From: Andrea Parri (Microsoft) <parri.andrea@xxxxxxxxx> Sent: Friday, March 11, 2022 5:38 AM > > Initialize the device's dma_{mask,parms} pointers before invoking > device_register(). Address the following trace with 5.17-rc7: > > [ 49.646839] WARNING: CPU: 0 PID: 189 at include/linux/dma-mapping.h:543 > netvsc_probe+0x37a/0x3a0 [hv_netvsc] > [ 49.646928] Call Trace: > [ 49.646930] <TASK> > [ 49.646935] vmbus_probe+0x40/0x60 [hv_vmbus] > [ 49.646942] really_probe+0x1ce/0x3b0 > [ 49.646948] __driver_probe_device+0x109/0x180 > [ 49.646952] driver_probe_device+0x23/0xa0 > [ 49.646955] __device_attach_driver+0x76/0xe0 > [ 49.646958] ? driver_allows_async_probing+0x50/0x50 > [ 49.646961] bus_for_each_drv+0x84/0xd0 > [ 49.646964] __device_attach+0xed/0x170 > [ 49.646967] device_initial_probe+0x13/0x20 > [ 49.646970] bus_probe_device+0x8f/0xa0 > [ 49.646973] device_add+0x41a/0x8e0 > [ 49.646975] ? hrtimer_init+0x28/0x80 > [ 49.646981] device_register+0x1b/0x20 > [ 49.646983] vmbus_device_register+0x5e/0xf0 [hv_vmbus] > [ 49.646991] vmbus_add_channel_work+0x12d/0x190 [hv_vmbus] > [ 49.646999] process_one_work+0x21d/0x3f0 > [ 49.647002] worker_thread+0x4a/0x3b0 > [ 49.647005] ? process_one_work+0x3f0/0x3f0 > [ 49.647007] kthread+0xff/0x130 > [ 49.647011] ? kthread_complete_and_exit+0x20/0x20 > [ 49.647015] ret_from_fork+0x22/0x30 > [ 49.647020] </TASK> > [ 49.647021] ---[ end trace 0000000000000000 ]--- > > Fixes: 743b237c3a7b0 ("scsi: storvsc: Add Isolation VM support for storvsc driver") > Signed-off-by: Andrea Parri (Microsoft) <parri.andrea@xxxxxxxxx> > --- > drivers/hv/vmbus_drv.c | 5 +++-- > 1 file changed, 3 insertions(+), 2 deletions(-) > > diff --git a/drivers/hv/vmbus_drv.c b/drivers/hv/vmbus_drv.c > index 12a2b37e87f30..65db5048b1763 100644 > --- a/drivers/hv/vmbus_drv.c > +++ b/drivers/hv/vmbus_drv.c > @@ -2097,6 +2097,9 @@ int vmbus_device_register(struct hv_device > *child_device_obj) > child_device_obj->device.parent = &hv_acpi_dev->dev; > child_device_obj->device.release = vmbus_device_release; > > + child_device_obj->device.dma_parms = &child_device_obj->dma_parms; > + child_device_obj->device.dma_mask = &child_device_obj->dma_mask; > + > /* > * Register with the LDM. This will kick off the driver/device > * binding...which will eventually call vmbus_match() and vmbus_probe() > @@ -2122,8 +2125,6 @@ int vmbus_device_register(struct hv_device > *child_device_obj) > } > hv_debug_add_dev_dir(child_device_obj); > > - child_device_obj->device.dma_parms = &child_device_obj->dma_parms; > - child_device_obj->device.dma_mask = &child_device_obj->dma_mask; > dma_set_mask(&child_device_obj->device, DMA_BIT_MASK(64)); Is there any reason to not move dma_set_mask() as well? That call is closely related to the previous two lines, so it is unexpected to have them separated. Michael