On Sun, May 20, 2012 at 05:40:50PM -0400, David Miller wrote: > From: Sam Ravnborg <sam@xxxxxxxxxxxx> > Date: Sun, 20 May 2012 23:34:52 +0200 > > > On Sun, May 20, 2012 at 05:24:01PM -0400, David Miller wrote: > >> Second, have you audited to make sure that none of this > >> code runs before the patchup code does? > > Yes - but a lot of other places that require patching will run > > long time before (wof.S, wuf.S etc). At least I assume so. > > The wof.S and wuf.S cases would only execute after we've started > userland, as that code is used to validate the user stack in the > register window store area. > > Therefore it won't run unless PSR_PS is clear in %psr or there > are user register windows in the cpu. > > Neither of which will be true during early boot. Thanks for the explanation, but wof.S and wuf.S was bad examples.. > > You should therefore be fine except for the bits in head_32.S > which you'll need to use tests and branches for anyways. We have ~27 places where we reference MMUREGS in kernel/* on top of head_32.S. Rather than trying to convince myself that they are not called in between start_kernel() and setup_arch() I prefer to patch is all up before calling start_kernel(). This also matches what we do on sparc64 for sun4v. Anyway - as the merge window is now opened I will put this on hold until -rc2 or so unless you say otherwise. Sam -- To unsubscribe from this list: send the line "unsubscribe sparclinux" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
