Date: Tue, 5 Jul 2016 11:26:43 -0700 From: Nathan Whitehorn <nwhitehorn@freebsd.org> To: Warner Losh <imp@bsdimp.com>, Adrian Chadd <adrian@freebsd.org> Cc: "freebsd-hackers@freebsd.org" <freebsd-hackers@freebsd.org>, outro pessoa <outro.pessoa@gmail.com> Subject: Re: Review request: sparse CPU ID maps Message-ID: <59222776-45b4-640c-b5e4-5f8b8d6c45e5@freebsd.org> In-Reply-To: <CANCZdfpJJLoKxB-ZdMRQyHq9eT1uihg4UGeBvRgBEOOC1pt_Yg@mail.gmail.com> References: <57761101.3030101@freebsd.org> <CAD9=5Xw-MmVVSSo6nRvSRvGaLbd1Z1YRyVKyF9JfmucNKMGBZg@mail.gmail.com> <5345fb94-91b8-5019-037e-d4825a694cfd@freebsd.org> <CAJ-Vmon4kRNc5LiwibtiPi_FQ1v5w_MQEjP%2BOfcC7J74iTKs0A@mail.gmail.com> <CANCZdfpJJLoKxB-ZdMRQyHq9eT1uihg4UGeBvRgBEOOC1pt_Yg@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On 07/03/16 13:11, Warner Losh wrote: > On Sun, Jul 3, 2016 at 1:37 PM, Adrian Chadd <adrian@freebsd.org> wrote: >> On 2 July 2016 at 17:08, Nathan Whitehorn <nwhitehorn@freebsd.org> wrote: >>> A reasonable first pass at checking for this kind of bug is doing grep -lR >>> '< mp_ncpus'. Running that on sys/arm and sys/arm64 shows the following >>> files: >>> arm/mv/armadaxp/armadaxp_mp.c >>> arm/include/counter.h >>> arm/broadcom/bcm2835/bcm2836.c >>> arm/broadcom/bcm2835/bcm2836_mp.c >>> arm/freescale/imx/imx6_mp.c >>> arm/allwinner/aw_mp.c >>> arm/rockchip/rk30xx_mp.c >>> arm/amlogic/aml8726/aml8726_mp.c >>> arm/samsung/exynos/exynos5_mp.c >>> arm/arm/mp_machdep.c >>> arm/nvidia/tegra124/tegra124_mp.c >>> arm64/include/counter.h >>> arm64/arm64/gic_v3.c >>> arm64/arm64/gic_v3_its.c >>> arm64/arm64/gicv3_its.c >>> >>> All of them should, in some sense, be CPU_FOREACH(), but it may not matter. >>> For example, it may not be possible to have sparse CPU IDs on some or all of >>> those SOCs. At least the generic ones (counter, mp_machdep.c, gic (why are >>> there both gic_v3_its.c and gicv3_its.c?)) should be changed, I think. >>> -Nathan >> I think converting all the users over to the CPU_FOREACH thing is the >> right way to go, even if the SOC doesn't require it. People do bring >> up new systems by copy/pasta'ing an existing similar system, so we're >> best served by having all the consumers migrated. >> >> But, I'd do it in head/12. Early in head/12. :-P > It is a mergeable change too, since it wouldn't change any APIs. > At least the conversion to CPU_FOREACH. We don't want too many > sweeping changes that can't be merged too early (that way leads to > lots of maintenance issues), but we can do something like this. Merging > would be optional, but possible, for those bits of the tree that need it. > Though, for something like this, there's little against doing a full merge > and a lot for it... > > Warner > That sounds like the right approach. Since the original patch fixes bugs in 11, rather than niceties, I will send it to re@ tomorrow. After the branch, I'll do a sweep of other obviously wrong code for 12 with an MFC timer. -Nathan
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?59222776-45b4-640c-b5e4-5f8b8d6c45e5>