Re: [PULL 00/28] Migration pull patches

[Juan Quintela]

Daniel P. Berrangé <address@hidden> wrote:
>> I also saw this on aarch32 host (more precisely, on the
>> aarch32-environment-in-aarch64-chroot setup I use for aarch32 build
>> and test):
>> 
>> malloc_consolidate(): invalid chunk size
>> Broken pipe
>> qemu-system-i386: check_section_footer: Read section footer failed: -5
>> qemu-system-i386: load of migration failed: Invalid argument
>> /home/peter.maydell/qemu/tests/libqtest.c:140: kill_qemu() tried to
>> terminate QEMU process but encountered exit status 1 (expected 0)
>> Aborted
>> ERROR - too few tests run (expected 14, got 13)
>> 
>> The memory corruption is reproducible running just the
>> /x86_64/migration/multifd/tcp subtest:
>> 
>> (armhf)pmaydell@mustang-maydell:~/qemu/build/all-a32$
>> QTEST_QEMU_BINARY=x86_64-softmmu/qemu-system-x86_64
>> tests/migration-test -p /x86_64/migration/multifd/tcp
>> /x86_64/migration/multifd/tcp: qemu-system-x86_64: -accel kvm: invalid
>> accelerator kvm
>> qemu-system-x86_64: falling back to tcg
>> qemu-system-x86_64: -accel kvm: invalid accelerator kvm
>> qemu-system-x86_64: falling back to tcg
>> qemu-system-x86_64: multifd_send_sync_main: multifd_send_pages fail
>> qemu-system-x86_64: failed to save SaveStateEntry with id(name): 3(ram)
>> double free or corruption (!prev)
>> Broken pipe
>> qemu-system-x86_64: Unknown combination of migration flags: 0
>> qemu-system-x86_64: error while loading state section id 3(ram)
>> qemu-system-x86_64: load of migration failed: Invalid argument
>> /home/peter.maydell/qemu/tests/libqtest.c:140: kill_qemu() tried to
>> terminate QEMU process but encountered exit status 1 (expected 0)
>> Aborted
>> 
>> Here's what a valgrind run in that aarch32 setup produces:
>> 
>
> Missing initialization of     MultiFDInit_t msg; to all zeros

I *thought* it was in.  Sorry.

>
>> ==12102==
>> ==12102== Thread 6 multifdsend_1:
>> ==12102== Invalid write of size 4
>> ==12102==    at 0x25CC08: multifd_send_fill_packet (ram.c:806)
>> ==12102==    by 0x25CC08: multifd_send_thread (ram.c:1157)
>> ==12102==    by 0x557551: qemu_thread_start (qemu-thread-posix.c:519)
>> ==12102==    by 0x53BE613: start_thread (pthread_create.c:463)
>> ==12102==    by 0x54767FB: ??? (clone.S:73)
>> ==12102==  Address 0x1d89c470 is 0 bytes after a block of size 832 alloc'd
>> ==12102==    at 0x4841BC4: calloc (vg_replace_malloc.c:711)
>> ==12102==    by 0x49EE269: g_malloc0 (in
>> /usr/lib/arm-linux-gnueabihf/libglib-2.0.so.0.5600.4)
>
> This is the same issue that was reported last time this mulitfd unit
> test was proposed for merge. Back then I pointed out the likely cause.
> We were allocating  ram_addr_t sized quantity for an array which is
> uint64_t, and ram_addr_t is probably 32-bit on this particular build.
>
>   https://lists.gnu.org/archive/html/qemu-devel/2019-07/msg03428.html
>
> That suggested fix doesn't seem to have been included

Thanks again.

And sorry for the disturbance.