[Qemu-devel] [PATCH 0/3 for 2.10] migration/i386 cleanup

Dr. David Alan Gilbert (git) posted 3 patches 6 years, 11 months ago
Patches applied successfully (tree, apply log)
git fetch https://github.com/patchew-project/qemu tags/patchew/20170405190024.27581-1-dgilbert@redhat.com
Test checkpatch passed
Test docker passed
Test s390x passed
target/i386/machine.c | 266 +++++++++++++-------------------------------------
1 file changed, 66 insertions(+), 200 deletions(-)
[Qemu-devel] [PATCH 0/3 for 2.10] migration/i386 cleanup
Posted by Dr. David Alan Gilbert (git) 6 years, 11 months ago
From: "Dr. David Alan Gilbert" <dgilbert@redhat.com>

Hi,
  This removes some qemu_get_ and qemu_put_ use from i386/machine.c
and cleans out some very old code.
  It breaks migration compatibility from prior to 0.12
for i386 and prior to ~0.14 for non-softfloat - it doesn't alter the
machine type, it just removes some old stream features.

(Has anyone got a good test of FP migration to make sure
I've not broken the FP/mmx/etc cases?)

Dave


Dr. David Alan Gilbert (3):
  migration/i386: Remove old non-softfloat 64bit FP support
  vmstatification: i386 FPReg
  migration/i386: Remove support for pre-0.12 formats

 target/i386/machine.c | 266 +++++++++++++-------------------------------------
 1 file changed, 66 insertions(+), 200 deletions(-)

-- 
2.9.3


Re: [Qemu-devel] [PATCH 0/3 for 2.10] migration/i386 cleanup
Posted by Eduardo Habkost 6 years, 11 months ago
On Wed, Apr 05, 2017 at 08:00:21PM +0100, Dr. David Alan Gilbert (git) wrote:
> From: "Dr. David Alan Gilbert" <dgilbert@redhat.com>
> 
> Hi,
>   This removes some qemu_get_ and qemu_put_ use from i386/machine.c
> and cleans out some very old code.
>   It breaks migration compatibility from prior to 0.12
> for i386 and prior to ~0.14 for non-softfloat - it doesn't alter the
> machine type, it just removes some old stream features.
> 
> (Has anyone got a good test of FP migration to make sure
> I've not broken the FP/mmx/etc cases?)

It would be nice if we could allow kvm-unit-tests test cases
trigger a migration operation at a specific instruction. I assume
we don't have an existing mechanism that could be used for that?

-- 
Eduardo

Re: [Qemu-devel] [PATCH 0/3 for 2.10] migration/i386 cleanup
Posted by Dr. David Alan Gilbert 6 years, 11 months ago
* Eduardo Habkost (ehabkost@redhat.com) wrote:
> On Wed, Apr 05, 2017 at 08:00:21PM +0100, Dr. David Alan Gilbert (git) wrote:
> > From: "Dr. David Alan Gilbert" <dgilbert@redhat.com>
> > 
> > Hi,
> >   This removes some qemu_get_ and qemu_put_ use from i386/machine.c
> > and cleans out some very old code.
> >   It breaks migration compatibility from prior to 0.12
> > for i386 and prior to ~0.14 for non-softfloat - it doesn't alter the
> > machine type, it just removes some old stream features.
> > 
> > (Has anyone got a good test of FP migration to make sure
> > I've not broken the FP/mmx/etc cases?)
> 
> It would be nice if we could allow kvm-unit-tests test cases
> trigger a migration operation at a specific instruction. I assume
> we don't have an existing mechanism that could be used for that?

Not that I know of, do we have any type of breakpoint type thing
we could use?

Dave

> -- 
> Eduardo
--
Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK

Re: [Qemu-devel] [PATCH 0/3 for 2.10] migration/i386 cleanup
Posted by Paolo Bonzini 6 years, 11 months ago

On 18/04/2017 13:00, Dr. David Alan Gilbert wrote:
>> It would be nice if we could allow kvm-unit-tests test cases
>> trigger a migration operation at a specific instruction. I assume
>> we don't have an existing mechanism that could be used for that?
> Not that I know of, do we have any type of breakpoint type thing
> we could use?

What about the gdbstub support?

Paolo

Re: [Qemu-devel] [PATCH 0/3 for 2.10] migration/i386 cleanup
Posted by Eduardo Habkost 6 years, 10 months ago
On Wed, Apr 05, 2017 at 08:00:21PM +0100, Dr. David Alan Gilbert (git) wrote:
> From: "Dr. David Alan Gilbert" <dgilbert@redhat.com>
> 
> Hi,
>   This removes some qemu_get_ and qemu_put_ use from i386/machine.c
> and cleans out some very old code.
>   It breaks migration compatibility from prior to 0.12
> for i386 and prior to ~0.14 for non-softfloat - it doesn't alter the
> machine type, it just removes some old stream features.
> 
> (Has anyone got a good test of FP migration to make sure
> I've not broken the FP/mmx/etc cases?)
> 
> Dave

Queued, thanks.

-- 
Eduardo