[PATCH v2 0/6] qom: Allow object to be aligned

Richard Henderson posted 6 patches 3 years, 7 months ago
Test docker-quick@centos7 failed
Test docker-mingw@fedora failed
Test checkpatch failed
Test FreeBSD failed
Patches applied successfully (tree, apply log)
git fetch https://github.com/patchew-project/qemu tags/patchew/20200916004638.2444147-1-richard.henderson@linaro.org
Maintainers: Peter Maydell <peter.maydell@linaro.org>, Alistair Francis <Alistair.Francis@wdc.com>, David Hildenbrand <david@redhat.com>, David Gibson <david@gibson.dropbear.id.au>, Bastian Koppelmann <kbastian@mail.uni-paderborn.de>, Palmer Dabbelt <palmer@dabbelt.com>, Thomas Huth <thuth@redhat.com>, Stefan Weil <sw@weilnetz.de>, Cornelia Huck <cohuck@redhat.com>, Richard Henderson <rth@twiddle.net>, Sagar Karandikar <sagark@eecs.berkeley.edu>
include/qom/object.h            |  5 +++++
qom/object.c                    | 36 ++++++++++++++++++++++++++++++---
target/arm/cpu.c                |  2 ++
target/riscv/cpu.c              |  1 +
target/s390x/cpu.c              |  1 +
util/oslib-win32.c              | 10 +++------
target/ppc/translate_init.c.inc |  1 +
7 files changed, 46 insertions(+), 10 deletions(-)
[PATCH v2 0/6] qom: Allow object to be aligned
Posted by Richard Henderson 3 years, 7 months ago
I've seen some failures on arm and s390x hosts after
enabling host vector support.  It turns out that the
malloc for these hosts does not provide 16-byte alignment.

We already have a function that can alloc with alignment,
but we need to pass this down from the structure.  We also
don't want to use this function unconditionally, because
the windows version does page allocation, which would be
overkill for the vast majority of the objects allocated.

Changes in v2:
  * Add _aligned_malloc patch for win32.  For what it's
    worth, this passes a gitlab cross-compile test.

  * Add and use qemu_max_align_t for choosing between
    g_malloc and qemu_memalign.

    I had been discussing extra checks for i386-linux with
    Eduardo, but then it occured to me that both linux libc
    posix_memalign is smart enough to not imply extra overhead.
    So qemu_memalign with alignment <= malloc alignment is
    handled easily by the system.


r~


Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: "Daniel P. Berrangé" <berrange@redhat.com>
Cc: Eduardo Habkost <ehabkost@redhat.com>
Cc: Peter Maydell <peter.maydell@linaro.org>
Cc: David Gibson <david@gibson.dropbear.id.au>
Cc: Alistair Francis <Alistair.Francis@wdc.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Cornelia Huck <cohuck@redhat.com>
Cc: qemu-arm@nongnu.org
Cc: qemu-ppc@nongnu.org
Cc: qemu-riscv@nongnu.org
Cc: qemu-s390x@nongnu.org


Richard Henderson (6):
  util/oslib-win32: Use _aligned_malloc for qemu_try_memalign
  qom: Allow objects to be allocated with increased alignment
  target/arm: Set instance_align on CPUARM TypeInfo
  target/ppc: Set instance_align on PowerPCCPU TypeInfo
  target/riscv: Set instance_align on RISCVCPU TypeInfo
  target/s390x: Set instance_align on S390CPU TypeInfo

 include/qom/object.h            |  5 +++++
 qom/object.c                    | 36 ++++++++++++++++++++++++++++++---
 target/arm/cpu.c                |  2 ++
 target/riscv/cpu.c              |  1 +
 target/s390x/cpu.c              |  1 +
 util/oslib-win32.c              | 10 +++------
 target/ppc/translate_init.c.inc |  1 +
 7 files changed, 46 insertions(+), 10 deletions(-)

-- 
2.25.1


Re: [PATCH v2 0/6] qom: Allow object to be aligned
Posted by Eduardo Habkost 3 years, 7 months ago
On Tue, Sep 15, 2020 at 05:46:32PM -0700, Richard Henderson wrote:
> I've seen some failures on arm and s390x hosts after
> enabling host vector support.  It turns out that the
> malloc for these hosts does not provide 16-byte alignment.
> 
> We already have a function that can alloc with alignment,
> but we need to pass this down from the structure.  We also
> don't want to use this function unconditionally, because
> the windows version does page allocation, which would be
> overkill for the vast majority of the objects allocated.

I'm queueing patches 2-6.  Thanks!

-- 
Eduardo