linux_dsm_epyc7002/arch/sparc64/kernel/ktlb.S
David S. Miller c9c1083074 [SPARC64]: Fix boot failures on SunBlade-150
The sequence to move over to the Linux trap tables from
the firmware ones needs to be more air tight.  It turns
out that to be %100 safe we do need to be able to translate
OBP mappings in our TLB miss handlers early.

In order not to eat up a lot of kernel image memory with
static page tables, just use the translations array in
the OBP TLB miss handlers.  That solves the bulk of the
problem.

Furthermore, to make sure the OBP TLB miss path will work
even before the fixed MMU globals are loaded, explicitly
load %g1 to TLB_SFSR at the beginning of the i-TLB and
d-TLB miss handlers.

To ease the OBP TLB miss walking of the prom_trans[] array,
we sort it then delete all of the non-OBP entries in there
(for example, there are entries for the kernel image itself
which we're not interested in at all).

We also save about 32K of kernel image size with this change.
Not a bad side effect :-)

There are still some reasons why trampoline.S can't use the
setup_trap_table() yet.  The most noteworthy are:

1) OBP boots secondary processors with non-bias'd stack for
   some reason.  This is easily fixed by using a small bootup
   stack in the kernel image explicitly for this purpose.

2) Doing a firmware call via the normal C call prom_set_trap_table()
   goes through the whole OBP enter/exit sequence that saves and
   restores OBP and Linux kernel state in the MMUs.  This path
   unfortunately does a "flush %g6" while loading up the OBP locked
   TLB entries for the firmware call.

   If we setup the %g6 in the trampoline.S code properly, that
   is in the PAGE_OFFSET linear mapping, but we're not on the
   kernel trap table yet so those addresses won't translate properly.

   One idea is to do a by-hand firmware call like we do in the
   early bootup code and elsewhere here in trampoline.S  But this
   fails as well, as aparently the secondary processors are not
   booted with OBP's special locked TLB entries loaded.  These
   are necessary for the firwmare to processes TLB misses correctly
   up until the point where we take over the trap table.

This does need to be resolved at some point.

Signed-off-by: David S. Miller <davem@davemloft.net>
2005-10-12 12:22:46 -07:00

195 lines
5.1 KiB
ArmAsm

/* arch/sparc64/kernel/ktlb.S: Kernel mapping TLB miss handling.
*
* Copyright (C) 1995, 1997, 2005 David S. Miller <davem@davemloft.net>
* Copyright (C) 1996 Eddie C. Dost (ecd@brainaid.de)
* Copyright (C) 1996 Miguel de Icaza (miguel@nuclecu.unam.mx)
* Copyright (C) 1996,98,99 Jakub Jelinek (jj@sunsite.mff.cuni.cz)
*/
#include <linux/config.h>
#include <asm/head.h>
#include <asm/asi.h>
#include <asm/page.h>
#include <asm/pgtable.h>
.text
.align 32
/*
* On a second level vpte miss, check whether the original fault is to the OBP
* range (note that this is only possible for instruction miss, data misses to
* obp range do not use vpte). If so, go back directly to the faulting address.
* This is because we want to read the tpc, otherwise we have no way of knowing
* the 8k aligned faulting address if we are using >8k kernel pagesize. This
* also ensures no vpte range addresses are dropped into tlb while obp is
* executing (see inherit_locked_prom_mappings() rant).
*/
sparc64_vpte_nucleus:
/* Note that kvmap below has verified that the address is
* in the range MODULES_VADDR --> VMALLOC_END already. So
* here we need only check if it is an OBP address or not.
*/
sethi %hi(LOW_OBP_ADDRESS), %g5
cmp %g4, %g5
blu,pn %xcc, kern_vpte
mov 0x1, %g5
sllx %g5, 32, %g5
cmp %g4, %g5
blu,pn %xcc, vpte_insn_obp
nop
/* These two instructions are patched by paginig_init(). */
kern_vpte:
sethi %hi(swapper_pgd_zero), %g5
lduw [%g5 + %lo(swapper_pgd_zero)], %g5
/* With kernel PGD in %g5, branch back into dtlb_backend. */
ba,pt %xcc, sparc64_kpte_continue
andn %g1, 0x3, %g1 /* Finish PMD offset adjustment. */
vpte_noent:
/* Restore previous TAG_ACCESS, %g5 is zero, and we will
* skip over the trap instruction so that the top level
* TLB miss handler will thing this %g5 value is just an
* invalid PTE, thus branching to full fault processing.
*/
mov TLB_SFSR, %g1
stxa %g4, [%g1 + %g1] ASI_DMMU
done
vpte_insn_obp:
/* Behave as if we are at TL0. */
wrpr %g0, 1, %tl
rdpr %tpc, %g4 /* Find original faulting iaddr */
srlx %g4, 13, %g4 /* Throw out context bits */
sllx %g4, 13, %g4 /* g4 has vpn + ctx0 now */
/* Restore previous TAG_ACCESS. */
mov TLB_SFSR, %g1
stxa %g4, [%g1 + %g1] ASI_IMMU
sethi %hi(prom_trans), %g5
or %g5, %lo(prom_trans), %g5
1: ldx [%g5 + 0x00], %g6 ! base
brz,a,pn %g6, longpath ! no more entries, fail
mov TLB_SFSR, %g1 ! and restore %g1
ldx [%g5 + 0x08], %g1 ! len
add %g6, %g1, %g1 ! end
cmp %g6, %g4
bgu,pt %xcc, 2f
cmp %g4, %g1
bgeu,pt %xcc, 2f
ldx [%g5 + 0x10], %g1 ! PTE
/* TLB load, restore %g1, and return from trap. */
sub %g4, %g6, %g6
add %g1, %g6, %g5
mov TLB_SFSR, %g1
stxa %g5, [%g0] ASI_ITLB_DATA_IN
retry
2: ba,pt %xcc, 1b
add %g5, (3 * 8), %g5 ! next entry
kvmap_do_obp:
sethi %hi(prom_trans), %g5
or %g5, %lo(prom_trans), %g5
srlx %g4, 13, %g4
sllx %g4, 13, %g4
1: ldx [%g5 + 0x00], %g6 ! base
brz,a,pn %g6, longpath ! no more entries, fail
mov TLB_SFSR, %g1 ! and restore %g1
ldx [%g5 + 0x08], %g1 ! len
add %g6, %g1, %g1 ! end
cmp %g6, %g4
bgu,pt %xcc, 2f
cmp %g4, %g1
bgeu,pt %xcc, 2f
ldx [%g5 + 0x10], %g1 ! PTE
/* TLB load, restore %g1, and return from trap. */
sub %g4, %g6, %g6
add %g1, %g6, %g5
mov TLB_SFSR, %g1
stxa %g5, [%g0] ASI_DTLB_DATA_IN
retry
2: ba,pt %xcc, 1b
add %g5, (3 * 8), %g5 ! next entry
/*
* On a first level data miss, check whether this is to the OBP range (note
* that such accesses can be made by prom, as well as by kernel using
* prom_getproperty on "address"), and if so, do not use vpte access ...
* rather, use information saved during inherit_prom_mappings() using 8k
* pagesize.
*/
.align 32
kvmap:
brgez,pn %g4, kvmap_nonlinear
nop
#ifdef CONFIG_DEBUG_PAGEALLOC
.globl kvmap_linear_patch
kvmap_linear_patch:
#endif
ba,pt %xcc, kvmap_load
xor %g2, %g4, %g5
#ifdef CONFIG_DEBUG_PAGEALLOC
sethi %hi(swapper_pg_dir), %g5
or %g5, %lo(swapper_pg_dir), %g5
sllx %g4, 64 - (PGDIR_SHIFT + PGDIR_BITS), %g6
srlx %g6, 64 - PAGE_SHIFT, %g6
andn %g6, 0x3, %g6
lduw [%g5 + %g6], %g5
brz,pn %g5, longpath
sllx %g4, 64 - (PMD_SHIFT + PMD_BITS), %g6
srlx %g6, 64 - PAGE_SHIFT, %g6
sllx %g5, 11, %g5
andn %g6, 0x3, %g6
lduwa [%g5 + %g6] ASI_PHYS_USE_EC, %g5
brz,pn %g5, longpath
sllx %g4, 64 - PMD_SHIFT, %g6
srlx %g6, 64 - PAGE_SHIFT, %g6
sllx %g5, 11, %g5
andn %g6, 0x7, %g6
ldxa [%g5 + %g6] ASI_PHYS_USE_EC, %g5
brz,pn %g5, longpath
nop
ba,a,pt %xcc, kvmap_load
#endif
kvmap_nonlinear:
sethi %hi(MODULES_VADDR), %g5
cmp %g4, %g5
blu,pn %xcc, longpath
mov (VMALLOC_END >> 24), %g5
sllx %g5, 24, %g5
cmp %g4, %g5
bgeu,pn %xcc, longpath
nop
kvmap_check_obp:
sethi %hi(LOW_OBP_ADDRESS), %g5
cmp %g4, %g5
blu,pn %xcc, kvmap_vmalloc_addr
mov 0x1, %g5
sllx %g5, 32, %g5
cmp %g4, %g5
blu,pn %xcc, kvmap_do_obp
nop
kvmap_vmalloc_addr:
/* If we get here, a vmalloc addr was accessed, load kernel VPTE. */
ldxa [%g3 + %g6] ASI_N, %g5
brgez,pn %g5, longpath
nop
kvmap_load:
/* PTE is valid, load into TLB and return from trap. */
stxa %g5, [%g0] ASI_DTLB_DATA_IN ! Reload TLB
retry