10:19 a.m.
On 7/16/19 3:54 PM, Peter Krempa wrote:
On Thu, Jul 11, 2019 at 17:53:59 +0200, Michal Privoznik wrote:
> This module will be used by virHostdevManager and it's inspired
> by virPCIDevice module. They are very similar except instead of
> what makes a NVMe device: PCI address AND namespace ID. This
> means that a NVMe device can appear in a domain multiple times,
> each time with a different namespace.
>
> Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com>
> ---
> src/libvirt_private.syms | 18 ++
> src/util/Makefile.inc.am | 2 +
> src/util/virnvme.c | 412 +++++++++++++++++++++++++++++++++++++++
> src/util/virnvme.h | 89 +++++++++
> 4 files changed, 521 insertions(+)
> create mode 100644 src/util/virnvme.c
> create mode 100644 src/util/virnvme.h
[...]
> diff --git a/src/util/virnvme.c b/src/util/virnvme.c
> new file mode 100644
> index 0000000000..53724b63f7
> --- /dev/null
> +++ b/src/util/virnvme.c
> @@ -0,0 +1,412 @@
> +/*
> + * virnvme.c: helper APIs for managing NVMe devices
> + *
> + * This library is free software; you can redistribute it and/or
> + * modify it under the terms of the GNU Lesser General Public
> + * License as published by the Free Software Foundation; either
> + * version 2.1 of the License, or (at your option) any later version.
> + *
> + * This library is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
> + * Lesser General Public License for more details.
> + *
> + * You should have received a copy of the GNU Lesser General Public
> + * License along with this library. If not, see
> + * <http://www.gnu.org/licenses/>;.
> + */
> +
> +#include <config.h>
> +
> +#include "virnvme.h"
> +#include "virobject.h"
> +#include "virpci.h"
> +#include "viralloc.h"
> +#include "virlog.h"
> +#include "virstring.h"
> +
> +VIR_LOG_INIT("util.pci");
please use a different log domain
> +#define VIR_FROM_THIS VIR_FROM_NONE
> +
> +struct _virNVMeDevice {
> + virPCIDeviceAddress address; /* PCI address of controller */
> + unsigned long namespace; /* Namespace ID */
unsinged int/unsigned long long
> + bool managed;
> +
> + char *drvname;
> + char *domname;
> +};
> +
> +
> +struct _virNVMeDeviceList {
> + virObjectLockable parent;
> +
> + size_t count;
> + virNVMeDevicePtr *devs;
> +};
> +
[...]
> +int
> +virNVMeDeviceListAdd(virNVMeDeviceListPtr list,
> + const virNVMeDevice *dev)
> +{
> + virNVMeDevicePtr tmp;
> +
> + if ((tmp = virNVMeDeviceListLookup(list, dev))) {
> + VIR_AUTOFREE(char *) addrStr =
virPCIDeviceAddressAsString(&tmp->address);
> + virReportError(VIR_ERR_INTERNAL_ERROR,
> + _("NVMe device %s namespace %lu is already on the
list"),
> + NULLSTR(addrStr), tmp->namespace);
> + return -1;
> + }
> +
> + if (!(tmp = virNVMeDeviceCopy(dev)) ||
> + VIR_APPEND_ELEMENT(list->devs, list->count, tmp) < 0) {
> + virNVMeDeviceFree(tmp);
> + return -1;
> + }
> +
> + return 0;
> +}
> +
> +
> +int
> +virNVMeDeviceListDel(virNVMeDeviceListPtr list,
> + const virNVMeDevice *dev)
> +{
> + ssize_t idx;
> + virNVMeDevicePtr tmp = NULL;
> +
> + if ((idx = virNVMeDeviceListLookupIndex(list, dev)) < 0) {
> + VIR_AUTOFREE(char *) addrStr =
virPCIDeviceAddressAsString(&dev->address);
> + virReportError(VIR_ERR_INTERNAL_ERROR,
> + _("NVMe device %s namespace %lu not found"),
> + NULLSTR(addrStr), dev->namespace);
> + return -1;
> + }
> +
> + tmp = list->devs[idx];
> + VIR_DELETE_ELEMENT(list->devs, idx, list->count);
> + virNVMeDeviceFree(tmp);
> + return 0;
> +}
> +
> +
> +virNVMeDevicePtr
> +virNVMeDeviceListGet(virNVMeDeviceListPtr list,
> + size_t i)
[1] (see below)
> +{
> + return i < list->count ? list->devs[i] : NULL;
> +}
> +
> +
> +virNVMeDevicePtr
> +virNVMeDeviceListLookup(virNVMeDeviceListPtr list,
> + const virNVMeDevice *dev)
> +{
> + ssize_t idx;
> +
> + if ((idx = virNVMeDeviceListLookupIndex(list, dev)) < 0)
> + return NULL;
> +
> + return list->devs[idx];
> +}
> +
> +
> +ssize_t
This function seems to be too unsafe to export as people might want to
store the index while not holding the lock and something would then
change it. Also [1] has the same issue.
This is not any different to other modules used by virhostdev. I agree
that they are unsafe, but at the same time I think rewriting them all
(to keep consistency) wouldn't result in cleaner code. Note that the
list is locked outside of this source file (is locked from within
virhostdev) - again, not something that complies with our rules, but it
makes sense IMO.
Note that all other APIs require locking from the caller (e.g.
virNVMeDeviceListAdd()).
I'll add a comment into the header file that virNVMeDeviceList is a
lockable object that requires caller to acquire the lock and hold it
throughout whole section involving it.
> +virNVMeDeviceListLookupIndex(virNVMeDeviceListPtr list,
> + const virNVMeDevice *dev)
> +{
> + size_t i;
> +
> + if (!list)
> + return -1;
> +
> + for (i = 0; i < list->count; i++) {
> + virNVMeDevicePtr other = list->devs[i];
> +
> + if (virPCIDeviceAddressEqual(&dev->address, &other->address)
&&
> + dev->namespace == other->namespace)
> + return i;
> + }
> +
> + return -1;
> +}
> +
> +
> +static virNVMeDevicePtr
> +virNVMeDeviceListLookupByPCIAddress(virNVMeDeviceListPtr list,
> + const virPCIDeviceAddress *address)
> +{
> + size_t i;
> +
> + if (!list)
> + return NULL;
> +
> + for (i = 0; i < list->count; i++) {
> + virNVMeDevicePtr other = list->devs[i];
> +
> + if (virPCIDeviceAddressEqual(address, &other->address))
> + return other;
> + }
> +
> + return NULL;
> +}
> +
> +
> +virPCIDeviceListPtr
> +virNVMeDeviceListCreateDetachList(virNVMeDeviceListPtr activeList,
> + virNVMeDeviceListPtr toDetachList)
> +{
> + VIR_AUTOUNREF(virPCIDeviceListPtr) pciDevices = NULL;
> + size_t i;
> +
> + if (!(pciDevices = virPCIDeviceListNew()))
> + return NULL;
> +
> + for (i = 0; i < toDetachList->count; i++) {
> + const virNVMeDevice *d = toDetachList->devs[i];
> + VIR_AUTOPTR(virPCIDevice) pci = NULL;
> +
> + /* If there is a NVMe device with the same PCI address on
> + * the activeList, the device is already detached. */
> + if (virNVMeDeviceListLookupByPCIAddress(activeList, &d->address))
> + continue;
> +
> + /* It may happen that we want to detach two namespaces
> + * from the same NVMe device. This will be represented as
> + * two different instances of virNVMeDevice, but
> + * obviously we want to put the PCI device on the detach
> + * list only once. */
> + if (virPCIDeviceListFindByIDs(pciDevices,
> + d->address.domain,
> + d->address.bus,
> + d->address.slot,
> + d->address.function))
> + continue;
> +
> + if (!(pci = virPCIDeviceNew(d->address.domain,
> + d->address.bus,
> + d->address.slot,
> + d->address.function)))
> + return NULL;
> +
> + /* NVMe devices must be bound to vfio */
> + virPCIDeviceSetStubDriver(pci, VIR_PCI_STUB_DRIVER_VFIO);
> + virPCIDeviceSetManaged(pci, d->managed);
> +
> + if (virPCIDeviceListAdd(pciDevices, pci) < 0)
> + return NULL;
> +
> + /* avoid freeing the device */
> + pci = NULL;
> + }
> +
> + VIR_RETURN_PTR(pciDevices);
> +}
> +
> +
> +virPCIDeviceListPtr
> +virNVMeDeviceListCreateReAttachList(virNVMeDeviceListPtr activeList,
This function is too complex to be without a comment describing it.
Especially since it's returning list of pci devices.
Okay, I'll add a comment here too.
Michal