David Brownell wrote: > On Wednesday 13 January 2010, Spencer Oliver wrote: >> - armv7m_run_algorithm now requires all algorithms to use >> a software breakpoint at their exit address >> - updated all algorithms to support this > > I like! Except, again, for a few overlong lines. :) > > But let's hold off for the 0.5 series for this; it's not a bugfix. > (I suspect that merge window will be open in a few weeks.) It's > a bit of a performance win for algorithm-heavy code, like flash > writing. > > $SUBJECT is just a bit off though, needs fixing ... this this is > ARMv7-M only, not for all ARMv7 targets. > > Covering the rest of ARMv7 would end up covering ARM9/ARM11 targets > too ... anything using the ARMv5 instruction set, or newer. (ARMv4 > didn't include BKPT.) And while those targets really should get the > benefit of this approach, IMO that'd best be a separate patch. >
I will make the changes and wait until merge window is open. > - Dave > > p.s. Random question ... have you looked at Cortex-M0 enough to have > a feeling for how many ARMv7-M algorithms should be reusable on > those new ARMv6-M chips? v6-M is basically a trimmed down v7-M. > Since it doesn't have many Thumb2 instructions, at least some of > the v7-M code won't run there. Looks to me like most of the rest > of the Cortex-M3 infrastructure will be fine with Cortex-M0. > Not really looked into it, but do not think it would be a problem. I have the src to the algo's, so will try building using arch=v6 Be more interested if nxp want to donate a board :) Cheers Spen _______________________________________________ Openocd-development mailing list [email protected] https://lists.berlios.de/mailman/listinfo/openocd-development
