static void kvm_mips_build_save_scratch(u32 **p, unsigned int tmp,
unsigned int frame)
{
/* Save the VCPU scratch register value in cp0_epc of the stack frame */
UASM_i_MFC0(p, tmp, scratch_vcpu[0], scratch_vcpu[1]);
UASM_i_SW(p, tmp, offsetof(struct pt_regs, cp0_epc), frame);
/* Save the temp scratch register value in cp0_cause of stack frame */
if (scratch_tmp[0] == 31) {
UASM_i_MFC0(p, tmp, scratch_tmp[0], scratch_tmp[1]);
UASM_i_SW(p, tmp, offsetof(struct pt_regs, cp0_cause), frame);
}
}
/**
* kvm_mips_build_exception() - Assemble first level guest exception handler.
* @addr: Address to start writing code.
* @handler: Address of common handler (within range of @addr).
*
* Assemble exception vector code for guest execution. The generated vector will
* branch to the common exception handler generated by kvm_mips_build_exit().
*
* Returns: Next address after end of written function.
*/
void *kvm_mips_build_exception(void *addr, void *handler)
{
u32 *p = addr;
struct uasm_label labels[2];
struct uasm_reloc relocs[2];
struct uasm_label *l = labels;
struct uasm_reloc *r = relocs;
memset(labels, 0, sizeof(labels));
memset(relocs, 0, sizeof(relocs));
/* Save guest k1 into scratch register */
UASM_i_MTC0(&p, K1, scratch_tmp[0], scratch_tmp[1]);
/* Get the VCPU pointer from the VCPU scratch register */
UASM_i_MFC0(&p, K1, scratch_vcpu[0], scratch_vcpu[1]);
UASM_i_ADDIU(&p, K1, K1, offsetof(struct kvm_vcpu, arch));
/* Save guest k0 into VCPU structure */
UASM_i_SW(&p, K0, offsetof(struct kvm_vcpu_arch, gprs[K0]), K1);
/* Branch to the common handler */
uasm_il_b(&p, &r, label_exit_common);
uasm_i_nop(&p);
uasm_l_exit_common(&l, handler);
uasm_resolve_relocs(relocs, labels);
return p;
}
static void __cpuinit build_r4000_tlb_load_handler(void)
{
u32 *p = handle_tlbl;
struct uasm_label *l = labels;
struct uasm_reloc *r = relocs;
memset(handle_tlbl, 0, sizeof(handle_tlbl));
memset(labels, 0, sizeof(labels));
memset(relocs, 0, sizeof(relocs));
if (bcm1250_m3_war()) {
UASM_i_MFC0(&p, K0, C0_BADVADDR);
UASM_i_MFC0(&p, K1, C0_ENTRYHI);
uasm_i_xor(&p, K0, K0, K1);
UASM_i_SRL(&p, K0, K0, PAGE_SHIFT + 1);
uasm_il_bnez(&p, &r, K0, label_leave);
/* No need for uasm_i_nop */
}
build_r4000_tlbchange_handler_head(&p, &l, &r, K0, K1);
build_pte_present(&p, &l, &r, K0, K1, label_nopage_tlbl);
if (m4kc_tlbp_war())
build_tlb_probe_entry(&p);
build_make_valid(&p, &r, K0, K1);
build_r4000_tlbchange_handler_tail(&p, &l, &r, K0, K1);
uasm_l_nopage_tlbl(&l, p);
uasm_i_j(&p, (unsigned long)tlb_do_page_fault_0 & 0x0fffffff);
uasm_i_nop(&p);
if ((p - handle_tlbl) > FASTPATH_SIZE)
panic("TLB load handler fastpath space exceeded");
uasm_resolve_relocs(relocs, labels);
pr_debug("Wrote TLB load handler fastpath (%u instructions).\n",
(unsigned int)(p - handle_tlbl));
dump_handler(handle_tlbl, ARRAY_SIZE(handle_tlbl));
}
/*
* R4000 style TLB load/store/modify handlers.
*/
static void __cpuinit
build_r4000_tlbchange_handler_head(u32 **p, struct uasm_label **l,
struct uasm_reloc **r, unsigned int pte,
unsigned int ptr)
{
#ifdef CONFIG_64BIT
build_get_pmde64(p, l, r, pte, ptr); /* get pmd in ptr */
#else
build_get_pgde32(p, pte, ptr); /* get pgd in ptr */
#endif
UASM_i_MFC0(p, pte, C0_BADVADDR);
UASM_i_LW(p, ptr, 0, ptr);
UASM_i_SRL(p, pte, pte, PAGE_SHIFT + PTE_ORDER - PTE_T_LOG2);
uasm_i_andi(p, pte, pte, (PTRS_PER_PTE - 1) << PTE_T_LOG2);
UASM_i_ADDU(p, ptr, ptr, pte);
#ifdef CONFIG_SMP
uasm_l_smp_pgtable_change(l, *p);
#endif
iPTE_LW(p, l, pte, ptr); /* get even pte */
if (!m4kc_tlbp_war())
build_tlb_probe_entry(p);
}
/**
* kvm_mips_build_exit() - Assemble common guest exit handler.
* @addr: Address to start writing code.
*
* Assemble the generic guest exit handling code. This is called by the
* exception vectors (generated by kvm_mips_build_exception()), and calls
* kvm_mips_handle_exit(), then either resumes the guest or returns to the host
* depending on the return value.
*
* Returns: Next address after end of written function.
*/
void *kvm_mips_build_exit(void *addr)
{
u32 *p = addr;
unsigned int i;
struct uasm_label labels[3];
struct uasm_reloc relocs[3];
struct uasm_label *l = labels;
struct uasm_reloc *r = relocs;
memset(labels, 0, sizeof(labels));
memset(relocs, 0, sizeof(relocs));
/*
* Generic Guest exception handler. We end up here when the guest
* does something that causes a trap to kernel mode.
*
* Both k0/k1 registers will have already been saved (k0 into the vcpu
* structure, and k1 into the scratch_tmp register).
*
* The k1 register will already contain the kvm_vcpu_arch pointer.
*/
/* Start saving Guest context to VCPU */
for (i = 0; i < 32; ++i) {
/* Guest k0/k1 saved later */
if (i == K0 || i == K1)
continue;
UASM_i_SW(&p, i, offsetof(struct kvm_vcpu_arch, gprs[i]), K1);
}
#ifndef CONFIG_CPU_MIPSR6
/* We need to save hi/lo and restore them on the way out */
uasm_i_mfhi(&p, T0);
UASM_i_SW(&p, T0, offsetof(struct kvm_vcpu_arch, hi), K1);
uasm_i_mflo(&p, T0);
UASM_i_SW(&p, T0, offsetof(struct kvm_vcpu_arch, lo), K1);
#endif
/* Finally save guest k1 to VCPU */
uasm_i_ehb(&p);
UASM_i_MFC0(&p, T0, scratch_tmp[0], scratch_tmp[1]);
UASM_i_SW(&p, T0, offsetof(struct kvm_vcpu_arch, gprs[K1]), K1);
/* Now that context has been saved, we can use other registers */
/* Restore vcpu */
UASM_i_MFC0(&p, A1, scratch_vcpu[0], scratch_vcpu[1]);
uasm_i_move(&p, S1, A1);
/* Restore run (vcpu->run) */
UASM_i_LW(&p, A0, offsetof(struct kvm_vcpu, run), A1);
/* Save pointer to run in s0, will be saved by the compiler */
uasm_i_move(&p, S0, A0);
/*
* Save Host level EPC, BadVaddr and Cause to VCPU, useful to process
* the exception
*/
UASM_i_MFC0(&p, K0, C0_EPC);
UASM_i_SW(&p, K0, offsetof(struct kvm_vcpu_arch, pc), K1);
UASM_i_MFC0(&p, K0, C0_BADVADDR);
UASM_i_SW(&p, K0, offsetof(struct kvm_vcpu_arch, host_cp0_badvaddr),
K1);
uasm_i_mfc0(&p, K0, C0_CAUSE);
uasm_i_sw(&p, K0, offsetof(struct kvm_vcpu_arch, host_cp0_cause), K1);
/* Now restore the host state just enough to run the handlers */
/* Switch EBASE to the one used by Linux */
/* load up the host EBASE */
uasm_i_mfc0(&p, V0, C0_STATUS);
uasm_i_lui(&p, AT, ST0_BEV >> 16);
uasm_i_or(&p, K0, V0, AT);
uasm_i_mtc0(&p, K0, C0_STATUS);
uasm_i_ehb(&p);
UASM_i_LA_mostly(&p, K0, (long)&ebase);
UASM_i_LW(&p, K0, uasm_rel_lo((long)&ebase), K0);
build_set_exc_base(&p, K0);
if (raw_cpu_has_fpu) {
/*
* If FPU is enabled, save FCR31 and clear it so that later
* ctc1's don't trigger FPE for pending exceptions.
*/
uasm_i_lui(&p, AT, ST0_CU1 >> 16);
uasm_i_and(&p, V1, V0, AT);
uasm_il_beqz(&p, &r, V1, label_fpu_1);
uasm_i_nop(&p);
uasm_i_cfc1(&p, T0, 31);
uasm_i_sw(&p, T0, offsetof(struct kvm_vcpu_arch, fpu.fcr31),
K1);
uasm_i_ctc1(&p, ZERO, 31);
uasm_l_fpu_1(&l, p);
}
if (cpu_has_msa) {
/*
* If MSA is enabled, save MSACSR and clear it so that later
* instructions don't trigger MSAFPE for pending exceptions.
*/
uasm_i_mfc0(&p, T0, C0_CONFIG5);
uasm_i_ext(&p, T0, T0, 27, 1); /* MIPS_CONF5_MSAEN */
uasm_il_beqz(&p, &r, T0, label_msa_1);
uasm_i_nop(&p);
uasm_i_cfcmsa(&p, T0, MSA_CSR);
uasm_i_sw(&p, T0, offsetof(struct kvm_vcpu_arch, fpu.msacsr),
K1);
uasm_i_ctcmsa(&p, MSA_CSR, ZERO);
uasm_l_msa_1(&l, p);
}
/* Now that the new EBASE has been loaded, unset BEV and KSU_USER */
uasm_i_addiu(&p, AT, ZERO, ~(ST0_EXL | KSU_USER | ST0_IE));
uasm_i_and(&p, V0, V0, AT);
uasm_i_lui(&p, AT, ST0_CU0 >> 16);
uasm_i_or(&p, V0, V0, AT);
uasm_i_mtc0(&p, V0, C0_STATUS);
uasm_i_ehb(&p);
/* Load up host GP */
UASM_i_LW(&p, GP, offsetof(struct kvm_vcpu_arch, host_gp), K1);
/* Need a stack before we can jump to "C" */
UASM_i_LW(&p, SP, offsetof(struct kvm_vcpu_arch, host_stack), K1);
/* Saved host state */
UASM_i_ADDIU(&p, SP, SP, -(int)sizeof(struct pt_regs));
/*
* XXXKYMA do we need to load the host ASID, maybe not because the
* kernel entries are marked GLOBAL, need to verify
*/
/* Restore host scratch registers, as we'll have clobbered them */
kvm_mips_build_restore_scratch(&p, K0, SP);
/* Restore RDHWR access */
UASM_i_LA_mostly(&p, K0, (long)&hwrena);
uasm_i_lw(&p, K0, uasm_rel_lo((long)&hwrena), K0);
uasm_i_mtc0(&p, K0, C0_HWRENA);
/* Jump to handler */
/*
* XXXKYMA: not sure if this is safe, how large is the stack??
* Now jump to the kvm_mips_handle_exit() to see if we can deal
* with this in the kernel
*/
UASM_i_LA(&p, T9, (unsigned long)kvm_mips_handle_exit);
uasm_i_jalr(&p, RA, T9);
UASM_i_ADDIU(&p, SP, SP, -CALLFRAME_SIZ);
uasm_resolve_relocs(relocs, labels);
p = kvm_mips_build_ret_from_exit(p);
return p;
}
static void __cpuinit build_r4000_tlb_refill_handler(void)
{
u32 *p = tlb_handler;
struct uasm_label *l = labels;
struct uasm_reloc *r = relocs;
u32 *f;
unsigned int final_len;
int i;
memset(tlb_handler, 0, sizeof(tlb_handler));
memset(labels, 0, sizeof(labels));
memset(relocs, 0, sizeof(relocs));
memset(final_handler, 0, sizeof(final_handler));
/*
* create the plain linear handler
*/
if (bcm1250_m3_war()) {
UASM_i_MFC0(&p, K0, C0_BADVADDR);
UASM_i_MFC0(&p, K1, C0_ENTRYHI);
uasm_i_xor(&p, K0, K0, K1);
UASM_i_SRL(&p, K0, K0, PAGE_SHIFT + 1);
uasm_il_bnez(&p, &r, K0, label_leave);
/* No need for uasm_i_nop */
}
#ifdef CONFIG_64BIT
build_get_pmde64(&p, &l, &r, K0, K1); /* get pmd in K1 */
#else
build_get_pgde32(&p, K0, K1); /* get pgd in K1 */
#endif
build_get_ptep(&p, K0, K1);
build_update_entries(&p, K0, K1);
build_tlb_write_entry(&p, &l, &r, tlb_random);
uasm_l_leave(&l, p);
uasm_i_eret(&p); /* return from trap */
#ifdef CONFIG_64BIT
build_get_pgd_vmalloc64(&p, &l, &r, K0, K1);
#endif
/*
* Overflow check: For the 64bit handler, we need at least one
* free instruction slot for the wrap-around branch. In worst
* case, if the intended insertion point is a delay slot, we
* need three, with the second nop'ed and the third being
* unused.
*/
/* Loongson2 ebase is different than r4k, we have more space */
#if defined(CONFIG_32BIT) || defined(CONFIG_CPU_LOONGSON2)
if ((p - tlb_handler) > 64)
panic("TLB refill handler space exceeded");
#else
if (((p - tlb_handler) > 63)
|| (((p - tlb_handler) > 61)
&& uasm_insn_has_bdelay(relocs, tlb_handler + 29)))
panic("TLB refill handler space exceeded");
#endif
/*
* Now fold the handler in the TLB refill handler space.
*/
#if defined(CONFIG_32BIT) || defined(CONFIG_CPU_LOONGSON2)
f = final_handler;
/* Simplest case, just copy the handler. */
uasm_copy_handler(relocs, labels, tlb_handler, p, f);
final_len = p - tlb_handler;
#else /* CONFIG_64BIT */
f = final_handler + 32;
if ((p - tlb_handler) <= 32) {
/* Just copy the handler. */
uasm_copy_handler(relocs, labels, tlb_handler, p, f);
final_len = p - tlb_handler;
} else {
u32 *split = tlb_handler + 30;
/*
* Find the split point.
*/
if (uasm_insn_has_bdelay(relocs, split - 1))
split--;
/* Copy first part of the handler. */
uasm_copy_handler(relocs, labels, tlb_handler, split, f);
f += split - tlb_handler;
/* Insert branch. */
uasm_l_split(&l, final_handler);
uasm_il_b(&f, &r, label_split);
if (uasm_insn_has_bdelay(relocs, split))
uasm_i_nop(&f);
else {
uasm_copy_handler(relocs, labels, split, split + 1, f);
uasm_move_labels(labels, f, f + 1, -1);
f++;
split++;
}
/* Copy the rest of the handler. */
uasm_copy_handler(relocs, labels, split, p, final_handler);
final_len = (f - (final_handler + 32)) + (p - split);
}
#endif /* CONFIG_64BIT */
uasm_resolve_relocs(relocs, labels);
pr_debug("Wrote TLB refill handler (%u instructions).\n",
final_len);
f = final_handler;
#if defined(CONFIG_64BIT) && !defined(CONFIG_CPU_LOONGSON2)
if (final_len > 32)
final_len = 64;
else
f = final_handler + 32;
#endif /* CONFIG_64BIT */
pr_debug("\t.set push\n");
pr_debug("\t.set noreorder\n");
for (i = 0; i < final_len; i++)
pr_debug("\t.word 0x%08x\n", f[i]);
pr_debug("\t.set pop\n");
memcpy((void *)ebase, final_handler, 0x100);
}
