> -----Original Message----- > From: Wood Scott-B07421 > Sent: Thursday, July 18, 2013 11:09 PM > To: Alexander Graf > Cc: Bhushan Bharat-R65777; kvm-ppc@xxxxxxxxxxxxxxx; kvm@xxxxxxxxxxxxxxx; Bhushan > Bharat-R65777 > Subject: Re: [PATCH 2/2 v2] kvm: powerpc: set cache coherency only for kernel > managed pages > > On 07/18/2013 12:32:18 PM, Alexander Graf wrote: > > > > On 18.07.2013, at 19:17, Scott Wood wrote: > > > > > On 07/18/2013 08:19:03 AM, Bharat Bhushan wrote: > > > Likewise, we want to make sure this matches the host entry. > > Unfortunately, this is a bit of a mess already. 64-bit booke appears > > to always set MAS2_M for TLB0 mappings. The initial KERNELBASE > > mapping on boot uses M_IF_SMP, and the settlbcam() that (IIRC) > > replaces it uses _PAGE_COHERENT. 32-bit always uses _PAGE_COHERENT, > > except that initial KERNELBASE mapping. _PAGE_COHERENT appears to be > > set based on CONFIG_SMP || CONFIG_PPC_STD_MMU (the latter config > > clears _PAGE_COHERENT in the non-CPU_FTR_NEED_COHERENT case). > > > > > > As for what we actually want to happen, there are cases when we > > want M to be set for non-SMP. One such case is AMP, where CPUs may be > > sharing memory even if the Linux instance only runs on one CPU (this > > is not hypothetical, BTW). It's also possible that we encounter a > > hardware bug that requires MAS2_M, similar to what some of our > > non-booke chips require. > > > > How about we always set M then for RAM? > > M is like I in that bad things happen if you mix them. I am trying to list the invalid mixing of WIMG: 1) I & M 2) W & I 3) W & M (Scott mentioned that he observed issues when mixing these two) 4) is there any other? So it mean it is safe to let guest control G and E. > So we really want to > match exactly what the rest of the kernel is doing. How the rest of kernel is doing is a bit complex. IIUC, if we forget about the boot state then this is how kernel set WIMG bits: 1) For Memory always set M if CONFIG_SMP set. - So KVM can do same. "M" will not be mixed with "W" and "I". G and E are guest control. 2) For I/O , drivers can pass flags to set M or "I + G". - For KVM; if not memory then it is I/O. For now we can always set "I + G". - Later we can design some mechanism in VFIO interface to let KVM somehow know whether to set "M" or "I+G". -Bharat > > Plus, the performance penalty on some single-core chips can be pretty bad. > > -Scott -- To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
