Wolfram! On Wed, Nov 30 2022 at 22:06, Wolfram Sang wrote: > Documentation for most CMTs say that we need to wait two input clocks > before changes propagate to the timer. This is especially relevant when > we stop the timer to change further settings. Implement the delays > according to the spec. To avoid unnecessary delays in atomic mode, we > also check if the to-be-written value actually differs. CMCNT is a bit > special because testing showed that we need to wait 3 cycles instead. > AFAIU, this is also true for all CMTs. Also, the WRFLAG needs to be > checked before writing. This fixes "cannot clear CMCNT" messages which > occur often on R-Car Gen4 SoCs, but only very rarely on older SoCs for > some reason. https://www.kernel.org/doc/html/latest/process/maintainer-tip.html#changelog I fixed it up for you this time... > Fixes: 81b3b2711072 ("clocksource: sh_cmt: Add support for multiple channels per device") > Signed-off-by: Wolfram Sang <wsa+renesas@xxxxxxxxxxxxxxxxxxxx> > --- > > Changes since RFC v2: > * use DIV_ROUND_UP also for caluclating cmcnt_delay > * remove a FIXME comment > > There were no further comments, so I dropped the RFC status and send > this as a regular patch. Here again the introduction from RFC v1: Instead of quoting, it would have been helpful to have the lore links for RFC v1/v2.... Thanks, tglx
