On Tue, Sep 25, 2018 at 11:28:24AM +0000, Ruhl, Michael J wrote: > >> >Signed-off-by: Jason Gunthorpe <jgg@xxxxxxxxxxxx> > >> > drivers/infiniband/hw/hfi1/init.c | 1 - > >> > drivers/infiniband/hw/hfi1/verbs.c | 4 ++- > >> > drivers/infiniband/hw/qib/qib_init.c | 1 - > >> > drivers/infiniband/hw/qib/qib_verbs.c | 4 ++- > >> >diff --git a/drivers/infiniband/hw/hfi1/init.c > >> >b/drivers/infiniband/hw/hfi1/init.c > >> >index 1e770a1337793e..e87ff7a544eb1a 100644 > >> >+++ b/drivers/infiniband/hw/hfi1/init.c > >> >@@ -1313,7 +1313,6 @@ static struct hfi1_devdata > >*hfi1_alloc_devdata(struct > >> >pci_dev *pdev, > >> > "Could not allocate unit ID: error %d\n", -ret); > >> > goto bail; > >> > } > >> >- rvt_set_ibdev_name(&dd->verbs_dev.rdi, "%s_%d", class_name(), > >> >dd->unit); > >> > >> If you remove this call, the HFI (and probably the QIB) dd_dev_xxx() macros will > >> not work correctly until we call ib_register(). Since we do a lot of work before > >> that, this will lose our ability to print out the device name. > > > >What prints exactly? > > > >From what I can see the hfi driver uses macros like this: > > > >#define dd_dev_warn(dd, fmt, ...) \ > > dev_warn(&(dd)->pcidev->dev, "%s: " fmt, \ > > You missed the important part of this macro: > > #define dd_dev_warn(dd, fmt, ...) \ > dev_warn(&(dd)->pcidev->dev, "%s: " fmt, \ > rvt_get_ibdev_name(&(dd)->verbs_dev.rdi), ##__VA_ARGS__) > > The unit name is retrieved with the rvt_get_ibdev_name() function. So lets just keep setting dev_name early in hfi1 then. This is really ugly BTW, drivers shouldn't be making so much stuff up on their own.. Jason
