> >> This is the really low level of guest entry/exit code. > >> > >> Book3s_64 has an SLB, which stores all ESID -> VSID mappings we're > >> currently aware of. > >> > >> The segments in the guest differ from the ones on the host, so we > >> need > >> to switch the SLB to tell the MMU that we're in a new context. > >> > >> So we store a shadow of the guest's SLB in the PACA, switch to that > >> on > >> entry and only restore bolted entries on exit, leaving the rest to > >> the > >> Linux SLB fault handler. > >> > >> That way we get a really clean way of switching the SLB. > >> > >> Signed-off-by: Alexander Graf <agraf@xxxxxxx> > >> --- > >> arch/powerpc/kvm/book3s_64_slb.S | 277 ++++++++++++++++++++++++++++ > >> ++++++++ > > ++ > >> 1 files changed, 277 insertions(+), 0 deletions(-) > >> create mode 100644 arch/powerpc/kvm/book3s_64_slb.S > >> > >> diff --git a/arch/powerpc/kvm/book3s_64_slb.S b/arch/powerpc/kvm/ > >> book3s_64_sl > > b.S > >> new file mode 100644 > >> index 0000000..00a8367 > >> --- /dev/null > >> +++ b/arch/powerpc/kvm/book3s_64_slb.S > >> @@ -0,0 +1,277 @@ > >> +/* > >> + * This program is free software; you can redistribute it and/or > >> modify > >> + * it under the terms of the GNU General Public License, version > >> 2, as > >> + * published by the Free Software Foundation. > >> + * > >> + * This program is distributed in the hope that it will be useful, > >> + * but WITHOUT ANY WARRANTY; without even the implied warranty of > >> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the > >> + * GNU General Public License for more details. > >> + * > >> + * You should have received a copy of the GNU General Public License > >> + * along with this program; if not, write to the Free Software > >> + * Foundation, 51 Franklin Street, Fifth Floor, Boston, MA > >> 02110-1301, USA. > >> + * > >> + * Copyright SUSE Linux Products GmbH 2009 > >> + * > >> + * Authors: Alexander Graf <agraf@xxxxxxx> > >> + */ > >> + > >> +/ > >> *** > >> *** > >> ********************************************************************* > > *** > >> + * > > * > >> + * Entry code > > * > >> + * > > * > >> + > >> *** > >> *** > >> ********************************************************************* > > **/ > >> + > >> +.global kvmppc_handler_trampoline_enter > >> +kvmppc_handler_trampoline_enter: > >> + > >> + /* Required state: > >> + * > >> + * MSR = ~IR|DR > >> + * R13 = PACA > >> + * R9 = guest IP > >> + * R10 = guest MSR > >> + * R11 = free > >> + * R12 = free > >> + * PACA[PACA_EXMC + EX_R9] = guest R9 > >> + * PACA[PACA_EXMC + EX_R10] = guest R10 > >> + * PACA[PACA_EXMC + EX_R11] = guest R11 > >> + * PACA[PACA_EXMC + EX_R12] = guest R12 > >> + * PACA[PACA_EXMC + EX_R13] = guest R13 > >> + * PACA[PACA_EXMC + EX_CCR] = guest CR > >> + * PACA[PACA_EXMC + EX_R3] = guest XER > >> + */ > >> + > >> + mtsrr0 r9 > >> + mtsrr1 r10 > >> + > >> + mtspr SPRN_SPRG_SCRATCH0, r0 > >> + > >> + /* Remove LPAR shadow entries */ > >> + > >> +#if SLB_NUM_BOLTED == 3 > > > > You could alternatively check the persistent entry in the slb_shawdow > > buffer. This would give you a run time check. Not sure what's best > > though. > > Well we're in the hot path here, so anything using as few registers as > possible and being simple is the best :-). I'd guess the more we are > clever at compile time the better. Yeah, I tend to agree. > > > > > > >> + > >> + ld r12, PACA_SLBSHADOWPTR(r13) > >> + ld r10, 0x10(r12) > >> + ld r11, 0x18(r12) > > > > Can you define something in asm-offsets.c for these magic constants > > 0x10 > > and 0x18. Similarly below. > > > >> + /* Invalid? Skip. */ > >> + rldicl. r0, r10, 37, 63 > >> + beq slb_entry_skip_1 > >> + xoris r9, r10, SLB_ESID_V@h > >> + std r9, 0x10(r12) > >> +slb_entry_skip_1: > >> + ld r9, 0x20(r12) > >> + /* Invalid? Skip. */ > >> + rldicl. r0, r9, 37, 63 > >> + beq slb_entry_skip_2 > >> + xoris r9, r9, SLB_ESID_V@h > >> + std r9, 0x20(r12) > >> +slb_entry_skip_2: > >> + ld r9, 0x30(r12) > >> + /* Invalid? Skip. */ > >> + rldicl. r0, r9, 37, 63 > >> + beq slb_entry_skip_3 > >> + xoris r9, r9, SLB_ESID_V@h > >> + std r9, 0x30(r12) > > > > Can these 3 be made into a macro? > > Phew - dynamically generating jump points sounds rather hard. I can > give it a try... > > > > >> +slb_entry_skip_3: > >> + > >> +#else > >> +#error unknown number of bolted entries > >> +#endif > >> + > >> + /* Flush SLB */ > >> + > >> + slbia > >> + > >> + /* r0 = esid & ESID_MASK */ > >> + rldicr r10, r10, 0, 35 > >> + /* r0 |= CLASS_BIT(VSID) */ > >> + rldic r12, r11, 56 - 36, 36 > >> + or r10, r10, r12 > >> + slbie r10 > >> + > >> + isync > >> + > >> + /* Fill SLB with our shadow */ > >> + > >> + lbz r12, PACA_KVM_SLB_MAX(r13) > >> + mulli r12, r12, 16 > >> + addi r12, r12, PACA_KVM_SLB > >> + add r12, r12, r13 > >> + > >> + /* for (r11 = kvm_slb; r11 < kvm_slb + kvm_slb_size; > >> r11+=slb_entry) */ > >> + li r11, PACA_KVM_SLB > >> + add r11, r11, r13 > >> + > >> +slb_loop_enter: > >> + > >> + ld r10, 0(r11) > >> + > >> + rldicl. r0, r10, 37, 63 > >> + beq slb_loop_enter_skip > >> + > >> + ld r9, 8(r11) > >> + slbmte r9, r10 > > > > If you're updating the first 3 slbs, you need to make sure the slb > > shadow is updated at the same time > > Well - what happens if we don't? We'd get a segment fault when phyp > stole our entry! So what? Let it fault, see the mapping is already > there and get back in again :-). The problem is you won't take the segment fault as PHYP may put a valid entry in there. PHYP will put back what's in the shadow buffer, which could be valid hence no segment fault. > > (BTW dumb question: can we run this > > under PHYP?) > > Yes, I tested it on bare metal, phyp and a PS3. Nice! Mikey -- 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
