2011/7/21 Michael Büsch <[email protected]>: > On Thu, 21 Jul 2011 19:34:40 +0000 > Michael Büsch <[email protected]> wrote: >> For historical reasons there is unused 64bit dma code in b43legacy. I'm not >> sure whether it would even work. Probably not. Larry's fixes were most >> likely not ported over. > > So I looked closer at the code now. > The b43legacy 64bit DMA code certainly is dead, unused and broken code. > One could remove it, if someone really cares (I don't). > > However, there is another issue. > b43legacy calls ssb_dma_translation() (also on 30/32bit code) for every > descriptor. > As our ssb patch adds a SSB register read to ssb_dma_translation(), this adds > an unacceptable > amount of overhead. This needs to be addressed before merging the ssb patch.
Good catch, thanks. As I can see, John has already took my patch. Some time has passed without anyone noticing real problems, so I'm feeling myself the one to blame. Don't you mind if I prepare improvement (AKA fix) for b43legacy and just post it today? Would that be acceptable? Or should we really revert my latest patches? Is that a really better solution? -- Rafał _______________________________________________ b43-dev mailing list [email protected] http://lists.infradead.org/mailman/listinfo/b43-dev
