| Message ID | 1489507263-21365-1-git-send-email-cavery@redhat.com (mailing list archive) |
|---|---|
| State | Changes Requested, archived |
| Headers | show |
On Tue, 14 Mar 2017 12:01:03 -0400 Cathy Avery <cavery@redhat.com> wrote: > #if IS_ENABLED(CONFIG_SCSI_FC_ATTRS) > if (host->transportt == fc_transport_template) { > + struct fc_rport_identifiers ids; > + > + ids.node_name = 0; > + ids.port_name = 0; > + ids.port_id = 0; > + ids.roles |= FC_PORT_ROLE_FCP_TARGET; Since the variable ids is on the stack, it is uninitialized data. Doing a OR with uninitialized data is not correct. Better off to use C99 style iniatializer and skip the zero fields. struct fc_rport_identifiers ids = { .roles = FC_PORT_ROLE_FCP_TARGET, };
Good catch. Thanks! On 03/14/2017 12:42 PM, Stephen Hemminger wrote: > On Tue, 14 Mar 2017 12:01:03 -0400 > Cathy Avery <cavery@redhat.com> wrote: > >> #if IS_ENABLED(CONFIG_SCSI_FC_ATTRS) >> if (host->transportt == fc_transport_template) { >> + struct fc_rport_identifiers ids; >> + >> + ids.node_name = 0; >> + ids.port_name = 0; >> + ids.port_id = 0; >> + ids.roles |= FC_PORT_ROLE_FCP_TARGET; > Since the variable ids is on the stack, it is uninitialized data. > Doing a OR with uninitialized data is not correct. > > Better off to use C99 style iniatializer and skip the zero fields. > > struct fc_rport_identifiers ids = { > .roles = FC_PORT_ROLE_FCP_TARGET, > };
diff --git a/drivers/scsi/storvsc_drv.c b/drivers/scsi/storvsc_drv.c index 638e5f4..c6c0316 100644 --- a/drivers/scsi/storvsc_drv.c +++ b/drivers/scsi/storvsc_drv.c @@ -478,6 +478,9 @@ struct storvsc_device { */ u64 node_name; u64 port_name; +#if IS_ENABLED(CONFIG_SCSI_FC_ATTRS) + struct fc_rport *rport; +#endif }; struct hv_host_device { @@ -1823,8 +1826,16 @@ static int storvsc_probe(struct hv_device *device, } #if IS_ENABLED(CONFIG_SCSI_FC_ATTRS) if (host->transportt == fc_transport_template) { + struct fc_rport_identifiers ids; + + ids.node_name = 0; + ids.port_name = 0; + ids.port_id = 0; + ids.roles |= FC_PORT_ROLE_FCP_TARGET; + fc_host_node_name(host) = stor_device->node_name; fc_host_port_name(host) = stor_device->port_name; + stor_device->rport = fc_remote_port_add(host, 0, &ids); } #endif return 0; @@ -1854,8 +1865,10 @@ static int storvsc_remove(struct hv_device *dev) struct Scsi_Host *host = stor_device->host; #if IS_ENABLED(CONFIG_SCSI_FC_ATTRS) - if (host->transportt == fc_transport_template) + if (host->transportt == fc_transport_template) { + fc_remote_port_delete(stor_device->rport); fc_remove_host(host); + } #endif scsi_remove_host(host); storvsc_dev_remove(dev);
Included in the current storvsc driver for Hyper-V is the ability to access luns on an FC fabric via a virtualized fiber channel adapter exposed by the Hyper-V host. The driver also attaches to the FC transport to allow host and port names to be published under /sys/class/fc_host/hostX. Current customer tools running on the VM require that these names be available in the well known standard location under fc_host/hostX. A problem arose when attaching to the FC transport. The scsi_scan code attempts to call fc_user_scan which has basically become a no-op due to the fact that the virtualized FC device does not expose rports. At this point you cannot refresh the scsi bus after mapping or unmapping luns on the SAN without a reboot of the VM. This patch stubs in an rport per fc_host in storvsc so that the requirement of a defined rport is now met within the fc_transport and echo "- - -" > /sys/class/scsi_host/hostX/scan now works. Signed-off-by: Cathy Avery <cavery@redhat.com> --- drivers/scsi/storvsc_drv.c | 15 ++++++++++++++- 1 file changed, 14 insertions(+), 1 deletion(-)