qemu-s390x
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [qemu-s390x] [PATCH v4 2/2] vfio-ccw: support async command subregio


From: Cornelia Huck
Subject: Re: [qemu-s390x] [PATCH v4 2/2] vfio-ccw: support async command subregion
Date: Mon, 20 May 2019 18:47:45 +0200

On Mon, 20 May 2019 12:29:56 -0400
Eric Farman <address@hidden> wrote:

> On 5/7/19 11:47 AM, Cornelia Huck wrote:
> > A vfio-ccw device may provide an async command subregion for
> > issuing halt/clear subchannel requests. If it is present, use
> > it for sending halt/clear request to the device; if not, fall
> > back to emulation (as done today).
> > 
> > Signed-off-by: Cornelia Huck <address@hidden>
> > ---
> >  hw/s390x/css.c              |  27 +++++++--
> >  hw/vfio/ccw.c               | 110 +++++++++++++++++++++++++++++++++++-
> >  include/hw/s390x/s390-ccw.h |   3 +
> >  3 files changed, 134 insertions(+), 6 deletions(-)

(...)

> > +int vfio_ccw_handle_clear(SubchDev *sch)
> > +{
> > +    S390CCWDevice *cdev = sch->driver_data;
> > +    VFIOCCWDevice *vcdev = DO_UPCAST(VFIOCCWDevice, cdev, cdev);
> > +    struct ccw_cmd_region *region = vcdev->async_cmd_region;
> > +    int ret;
> > +
> > +    if (!vcdev->async_cmd_region) {
> > +        /* Async command region not available, fall back to emulation */
> > +        return -ENOSYS;
> > +    }
> > +
> > +    memset(region, 0, sizeof(*region));
> > +    region->command = VFIO_CCW_ASYNC_CMD_CSCH;  
> 
> Considering the serialization you added on the kernel side, what happens
> if another vcpu runs this code (or _halt) and clears the async region
> before the kernel code gains control from the pwrite() call below?
> Asked another way, there's nothing preventing us from issuing more than
> one asynchronous command concurrently, so how do we make sure the
> command gets to the kernel rather than "current command wins"  ?

Hm... good question. But unfortunately not one I can answer quickly, so
I'll put off queuing this patch and just send a pull request without
it. It's not like we're in a hurry :)

> 
> That possibly worrisome question aside, this seems generally fine.

Thanks for looking!

> > +
> > +again:
> > +    ret = pwrite(vcdev->vdev.fd, region,
> > +                 vcdev->async_cmd_region_size, 
> > vcdev->async_cmd_region_offset);
> > +    if (ret != vcdev->async_cmd_region_size) {
> > +        if (errno == EAGAIN) {
> > +            goto again;
> > +        }
> > +        error_report("vfio-ccw: write cmd region failed with errno=%d", 
> > errno);
> > +        ret = -errno;
> > +    } else {
> > +        ret = region->ret_code;
> > +    }
> > +    switch (ret) {
> > +    case 0:
> > +    case -ENODEV:
> > +    case -EACCES:
> > +        return 0;
> > +    case -EFAULT:
> > +    default:
> > +        sch_gen_unit_exception(sch);
> > +        css_inject_io_interrupt(sch);
> > +        return 0;
> > +    }



reply via email to

[Prev in Thread] Current Thread [Next in Thread]