/**
* 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_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);
}
static void * __init cps_gen_entry_code(unsigned cpu, enum cps_pm_state state)
{
struct uasm_label *l = labels;
struct uasm_reloc *r = relocs;
u32 *buf, *p;
const unsigned r_online = a0;
const unsigned r_nc_count = a1;
const unsigned r_pcohctl = t7;
const unsigned max_instrs = 256;
unsigned cpc_cmd;
enum {
lbl_incready = 1,
lbl_poll_cont,
lbl_secondary_hang,
lbl_disable_coherence,
lbl_flush_fsb,
lbl_invicache,
lbl_flushdcache,
lbl_hang,
lbl_set_cont,
lbl_secondary_cont,
lbl_decready,
};
/* Allocate a buffer to hold the generated code */
p = buf = kcalloc(max_instrs, sizeof(u32), GFP_KERNEL);
if (!buf)
return NULL;
/* Clear labels & relocs ready for (re)use */
memset(labels, 0, sizeof(labels));
memset(relocs, 0, sizeof(relocs));
if (state == CPS_PM_POWER_GATED) {
/* Power gating relies upon CPS SMP */
if (!mips_cps_smp_in_use())
goto out_err;
/*
* Save CPU state. Note the non-standard calling convention
* with the return address placed in v0 to avoid clobbering
* the ra register before it is saved.
*/
UASM_i_LA(&p, t0, (long)mips_cps_pm_save);
uasm_i_jalr(&p, v0, t0);
uasm_i_nop(&p);
}
/*
* Load addresses of required CM & CPC registers. This is done early
* because they're needed in both the enable & disable coherence steps
* but in the coupled case the enable step will only run on one VPE.
*/
UASM_i_LA(&p, r_pcohctl, (long)_gcmp_base + GCMPCLCBOFS(COHCTL));
if (coupled_coherence) {
/* Increment ready_count */
uasm_i_sync(&p, stype_ordering);
uasm_build_label(&l, p, lbl_incready);
uasm_i_ll(&p, t1, 0, r_nc_count);
uasm_i_addiu(&p, t2, t1, 1);
uasm_i_sc(&p, t2, 0, r_nc_count);
uasm_il_beqz(&p, &r, t2, lbl_incready);
uasm_i_addiu(&p, t1, t1, 1);
/* Ordering barrier */
uasm_i_sync(&p, stype_ordering);
/*
* If this is the last VPE to become ready for non-coherence
* then it should branch below.
*/
uasm_il_beq(&p, &r, t1, r_online, lbl_disable_coherence);
uasm_i_nop(&p);
if (state < CPS_PM_POWER_GATED) {
/*
* Otherwise this is not the last VPE to become ready
* for non-coherence. It needs to wait until coherence
* has been disabled before proceeding, which it will do
* by polling for the top bit of ready_count being set.
*/
uasm_i_addiu(&p, t1, zero, -1);
uasm_build_label(&l, p, lbl_poll_cont);
uasm_i_lw(&p, t0, 0, r_nc_count);
uasm_il_bltz(&p, &r, t0, lbl_secondary_cont);
uasm_i_ehb(&p);
uasm_i_yield(&p, zero, t1);
uasm_il_b(&p, &r, lbl_poll_cont);
uasm_i_nop(&p);
} else {
/*
* The core will lose power & this VPE will not continue
* so it can simply halt here.
*/
uasm_i_addiu(&p, t0, zero, TCHALT_H);
uasm_i_mtc0(&p, t0, 2, 4);
uasm_build_label(&l, p, lbl_secondary_hang);
uasm_il_b(&p, &r, lbl_secondary_hang);
uasm_i_nop(&p);
}
}
/*
* This is the point of no return - this VPE will now proceed to
* disable coherence. At this point we *must* be sure that no other
* VPE within the core will interfere with the L1 dcache.
*/
uasm_build_label(&l, p, lbl_disable_coherence);
/* Invalidate the L1 icache */
cps_gen_cache_routine(&p, &l, &r, &cpu_data[cpu].icache,
Index_Invalidate_I, lbl_invicache);
/* Writeback & invalidate the L1 dcache */
cps_gen_cache_routine(&p, &l, &r, &cpu_data[cpu].dcache,
Index_Writeback_Inv_D, lbl_flushdcache);
/* Completion barrier */
uasm_i_sync(&p, stype_memory);
uasm_i_ehb(&p);
/*
* Disable all but self interventions. The load from COHCTL is defined
* by the interAptiv & proAptiv SUMs as ensuring that the operation
* resulting from the preceeding store is complete.
*/
uasm_i_addiu(&p, t0, zero, 1 << cpu_data[cpu].core);
uasm_i_sw(&p, t0, 0, r_pcohctl);
uasm_i_lw(&p, t0, 0, r_pcohctl);
/* Sync to ensure previous interventions are complete */
uasm_i_sync(&p, stype_intervention);
uasm_i_ehb(&p);
/* Disable coherence */
uasm_i_sw(&p, zero, 0, r_pcohctl);
uasm_i_lw(&p, t0, 0, r_pcohctl);
if (state >= CPS_PM_CLOCK_GATED) {
/* TODO: determine whether required based on CPC version */
cps_gen_flush_fsb(&p, &l, &r, &cpu_data[cpu].dcache,
lbl_flush_fsb);
/* Determine the CPC command to issue */
switch (state) {
case CPS_PM_CLOCK_GATED:
cpc_cmd = CPC_Cx_CMD_CLOCKOFF;
break;
case CPS_PM_POWER_GATED:
cpc_cmd = CPC_Cx_CMD_PWRDOWN;
break;
default:
BUG();
goto out_err;
}
/* Issue the CPC command */
UASM_i_LA(&p, t0, (long)addr_cpc_cl_cmd());
uasm_i_addiu(&p, t1, zero, cpc_cmd);
uasm_i_sw(&p, t1, 0, t0);
if (state == CPS_PM_POWER_GATED) {
/* If anything goes wrong just hang */
uasm_build_label(&l, p, lbl_hang);
uasm_il_b(&p, &r, lbl_hang);
uasm_i_nop(&p);
/*
* There's no point generating more code, the core is
* powered down & if powered back up will run from the
* reset vector not from here.
*/
goto gen_done;
}
/* Completion barrier */
uasm_i_sync(&p, stype_memory);
uasm_i_ehb(&p);
}
if (state == CPS_PM_NC_WAIT) {
/*
* At this point it is safe for all VPEs to proceed with
* execution. This VPE will set the top bit of ready_count
* to indicate to the other VPEs that they may continue.
*/
if (coupled_coherence)
cps_gen_set_top_bit(&p, &l, &r, r_nc_count,
lbl_set_cont);
/*
* VPEs which did not disable coherence will continue
* executing, after coherence has been disabled, from this
* point.
*/
uasm_build_label(&l, p, lbl_secondary_cont);
/* Now perform our wait */
uasm_i_wait(&p, 0);
}
/*
* Re-enable coherence. Note that for CPS_PM_NC_WAIT all coupled VPEs
* will run this. The first will actually re-enable coherence & the
* rest will just be performing a rather unusual nop.
*/
uasm_i_addiu(&p, t0, zero, GCMP_CCB_COHCTL_DOMAIN_MSK);
uasm_i_sw(&p, t0, 0, r_pcohctl);
uasm_i_lw(&p, t0, 0, r_pcohctl);
/* Completion barrier */
uasm_i_sync(&p, stype_memory);
uasm_i_ehb(&p);
if (coupled_coherence && (state == CPS_PM_NC_WAIT)) {
/* Decrement ready_count */
uasm_build_label(&l, p, lbl_decready);
uasm_i_sync(&p, stype_ordering);
uasm_i_ll(&p, t1, 0, r_nc_count);
uasm_i_addiu(&p, t2, t1, -1);
uasm_i_sc(&p, t2, 0, r_nc_count);
uasm_il_beqz(&p, &r, t2, lbl_decready);
uasm_i_andi(&p, v0, t1, (1 << fls(smp_num_siblings)) - 1);
/* Ordering barrier */
uasm_i_sync(&p, stype_ordering);
}
if (coupled_coherence && (state == CPS_PM_CLOCK_GATED)) {
/*
* At this point it is safe for all VPEs to proceed with
* execution. This VPE will set the top bit of ready_count
* to indicate to the other VPEs that they may continue.
*/
cps_gen_set_top_bit(&p, &l, &r, r_nc_count, lbl_set_cont);
/*
* This core will be reliant upon another core sending a
* power-up command to the CPC in order to resume operation.
* Thus an arbitrary VPE can't trigger the core leaving the
* idle state and the one that disables coherence might as well
* be the one to re-enable it. The rest will continue from here
* after that has been done.
*/
uasm_build_label(&l, p, lbl_secondary_cont);
/* Ordering barrier */
uasm_i_sync(&p, stype_ordering);
}
/* The core is coherent, time to return to C code */
uasm_i_jr(&p, ra);
uasm_i_nop(&p);
gen_done:
/* Ensure the code didn't exceed the resources allocated for it */
BUG_ON((p - buf) > max_instrs);
BUG_ON((l - labels) > ARRAY_SIZE(labels));
BUG_ON((r - relocs) > ARRAY_SIZE(relocs));
/* Patch branch offsets */
uasm_resolve_relocs(relocs, labels);
/* Flush the icache */
local_flush_icache_range((unsigned long)buf, (unsigned long)p);
return buf;
out_err:
kfree(buf);
return NULL;
}
/*
* BVADDR is the faulting address, PTR is scratch.
* PTR will hold the pgd for vmalloc.
*/
static __init void
build_get_pgd_vmalloc64(u32 **p, struct uasm_label **l, struct uasm_reloc **r,
unsigned int bvaddr, unsigned int ptr)
{
long swpd = (long)swapper_pg_dir;
#ifdef MODULE_START
long modd = (long)module_pg_dir;
uasm_l_module_alloc(l, *p);
/*
* Assumption:
* VMALLOC_START >= 0xc000000000000000UL
* MODULE_START >= 0xe000000000000000UL
*/
UASM_i_SLL(p, ptr, bvaddr, 2);
uasm_il_bgez(p, r, ptr, label_vmalloc);
if (uasm_in_compat_space_p(MODULE_START) &&
!uasm_rel_lo(MODULE_START)) {
uasm_i_lui(p, ptr, uasm_rel_hi(MODULE_START)); /* delay slot */
} else {
/* unlikely configuration */
uasm_i_nop(p); /* delay slot */
UASM_i_LA(p, ptr, MODULE_START);
}
uasm_i_dsubu(p, bvaddr, bvaddr, ptr);
if (uasm_in_compat_space_p(modd) && !uasm_rel_lo(modd)) {
uasm_il_b(p, r, label_vmalloc_done);
uasm_i_lui(p, ptr, uasm_rel_hi(modd));
} else {
UASM_i_LA_mostly(p, ptr, modd);
uasm_il_b(p, r, label_vmalloc_done);
if (uasm_in_compat_space_p(modd))
uasm_i_addiu(p, ptr, ptr, uasm_rel_lo(modd));
else
uasm_i_daddiu(p, ptr, ptr, uasm_rel_lo(modd));
}
uasm_l_vmalloc(l, *p);
if (uasm_in_compat_space_p(MODULE_START) &&
!uasm_rel_lo(MODULE_START) &&
MODULE_START << 32 == VMALLOC_START)
uasm_i_dsll32(p, ptr, ptr, 0); /* typical case */
else
UASM_i_LA(p, ptr, VMALLOC_START);
#else
uasm_l_vmalloc(l, *p);
UASM_i_LA(p, ptr, VMALLOC_START);
#endif
uasm_i_dsubu(p, bvaddr, bvaddr, ptr);
if (uasm_in_compat_space_p(swpd) && !uasm_rel_lo(swpd)) {
uasm_il_b(p, r, label_vmalloc_done);
uasm_i_lui(p, ptr, uasm_rel_hi(swpd));
} else {
UASM_i_LA_mostly(p, ptr, swpd);
uasm_il_b(p, r, label_vmalloc_done);
if (uasm_in_compat_space_p(swpd))
uasm_i_addiu(p, ptr, ptr, uasm_rel_lo(swpd));
else
uasm_i_daddiu(p, ptr, ptr, uasm_rel_lo(swpd));
}
}