Re: [PATCH v2 2/3] s390x/pci: Honor DMA limits set by vfio

[Cornelia Huck]

On Tue, 15 Sep 2020 10:16:23 -0400
Matthew Rosato <mjrosato@linux.ibm.com> wrote:

> On 9/15/20 7:28 AM, Cornelia Huck wrote:
> > On Mon, 14 Sep 2020 18:29:29 -0400
> > Matthew Rosato <mjrosato@linux.ibm.com> wrote:
> >   
> >> When an s390 guest is using lazy unmapping, it can result in a very
> >> large number of oustanding DMA requests, far beyond the default
> >> limit configured for vfio.  Let's track DMA usage similar to vfio
> >> in the host, and trigger the guest to flush their DMA mappings
> >> before vfio runs out.
> >>
> >> Signed-off-by: Matthew Rosato <mjrosato@linux.ibm.com>
> >> ---
> >>   hw/s390x/s390-pci-bus.c  | 99 
> >> +++++++++++++++++++++++++++++++++++++++++++++---
> >>   hw/s390x/s390-pci-bus.h  |  9 +++++
> >>   hw/s390x/s390-pci-inst.c | 29 +++++++++++---
> >>   hw/s390x/s390-pci-inst.h |  3 ++
> >>   4 files changed, 129 insertions(+), 11 deletions(-)
> >>  
> > 
> > (...)
> >   
> >> @@ -737,6 +742,82 @@ static void s390_pci_iommu_free(S390pciState *s, 
> >> PCIBus *bus, int32_t devfn)
> >>       object_unref(OBJECT(iommu));
> >>   }
> >>   
> >> +static bool s390_sync_dma_avail(int fd, unsigned int *avail)  
> > 
> > Not sure I like the name. It sounds like the function checks whether
> > "sync dma" is available. Maybe s390_update_dma_avail()?
> >   
> 
> Sounds fine to me.
> 
> >> +{
> >> +    struct vfio_iommu_type1_info *info;
> >> +    uint32_t argsz;
> >> +    bool rval = false;
> >> +    int ret;
> >> +
> >> +    if (avail == NULL) {
> >> +        return false;
> >> +    }
> >> +
> >> +    argsz = sizeof(struct vfio_iommu_type1_info);
> >> +    info = g_malloc0(argsz);
> >> +    info->argsz = argsz;
> >> +    /*
> >> +     * If the specified argsz is not large enough to contain all
> >> +     * capabilities it will be updated upon return.  In this case
> >> +     * use the updated value to get the entire capability chain.
> >> +     */
> >> +    ret = ioctl(fd, VFIO_IOMMU_GET_INFO, info);
> >> +    if (argsz != info->argsz) {
> >> +        argsz = info->argsz;
> >> +        info = g_realloc(info, argsz);
> >> +        info->argsz = argsz;
> >> +        ret = ioctl(fd, VFIO_IOMMU_GET_INFO, info);
> >> +    }
> >> +
> >> +    if (ret) {
> >> +        goto out;
> >> +    }
> >> +
> >> +    /* If the capability exists, update with the current value */
> >> +    rval = vfio_get_info_dma_avail(info, avail);  
> > 
> > Adding vfio specific things into the generic s390 pci emulation code
> > looks a bit ugly... I'd prefer to factor that out into a separate file,
> > especially if you plan to add more vfio-specific things in the future.
> >   
> 
> Fair.   hw/s390x/s390-pci-vfio.* ?

Sounds good to me.