Date: Sat, 3 Aug 2013 12:49:41 -0700 From: Adrian Chadd <adrian@freebsd.org> To: Cedric GROSS <cg@cgross.info> Cc: freebsd-wireless@freebsd.org Subject: Re: [IWN] Reviw split 2 Message-ID: <CAJ-Vmo=YBzkcySqJxWhn0WWtbSrCsT41GO%2BBwYOvDHu2_A7KBQ@mail.gmail.com> In-Reply-To: <003201ce9081$b401a270$1c04e750$@info> References: <51f3f0ce.055a420a.2e1e.fffff220SMTPIN_ADDED_BROKEN@mx.google.com> <CAJ-VmokCVB5kNY44hJLbAfOb1DMSHmJAG3QTUZYhmPL1gHwMwA@mail.gmail.com> <002d01ce8c46$a13b23d0$e3b16b70$@info> <CAJ-Vmon4hMbgFKaWva3-HhcJv=eUXKwX7s0uPcD9Nu9g86QEbA@mail.gmail.com> <002701ce8e03$c033f640$409be2c0$@info> <CAJ-Vmo=yZXdKuXZ85bXs-uG2tAmcZFMAgFXCswnVBk2PUmaXfQ@mail.gmail.com> <002401ce8f5f$fc5ad780$f5108680$@info> <CAJ-VmoniUozz48U0MHhF4sAsrJt6sd06Q9UESRFG9kOXSB2ObQ@mail.gmail.com> <001001ce903b$e77a5f70$b66f1e50$@info> <CAJ-VmonRBrVCZu9dshSEiVxH9=0LhHdxr5tew4tsN1A5R9f0Sw@mail.gmail.com> <001e01ce907e$f1daf220$d590d660$@info> <CAJ-VmokSjs1gWJa0h6evb71Nf0=jNSLBMCW00EtAjinOmT69ig@mail.gmail.com> <003201ce9081$b401a270$1c04e750$@info>
next in thread | previous in thread | raw e-mail | index | archive | help
On 3 August 2013 12:43, Cedric GROSS <cg@cgross.info> wrote: >> Ok, why'd you change the debug print macro to check if the debug flags >> match the check, rather than if the debug flags are set in the check? >> >> ie >> >> (f) & (v) >> >> versus >> ( (f) & (v) == (v) ) >> >> ? > > It's for reducing tracing verbosity and just do trace when associate with > another IWN_DEBUG_* > So if you wish to debug only XMIT, trace also print only associate with that > level (ie IWN_DEBUG_TRACE | IWN_DEBUG_XMIT) Ok. I like the general idea, but I think overloading that for the general case is against POLA. Eg, ath(4), ath_hal(4), net80211(4) all have the mask idea, rather than the exact match idea. So there are cases where multiple bits are set in a debug mask (eg some INPUT and 11N flags in net80211) since they're relevant for both. So I'd like to come up with an alternative way to do trace debugging like you ask. Maybe what we should do is add a DPRINTF_TRACE() macro for things that are specifically _trace_ events, then have a separate trace bitmap for trace debugging. -adrian
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAJ-Vmo=YBzkcySqJxWhn0WWtbSrCsT41GO%2BBwYOvDHu2_A7KBQ>