Re: [Qemu-devel] Status of my hacks on the MTTCG WIP branch

[Alex Bennée]

Alex Bennée <address@hidden> writes:

> Pranith Kumar <address@hidden> writes:
>
>> Hi Alex,
>>
>> On Tue, Jan 12, 2016 at 12:29 PM, Alex Bennée <address@hidden>
>> wrote:
>>
>>>
>> https://github.com/stsquad/qemu/tree/mttcg/multi_tcg_v8_wip_ajb_fix_locks
>>>
>>
>> I built this branch and ran an arm64 guest. It seems to be failing
>> similarly to what I reported earlier:
>>
>> #0  0x00007ffff2211cc9 in __GI_raise (address@hidden) at
>> ../nptl/sysdeps/unix/sysv/linux/raise.c:56
>> #1  0x00007ffff22150d8 in __GI_abort () at abort.c:89
>> #2  0x000055555572014c in qemu_ram_addr_from_host_nofail
>> (ptr=0xffffffc000187863) at /home/pranith/devops/code/qemu/cputlb.c:357
>> #3  0x00005555557209dd in get_page_addr_code (env1=0x555556702058,
>> addr=18446743798833248356) at /home/pranith/devops/code/qemu/cputlb.c:568
>> #4  0x00005555556db98c in tb_find_physical (cpu=0x5555566f9dd0,
>> pc=18446743798833248356, cs_base=0, flags=18446744071830503424) at
>> /home/pranith/devops/code/qemu/cpu-exec.c:224
>> #5  0x00005555556dbaf4 in tb_find_slow (cpu=0x5555566f9dd0,
>> pc=18446743798833248356, cs_base=0, flags=18446744071830503424) at
>> /home/pranith/devops/code/qemu/cpu-exec.c:268
>> #6  0x00005555556dbc77 in tb_find_fast (cpu=0x5555566f9dd0) at
>> /home/pranith/devops/code/qemu/cpu-exec.c:311
>> #7  0x00005555556dc0f1 in cpu_arm_exec (cpu=0x5555566f9dd0) at
>> /home/pranith/devops/code/qemu/cpu-exec.c:492
>> #8  0x00005555557050ee in tcg_cpu_exec (cpu=0x5555566f9dd0) at
>> /home/pranith/devops/code/qemu/cpus.c:1486
>> #9  0x00005555557051af in tcg_exec_all (cpu=0x5555566f9dd0) at
>> /home/pranith/devops/code/qemu/cpus.c:1515
>> #10 0x0000555555704800 in qemu_tcg_cpu_thread_fn (arg=0x5555566f9dd0) at
>> /home/pranith/devops/code/qemu/cpus.c:1187
>> #11 0x00007ffff25a8182 in start_thread (arg=0x7fffd20c8700) at
>> pthread_create.c:312
>> #12 0x00007ffff22d547d in clone () at
>> ../sysdeps/unix/sysv/linux/x86_64/clone.S:111
<snip>

Having seen a backtrace of a crash while the other thread was flushing
the TLB entries I sprinkled a bunch of:

  g_assert(cpu == current_cpu);

In all public functions in cputlb that took a CPU. There are a bunch of
cases that don't defer actions across CPUs which need to be fixed up. I
suspect they don't hit in the arm case because the type of TLB flushing
pattern is different. In aarch64 it my backtrace it was triggered by
tlbi_aa64_vae1is_write:

  7    Thread 0x7ffe777fe700 (LWP 32705) "worker" sem_timedwait () at 
../nptl/sysdeps/unix/sysv/linux/x86_64/sem_timedwait.S:101
  6    Thread 0x7ffe77fff700 (LWP 32704) "worker" sem_timedwait () at 
../nptl/sysdeps/unix/sysv/linux/x86_64/sem_timedwait.S:101
  5    Thread 0x7fff8d9d0700 (LWP 32703) "CPU 1/TCG" 0x000055555572cc18 in 
memcpy (__len=8, __src=<synthetic pointer>, __dest=<optimised out>)
    at /usr/include/x86_64-linux-gnu/bits/string3.h:51
* 4    Thread 0x7fff8e1d1700 (LWP 32702) "CPU 0/TCG" memset () at 
../sysdeps/x86_64/memset.S:94
  3    Thread 0x7fff8f1cb700 (LWP 32701) "worker" sem_timedwait () at 
../nptl/sysdeps/unix/sysv/linux/x86_64/sem_timedwait.S:101
  2    Thread 0x7fffe45c8700 (LWP 32700) "qemu-system-aar" syscall () at 
../sysdeps/unix/sysv/linux/x86_64/syscall.S:38
  1    Thread 0x7ffff7f98c00 (LWP 32696) "qemu-system-aar" 0x00007ffff0ba01ef 
in __GI_ppoll (fds=0x5555575cb5b0, nfds=8, timeout=<optimised out>,
    address@hidden, address@hidden) at ../sysdeps/unix/sysv/linux/ppoll.c:56
#0  memset () at ../sysdeps/x86_64/memset.S:94
#1  0x0000555555728bee in memset (__len=32768, __ch=0, __dest=0x555556632568) 
at /usr/include/x86_64-linux-gnu/bits/string3.h:84
#2  v_tlb_flush_by_mmuidx (argp=0x7fff8e1d0430, cpu=0x555556632380) at 
/home/alex/lsrc/qemu/qemu.git/cputlb.c:136
#3  tlb_flush_page_by_mmuidx (address@hidden, address@hidden) at 
/home/alex/lsrc/qemu/qemu.git/cputlb.c:243
#4  0x00005555557fcb4a in tlbi_aa64_vae1is_write (env=<optimised out>, 
ri=<optimised out>, value=<optimised out>)
    at /home/alex/lsrc/qemu/qemu.git/target-arm/helper.c:2757
#5  0x00007fffa441dac5 in code_gen_buffer ()
#6  0x00005555556eef4b in cpu_tb_exec (tb_ptr=<optimised out>, 
cpu=0x5555565eddd0) at /home/alex/lsrc/qemu/qemu.git/cpu-exec.c:157
#7  cpu_arm_exec (address@hidden) at 
/home/alex/lsrc/qemu/qemu.git/cpu-exec.c:520
#8  0x00005555557108e8 in tcg_cpu_exec (cpu=0x5555565eddd0) at 
/home/alex/lsrc/qemu/qemu.git/cpus.c:1486
#9  tcg_exec_all (cpu=0x5555565eddd0) at 
/home/alex/lsrc/qemu/qemu.git/cpus.c:1515
#10 qemu_tcg_cpu_thread_fn (arg=0x5555565eddd0) at 
/home/alex/lsrc/qemu/qemu.git/cpus.c:1187
#11 0x00007ffff0e80182 in start_thread (arg=0x7fff8e1d1700) at 
pthread_create.c:312
#12 0x00007ffff0bad47d in clone () at 
../sysdeps/unix/sysv/linux/x86_64/clone.S:111
[Switching to thread 5 (Thread 0x7fff8d9d0700 (LWP 32703))]
#0  0x000055555572cc18 in memcpy (__len=8, __src=<synthetic pointer>, 
__dest=<optimised out>) at /usr/include/x86_64-linux-gnu/bits/string3.h:51
51        return __builtin___memcpy_chk (__dest, __src, __len, __bos0 (__dest));
#0  0x000055555572cc18 in memcpy (__len=8, __src=<synthetic pointer>, 
__dest=<optimised out>) at /usr/include/x86_64-linux-gnu/bits/string3.h:51
#1  stq_he_p (v=<optimised out>, ptr=<optimised out>) at 
/home/alex/lsrc/qemu/qemu.git/include/qemu/bswap.h:292
#2  stq_le_p (v=547973099520, ptr=<optimised out>) at 
/home/alex/lsrc/qemu/qemu.git/include/qemu/bswap.h:327
#3  helper_le_stq_mmu (env=0x55555663a608, addr=18446743801961580216, 
val=547973099520, oi=<optimised out>, retaddr=140735948385557)
    at /home/alex/lsrc/qemu/qemu.git/softmmu_template.h:455
#4  0x00007fffa435ed17 in code_gen_buffer ()
#5  0x00005555556eef4b in cpu_tb_exec (tb_ptr=<optimised out>, 
cpu=0x555556632380) at /home/alex/lsrc/qemu/qemu.git/cpu-exec.c:157
#6  cpu_arm_exec (address@hidden) at 
/home/alex/lsrc/qemu/qemu.git/cpu-exec.c:520
#7  0x00005555557108e8 in tcg_cpu_exec (cpu=0x555556632380) at 
/home/alex/lsrc/qemu/qemu.git/cpus.c:1486
#8  tcg_exec_all (cpu=0x555556632380) at 
/home/alex/lsrc/qemu/qemu.git/cpus.c:1515
#9  qemu_tcg_cpu_thread_fn (arg=0x555556632380) at 
/home/alex/lsrc/qemu/qemu.git/cpus.c:1187
#10 0x00007ffff0e80182 in start_thread (arg=0x7fff8d9d0700) at 
pthread_create.c:312
#11 0x00007ffff0bad47d in clone () at 
../sysdeps/unix/sysv/linux/x86_64/clone.S:111
A debugging session is active.

Needless to say anything messing with structures used by the other
threads needs to take great care or doom will occur ;-)

I'll look at fixing them up in my tree while Fred finishes his re-base.

--
Alex Bennée