Arend van Spriel <[email protected]> writes: > On 7/27/2017 12:17 PM, Kalle Valo wrote: >> Larry Finger <[email protected]> writes: >> >>> From: Ping-Ke Shih <[email protected]> >>> >>> Use debugfs to dump register and btcoex status. >> >> The title is useless and the commit log does not mention anything about >> what files are added and to which location. >> >>> The kbuild test robot <[email protected]> suggested a change to use >>> simple_open() instead of a local routine. >> >> This comment doesn't belong to commit log, it should be in the changelog. >> >>> +void rtl_debugfs_add_topdir(void) >>> +{ >>> + debugfs_topdir = debugfs_create_dir("rtlwifi", NULL); >>> +} >>> + >>> +void rtl_debugfs_remove_topdir(void) >>> +{ >>> + debugfs_remove_recursive(debugfs_topdir); >>> +} >> >> I'm surprised to see that rtlwifi creates it's own top level debugfs >> directory and does not use wiphy->debugfsdir. So how is this supposed to >> work when we have multiple rtlwifi devices on the same system? > > In brcmfmac we also have our own top level debugfs dir because we do > wiphy_new()/wiphy_register() kinda late. To cover the multiple devices > issue we create a subdir per device, ie. > /sys/kernel/debug/brcmfmac/mmc0:0001:1/.
Yeah, that's fine. If rtlwifi does the same that's good but document that in the commit log. -- Kalle Valo
