On Wed, Jun 15, 2016 at 10:51:06AM -0600, Jason Gunthorpe wrote: > On Wed, Jun 15, 2016 at 11:20:49AM -0400, ira.weiny wrote: > > The default here is that a device does not provide a firmware string. In that > > case the assumption is that there is no firmware. Hence the empty string. I > > don't think that reporting 0's or F's are appropriate if a device does not have > > firmware. > > > > In the cases you speak of I would expect the device to supply the firmware > > function and report all 0s or all Fs for whatever version that may represent. > > (I'm assuming that would be some sort of debug or unreleased firmware...) > > > > I could just as well done something like. > > > > snprintf(str, str_len, "<none>"); > > Just don't provide a 'fw' sysfs file at all if the driver cannot > support it. This function is supporting both sysfs and ethtool. ethtool_drvinfo->fw_version is normally up to the device. I've just provided a default. For example I see "N/A", "n/a", and "none" all used in the network stack. ./ethernet/freescale/gianfar_ethtool.c: strlcpy(drvinfo->fw_version, "N/A", sizeof(drvinfo->fw_version)); ./rionet.c: strlcpy(info->fw_version, "n/a", sizeof(info->fw_version)); ./fjes/fjes_ethtool.c: strlcpy(drvinfo->fw_version, "none", sizeof(drvinfo->fw_version)); I really think we are in the weeds here and are fine to leave it blank. But if you wish, I can force all drivers to provide the string and then they can put something in like the above. Ira -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html