Re: [Qemu-devel] [PATCH V9 4/4] tests: Add migration test for aarch64

[Dr. David Alan Gilbert]

* Dr. David Alan Gilbert (address@hidden) wrote:
> * Wei Huang (address@hidden) wrote:
> > 
> > 
> > On 09/28/2018 09:04 AM, Dr. David Alan Gilbert wrote:
> > > * Wei Huang (address@hidden) wrote:
> > >>
> > >>
> > >> On 09/26/2018 11:31 AM, Dr. David Alan Gilbert wrote:
> > >>> * Wei Huang (address@hidden) wrote:
> > >>>> This patch adds migration test support for aarch64. The test code, 
> > >>>> which
> > >>>> implements the same functionality as x86, is booted as a kernel in 
> > >>>> qemu.
> > >>>> Here are the design choices we make for aarch64:
> > >>>>
> > >>>>  * We choose this -kernel approach because aarch64 QEMU doesn't 
> > >>>> provide a
> > >>>>    built-in fw like x86 does. So instead of relying on a boot loader, 
> > >>>> we
> > >>>>    use -kernel approach for aarch64.
> > >>>>  * The serial output is sent to PL011 directly.
> > >>>>  * The physical memory base for mach-virt machine is 0x40000000. We 
> > >>>> change
> > >>>>    the start_address and end_address for aarch64.
> > >>>>
> > >>>> In addition to providing the binary, this patch also includes the 
> > >>>> source
> > >>>> code and the build script in tests/migration/aarch64. So users can 
> > >>>> change
> > >>>> the source and/or re-compile the binary as they wish.
> > >>>>
> > >>>> Reviewed-by: Juan Quintela <address@hidden>
> > >>>> Signed-off-by: Wei Huang <address@hidden>
> > >>>
> > >>> I've had to bounce this patch (kept the other 3) because the test
> > >>> is actually failing on aarch64 kvm.
> > >>> It's about 1 in 5 for me on one of the larger (40ish core) boxes which
> > >>> is otherwise idle.
> > >>>
> > >>> /aarch64/migration/precopy/unix: Memory content inconsistency at 
> > >>> 43cf1000 first_byte = 5 last_byte = 4 current = 5 hit_edge = 1
> > >>> Memory content inconsistency at 43cf2000 first_byte = 5 last_byte = 4 
> > >>> current = 5 hit_edge = 1
> > >>> Memory content inconsistency at 43cf3000 first_byte = 5 last_byte = 4 
> > >>> current = 5 hit_edge = 1
> > >>> Memory content inconsistency at 43cf4000 first_byte = 5 last_byte = 4 
> > >>> current = 5 hit_edge = 1
> > >>> Memory content inconsistency at 43cf5000 first_byte = 5 last_byte = 4 
> > >>> current = 5 hit_edge = 1
> > >>> Memory content inconsistency at 43cf6000 first_byte = 5 last_byte = 4 
> > >>> current = 5 hit_edge = 1
> > >>> Memory content inconsistency at 43cf7000 first_byte = 5 last_byte = 4 
> > >>> current = 5 hit_edge = 1
> > >>
> > >> Thanks, Dave. The root cause is the wrong place of cache invalidation.
> > >> Here is the fix which has been tested on Amberwing. Could you please
> > >> verify it from your side? After that, I can either send in a new patch
> > >> or you can help me revise the existing one.
> > > 
> > > Thanks, I'll grab a box and try it.
> > > If it works I'd prefer if you can resubmit the patch with the fix in.
> > > 
> > > However, this fix looks suspicious to me.  We check the guest ram
> > > contents after we stop the guest;  if the guests RAM is inconsistent as
> > > seen by QEMU without the 'dc' then how will this work with normal
> > > applications?
> > 
> > Such cases happened before. Here is the one presentation given at KVM
> > Forum:
> > https://events.static.linuxfound.org/sites/events/files/slides/slides_10.pdf.
> 
> I have seen that presentation but I don't understand how it relates to
> your case.   Imagine that we're migrating some totally arbitrary
> application - how do we know we get consistent memory  - if you're
> having to put cache clears in the application that sounds like it's a
> real kvm/arm bug that needs fixing.

The change does seem to be working; but I really would like to
understand what's going on with the caches here.

Dave

> Dave
> 
> > > 
> > > Shouldn't this be something qemu or kvm is doing when it stops the
> > > guest?
> > > 
> > > Dave
> > > 
> > >> Thanks,
> > >> -Wei
> > >>
> > >>
> > >> diff --git a/tests/migration/aarch64/a-b-kernel.S
> > >> b/tests/migration/aarch64/a-b-
> > >> index 507af30..c8d2720 100644
> > >> --- a/tests/migration/aarch64/a-b-kernel.S
> > >> +++ b/tests/migration/aarch64/a-b-kernel.S
> > >> @@ -50,15 +50,15 @@ mainloop:
> > >>          mov     x4, x0
> > >>
> > >>  innerloop:
> > >> -        /* clean cache because el2 might still cache guest data under
> > >> KVM */
> > >> -        dc      civac, x4
> > >> -
> > >>          /* increment the first byte of each page by 1 */
> > >>          ldrb    w3, [x4]
> > >>          add     w3, w3, #1
> > >>          and     w3, w3, #0xff
> > >>          strb    w3, [x4]
> > >>
> > >> +        /* clean cache because el2 might still cache guest data under
> > >> KVM */
> > >> +        dc      civac, x4
> > >> +
> > >>          add     x4, x4, #TEST_MEM_PAGE_SIZE
> > >>          cmp     x4, x1
> > >>          blt     innerloop
> > >>
> > >>
> > >>>
> > >>> Dave
> > >>>
> > >>>> ---
> > >>>>  tests/Makefile.include               |  1 +
> > >>>>  tests/migration-test.c               | 27 +++++++++++--
> > >>>>  tests/migration/Makefile             |  2 +-
> > >>>>  tests/migration/aarch64/Makefile     | 20 ++++++++++
> > >>>>  tests/migration/aarch64/a-b-kernel.S | 75 
> > >>>> ++++++++++++++++++++++++++++++++++++
> > >>>>  tests/migration/aarch64/a-b-kernel.h | 19 +++++++++
> > >>>>  tests/migration/migration-test.h     |  9 +++++
> > >>>>  7 files changed, 148 insertions(+), 5 deletions(-)
> > >>>>  create mode 100644 tests/migration/aarch64/Makefile
> > >>>>  create mode 100644 tests/migration/aarch64/a-b-kernel.S
> > >>>>  create mode 100644 tests/migration/aarch64/a-b-kernel.h
> > >>>>
> > >>>> diff --git a/tests/Makefile.include b/tests/Makefile.include
> > >>>> index 87c81d1..fab8fb9 100644
> > >>>> --- a/tests/Makefile.include
> > >>>> +++ b/tests/Makefile.include
> > >>>> @@ -390,6 +390,7 @@ check-qtest-arm-y += tests/hexloader-test$(EXESUF)
> > >>>>  check-qtest-aarch64-y = tests/numa-test$(EXESUF)
> > >>>>  check-qtest-aarch64-$(CONFIG_SDHCI) += tests/sdhci-test$(EXESUF)
> > >>>>  check-qtest-aarch64-y += tests/boot-serial-test$(EXESUF)
> > >>>> +check-qtest-aarch64-y += tests/migration-test$(EXESUF)
> > >>>>  
> > >>>>  check-qtest-microblazeel-y = $(check-qtest-microblaze-y)
> > >>>>  
> > >>>> diff --git a/tests/migration-test.c b/tests/migration-test.c
> > >>>> index 17c6896..ecfae0b 100644
> > >>>> --- a/tests/migration-test.c
> > >>>> +++ b/tests/migration-test.c
> > >>>> @@ -86,12 +86,13 @@ static const char *tmpfs;
> > >>>>   * repeatedly. It outputs a 'B' at a fixed rate while it's still 
> > >>>> running.
> > >>>>   */
> > >>>>  #include "tests/migration/i386/a-b-bootblock.h"
> > >>>> +#include "tests/migration/aarch64/a-b-kernel.h"
> > >>>>  
> > >>>> -static void init_bootfile_x86(const char *bootpath)
> > >>>> +static void init_bootfile(const char *bootpath, void *content)
> > >>>>  {
> > >>>>      FILE *bootfile = fopen(bootpath, "wb");
> > >>>>  
> > >>>> -    g_assert_cmpint(fwrite(x86_bootsect, 512, 1, bootfile), ==, 1);
> > >>>> +    g_assert_cmpint(fwrite(content, 512, 1, bootfile), ==, 1);
> > >>>>      fclose(bootfile);
> > >>>>  }
> > >>>>  
> > >>>> @@ -428,7 +429,7 @@ static int test_migrate_start(QTestState **from, 
> > >>>> QTestState **to,
> > >>>>      got_stop = false;
> > >>>>  
> > >>>>      if (strcmp(arch, "i386") == 0 || strcmp(arch, "x86_64") == 0) {
> > >>>> -        init_bootfile_x86(bootpath);
> > >>>> +        init_bootfile(bootpath, x86_bootsect);
> > >>>>          cmd_src = g_strdup_printf("-machine accel=%s -m 150M"
> > >>>>                                    " -name source,debug-threads=on"
> > >>>>                                    " -serial file:%s/src_serial"
> > >>>> @@ -459,6 +460,24 @@ static int test_migrate_start(QTestState **from, 
> > >>>> QTestState **to,
> > >>>>  
> > >>>>          start_address = PPC_TEST_MEM_START;
> > >>>>          end_address = PPC_TEST_MEM_END;
> > >>>> +    } else if (strcmp(arch, "aarch64") == 0) {
> > >>>> +        init_bootfile(bootpath, aarch64_kernel);
> > >>>> +        cmd_src = g_strdup_printf("-machine 
> > >>>> virt,accel=%s,gic-version=max "
> > >>>> +                                  "-name vmsource,debug-threads=on 
> > >>>> -cpu max "
> > >>>> +                                  "-m 150M -serial file:%s/src_serial 
> > >>>> "
> > >>>> +                                  "-kernel %s ",
> > >>>> +                                  accel, tmpfs, bootpath);
> > >>>> +        cmd_dst = g_strdup_printf("-machine 
> > >>>> virt,accel=%s,gic-version=max "
> > >>>> +                                  "-name vmdest,debug-threads=on -cpu 
> > >>>> max "
> > >>>> +                                  "-m 150M -serial 
> > >>>> file:%s/dest_serial "
> > >>>> +                                  "-kernel %s "
> > >>>> +                                  "-incoming %s ",
> > >>>> +                                  accel, tmpfs, bootpath, uri);
> > >>>> +
> > >>>> +        start_address = ARM_TEST_MEM_START;
> > >>>> +        end_address = ARM_TEST_MEM_END;
> > >>>> +
> > >>>> +        g_assert(sizeof(aarch64_kernel) <= ARM_TEST_MAX_KERNEL_SIZE);
> > >>>>      } else {
> > >>>>          g_assert_not_reached();
> > >>>>      }
> > >>>> @@ -545,7 +564,7 @@ static void test_deprecated(void)
> > >>>>  {
> > >>>>      QTestState *from;
> > >>>>  
> > >>>> -    from = qtest_start("");
> > >>>> +    from = qtest_start("-machine none");
> > >>>>  
> > >>>>      deprecated_set_downtime(from, 0.12345);
> > >>>>      deprecated_set_speed(from, 12345);
> > >>>> diff --git a/tests/migration/Makefile b/tests/migration/Makefile
> > >>>> index a9ed875..13af934 100644
> > >>>> --- a/tests/migration/Makefile
> > >>>> +++ b/tests/migration/Makefile
> > >>>> @@ -5,7 +5,7 @@
> > >>>>  # See the COPYING file in the top-level directory.
> > >>>>  #
> > >>>>  
> > >>>> -TARGET_LIST = i386
> > >>>> +TARGET_LIST = i386 aarch64
> > >>>>  
> > >>>>  SRC_PATH = ../..
> > >>>>  
> > >>>> diff --git a/tests/migration/aarch64/Makefile 
> > >>>> b/tests/migration/aarch64/Makefile
> > >>>> new file mode 100644
> > >>>> index 0000000..d440fa8
> > >>>> --- /dev/null
> > >>>> +++ b/tests/migration/aarch64/Makefile
> > >>>> @@ -0,0 +1,20 @@
> > >>>> +# To specify cross compiler prefix, use CROSS_PREFIX=
> > >>>> +#   $ make CROSS_PREFIX=aarch64-linux-gnu-
> > >>>> +
> > >>>> +.PHONY: all clean
> > >>>> +all: a-b-kernel.h
> > >>>> +
> > >>>> +a-b-kernel.h: aarch64.kernel
> > >>>> +      echo "$$__note" > header.tmp
> > >>>> +      xxd -i $< | sed -e 's/.*int.*//' >> header.tmp
> > >>>> +      mv header.tmp $@
> > >>>> +
> > >>>> +aarch64.kernel: aarch64.elf
> > >>>> +      $(CROSS_PREFIX)objcopy -O binary $< $@
> > >>>> +
> > >>>> +aarch64.elf: a-b-kernel.S
> > >>>> +      $(CROSS_PREFIX)gcc -o $@ -nostdlib -Wl,--build-id=none $<
> > >>>> +
> > >>>> +clean:
> > >>>> +      @rm -rf *.kernel *.elf
> > >>>> +
> > >>>> diff --git a/tests/migration/aarch64/a-b-kernel.S 
> > >>>> b/tests/migration/aarch64/a-b-kernel.S
> > >>>> new file mode 100644
> > >>>> index 0000000..507af30
> > >>>> --- /dev/null
> > >>>> +++ b/tests/migration/aarch64/a-b-kernel.S
> > >>>> @@ -0,0 +1,75 @@
> > >>>> +#
> > >>>> +# Copyright (c) 2018 Red Hat, Inc. and/or its affiliates
> > >>>> +#
> > >>>> +# Author:
> > >>>> +#   Wei Huang <address@hidden>
> > >>>> +#
> > >>>> +# This work is licensed under the terms of the GNU GPL, version 2 or 
> > >>>> later.
> > >>>> +# See the COPYING file in the top-level directory.
> > >>>> +#
> > >>>> +# Note: Please make sure the compiler compiles the assembly code 
> > >>>> below with
> > >>>> +# pc-relative address. Also the branch instructions should use 
> > >>>> relative
> > >>>> +# addresses only.
> > >>>> +
> > >>>> +#include "../migration-test.h"
> > >>>> +
> > >>>> +.section .text
> > >>>> +
> > >>>> +        .globl  _start
> > >>>> +
> > >>>> +_start:
> > >>>> +        /* disable MMU to use phys mem address */
> > >>>> +        mrs     x0, sctlr_el1
> > >>>> +        bic     x0, x0, #(1<<0)
> > >>>> +        msr     sctlr_el1, x0
> > >>>> +        isb
> > >>>> +
> > >>>> +        /* traverse test memory region */
> > >>>> +        mov     x0, #ARM_TEST_MEM_START
> > >>>> +        mov     x1, #ARM_TEST_MEM_END
> > >>>> +
> > >>>> +        /* output char 'A' to PL011 */
> > >>>> +        mov     w3, 'A'
> > >>>> +        mov     x2, #ARM_MACH_VIRT_UART
> > >>>> +        strb    w3, [x2]
> > >>>> +
> > >>>> +        /* clean up memory */
> > >>>> +        mov     w3, #0
> > >>>> +        mov     x4, x0
> > >>>> +clean:
> > >>>> +        strb    w3, [x4]
> > >>>> +        add     x4, x4, #TEST_MEM_PAGE_SIZE
> > >>>> +        cmp     x4, x1
> > >>>> +        ble     clean
> > >>>> +
> > >>>> +        /* w5 keeps a counter so we can limit the output speed */
> > >>>> +        mov     w5, #0
> > >>>> +
> > >>>> +        /* main body */
> > >>>> +mainloop:
> > >>>> +        mov     x4, x0
> > >>>> +
> > >>>> +innerloop:
> > >>>> +        /* clean cache because el2 might still cache guest data under 
> > >>>> KVM */
> > >>>> +        dc      civac, x4
> > >>>> +
> > >>>> +        /* increment the first byte of each page by 1 */
> > >>>> +        ldrb    w3, [x4]
> > >>>> +        add     w3, w3, #1
> > >>>> +        and     w3, w3, #0xff
> > >>>> +        strb    w3, [x4]
> > >>>> +
> > >>>> +        add     x4, x4, #TEST_MEM_PAGE_SIZE
> > >>>> +        cmp     x4, x1
> > >>>> +        blt     innerloop
> > >>>> +
> > >>>> +        add     w5, w5, #1
> > >>>> +        and     w5, w5, #0xff
> > >>>> +        cmp     w5, #0
> > >>>> +        bne     mainloop
> > >>>> +
> > >>>> +        /* output char 'B' to PL011 */
> > >>>> +        mov     w3, 'B'
> > >>>> +        strb    w3, [x2]
> > >>>> +
> > >>>> +        b       mainloop
> > >>>> diff --git a/tests/migration/aarch64/a-b-kernel.h 
> > >>>> b/tests/migration/aarch64/a-b-kernel.h
> > >>>> new file mode 100644
> > >>>> index 0000000..521125e
> > >>>> --- /dev/null
> > >>>> +++ b/tests/migration/aarch64/a-b-kernel.h
> > >>>> @@ -0,0 +1,19 @@
> > >>>> +/* This file is automatically generated from the assembly file in
> > >>>> + * tests/migration/aarch64. Edit that file and then run "make all"
> > >>>> + * inside tests/migration to update, and then remember to send both
> > >>>> + * the header and the assembler differences in your patch submission.
> > >>>> + */
> > >>>> +unsigned char aarch64_kernel[] = {
> > >>>> +  0x00, 0x10, 0x38, 0xd5, 0x00, 0xf8, 0x7f, 0x92, 0x00, 0x10, 0x18, 
> > >>>> 0xd5,
> > >>>> +  0xdf, 0x3f, 0x03, 0xd5, 0x00, 0x02, 0xa8, 0xd2, 0x01, 0xc8, 0xa8, 
> > >>>> 0xd2,
> > >>>> +  0x23, 0x08, 0x80, 0x52, 0x02, 0x20, 0xa1, 0xd2, 0x43, 0x00, 0x00, 
> > >>>> 0x39,
> > >>>> +  0x03, 0x00, 0x80, 0x52, 0xe4, 0x03, 0x00, 0xaa, 0x83, 0x00, 0x00, 
> > >>>> 0x39,
> > >>>> +  0x84, 0x04, 0x40, 0x91, 0x9f, 0x00, 0x01, 0xeb, 0xad, 0xff, 0xff, 
> > >>>> 0x54,
> > >>>> +  0x05, 0x00, 0x80, 0x52, 0xe4, 0x03, 0x00, 0xaa, 0x24, 0x7e, 0x0b, 
> > >>>> 0xd5,
> > >>>> +  0x83, 0x00, 0x40, 0x39, 0x63, 0x04, 0x00, 0x11, 0x63, 0x1c, 0x00, 
> > >>>> 0x12,
> > >>>> +  0x83, 0x00, 0x00, 0x39, 0x84, 0x04, 0x40, 0x91, 0x9f, 0x00, 0x01, 
> > >>>> 0xeb,
> > >>>> +  0x2b, 0xff, 0xff, 0x54, 0xa5, 0x04, 0x00, 0x11, 0xa5, 0x1c, 0x00, 
> > >>>> 0x12,
> > >>>> +  0xbf, 0x00, 0x00, 0x71, 0x81, 0xfe, 0xff, 0x54, 0x43, 0x08, 0x80, 
> > >>>> 0x52,
> > >>>> +  0x43, 0x00, 0x00, 0x39, 0xf1, 0xff, 0xff, 0x17
> > >>>> +};
> > >>>> +
> > >>>> diff --git a/tests/migration/migration-test.h 
> > >>>> b/tests/migration/migration-test.h
> > >>>> index c4c0c52..6939a13 100644
> > >>>> --- a/tests/migration/migration-test.h
> > >>>> +++ b/tests/migration/migration-test.h
> > >>>> @@ -18,4 +18,13 @@
> > >>>>  #define PPC_TEST_MEM_START (1 * 1024 * 1024)
> > >>>>  #define PPC_TEST_MEM_END   (100 * 1024 * 1024)
> > >>>>  
> > >>>> +/* ARM */
> > >>>> +#define ARM_TEST_MEM_START (0x40000000 + 1 * 1024 * 1024)
> > >>>> +#define ARM_TEST_MEM_END   (0x40000000 + 100 * 1024 * 1024)
> > >>>> +#define ARM_MACH_VIRT_UART 0x09000000
> > >>>> +/* AArch64 kernel load address is 0x40080000, and the test memory 
> > >>>> starts at
> > >>>> + * 0x40100000. So the maximum allowable kernel size is 512KB.
> > >>>> + */
> > >>>> +#define ARM_TEST_MAX_KERNEL_SIZE (512 * 1024)
> > >>>> +
> > >>>>  #endif /* _TEST_MIGRATION_H_ */
> > >>>> -- 
> > >>>> 1.8.3.1
> > >>>>
> > >>>>
> > >>> --
> > >>> Dr. David Alan Gilbert / address@hidden / Manchester, UK
> > >>>
> > > --
> > > Dr. David Alan Gilbert / address@hidden / Manchester, UK
> > > 
> --
> Dr. David Alan Gilbert / address@hidden / Manchester, UK
--
Dr. David Alan Gilbert / address@hidden / Manchester, UK