[U-Boot] [PATCH 12/16] efi_loader: Add DCACHE_OFF support for arm64
Mark Rutland
mark.rutland at arm.com
Tue Feb 2 16:55:17 CET 2016
On Tue, Feb 02, 2016 at 03:45:10AM +0100, Alexander Graf wrote:
> On arm64, boards can declare that they want to run with dcache disabled.
>
> However, uEFI guarantees to payloads that they're running with the dcache
> enabled which on arm64 means that they can do unaligned accesses.
>
> To not leave those systems out of the door, let's handle the unaligned traps.
> In the typical boot case, the OS will set up page tables and dcache itself
> early on anyway once it's done talking with uEFI.
This is not sufficient to emulate having caches enabled.
There are other things which operate differently with the caches on
(e.g. exclusives and/or atomics, which a compiler might generate
implicitly).
Likewise, cache-maintenance by Va (which you may require from the
I-side) implicitly hazards against cacheable accesses, but not against
non-cacheable accesses.
There are almsot certainly other differences.
Due to that, I don't think this is a good approach.
Why can we not map memory using cacheable attributes in all cases?
Mark.
> Signed-off-by: Alexander Graf <agraf at suse.de>
> ---
> arch/arm/cpu/armv8/exceptions.S | 10 ++
> arch/arm/lib/Makefile | 3 +
> arch/arm/lib/interrupts_64.c | 19 +++
> arch/arm/lib/unaligned_64.c | 284 ++++++++++++++++++++++++++++++++++++++++
> cmd/bootefi.c | 5 +
> 5 files changed, 321 insertions(+)
> create mode 100644 arch/arm/lib/unaligned_64.c
>
> diff --git a/arch/arm/cpu/armv8/exceptions.S b/arch/arm/cpu/armv8/exceptions.S
> index 4f4f526..97101c3 100644
> --- a/arch/arm/cpu/armv8/exceptions.S
> +++ b/arch/arm/cpu/armv8/exceptions.S
> @@ -144,3 +144,13 @@ exception_exit:
> ldp x27, x28, [sp],#16
> ldp x29, x30, [sp],#16
> eret
> +
> +.global read_far
> +read_far:
> + switch_el x1, 3f, 2f, 1f
> +3: mrs x0, far_el3
> + ret
> +2: mrs x0, far_el2
> + ret
> +1: mrs x0, far_el1
> + ret
> diff --git a/arch/arm/lib/Makefile b/arch/arm/lib/Makefile
> index f3db7b5..ce5ed99 100644
> --- a/arch/arm/lib/Makefile
> +++ b/arch/arm/lib/Makefile
> @@ -42,6 +42,9 @@ else ifdef CONFIG_ARM64
> obj-y += ccn504.o
> obj-y += gic_64.o
> obj-y += interrupts_64.o
> +ifeq ($(CONFIG_SYS_DCACHE_OFF),y)
> +obj-$(CONFIG_EFI_LOADER) += unaligned_64.o
> +endif
> else
> obj-y += interrupts.o
> endif
> diff --git a/arch/arm/lib/interrupts_64.c b/arch/arm/lib/interrupts_64.c
> index 7c9cfce..4aa36de 100644
> --- a/arch/arm/lib/interrupts_64.c
> +++ b/arch/arm/lib/interrupts_64.c
> @@ -81,12 +81,31 @@ void do_bad_error(struct pt_regs *pt_regs, unsigned int esr)
> panic("Resetting CPU ...\n");
> }
>
> +#if defined(CONFIG_EFI_LOADER) && defined(CONFIG_SYS_DCACHE_OFF)
> +int do_unaligned_data(struct pt_regs *pt_regs, unsigned int esr);
> +#else
> +static int do_unaligned_data(struct pt_regs *pt_regs, unsigned int esr)
> +{
> + return -1;
> +}
> +#endif
> +
> /*
> * do_sync handles the Synchronous Abort exception.
> */
> void do_sync(struct pt_regs *pt_regs, unsigned int esr)
> {
> efi_restore_gd();
> +
> + /*
> + * EFI guarantees that unaligned accesses do succeed, so while we
> + * still need hardware access and thus are unsure whether we can
> + * enable the dcache to have the CPU deal with them, we fix unaligned
> + * accesses up ourselves.
> + */
> + if (!do_unaligned_data(pt_regs, esr))
> + return;
> +
> printf("\"Synchronous Abort\" handler, esr 0x%08x\n", esr);
> show_regs(pt_regs);
> panic("Resetting CPU ...\n");
> diff --git a/arch/arm/lib/unaligned_64.c b/arch/arm/lib/unaligned_64.c
> new file mode 100644
> index 0000000..b307b7e
> --- /dev/null
> +++ b/arch/arm/lib/unaligned_64.c
> @@ -0,0 +1,284 @@
> +/*
> + * (C) Copyright 2016
> + * Alexander Graf <agraf at suse.de>
> + *
> + * SPDX-License-Identifier: GPL-2.0+
> + */
> +
> +#include <common.h>
> +#include <linux/compiler.h>
> +#include <efi_loader.h>
> +
> +#define ESR_EC_MASK 0xFC000000
> +#define ESR_EC_SHIFT 26
> +#define ESR_IL_MASK 0x02000000
> +#define ESR_IL_SHIFT 25
> +#define ESR_ISS_MASK 0x01FFFFFF
> +#define ESR_ISS_SHIFT 0
> +
> +#define EC_DATA_SL 0x25
> +
> +#define ISS_ISV_MASK 0x01000000
> +#define ISS_ISV_SHIFT 24
> +#define ISS_SAS_MASK 0x00C00000
> +#define ISS_SAS_SHIFT 22
> +#define ISS_SSE_MASK 0x00200000
> +#define ISS_SSE_SHIFT 21
> +#define ISS_SRT_MASK 0x000F0000
> +#define ISS_SRT_SHIFT 16
> +#define ISS_SF_MASK 0x00008000
> +#define ISS_SF_SHIFT 15
> +#define ISS_AR_MASK 0x00004000
> +#define ISS_AR_SHIFT 14
> +#define ISS_EA_MASK 0x00000200
> +#define ISS_EA_SHIFT 9
> +#define ISS_CM_MASK 0x00000100
> +#define ISS_CM_SHIFT 8
> +#define ISS_S1PTW_MASK 0x00000080
> +#define ISS_S1PTW_SHIFT 7
> +#define ISS_WNR_MASK 0x00000040
> +#define ISS_WNR_SHIFT 6
> +#define WNR_READ 0
> +#define WNR_WRITE 1
> +#define ISS_DFSC_MASK 0x0000003F
> +#define ISS_DFSC_SHIFT 0
> +
> +#define ISV_VALID 1
> +#define DFSC_ALIGN 0x21
> +
> +ulong read_far(void);
> +int do_unaligned_data(struct pt_regs *pt_regs, unsigned int esr);
> +
> +static inline int32_t sextract32(uint32_t value, int start, int length)
> +{
> + return ((int32_t)(value << (32 - length - start))) >> (32 - length);
> +}
> +
> +static inline uint32_t extract32(uint32_t value, int start, int length)
> +{
> + return (value >> start) & (~0U >> (32 - length));
> +}
> +
> +static int insn_iss_ldst(uint32_t insn, int iss, int *wb_reg, ulong *wb_val)
> +{
> + int rt = extract32(insn, 0, 5);
> + int rn = extract32(insn, 5, 5);
> + int idx = extract32(insn, 10, 2);
> + int imm9 = sextract32(insn, 12, 9);
> + int opc = extract32(insn, 22, 2);
> + int size = extract32(insn, 30, 2);
> + bool is_signed = false;
> + bool is_store = false;
> + bool is_extended = false;
> + bool is_vector = extract32(insn, 26, 1);
> +
> + switch (extract32(insn, 25, 4)) {
> + case 0x4:
> + case 0x6:
> + case 0xc:
> + case 0xe: /* Loads and stores */
> + break;
> + default:
> + return iss;
> + }
> +
> + switch (extract32(insn, 24, 6)) {
> + case 0x38: case 0x39:
> + case 0x3c: case 0x3d: /* Load/store register (all forms) */
> + break;
> + default:
> + return iss;
> + }
> +
> + switch (extract32(insn, 24, 2)) {
> + case 0:
> + if (extract32(insn, 21, 1) != 1 &&
> + extract32(insn, 10, 2) != 2) {
> + /* Write back */
> + if (idx & 1) {
> + ulong far = read_far();
> + *wb_reg = rn;
> + *wb_val = (idx & 2) ? far : far + imm9;
> + break;
> + }
> + }
> + break;
> + case 1:
> + break;
> + default:
> + return iss;
> + }
> +
> + if (is_vector) {
> + return iss;
> + }
> +
> + is_store = (opc == 0);
> + is_signed = extract32(opc, 1, 1);
> + is_extended = (size < 3) && extract32(opc, 0, 1);
> +
> + iss |= ISS_ISV_MASK;
> + iss |= size << ISS_SAS_SHIFT;
> + iss |= (is_extended && is_signed) ? ISS_SSE_MASK : 0;
> + iss |= rt << ISS_SRT_SHIFT;
> + iss |= ISS_SF_MASK;
> + iss |= is_store ? ISS_WNR_MASK : 0;
> +
> + return iss;
> +}
> +
> +static void do_unaligned_access(struct pt_regs *pt_regs, int wnr, int rt,
> + ulong addr, int sas, int sse)
> +{
> + void *rr = (void*)&pt_regs->regs[rt];
> + void *r = rr;
> + int len = 1 << sas;
> +
> +#if __BYTE_ORDER == __BIG_ENDIAN
> + /* On BE registers get filled from the back */
> + r = (char*)r + (8 - len);
> +#endif
> +
> + if (wnr == WNR_READ) {
> + /* Read with zero pad */
> + pt_regs->regs[rt] = 0;
> + memcpy(r, (void*)addr, len);
> +
> + /* Sign extend */
> + if (sse) {
> + switch (sas) {
> + case 0: *(long*)rr = *(char*)r; break;
> + case 1: *(long*)rr = *(short*)r; break;
> + case 2: *(long*)rr = *(int*)r; break;
> + }
> + }
> + } else if (wnr == WNR_WRITE) {
> + memcpy((void*)addr, r, len);
> + }
> +}
> +
> +static int do_ldst_pair(u32 insn, struct pt_regs *pt_regs)
> +{
> + int rt = extract32(insn, 0, 5);
> + int rn = extract32(insn, 5, 5);
> + int rt2 = extract32(insn, 10, 5);
> + int index = extract32(insn, 23, 2);
> + bool is_vector = extract32(insn, 26, 1);
> + bool is_load = extract32(insn, 22, 1);
> + int opc = extract32(insn, 30, 2);
> + ulong far = read_far();
> + ulong old_far = far;
> + int wnr = is_load ? WNR_READ : WNR_WRITE;
> +
> + bool is_signed = extract32(opc, 0, 1);
> + bool postindex = false;
> + bool wback = false;
> +
> + int size = 2 + extract32(opc, 1, 1);
> +
> + switch (extract32(insn, 25, 4)) {
> + case 0x4:
> + case 0x6:
> + case 0xc:
> + case 0xe: /* Loads and stores */
> + break;
> + default:
> + return -1;
> + }
> +
> + switch (extract32(insn, 24, 6)) {
> + case 0x28: case 0x29:
> + case 0x2c: case 0x2d: /* Load/store pair (all forms) */
> + break;
> + default:
> + return -1;
> + }
> +
> + if (is_vector)
> + return -1;
> +
> + switch (index) {
> + case 1: /* post-index */
> + postindex = true;
> + wback = true;
> + break;
> + case 0: /* signed offset with "non-temporal" hint. */
> + postindex = false;
> + break;
> + case 2: /* signed offset, rn not updated */
> + postindex = false;
> + break;
> + case 3: /* pre-index */
> + postindex = false;
> + wback = true;
> + break;
> + }
> +
> + do_unaligned_access(pt_regs, wnr, rt, far, size, is_signed);
> + far += 1 << size;
> + do_unaligned_access(pt_regs, wnr, rt2, far, size, is_signed);
> + far += 1 << size;
> +
> + if (wback)
> + pt_regs->regs[rn] = postindex ? far : old_far;
> +
> + return 0;
> +}
> +
> +int do_unaligned_data(struct pt_regs *pt_regs, unsigned int esr)
> +{
> + int insn = *(u32*)(void*)pt_regs->elr;
> + int ec = (esr & ESR_EC_MASK) >> ESR_EC_SHIFT;
> + int iss = (esr & ESR_ISS_MASK) >> ESR_ISS_SHIFT;
> + int isv = (iss & ISS_ISV_MASK) >> ISS_ISV_SHIFT;
> + int dfsc = (iss & ISS_DFSC_MASK) >> ISS_DFSC_SHIFT;
> + int sas, sse, srt, sf, ar, cm, s1ptw, wnr;
> + int wb_reg = -1;
> + ulong wb_val;
> +
> + /* Check whether we have an alignment fault */
> + if ((ec != EC_DATA_SL) || (dfsc != DFSC_ALIGN))
> + return -1;
> +
> + /* Fix up instruction decoding */
> + if (!isv) {
> + iss = insn_iss_ldst(insn, iss, &wb_reg, &wb_val);
> + }
> +
> + isv = (iss & ISS_ISV_MASK) >> ISS_ISV_SHIFT;
> + sas = (iss & ISS_SAS_MASK) >> ISS_SAS_SHIFT;
> + sse = (iss & ISS_SSE_MASK) >> ISS_SSE_SHIFT;
> + srt = (iss & ISS_SRT_MASK) >> ISS_SRT_SHIFT;
> + sf = (iss & ISS_SF_MASK) >> ISS_SF_SHIFT;
> + ar = (iss & ISS_AR_MASK) >> ISS_AR_SHIFT;
> + cm = (iss & ISS_CM_MASK) >> ISS_CM_SHIFT;
> + s1ptw = (iss & ISS_S1PTW_MASK) >> ISS_S1PTW_SHIFT;
> + wnr = (iss & ISS_WNR_MASK) >> ISS_WNR_SHIFT;
> +
> + /* Check whether we have an easily fixable alignment fault */
> + if (isv && sf && !ar && !cm && !s1ptw) {
> + ulong far = read_far();
> +
> + do_unaligned_access(pt_regs, wnr, srt, far, sas, sse);
> +
> + /* Jump across the offending instruction */
> + pt_regs->elr += 4;
> +
> + /* Do writebacks if required */
> + if (wb_reg != -1)
> + pt_regs->regs[wb_reg] = wb_val;
> +
> + /* And return from the exception */
> + return 0;
> + }
> +
> + if (!do_ldst_pair(insn, pt_regs)) {
> + pt_regs->elr += 4;
> + return 0;
> + }
> +
> + /* Couldn't fix it, panic */
> + printf("Alignment handler couldn't decode insn %08x\n",
> + *(u32*)(void*)pt_regs->elr);
> + return -1;
> +}
> diff --git a/cmd/bootefi.c b/cmd/bootefi.c
> index e3e51d4..5968ecc 100644
> --- a/cmd/bootefi.c
> +++ b/cmd/bootefi.c
> @@ -108,6 +108,11 @@ static unsigned long do_bootefi_exec(void *efi)
> systab.nr_tables = 0;
> }
>
> +#ifdef CONFIG_SYS_DCACHE_OFF
> + printf("WARNING: You're running with MMU disabled, expect crashes\n"
> + " and send bug reports to your firmware vendor!\n");
> +#endif
> +
> /* Load the EFI payload */
> entry = efi_load_pe(efi, &loaded_image_info);
> if (!entry)
> --
> 2.6.2
>
> _______________________________________________
> U-Boot mailing list
> U-Boot at lists.denx.de
> http://lists.denx.de/mailman/listinfo/u-boot
>
More information about the U-Boot
mailing list