Messages in this thread | | | Date | Thu, 15 Oct 2009 18:27:37 +0400 | From | Cyrill Gorcunov <> | Subject | Re: [patch 3/3] x86,apic: limit apic dumping, introduce show_lapic setup option |
| |
[Cyrill Gorcunov - Thu, Oct 15, 2009 at 12:29:41AM +0400] | [Ingo Molnar - Wed, Oct 14, 2009 at 07:56:16PM +0200] | | | | btw., i'm getting this warning on a test-box: | ... | | | | [ 0.400000] Checking 'hlt' instruction... OK. | | [ 0.440000] SMP alternatives: switching to UP code | | [ 0.450000] debug: unmapping init memory 79c7a000..79c80000 | | | | noop_default_phys_pkg_id() should return 0 i guess, and not generate a | | warning? | | | | Ingo | | | | Ingo, I think the better would be to just make NOOP plain | NOOP without any kind of checking. Something like below | (I still can't fetch last/fresh -tip repo /remote point hangs/ | for this so I've used interdiff). | | Comments? | | Btw, the base problem from dump you posted is that we've been | cheking only for read/write operations allowing all other ops | be called without any kind of check (and retrieving pkgid happened | there all the time just didn't trigger warn until we've substituted | apic driver completely). Using NOOP without warn should be more | convenient (all previous "dark" places in apic code already covered | with cpu_has_apic/disable_apic checks so the only risk we have -- | is code that will appear in future). | | -- Cyrill | ---
Ingo, please don't apply it yet. I think there is a nit. Checking now. Will resend later the proper patch.
-- Cyrill
| |