On Tue, Jun 09, 2020 at 05:39:45PM +0200, refactormyself@xxxxxxxxx wrote:
> From: Bolarinwa Olayemi Saheed <refactormyself@xxxxxxxxx>
>
> restore_pci_variables() and save_pci_variables() return PCIBIOS_ error
> codes from pcie capability accessors.
>
> PCIBIOS_ error codes have positive values. Passing on these values is
> inconsistent with functions which return only a negative value on failure.
>
> Before passing on return value of pcie capability accessors, call
> pcibios_err_to_errno() to convert any positive PCIBIOS_ error codes to
> negative non-PCI generic error values.
>
> Signed-off-by: Bolarinwa Olayemi Saheed <refactormyself@xxxxxxxxx>
> Suggested-by: Bjorn Helgaas <bjorn@xxxxxxxxxxx>
> ---
> drivers/infiniband/hw/hfi1/pcie.c | 16 ++++++++++++----
> 1 file changed, 12 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/infiniband/hw/hfi1/pcie.c b/drivers/infiniband/hw/hfi1/pcie.c
> index eb53781d0c6a..eb2790e9db36 100644
> --- a/drivers/infiniband/hw/hfi1/pcie.c
> +++ b/drivers/infiniband/hw/hfi1/pcie.c
> @@ -334,7 +334,11 @@ int pcie_speeds(struct hfi1_devdata *dd)
> return 0;
> }
>
> -/* restore command and BARs after a reset has wiped them out */
> +/**
> + * restore command and BARs after a reset has wiped them out
> + *
> + * Returns 0 on success, otherwise a negative error value
> + */
> int restore_pci_variables(struct hfi1_devdata *dd)
> {
> int ret = 0;
These initializations are obviously superfluous. Generally
I wouldn't encourage mixing fixes into a single patch, but
removing the initialization and adding pcibios_err_to_errno()
are both so trivial that I think it would be OK here.
Also, I didn't notice before, but these driver patches need to be cc'd
to the driver maintainers. In this case, get_maintainers.pl says:
Mike Marciniszyn <mike.marciniszyn@xxxxxxxxx> (supporter:HFI1 DRIVER)
Dennis Dalessandro <dennis.dalessandro@xxxxxxxxx> (supporter:HFI1 DRIVER)
Doug Ledford <dledford@xxxxxxxxxx> (supporter:INFINIBAND SUBSYSTEM)
Jason Gunthorpe <jgg@xxxxxxxx> (supporter:INFINIBAND SUBSYSTEM)
linux-rdma@xxxxxxxxxxxxxxx (open list:HFI1 DRIVER)
linux-kernel@xxxxxxxxxxxxxxx (open list)
I would also cc *all* the driver maintainers for all the patches on
the cover letter and add a note about how we might merge these.
They're all independent, so they *could* go individually via the
separate driver trees, but I think it might make sense to merge them
all together via the PCI tree just so the collection of similar fixes
is merged all at once.
> @@ -386,10 +390,14 @@ int restore_pci_variables(struct hfi1_devdata *dd)
>
> error:
> dd_dev_err(dd, "Unable to write to PCI config\n");
> - return ret;
> + return pcibios_err_to_errno(ret);
> }
>
> -/* Save BARs and command to rewrite after device reset */
> +/**
> + * Save BARs and command to rewrite after device reset
> + *
> + * Returns 0 on success, otherwise a negative error value
> + */
> int save_pci_variables(struct hfi1_devdata *dd)
> {
> int ret = 0;
> @@ -441,7 +449,7 @@ int save_pci_variables(struct hfi1_devdata *dd)
>
> error:
> dd_dev_err(dd, "Unable to read from PCI config\n");
> - return ret;
> + return pcibios_err_to_errno(ret);
> }
>
> /*
> --
> 2.18.2
>
