[Qemu-devel] [PATCH v4 0/8] target-arm: enable qmp-dump-guest-memory

[Andrew Jones <drjones@redhat.com>]

This series brings qmp-dump-guest-memory to arm and aarch64
targets. I've detailed my testing and the results in the
following table.

v4:
 - drop aarch64_compat_sp in favor of fixing crash[*] [Peter]
 - added more Peter r-b's (fix 1 nit and 1 typo)

[*] new crash fix: commit 4641ea1f6 https://github.com/crash-utility/crash

v3:
 - more changes thanks to Peter's review comments
   - fix 32-to-64 shoehorning to work better with EL2/EL3
   - stole comment from Peter for best-effort at endianness selection
   - only make ARM VFP notes when the guest has VFP registers
   - some more patch squashing and code commenting
   - another r-b added
 - collapsed a level of indirection, because if we're writing elf64 notes
   then we'll always call the aarch64 function [drew]
 - fixed the note type for ARM VFP notes (+ a tiny bit of refactoring) [drew]
 - set prstatus.pr_fpvalid [drew]
 - tested on tcg [drew]

v2: changes thanks to Peter's review questions
 - Threw in the FP registers. Can view them with gdb on elf
   formatted dumps.
 - Added comments stating where the register structs come from.
 - Fixed determination of 32 vs. 64-bit and LE vs. BE formats.
   Added a new config to the test matrix below to test this fix.
 - Added a couple R-b's


arm/aarch64 kvm guest kdump testing (P - PASS, F - FAIL). Testing done
with a latest mainline crash utility (all new patches needed are now in
master).

.-------------------------------------------------------------------------------.
|                               Host    | arm32 | arm64 | arm64 | arm64 | arm64 |
|---------------------------------------|-------|-------|-------|-------|-------|
|                               Guest   | arm32 | arm64 | arm64 | arm64 | arm32 |
|---------------------------------------|-------|-------|-------|-------|-------|
|                            User binary| arm32 | arm64 | arm32 | arm64 | arm32 |
|---------------------------------------|-------|-------|-------|-------|-------|
|                               Pagesize| 4K    | 4K    | 4K    | 64K   | 4K    |
|===============================================================================|
| kdump in guest                        | F[1]  | P[2]  | P[2]  | P[3]  | F[1]  |
|---------------------------------------|-------|-------|-------|-------|-------|
| qmp-dump-guest-memory <filename>[4]   | P     | P     | P     | P     | P     |
|---------------------------------------|-------|-------|-------|-------|-------|
| qmp-dump-guest-memory -z <filename>[5]| F[8]  | P     | P     | P     | F[8]  |
|---------------------------------------|-------|-------|-------|-------|-------|
| qmp-dump-guest-memory -l <filename>[6]| F[8]  | P     | P     | P     | F[8]  |
|---------------------------------------|-------|-------|-------|-------|-------|
| qmp-dump-guest-memory -s <filename>[7]| F[8]  | P     | P     | P     | F[8]  |
.-------------------------------------------------------------------------------.

[1] Kernel v4.4-rc1 crashes with a NULL pointer dereference at virtual
    address 00000000 in a memcpy (crash_kexec/machine_kexec/fncpy/memcpy).
    Needs kernel debugging.
[2] Not sure about mainline, but works with the RHEL kernel,
    makedumpfile does not yet support arm64 with 4K pages, but using
    'core_collector cp' in /etc/kdump.conf allows saving an uncompressed
    elf file.
[3] Not sure about mainline, but works with the RHEL kernel,
    uses makedumpfile, thus generates a makedumpfile formatted file
    using zlib compression.
[4] No format specified, creates an uncompressed elf formatted file.
[5] makedumpfile format, with zlib compression
[6] makedumpfile format, with lzo compression
[7] makedumpfile format, with snappy compression
[8] The crash utility doesn't seem to like arm32 dumps in makedumpfile
    format. Looks like the physical page bitmap is all zeros? Needs
    qemu and crash debugging.

Additional notes:
1) QEMU also has scripts/dump-guest-memory.py, which can and should be
   updated to support multiple architectures, pagesizes, and physbases.
   This is currently left as future work.


Andrew Jones (8):
  qapi-schema: dump-guest-memory: Improve text
  dump: qemunotes aren't commonly needed
  dump: allow target to set the page size
  dump: allow target to set the physical base
  target-arm: support QMP dump-guest-memory
  target-arm: dump-guest-memory: add prfpreg notes for aarch64
  elf: add arm note types
  target-arm: dump-guest-memory: add vfp notes for arm

 dump.c                      | 131 ++++++++++-------
 include/elf.h               |   5 +
 include/sysemu/dump-arch.h  |   9 +-
 include/sysemu/dump.h       |  11 +-
 qapi-schema.json            |   4 +-
 qom/cpu.c                   |   4 +-
 target-arm/Makefile.objs    |   3 +-
 target-arm/arch_dump.c      | 336 ++++++++++++++++++++++++++++++++++++++++++++
 target-arm/cpu-qom.h        |   5 +
 target-arm/cpu.c            |   3 +
 target-ppc/arch_dump.c      |   6 -
 target-ppc/cpu-qom.h        |   2 -
 target-ppc/translate_init.c |   1 -
 target-s390x/arch_dump.c    |   6 -
 target-s390x/cpu-qom.h      |   2 -
 target-s390x/cpu.c          |   1 -
 16 files changed, 442 insertions(+), 87 deletions(-)
 create mode 100644 target-arm/arch_dump.c

-- 
2.4.3