On 2021-05-25 4:32 a.m., Krzysztof Wilczyński wrote: > Hi Bjorn, > > [...] >>> I really like this, but I would like it even better if the >>> sysfs_emit() change were easier to review. >>> >>> It seems pointless that the current code uses strlen() when >>> scnprintf() and dsm_label_utf16s_to_utf8s() both have that >>> information and we just throw it away. >>> >>> I think it should be possible to split the len and >>> dsm_label_utf16s_to_utf8s() changes to a separate patch, which would >>> remove the need for the strlen, and then the conversion to >>> sysfs_emit() would be completely trivial like all the rest of them. >>> >>> My goal is to make all the sysfs_emit() changes look almost >>> mechanical, with the non-trivial parts separated out. >> >> And BTW, when all the sysfs_emit() changes are trivial like that, I >> would probably squash them all into one patch that converts all of >> drivers/pci/ at once. >> >> That would still leave a few separate patches: >> >> - This dsm_label_utf16s_to_utf8s() change >> - The resource_alignment newline change >> - The devspec_show newline change >> - The driver_override change > > Got it! I will send v4 updated as per the above suggestion. Also, if > Logan does not mind, I will carry his "Reviewed-by" over as there will > be no changes to the actual code, just how the patch will be arranged. Yup, I'm good with that. Still looks fine by me. Logan
