3519, ...}) = 0
uname({sys="Linux", node="suse-ppc", ...}) = 0
write(3, "2013-03-25 10:33:38 <1> suse-ppc"..., 137   <------------ QEMU window becomes unresponsive



On Mon, Mar 25, 2013 at 08:57:46PM -0000, Kenneth Salerno wrote:
> Result of git bisect:
> 
> commit 837d1f978224f7e7b020c71ffb10b291952cc596
> Merge: a6fc23e 2b35e93
> Author: Blue Swirl <email address hidden>
> Date:   Sat Jan 12 12:46:57 2013 +0000
> 
>     Merge branch 's390-reorg' of git://repo.or.cz/qemu/rth
>     
>     * 's390-reorg' of git://repo.or.cz/qemu/rth: (149 commits)
>       target-s390: Claim maintainership
>       target-s390: Use noreturn for exception and load_psw
>       target-s390: Use TCG_CALL_NO_WG for misc helpers
>       target-s390: Use TCG_CALL_NO_WG for integer helpers
>       target-s390: Use TCG_CALL_NO_WG for floating-point helpers
>       target-s390: Use TCG_CALL_NO_WG for memory helpers
>       target-s390: Perform COMPARE AND SWAP inline
>       target-s390: Optimize get_address
>       target-s390: Optimize ADDC/SUBB
>       target-s390: Optimize ADDU/SUBU CC testing
>       target-s390: Tidy comparisons
>       target-s390: Optmize emitting discards
>       target-s390: Optimize XC
>       target-s390: Fix cpu_clone_regs
>       target-s390: Implement LOAD/SET FP AND SIGNAL
>       target-s390: Implement SET ROUNDING MODE
>       target-s390: Use uint64_to_float128
>       target-s390: Implement LCDFR
>       target-s390: Check insn operand specifications
>       target-s390: Implement CPSDR
>       ...
> 
> 
> There are 1,484 files affected by this commit - Is there a way I can narrow this down even further within this commit?

Looks like git-bisect(1) got stuck on a merge - maybe it followed the
wrong parent commit.

Please manually check whether a6fc23e hangs:

  $ cd qemu
  $ git checkout a6fc23e
  $ ...build and test hang...

If a6fc23e is indeed broken then you can start a new git bisect between
v1.3.0 (good) and a6fc23e (bad).

If a6fc23e is not broken then that really suggestes the 's390-reorg'
merge broke SLIRP, which is unlikely.

Hope this helps.

Regarding the QEMU backtrace, it shows SLIRP is waiting in a blocking
accept() call:

Thread 1 (Thread 416.0x228):
#0  0x7c90e514 in ntdll!KiFastSystemCallRet ()
   from /cygdrive/c/WINDOWS/system32/ntdll.dll
#1  0x7c90df5a in ntdll!ZwWaitForSingleObject ()
   from /cygdrive/c/WINDOWS/system32/ntdll.dll
#2  0x71a5402b in ?? () from /cygdrive/c/WINDOWS/system32/mswsock.dll
#3  0x71a6b858 in StartWsdpService ()
   from /cygdrive/c/WINDOWS/system32/mswsock.dll
#4  0x71ac0e46 in WSAAccept () from /cygdrive/c/WINDOWS/system32/WS2_32.dll
#5  0x71ac1057 in accept () from /cygdrive/c/WINDOWS/system32/WS2_32.dll
#6  0x0057caff in tcp_connect (inso=0x1c07c68) at slirp/tcp_subr.c:423
#7  0x005763d0 in slirp_select_poll (readfds=0xa19a40 <rfds>, 
    writefds=0xa19b60 <wfds>, xfds=0xa19c80 <xfds>, select_error=0)
    at slirp/slirp.c:485
#8  0x0053688b in main_loop_wait (nonblocking=0) at main-loop.c:419
#9  0x005a8333 in main_loop () at vl.c:2001
#10 0x005af134 in qemu_main (argc=34, argv=0x1c08048, envp=0x0) at vl.c:4326
#11 0x005a4d84 in SDL_main (argc=34, argv=0x1c08048) at vl.c:102
#12 0x0073a951 in console_main ()
#13 0x00000022 in ?? ()
#14 0x01c08048 in ?? ()

Since the main loop holds the global mutex the guest is unable to make
progress and appears to hang.

Stefan


Sorry for the confusion, I was impatient for the first bisect run to complete - this time I figured out how to automate the testing portion of the git bisect run script so I could walk away and let it run until full completion.

Here is the result:

acbb090b2400f627a801074c4e3e006c7501bb26 is the first bad commit
commit acbb090b2400f627a801074c4e3e006c7501bb26
Author: Andreas Färber <email address hidden>
Date:   Wed Aug 15 14:15:41 2012 +0200

    prep: Include devices for ppc64 as well
    
    Allows running qemu-system-ppc64 -M prep for consistency.
    
    Reported-by: Markus Armbruster <email address hidden>
    Signed-off-by: Andreas Färber <email address hidden>
    Acked-by: Hervé Poussineau <email address hidden>

:040000 040000 efe2ed40eeef1863d210ab089033fdb0ce1eaea5 05c5174a00f99176a57398b32c5af659b8b0096c M default-configs
bisect run success

Triaging old bug tickets... can you still reproduce this issue with the latest version of QEMU? Or could we close this ticket nowadays?

Hi, you can close this bug.
Ken

Sent from Yahoo Mail on Android 
 
  On Tue, Jan 2, 2018 at 11:19 AM, Thomas Huth<email address hidden> wrote:   Triaging old bug tickets... can you still reproduce this issue with the
latest version of QEMU? Or could we close this ticket nowadays?

** Changed in: qemu
      Status: New => Incomplete

-- 
You received this bug notification because you are subscribed to the bug
report.
https://bugs.launchpad.net/bugs/1158912

Title:
  QEMU Version 1.4.0 - SLIRP hangs VM

Status in QEMU:
  Incomplete

Bug description:
  (Note: problem is not present in version 1.3.0)

  Stacktrace: please see attached gdb log file.

  Steps to reproduce:

  1. gdb -x debug-qemu.gdb testing/qemu-1.4.0/ppc64-softmmu/qemu-system-
  ppc64

  Contents of debug-qemu.gdb:

  run -L ./testing/qemu-1.4.0/pc-bios  -name "[DEBUG] Software und
  System-Entwicklung IBM POWER7" -cpu POWER7_v2.3 -M pseries -m 1024
  -rtc base=utc -nodefaults -vga std -monitor vc -serial vc -netdev
  type=user,id=mynet0,hostfwd=tcp:127.0.0.1:9011-10.0.2.11:22 -device
  virtio-net-pci,netdev=mynet0 -drive file=images/suse-
  ppc.img,if=virtio,index=0,media=disk,cache=unsafe -kernel
  images/iso/suseboot/vmlinux -append "root=/dev/mapper/system-root ro
  audit=0 selinux=0 apparmor=0" -initrd images/iso/suseboot/initrd.img
  -drive if=scsi,bus=0,unit=0,media=cdrom

  
  2. build information
      QEMU: ppc64 full emulation, version 1.4.0
      Host OS: Windows XP
      Guest OS: openSUSE 12.2 kernel 3.4.6-2.10-ppc64

      PATH=/usr/i686-pc-mingw32/sys-root/mingw/bin:$PATH
      PKG_CONFIG_LIBDIR=/usr/i686-pc-mingw32/sys-root/mingw/lib/pkgconfig
      THREADS=4
      export PKG_CONFIG_LIBDIR
      
      sed -i 's/--static-libs/--static --libs/' configure
      CC=i686-pc-mingw32-gcc ./configure \
            --target-list=ppc64-softmmu \
            --enable-debug \
            --enable-sdl \
            --static \
            --enable-fdt && \
      sed -i 's/ -Wl,--dynamicbase//g; s/-Wl,--nxcompat //g;'  config-host.mak && \
      make -j$THREADS && {
            echo "renaming binw.exe to bin.exe..."
            for i in `echo $TARGET_LIST | tr ',' ' '`; do
                  BINARCH=`echo $i | sed 's/-softmmu//'`
                  mv $i/qemu-system-${BINARCH}w.exe \
                          $i/qemu-system-$BINARCH.exe
            done
      }

    
  3. From VM: 
      Command to hang VM: zypper dup
      Last message before VM hang:  Retrieving repository 'openSUSE-12.2-12.2-0' metadata -----------------------[|]

To manage notifications about this bug go to:
https://bugs.launchpad.net/qemu/+bug/1158912/+subscriptions