arm64: Add EFI stub and runtime services support
Subsystems (3)
cover arm64: Add EFI stub and runtime services support
This patch series adds EFI support to the arm64 kernel. This support has two main parts: an EFI stub and runtime support. The EFI stub support has the kernel masquerade as a PE/COFF application which can be directly booted by EFI firmware (or by secondary loaders with EFI support). The runtime services support provides access to various EFI firmware services such as reboot, real-time clock, boot variables, and others.
These patches have dependencies on other patches which are not yet in the kernel but have been posted and are currently under review. In particular:
- Generic fixmap support being discussed here:
http://lkml.org/lkml/2013/11/25/474
- early_ioremap support being discussed here:
https://lkml.org/lkml/2013/11/27/621
- shared EFI update_fdt() function being discussed here:
https://lkml.org/lkml/2013/11/27/565
A repo with this patch series and the prerequisite patches is at:
git://github.com/mosalter/linux.git (arm64-efi-patches branch)
Mark Salter (3):
arm64: add EFI stub
doc: arm64: add description of EFI stub support
arm64: add EFI runtime services
Documentation/arm64/booting.txt | 4 + Documentation/efi-stub.txt | 10 +- arch/arm64/Kconfig | 25 ++ arch/arm64/include/asm/efi.h | 18 ++ arch/arm64/kernel/Makefile | 4 + arch/arm64/kernel/efi-entry.S | 81 +++++++ arch/arm64/kernel/efi-stub.c | 280 ++++++++++++++++++++++ arch/arm64/kernel/efi.c | 507 ++++++++++++++++++++++++++++++++++++++++ arch/arm64/kernel/head.S | 112 +++++++++ arch/arm64/kernel/setup.c | 6 + include/linux/efi.h | 2 +- 11 files changed, 1046 insertions(+), 3 deletions(-) create mode 100644 arch/arm64/include/asm/efi.h create mode 100644 arch/arm64/kernel/efi-entry.S create mode 100644 arch/arm64/kernel/efi-stub.c create mode 100644 arch/arm64/kernel/efi.c
Signature · 1 line
1.8.3.1
PATCH 1/3 arm64: add EFI stub
This patch adds PE/COFF header fields to the start of the Image so that it appears as an EFI application to EFI firmware. An EFI stub is included to allow direct booting of the kernel Image. Due to EFI firmware limitations, only little endian kernels with 4K page sizes are supported at this time. Support in the COFF header for signed images was provided by Ard Biesheuvel.
--- arch/arm64/Kconfig | 10 ++ arch/arm64/kernel/Makefile | 3 + arch/arm64/kernel/efi-entry.S | 81 ++++++++++++ arch/arm64/kernel/efi-stub.c | 280 ++++++++++++++++++++++++++++++++++++++++++ arch/arm64/kernel/head.S | 112 +++++++++++++++++ 5 files changed, 486 insertions(+) create mode 100644 arch/arm64/kernel/efi-entry.S create mode 100644 arch/arm64/kernel/efi-stub.c
arch/arm64/Kconfig +10 -0
@@ -250,6 +250,16 @@ config CMDLINE_FORCE This is useful if you cannot or don't want to change the command-line options your boot loader passes to the kernel. +config EFI_STUB + bool "EFI stub support" + depends on !CPU_BIG_ENDIAN && !ARM64_64K_PAGES && OF + select LIBFDT + default y + help + This kernel feature allows an Image to be loaded directly + by EFI firmware without the use of a bootloader. + See Documentation/efi-stub.txt for more information. + endmenu menu "Userspace binary formats"
arch/arm64/kernel/Makefile +3 -0
@@ -4,6 +4,8 @@ CPPFLAGS_vmlinux.lds := -DTEXT_OFFSET=$(TEXT_OFFSET) AFLAGS_head.o := -DTEXT_OFFSET=$(TEXT_OFFSET) +CFLAGS_efi-stub.o := -DTEXT_OFFSET=$(TEXT_OFFSET) \ + -I$(src)/../../../scripts/dtc/libfdt # Object file lists. arm64-obj-y := cputable.o debug-monitors.o entry.o irq.o fpsimd.o \
@@ -18,6 +20,7 @@ arm64-obj-$(CONFIG_SMP) += smp.o smp_spin_table.o arm64-obj-$(CONFIG_HW_PERF_EVENTS) += perf_event.o arm64-obj-$(CONFIG_HAVE_HW_BREAKPOINT)+= hw_breakpoint.o arm64-obj-$(CONFIG_EARLY_PRINTK) += early_printk.o +arm64-obj-$(CONFIG_EFI_STUB) += efi-stub.o efi-entry.o obj-y += $(arm64-obj-y) vdso/ obj-m += $(arm64-obj-m)
arch/arm64/kernel/efi-entry.S new file +81 -0
81 lines not on this page · raw message
arch/arm64/kernel/efi-stub.c new file +280 -0
280 lines not on this page · raw message
arch/arm64/kernel/head.S +112 -0
127 lines not on this page · raw message
Signature · 1 line
1.8.3.1
Re: [PATCH 1/3] arm64: add EFI stub
Hi Mark, Roy,
Quoted 6 lines · Mark Salter
This patch adds PE/COFF header fields to the start of the Image so that it appears as an EFI application to EFI firmware. An EFI stub is included to allow direct booting of the kernel Image. Due to EFI firmware limitations, only little endian kernels with 4K page sizes are supported at this time. Support in the COFF header for signed images was provided by Ard Biesheuvel.
I haven't really jumped into this but, whilst I see the use of EFI_STUB on both arm and arm64, there seems to be some duplication/reinvention between the two series you've put together.
Maybe I'm just being ignorant, but the stuff in efi-stub.c really looks to be doing the same thing on both architectures. Would you guys be able to work to together to produce an independent series containing the common parts, then add arm/arm64 backends on top of that please? In particular, factoring out the device-tree parts ensures that we don't introduce subtle differences between the two architectures when there's no real need to do so...
...or shout at me because I didn't understand what you were doing!
Cheers,
Will
Re: [PATCH 1/3] arm64: add EFI stub
Quoted 22 lines · Will Deacon
Hi Mark, Roy,
Quoted 6 lines · Mark Salter
This patch adds PE/COFF header fields to the start of the Image so that it appears as an EFI application to EFI firmware. An EFI stub is included to allow direct booting of the kernel Image. Due to EFI firmware limitations, only little endian kernels with 4K page sizes are supported at this time. Support in the COFF header for signed images was provided by Ard Biesheuvel.
I haven't really jumped into this but, whilst I see the use of
EFI_STUBon both arm and arm64, there seems to be some duplication/reinvention between the two series you've put together.Maybe I'm just being ignorant, but the stuff in
efi-stub.creally looks to be doing the same thing on both architectures. Would you guys be able to work to together to produce an independent series containing the common parts, then add arm/arm64 backends on top of that please? In particular, factoring out the device-tree parts ensures that we don't introduce subtle differences between the two architectures when there's no real need to do so...
Quoted 5 lines
...or shout at me because I didn't understand what you were doing!
Cheers,
Will
Hi Will,
Some of the device tree code is already factored out indrivers/firmware/efi/fdt.c, but
looking at this again I think that there is more that can be moved to common code. The
main difference between the arm/arm64 stubs is the restrictions on where the kernel binary
can be loaded. On arm64 it is the kernel image itself, and on arm it is the zImage that is the
next stage. This also affects the build environment - the arm64 stub is part of the kernel itself,
and the arm stub is part of the decompressor.
I'll work with Mark to see how much of the two stubs we can refactor into shared code.
Roy
Re: [PATCH 1/3] arm64: add EFI stub
In reply to Will DeaconQuoted 10 lines · Will Deacon
Hi Mark, Roy,
Quoted 6 lines · Mark Salter
This patch adds PE/COFF header fields to the start of the Image so that it appears as an EFI application to EFI firmware. An EFI stub is included to allow direct booting of the kernel Image. Due to EFI firmware limitations, only little endian kernels with 4K page sizes are supported at this time. Support in the COFF header for signed images was provided by Ard Biesheuvel.
Quoted 3 lines
I haven't really jumped into this but, whilst I see the use of
EFI_STUBon both arm and arm64, there seems to be some duplication/reinvention between the two series you've put together.
Indeed. As the file banner says, the arm64 stub started out from the arm code.
Quoted 10 lines
Maybe I'm just being ignorant, but the stuff in
efi-stub.creally looks to be doing the same thing on both architectures. Would you guys be able to work to together to produce an independent series containing the common parts, then add arm/arm64 backends on top of that please? In particular, factoring out the device-tree parts ensures that we don't introduce subtle differences between the two architectures when there's no real need to do so......or shout at me because I didn't understand what you were doing!
Along the way, Roy has pulled out common bits into:
drivers/firmware/efi-stub-helper.c
and
drivers/firmware/fdt.c
There are arguably more bits which could be made common but there's not really a lot left. There are differences between the two stubs which limit what we can do. The arm stub is not part of the kernel (it is part of the zImage wrapper) but the arm64 stub is. So arm64 has a little more flexibility for using kernel facilities (__init attribute, LIBFDT, etc).
--Mark
Catalin Marinas
maintainerRe: [PATCH 1/3] arm64: add EFI stub
In reply to Mark SalterHi Mark,
Quoted 5 lines · Mark Salter
This patch adds PE/COFF header fields to the start of the Image so that it appears as an EFI application to EFI firmware. An EFI stub is included to allow direct booting of the kernel Image. Due to EFI firmware limitations, only little endian kernels with 4K page sizes are supported at this time.
I don't fully understand the EFI firmware limitations but for big endian we could have the EFI_STUB wrapper in little endian and get the kernel to switch to big endian once booted. The image header should always be little endian.
And I have to dig further into the 4K limitation (or you could give a quick summary ;)).
Quoted 13 lines
arch/arm64/kernel/efi-entry.S new file +7
-0@@ -0,0 +1,81 @@ +/* + * EFI entry point. + * + * Copyright (C) 2013 Red Hat, Inc. + * Author: Mark Salter <[email protected]> + * + * This program is free software; you can redistribute it and/or modify
arch/arm64/kernel/efi-entry.S line 17
Quoted 10 lines
arch/arm64/kernel/efi-entry.S new file +10
-0+ * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + * + */ +#include <linux/linkage.h> +#include <linux/init.h> + +#include <asm/assembler.h> + +#define EFI_LOAD_ERROR 0x8000000000000001
It's defined already but I see why you can't include efi.h here. Maybe a comment.
Quoted 17 lines
arch/arm64/kernel/efi-entry.S new file +17
-0+ + __INIT + + /* + * We arrive here from the EFI boot manager with: + * + * * MMU on with identity-mapped RAM. + * * Icache and Dcache on + * + * We will most likely be running from some place other than where + * we want to be. The kernel image wants to be placed at TEXT_OFFSET + * from start of RAM. + */ +ENTRY(efi_stub_entry) + stp x29, x30, [sp, #-32]! + + /*
arch/arm64/kernel/efi-entry.S line 44
Quoted 10 lines
arch/arm64/kernel/efi-entry.S new file +10
-0+ * Call efi_entry to do the real work. + * x0 and x1 are already set up by firmware. Current runtime + * address of image is calculated and passed via *image_addr. + * + * unsigned long efi_entry(void *handle, + * efi_system_table_t *sys_table, + * unsigned long *image_addr) ; + */ + adrp x8, _text + add x8, x8, #:lo12:_text
Minor: some wrong whitespace (but I don't trust our incoming mail server either, it corrupts patches usually).
Quoted 7 lines
arch/arm64/kernel/efi-entry.S new file +7
-0+ add x2, sp, 16 + str x8, [x2] + bl efi_entry + cmn x0, #1 + b.eq efi_load_fail + + /*
arch/arm64/kernel/efi-entry.S line 61
Quoted 10 lines
arch/arm64/kernel/efi-entry.S new file +10
-0+ * efi_entry() will have relocated the kernel image if necessary + * and we return here with device tree address in x0 and the kernel + * entry point stored at *image_addr. Save those values in registers + * which are preserved by __flush_dcache_all. + */ + ldr x1, [sp, #16] + mov x20, x0 + mov x21, x1 + + bl __flush_dcache_all
Regarding __flush_dcache_all, I plan to remove it for all cases apart from power management with the MMU disabled. With MMU enabled, there is no guarantee that this function does the right thing. It's even worse in the guest context.
arch/arm64/kernel/efi-entry.S line 67
Quoted 6 lines
arch/arm64/kernel/efi-entry.S new file +6
-0+ /* Turn off Dcache and MMU */ + mrs x0, sctlr_el1 + bic x0, x0, #1 << 0 // clear SCTLR.M + bic x0, x0, #1 << 2 // clear SCTLR.C + msr sctlr_el1, x0 + isb
I assume an EFI app is running with the MMU enabled (and UP only). Do we always run it in EL1? What about EL2 mode (needed by KVM and Xen)?
arch/arm64/kernel/efi-entry.S line 77
Quoted 10 lines
arch/arm64/kernel/efi-entry.S new file +10
-0+ + /* Jump to real entry point */ + mov x0, x20 + mov x1, xzr + mov x2, xzr + mov x3, xzr + br x21 + +efi_load_fail: + mov x0, EFI_LOAD_ERROR
Needs #EFI_LOAD_ERROR (strange that gas doesn't complain).
Quoted 47 lines
arch/arm64/kernel/efi-entry.S new file +4
-0+ ldp x29, x30, [sp], #32 + ret + +ENDPROC(efi_stub_entry)arch/arm64/kernel/efi-stub.c new file +37
-0@@ -0,0 +1,280 @@ +/* +* linux/arch/arm/boot/compressed/efi-stub.c +* +* Copyright (C) 2013 Linaro Ltd; <[email protected]> +* +* This file implements the EFI boot stub for the arm64 kernel. +* Adapted from ARM version by Mark Salter <[email protected]> +* +* 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. +* +*/ +#include <linux/efi.h> +#include <linux/libfdt.h> +#include <asm/sections.h> +#include <generated/compile.h> +#include <linux/uts.h> +#include <linux/utsname.h> +#include <generated/utsrelease.h> +#include <linux/version.h> + +/* error code which can't be mistaken for valid address */ +#define EFI_ERROR (~0UL) + +/* +* EFI function call wrappers. These are not required for arm64, but wrappers +* are required for X86 to convert between ABIs. These wrappers are provided +* to allow code sharing between X86 and other architectures. Since these +* wrappers directly invoke the EFI function pointer, the function pointer +* type must be properly defined, which is not the case for X86. One advantage +* of this is it allows for type checking of arguments, which is not possible +* with the X86 wrappers. +*/ +#define efi_call_phys0 (f) f() +#define efi_call_phys1 (f, a1) f(a1) +#define efi_call_phys2 (f, a1, a2) f(a1, a2)
arch/arm64/kernel/efi-stub.c line 47
Quoted 10 lines
arch/arm64/kernel/efi-stub.c new file +10
-0+#define efi_call_phys3 (f, a1, a2, a3) f(a1, a2, a3) +#define efi_call_phys4 (f, a1, a2, a3, a4) f(a1, a2, a3, a4) +#define efi_call_phys5 (f, a1, a2, a3, a4, a5) f(a1, a2, a3, a4, a5) + +/* +* AArch64 requires the DTB to be 8-byte aligned in the first 512MiB from +* start of kernel and may not cross a 2MiB boundary. We set alignment to +* equal max size so we know it won't cross a 2MiB boudary. +*/ +#define MAX_DTB_SIZE 0x40000
2MB is 0x200000 (or I don't understand the comment).
Quoted 4 lines
arch/arm64/kernel/efi-stub.c new file +4
-0+#define DTB_ALIGN MAX_DTB_SIZE +#define MAX_DTB_OFFSET 0x20000000 + +#define pr_efi (msg) efi_printk(sys_table, "EFI stub: "msg)
arch/arm64/kernel/efi-stub.c line 61
Quoted 10 lines
arch/arm64/kernel/efi-stub.c new file +10
-0+#define pr_efi_err (msg) efi_printk(sys_table, "EFI stub: ERROR: "msg) + +struct fdt_region { +u64 base; +u64 size; +}; + +/* Include shared EFI stub code */ +#include "../../../drivers/firmware/efi/efi-stub-helper.c" +#include "../../../drivers/firmware/efi/fdt.c"
I don't particularly like .c files inclusion but it looks like x86 does the same.
arch/arm64/kernel/efi-stub.c line 73
Quoted 12 lines
arch/arm64/kernel/efi-stub.c new file +12
-0+ +static unsigned long __init get_dram_base (efi_system_table_t *sys_table) +{ +efi_status_t status; +unsigned long map_size, desc_size; +unsigned long membase =EFI_ERROR ; +efi_memory_desc_t *memory_map; +int i; + + status =efi_get_memory_map (sys_table, &memory_map, &map_size, + &desc_size,NULL ,NULL ); +if (status ==EFI_SUCCESS ) {
Can you exit earlier here if !EFI_SUCCESS? It reduces the indentation level.
Signature · 1 line
Catalin
Re: [PATCH 1/3] arm64: add EFI stub
Quoted 11 lines · Catalin Marinas
Quoted 5 lines · Mark Salter
This patch adds PE/COFF header fields to the start of the Image so that it appears as an EFI application to EFI firmware. An EFI stub is included to allow direct booting of the kernel Image. Due to EFI firmware limitations, only little endian kernels with 4K page sizes are supported at this time.
I don't fully understand the EFI firmware limitations but for big endian we could have the
EFI_STUBwrapper in little endian and get the kernel to switch to big endian once booted. The image header should always be little endian.
That would be fun. :) You'd also have to switch back and forth to make EFI runtime services calls.
Quoted 3 lines
And I have to dig further into the 4K limitation (or you could give a quick summary ;)).
Just that the current UEFI spec mandates 4K pages for UEFI. So if UEFI maps two 4k pages with different attributes and those two pages are within the same 64k kernel page, there'd be a problem. I'd be better if UEFI used 64k pages, then the kernel could easily use 4k or 64k.
--Mark
Catalin Marinas
maintainerRe: [PATCH 1/3] arm64: add EFI stub
Quoted 8 lines · Mark Salter
Quoted 7 lines · Catalin Marinas
Quoted 5 lines · Mark Salter
This patch adds PE/COFF header fields to the start of the Image so that it appears as an EFI application to EFI firmware. An EFI stub is included to allow direct booting of the kernel Image. Due to EFI firmware limitations, only little endian kernels with 4K page sizes are supported at this time.
Quoted 7 lines
Quoted 4 lines
I don't fully understand the EFI firmware limitations but for big endian we could have the
EFI_STUBwrapper in little endian and get the kernel to switch to big endian once booted. The image header should always be little endian.That would be fun. :) You'd also have to switch back and forth to make EFI runtime services calls.
OK, we'll have to live with this restriction.
Quoted 7 lines
Quoted 2 lines
And I have to dig further into the 4K limitation (or you could give a quick summary ;)).
Just that the current UEFI spec mandates 4K pages for UEFI. So if UEFI maps two 4k pages with different attributes and those two pages are within the same 64k kernel page, there'd be a problem. I'd be better if UEFI used 64k pages, then the kernel could easily use 4k or 64k.
For server space, we may see some people asking for 64K pages. But I don't know whether there are any UEFI plans here.
Thanks.
Signature · 1 line
Catalin
Re: [PATCH 1/3] arm64: add EFI stub
Quoted 14 lines · Catalin Marinas
Quoted 13 lines · Mark Salter
Quoted 11 lines · Catalin Marinas
Quoted 5 lines · Mark Salter
This patch adds PE/COFF header fields to the start of the Image so that it appears as an EFI application to EFI firmware. An EFI stub is included to allow direct booting of the kernel Image. Due to EFI firmware limitations, only little endian kernels with 4K page sizes are supported at this time.
I don't fully understand the EFI firmware limitations but for big endian we could have the
EFI_STUBwrapper in little endian and get the kernel to switch to big endian once booted. The image header should always be little endian.
Quoted 4 lines
Quoted 2 lines
That would be fun. :) You'd also have to switch back and forth to make EFI runtime services calls.
OK, we'll have to live with this restriction.
Or just disable runtime services on the switch to big ending. Big endian should not disable the stub (but getting it to work could be a follow-up patch)
g.
Re: [PATCH 1/3] arm64: add EFI stub
Quoted 15 lines · Grant Likely
Quoted 14 lines · Catalin Marinas
Quoted 13 lines · Mark Salter
Quoted 11 lines · Catalin Marinas
Quoted 5 lines · Mark Salter
This patch adds PE/COFF header fields to the start of the Image so that it appears as an EFI application to EFI firmware. An EFI stub is included to allow direct booting of the kernel Image. Due to EFI firmware limitations, only little endian kernels with 4K page sizes are supported at this time.
I don't fully understand the EFI firmware limitations but for big endian we could have the
EFI_STUBwrapper in little endian and get the kernel to switch to big endian once booted. The image header should always be little endian.
Quoted 9 lines
Quoted 4 lines
Quoted 2 lines
That would be fun. :) You'd also have to switch back and forth to make EFI runtime services calls.
OK, we'll have to live with this restriction.
Or just disable runtime services on the switch to big ending. Big endian should not disable the stub (but getting it to work could be a follow-up patch)
The other problem with BE is that the PE/COFF masquerading is built into head.S so the same Image can be used for EFI and non-EFI. I don't see a BE opcode which we could us to provide the magic "MZ" at the start of a BE kernel Image.
--Mark
Re: [PATCH 1/3] arm64: add EFI stub
Quoted 8 lines · Mark Salter
Quoted 4 lines
Or just disable runtime services on the switch to big ending. Big endian should not disable the stub (but getting it to work could be a follow-up patch)
The other problem with BE is that the PE/COFF masquerading is built into
head.Sso the same Image can be used for EFI and non-EFI. I don't see a BE opcode which we could us to provide the magic "MZ" at the start of a BE kernel Image.
That's actually not an issue.
Instructions are always LE - endinaness affects only data.
/
Leif
Re: [PATCH 1/3] arm64: add EFI stub
Quoted 5 lines · Leif Lindholm
Quoted 4 lines · Mark Salter
Quoted 4 lines
Or just disable runtime services on the switch to big ending. Big endian should not disable the stub (but getting it to work could be a follow-up patch)
Quoted 8 lines
Quoted 4 lines
The other problem with BE is that the PE/COFF masquerading is built into
head.Sso the same Image can be used for EFI and non-EFI. I don't see a BE opcode which we could us to provide the magic "MZ" at the start of a BE kernel Image.That's actually not an issue.
Instructions are always LE - endinaness affects only data.
Oh right, I knew that. Time for coffee.
Re: [PATCH 1/3] arm64: add EFI stub
In reply to Catalin MarinasQuoted 12 lines · Catalin Marinas
Hi Mark,
arch/arm64/kernel/efi-entry.S line 17
Quoted 6 lines · Mark Salter
arch/arm64/kernel/efi-entry.S new file +6
-0+#include <linux/linkage.h> +#include <linux/init.h> + +#include <asm/assembler.h> + +#define EFI_LOAD_ERROR 0x8000000000000001It's defined already but I see why you can't include
efi.hhere. Maybe a comment.
okay
Quoted 23 lines
Quoted 23 lines
arch/arm64/kernel/efi-entry.S new file +22
-0+ + __INIT + + /* + * We arrive here from the EFI boot manager with: + * + * * MMU on with identity-mapped RAM. + * * Icache and Dcache on + * + * We will most likely be running from some place other than where + * we want to be. The kernel image wants to be placed at TEXT_OFFSET + * from start of RAM. + */ +ENTRY(efi_stub_entry) + stp x29, x30, [sp, #-32]! + + /* + * Call efi_entry to do the real work. + * x0 and x1 are already set up by firmware. Current runtime + * address of image is calculated and passed via *image_addr. + * + * unsigned long efi_entry(void *handle,
Quoted 8 lines
arch/arm64/kernel/efi-entry.S line 44
Quoted 5 lines
arch/arm64/kernel/efi-entry.S new file +5
-0+ * efi_system_table_t *sys_table, + * unsigned long *image_addr) ; + */ + adrp x8, _text + add x8, x8, #:lo12:_textMinor: some wrong whitespace (but I don't trust our incoming mail server either, it corrupts patches usually).
I will fix it. (the whitespace, not your mail server)
Quoted 15 lines
Quoted 15 lines
arch/arm64/kernel/efi-entry.S new file +14
-0+ add x2, sp, 16 + str x8, [x2] + bl efi_entry + cmn x0, #1 + b.eq efi_load_fail + + /* + * efi_entry() will have relocated the kernel image if necessary + * and we return here with device tree address in x0 and the kernel + * entry point stored at *image_addr. Save those values in registers + * which are preserved by __flush_dcache_all. + */ + ldr x1, [sp, #16] + mov x20, x0
Quoted 8 lines
arch/arm64/kernel/efi-entry.S line 61
Quoted 3 lines
arch/arm64/kernel/efi-entry.S new file +3
-0+ mov x21, x1 + + bl __flush_dcache_allRegarding
__flush_dcache_all, I plan to remove it for all cases apart from power management with the MMU disabled. With MMU enabled, there is no guarantee that this function does the right thing. It's even worse in the guest context.
According to booting.txt, the dcache needs to be invalidated. Is there something existing I can use or do I need to write it?
Quoted 10 lines
arch/arm64/kernel/efi-entry.S line 67
Quoted 7 lines
arch/arm64/kernel/efi-entry.S new file +6
-0+ /* Turn off Dcache and MMU */ + mrs x0, sctlr_el1 + bic x0, x0, #1 << 0 // clear SCTLR.M + bic x0, x0, #1 << 2 // clear SCTLR.C + msr sctlr_el1, x0 + isbI assume an EFI app is running with the MMU enabled (and UP only). Do we always run it in EL1? What about EL2 mode (needed by KVM and Xen)?
Good point. It could be non-secure EL2.
Quoted 13 lines
arch/arm64/kernel/efi-entry.S line 77
Quoted 11 lines
arch/arm64/kernel/efi-entry.S new file +10
-0+ + /* Jump to real entry point */ + mov x0, x20 + mov x1, xzr + mov x2, xzr + mov x3, xzr + br x21 + +efi_load_fail: + mov x0, EFI_LOAD_ERRORNeeds #EFI_LOAD_ERROR (strange that gas doesn't complain).
Hmm, no complaint but it DTRT.
Quoted 8 lines
arch/arm64/kernel/efi-stub.c line 47
Quoted 6 lines
arch/arm64/kernel/efi-stub.c new file +6
-0+/* +* AArch64 requires the DTB to be 8-byte aligned in the first 512MiB from +* start of kernel and may not cross a 2MiB boundary. We set alignment to +* equal max size so we know it won't cross a 2MiB boudary. +*/ +#define MAX_DTB_SIZE 0x400002MB is
0x200000(or I don't understand the comment).
I had a little trouble with it myself. :) The size was left over from older code which used it directly in an allocation. I'll fix the comment, drop MAX_DTB_SIZE, and fix DTB_ALIGN to be 2MiB.
Quoted 7 lines
Quoted 7 lines
arch/arm64/kernel/efi-stub.c new file +7
-0+ +static unsigned long __init get_dram_base (efi_system_table_t *sys_table) +{ +efi_status_t status; +unsigned long map_size, desc_size; +unsigned long membase =EFI_ERROR ; +efi_memory_desc_t *memory_map;
Quoted 9 lines
arch/arm64/kernel/efi-stub.c line 73
Quoted 5 lines
arch/arm64/kernel/efi-stub.c new file +5
-0+int i; + + status =efi_get_memory_map (sys_table, &memory_map, &map_size, + &desc_size,NULL ,NULL ); +if (status ==EFI_SUCCESS ) {Can you exit earlier here if !EFI_SUCCESS? It reduces the indentation level.
Yes.
Catalin Marinas
maintainerRe: [PATCH 1/3] arm64: add EFI stub
Quoted 25 lines · Mark Salter
Quoted 15 lines · Catalin Marinas
Quoted 14 lines · Mark Salter
arch/arm64/kernel/efi-entry.S new file +14
-0+ add x2, sp, 16 + str x8, [x2] + bl efi_entry + cmn x0, #1 + b.eq efi_load_fail + + /* + * efi_entry() will have relocated the kernel image if necessary + * and we return here with device tree address in x0 and the kernel + * entry point stored at *image_addr. Save those values in registers + * which are preserved by __flush_dcache_all. + */ + ldr x1, [sp, #16] + mov x20, x0Quoted 8 lines
arch/arm64/kernel/efi-entry.S line 61
Quoted 3 lines
arch/arm64/kernel/efi-entry.S new file +3
-0+ mov x21, x1 + + bl __flush_dcache_allRegarding
__flush_dcache_all, I plan to remove it for all cases apart from power management with the MMU disabled. With MMU enabled, there is no guarantee that this function does the right thing. It's even worse in the guest context.
Quoted 2 lines
According to booting.txt, the dcache needs to be invalidated. Is there something existing I can use or do I need to write it?
The function will stay for a few cases where needed. But here the D-cache and MMU are still on at this point and there is a slight chance of speculative loads after the flush (though only clean lines). You could move this after he MMU disabling (but still keep the I-cache on for performance).
Signature · 1 line
Catalin
Re: [PATCH 1/3] arm64: add EFI stub
In reply to Mark SalterQuoted 9 lines · Mark Salter
This patch adds PE/COFF header fields to the start of the Image so that it appears as an EFI application to EFI firmware. An EFI stub is included to allow direct booting of the kernel Image. Due to EFI firmware limitations, only little endian kernels with 4K page sizes are supported at this time. Support in the COFF header for signed images was provided by Ard Biesheuvel.
Signed-off-by Mark Salter <[email protected]>Signed-off-by Ard Biesheuvel <[email protected]>
I've already made comments on Roy's arm32 version of this code. I don't like the duplication and it needs to be consolidated, but I would be fine with consolidation being done as follow-on patches if that expedites getting the code in.
g.
Quoted 573 lines
CC Catalin Marinas <[email protected]>CC Will Deacon <[email protected]>CC linux-arm-kernel at lists.infradead.orgCC matt.fleming at intel.comCC linux-efi at vger.kernel.orgCC Leif Lindholm <[email protected]>CC roy.franz at linaro.org--- arch/arm64/Kconfig | 10 ++ arch/arm64/kernel/Makefile | 3 + arch/arm64/kernel/efi-entry.S | 81 ++++++++++++ arch/arm64/kernel/efi-stub.c | 280 ++++++++++++++++++++++++++++++++++++++++++ arch/arm64/kernel/head.S | 112 +++++++++++++++++ 5 files changed, 486 insertions(+) create mode 100644 arch/arm64/kernel/efi-entry.S create mode 100644 arch/arm64/kernel/efi-stub.carch/arm64/Kconfig +10
-0@@ -250,6 +250,16 @@ config CMDLINE_FORCE This is useful if you cannot or don't want to change the command-line options your boot loader passes to the kernel. +config EFI_STUB + bool "EFI stub support" + depends on !CPU_BIG_ENDIAN && !ARM64_64K_PAGES && OF + select LIBFDT + default y + help + This kernel feature allows an Image to be loaded directly + by EFI firmware without the use of a bootloader. + See Documentation/efi-stub.txt for more information. + endmenu menu "Userspace binary formats"arch/arm64/kernel/Makefile +3
-0@@ -4,6 +4,8 @@ CPPFLAGS_vmlinux.lds := -DTEXT_OFFSET=$(TEXT_OFFSET) AFLAGS_head.o := -DTEXT_OFFSET=$(TEXT_OFFSET) +CFLAGS_efi-stub.o := -DTEXT_OFFSET=$(TEXT_OFFSET) \ + -I$(src)/../../../scripts/dtc/libfdt # Object file lists. arm64-obj-y := cputable.o debug-monitors.o entry.o irq.o fpsimd.o \@@ -18,6 +20,7 @@ arm64-obj-$(CONFIG_SMP) += smp.o smp_spin_table.o arm64-obj-$(CONFIG_HW_PERF_EVENTS) += perf_event.o arm64-obj-$(CONFIG_HAVE_HW_BREAKPOINT)+= hw_breakpoint.o arm64-obj-$(CONFIG_EARLY_PRINTK) += early_printk.o +arm64-obj-$(CONFIG_EFI_STUB) += efi-stub.o efi-entry.o obj-y += $(arm64-obj-y) vdso/ obj-m += $(arm64-obj-m)arch/arm64/kernel/efi-entry.S new file +81
-0@@ -0,0 +1,81 @@ +/* + * EFI entry point. + * + * Copyright (C) 2013 Red Hat, Inc. + * Author: Mark Salter <[email protected]> + * + * 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. + * + */ +#include <linux/linkage.h> +#include <linux/init.h> + +#include <asm/assembler.h> + +#define EFI_LOAD_ERROR 0x8000000000000001 + + __INIT + + /* + * We arrive here from the EFI boot manager with: + * + * * MMU on with identity-mapped RAM. + * * Icache and Dcache on + * + * We will most likely be running from some place other than where + * we want to be. The kernel image wants to be placed at TEXT_OFFSET + * from start of RAM. + */ +ENTRY(efi_stub_entry) + stp x29, x30, [sp, #-32]! + + /* + * Call efi_entry to do the real work. + * x0 and x1 are already set up by firmware. Current runtime + * address of image is calculated and passed via *image_addr. + * + * unsigned long efi_entry(void *handle, + * efi_system_table_t *sys_table, + * unsigned long *image_addr) ; + */ + adrp x8, _text + add x8, x8, #:lo12:_text + add x2, sp, 16 + str x8, [x2] + bl efi_entry + cmn x0, #1 + b.eq efi_load_fail + + /* + * efi_entry() will have relocated the kernel image if necessary + * and we return here with device tree address in x0 and the kernel + * entry point stored at *image_addr. Save those values in registers + * which are preserved by __flush_dcache_all. + */ + ldr x1, [sp, #16] + mov x20, x0 + mov x21, x1 + + bl __flush_dcache_all + /* Turn off Dcache and MMU */ + mrs x0, sctlr_el1 + bic x0, x0, #1 << 0 // clear SCTLR.M + bic x0, x0, #1 << 2 // clear SCTLR.C + msr sctlr_el1, x0 + isb + + /* Jump to real entry point */ + mov x0, x20 + mov x1, xzr + mov x2, xzr + mov x3, xzr + br x21 + +efi_load_fail: + mov x0, EFI_LOAD_ERROR + ldp x29, x30, [sp], #32 + ret + +ENDPROC(efi_stub_entry)arch/arm64/kernel/efi-stub.c new file +280
-0@@ -0,0 +1,280 @@ +/* +* linux/arch/arm/boot/compressed/efi-stub.c +* +* Copyright (C) 2013 Linaro Ltd; <[email protected]> +* +* This file implements the EFI boot stub for the arm64 kernel. +* Adapted from ARM version by Mark Salter <[email protected]> +* +* 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. +* +*/ +#include <linux/efi.h> +#include <linux/libfdt.h> +#include <asm/sections.h> +#include <generated/compile.h> +#include <linux/uts.h> +#include <linux/utsname.h> +#include <generated/utsrelease.h> +#include <linux/version.h> + +/* error code which can't be mistaken for valid address */ +#define EFI_ERROR (~0UL) + +/* +* EFI function call wrappers. These are not required for arm64, but wrappers +* are required for X86 to convert between ABIs. These wrappers are provided +* to allow code sharing between X86 and other architectures. Since these +* wrappers directly invoke the EFI function pointer, the function pointer +* type must be properly defined, which is not the case for X86. One advantage +* of this is it allows for type checking of arguments, which is not possible +* with the X86 wrappers. +*/ +#define efi_call_phys0 (f) f() +#define efi_call_phys1 (f, a1) f(a1) +#define efi_call_phys2 (f, a1, a2) f(a1, a2) +#define efi_call_phys3 (f, a1, a2, a3) f(a1, a2, a3) +#define efi_call_phys4 (f, a1, a2, a3, a4) f(a1, a2, a3, a4) +#define efi_call_phys5 (f, a1, a2, a3, a4, a5) f(a1, a2, a3, a4, a5) + +/* +* AArch64 requires the DTB to be 8-byte aligned in the first 512MiB from +* start of kernel and may not cross a 2MiB boundary. We set alignment to +* equal max size so we know it won't cross a 2MiB boudary. +*/ +#define MAX_DTB_SIZE 0x40000 +#define DTB_ALIGN MAX_DTB_SIZE +#define MAX_DTB_OFFSET 0x20000000 + +#define pr_efi (msg) efi_printk(sys_table, "EFI stub: "msg) +#define pr_efi_err (msg) efi_printk(sys_table, "EFI stub: ERROR: "msg) + +struct fdt_region { +u64 base; +u64 size; +}; + +/* Include shared EFI stub code */ +#include "../../../drivers/firmware/efi/efi-stub-helper.c" +#include "../../../drivers/firmware/efi/fdt.c" + +static unsigned long __init get_dram_base (efi_system_table_t *sys_table) +{ +efi_status_t status; +unsigned long map_size, desc_size; +unsigned long membase =EFI_ERROR ; +efi_memory_desc_t *memory_map; +int i; + + status =efi_get_memory_map (sys_table, &memory_map, &map_size, + &desc_size,NULL ,NULL ); +if (status ==EFI_SUCCESS ) { +for (i =0 ; i < (map_size /sizeof (efi_memory_desc_t)); i++) { +efi_memory_desc_t *desc; +unsigned long m = (unsigned long )memory_map; + + desc = (efi_memory_desc_t *)(m + (i * desc_size)); + +if (desc->num_pages ==0 ) +break ; + +if (desc->type ==EFI_CONVENTIONAL_MEMORY ) { +unsigned long base = desc->phys_addr; + + base &= ~((unsigned long )(TEXT_OFFSET -1 )); + +if (membase > base) + membase = base; + } + } + } +return membase; +} + +unsigned long __init efi_entry (void *handle,efi_system_table_t *sys_table, +unsigned long *image_addr) +{ +efi_loaded_image_t *image; +efi_status_t status; +unsigned long image_size, image_memsize =0 ; +unsigned long dram_base; +/* addr/point and size pairs for memory management*/ +u64 initrd_addr; +u64 initrd_size =0 ; +u64 fdt_addr;/* Original DTB */ +u64 fdt_size =0 ; +unsigned long new_fdt_size; +char *cmdline_ptr; +int cmdline_size =0 ; +unsigned long new_fdt_addr; +unsigned long map_size, desc_size; +unsigned long mmap_key; +efi_memory_desc_t *memory_map; +u32 desc_ver; +efi_guid_t proto =LOADED_IMAGE_PROTOCOL_GUID ; + +/* Check if we were booted by the EFI firmware */ +if (sys_table->hdr.signature !=EFI_SYSTEM_TABLE_SIGNATURE ) + goto fail; + +pr_efi ("Booting Linux Kernel...\n" ); + +/* get the command line from EFI, using the LOADED_IMAGE protocol */ + status =efi_call_phys3 (sys_table->boottime->handle_protocol, + handle, &proto, (void *)&image); +if (status !=EFI_SUCCESS ) { +pr_efi_err ("Failed to get handle for LOADED_IMAGE_PROTOCOL\n" ); + goto fail; + } + +/* +* We are going to copy this into device tree, so we don't care where +* in memory it is. +*/ + cmdline_ptr =efi_convert_cmdline_to_ascii (sys_table, image, + &cmdline_size); +if (!cmdline_ptr) { +pr_efi_err ("Failed to convert command line to ascii\n" ); + goto fail; + } + + status =handle_cmdline_files (sys_table, image, cmdline_ptr,"dtb=" , + ~0UL , (unsigned long *)&fdt_addr, + (unsigned long *)&fdt_size); +if (status !=EFI_SUCCESS ) { +pr_efi_err ("Failed to load device tree blob\n" ); + goto fail_free_cmdline; + } + +if (fdt_check_header ((void *)fdt_addr)) { +pr_efi_err ("Device Tree header not valid\n" ); + goto fail_free_dtb; + } +if (fdt_totalsize ((void *)fdt_addr) > fdt_size) { +pr_efi_err ("Incomplete device tree\n" ); + goto fail_free_dtb; + } + + dram_base =get_dram_base (sys_table); +if (dram_base ==EFI_ERROR ) { +pr_efi_err ("Failed to get DRAM base\n" ); + goto fail_free_dtb; + } + +/* Relocate the image, if required. */ + image_size = image->image_size; +if (*image_addr != (dram_base +TEXT_OFFSET )) { + image_memsize = image_size + (_end - _edata); + status =efi_relocate_kernel (sys_table, image_addr, + image_size, image_memsize, + dram_base +TEXT_OFFSET , +PAGE_SIZE ); +if (status !=EFI_SUCCESS ) { +pr_efi_err ("Failed to relocate kernel\n" ); + goto fail_free_dtb; + } +if (*image_addr != (dram_base +TEXT_OFFSET )) { +pr_efi_err ("Failed to alloc kernel memory\n" ); + goto fail_free_image; + } + } + + status =handle_cmdline_files (sys_table, image, cmdline_ptr,"initrd=" , + dram_base +0x20000000 , + (unsigned long *)&initrd_addr, + (unsigned long *)&initrd_size); +if (status !=EFI_SUCCESS ) +pr_efi ("No initrd found\n" ); + +/* +* Estimate size of new FDT, and allocate memory for it. We +* will allocate a bigger buffer if this ends up being too +* small, so a rough guess is OK here. We increment the size +* by PAGE_SIZE since the firmware allocates by pages anyway. +*/ + new_fdt_size = fdt_size +EFI_PAGE_SIZE ; +while (1 ) { + status =efi_high_alloc (sys_table, new_fdt_size,DTB_ALIGN , + &new_fdt_addr, + dram_base +MAX_DTB_OFFSET ); +if (status !=EFI_SUCCESS ) { +pr_efi_err ("No memory for new device tree\n" ); + goto fail_free_initrd; + } + +/* +* Now that we have done our final memory allocation, we can +* get the memory map key needed for exit_boot_services(). +*/ + status =efi_get_memory_map (sys_table, &memory_map, &map_size, + &desc_size, &desc_ver, &mmap_key); +if (status !=EFI_SUCCESS ) + goto fail_free_new_fdt; + + status =update_fdt (sys_table, + (void *)fdt_addr, (void *)new_fdt_addr, + new_fdt_size, cmdline_ptr, + initrd_addr, initrd_size, + memory_map, map_size, desc_size, desc_ver); + +/* Succeeding the first time is the expected case. */ +if (status ==EFI_SUCCESS ) +break ; + +if (status ==EFI_BUFFER_TOO_SMALL ) { +/* +* We need to allocate more space for the new +* device tree, so free existing buffer that is +* too small. Also free memory map, as we will need +* to get new one that reflects the free/alloc we do +* on the device tree buffer. +*/ +efi_free (sys_table, new_fdt_size, new_fdt_addr); +efi_call_phys1 (sys_table->boottime->free_pool, + memory_map); + new_fdt_size +=EFI_PAGE_SIZE ; + }else { +pr_efi_err ("Unable to constuct new device tree\n" ); + goto fail_free_mmap; + } + } + +/* Now we are ready to exit_boot_services.*/ + status =efi_call_phys2 (sys_table->boottime->exit_boot_services, + handle, mmap_key); + +if (status !=EFI_SUCCESS ) { +pr_efi_err ("Exit boot services failed\n" ); + goto fail_free_mmap; + } + +/* +* Now we need to return the FDT address to the calling +* function so it can be used as part of normal boot. +*/ +return new_fdt_addr; + +fail_free_mmap: +efi_call_phys1 (sys_table->boottime->free_pool, memory_map); + +fail_free_new_fdt: +efi_free (sys_table, new_fdt_size, new_fdt_addr); + +fail_free_initrd: +efi_free (sys_table, initrd_size, initrd_addr); + +fail_free_image: +efi_free (sys_table, image_memsize, *image_addr); + +fail_free_dtb: +if (fdt_addr) +efi_free (sys_table, fdt_size, fdt_addr); + +fail_free_cmdline: +efi_free (sys_table, cmdline_size, (u64 )cmdline_ptr); + +fail: +return EFI_ERROR ; +}arch/arm64/kernel/head.S +112
-0@@ -107,8 +107,18 @@ /* * DO NOT MODIFY. Image header expected by Linux boot-loaders. */ +#ifdef CONFIG_EFI_STUB + /* + * Magic "MZ" signature for PE/COFF + * Little Endian: add x13, x18, #0x16 + */ +efi_head: + .long 0x91005a4d + b stext +#else b stext // branch to kernel start, magic .long 0 // reserved +#endif .quad TEXT_OFFSET // Image load offset from start of RAM .quad 0 // reserved .quad 0 // reserved@@ -119,7 +129,109 @@ .byte 0x52 .byte 0x4d .byte 0x64 +#ifdef CONFIG_EFI_STUB + .long pe_header - efi_head // Offset to the PE header. +#else .word 0 // reserved +#endif + +#ifdef CONFIG_EFI_STUB + .align 3 +pe_header: + .ascii "PE" + .short 0 +coff_header: + .short 0xaa64 // AArch64 + .short 2 // nr_sections + .long 0 // TimeDateStamp + .long 0 // PointerToSymbolTable + .long 1 // NumberOfSymbols + .short section_table - optional_header // SizeOfOptionalHeader + .short 0x206 // Characteristics. + // IMAGE_FILE_DEBUG_STRIPPED | + // IMAGE_FILE_EXECUTABLE_IMAGE | + // IMAGE_FILE_LINE_NUMS_STRIPPED +optional_header: + .short 0x20b // PE32+ format + .byte 0x02 // MajorLinkerVersion + .byte 0x14 // MinorLinkerVersion + .long _edata - stext // SizeOfCode + .long 0 // SizeOfInitializedData + .long 0 // SizeOfUninitializedData + .long efi_stub_entry - efi_head // AddressOfEntryPoint + .long stext - efi_head // BaseOfCode + +extra_header_fields: + .quad 0 // ImageBase + .long 0x20 // SectionAlignment + .long 0x8 // FileAlignment + .short 0 // MajorOperatingSystemVersion + .short 0 // MinorOperatingSystemVersion + .short 0 // MajorImageVersion + .short 0 // MinorImageVersion + .short 0 // MajorSubsystemVersion + .short 0 // MinorSubsystemVersion + .long 0 // Win32VersionValue + + .long _edata - efi_head // SizeOfImage + + // Everything before the kernel image is considered part of the header + .long stext - efi_head // SizeOfHeaders + .long 0 // CheckSum + .short 0xa // Subsystem (EFI application) + .short 0 // DllCharacteristics + .quad 0 // SizeOfStackReserve + .quad 0 // SizeOfStackCommit + .quad 0 // SizeOfHeapReserve + .quad 0 // SizeOfHeapCommit + .long 0 // LoaderFlags + .long 0x6 // NumberOfRvaAndSizes + + .quad 0 // ExportTable + .quad 0 // ImportTable + .quad 0 // ResourceTable + .quad 0 // ExceptionTable + .quad 0 // CertificationTable + .quad 0 // BaseRelocationTable + + // Section table +section_table: + + /* + * The EFI application loader requires a relocation section + * because EFI applications must be relocatable. This is a + * dummy section as far as we are concerned. + */ + .ascii ".reloc" + .byte 0 + .byte 0 // end of 0 padding of section name + .long 0 + .long 0 + .long 0 // SizeOfRawData + .long 0 // PointerToRawData + .long 0 // PointerToRelocations + .long 0 // PointerToLineNumbers + .short 0 // NumberOfRelocations + .short 0 // NumberOfLineNumbers + .long 0x42100040 // Characteristics (section flags) + + + .ascii ".text" + .byte 0 + .byte 0 + .byte 0 // end of 0 padding of section name + .long _edata - stext // VirtualSize + .long stext - efi_head // VirtualAddress + .long _edata - stext // SizeOfRawData + .long stext - efi_head // PointerToRawData + + .long 0 // PointerToRelocations (0 for executables) + .long 0 // PointerToLineNumbers (0 for executables) + .short 0 // NumberOfRelocations (0 for executables) + .short 0 // NumberOfLineNumbers (0 for executables) + .long 0xe0500020 // Characteristics (section flags) + .align 5 +#endif ENTRY(stext) mov x21, x0 // x21=FDTSignature · 7 lines
1.8.3.1 -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo at vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
PATCH 2/3 doc: arm64: add description of EFI stub support
In reply to Mark Salter--- Documentation/arm64/booting.txt | 4 ++++ Documentation/efi-stub.txt | 10 ++++++++-- 2 files changed, 12 insertions(+), 2 deletions(-)
Documentation/arm64/booting.txt +4 -0
@@ -85,6 +85,10 @@ The decompressed kernel image contains a 64-byte header as follows: Header notes: - code0/code1 are responsible for branching to stext. +- when booting through EFI, code0/code1 are initially skipped. + res5 is an offset to the PE header and the PE header has the EFI + entry point (efi_stub_entry). When the stub has done its work, it + jumps to code0 to resume the normal boot process. The image must be placed at the specified offset (currently 0x80000) from the start of the system RAM and called there. The start of the
Documentation/efi-stub.txt +8 -2
@@ -12,6 +12,11 @@ arch/arm/boot/compressed/efi-header.S and arch/arm/boot/compressed/efi-stub.c. EFI stub code that is shared between architectures is in drivers/firmware/efi/efi-stub-helper.c. +For arm64, there is no compressed kernel support, so the Image itself +masquerades as a PE/COFF image and the EFI stub is linked into the +kernel. The arm64 EFI stub lives in arch/arm64/kernel/efi-entry.S +and arch/arm64/kernel/efi-stub.c. + By using the EFI boot stub it's possible to boot a Linux kernel without the use of a conventional EFI boot loader, such as grub or elilo. Since the EFI boot stub performs the jobs of a boot loader, in
@@ -28,7 +33,8 @@ the extension the EFI firmware loader will refuse to execute it. It's not possible to execute bzImage.efi from the usual Linux file systems because EFI firmware doesn't have support for them. For ARM the arch/arm/boot/zImage should be copied to the system partition, and it -may not need to be renamed. +may not need to be renamed. Similarly for arm64, arch/arm64/boot/Image +should be copied but not necessarily renamed. **** Passing kernel parameters from the EFI shell
@@ -72,7 +78,7 @@ is passed to bzImage.efi. **** The "dtb=" option -For the ARM architecture, we also need to be able to provide a device +For ARM and arm64 architecture, we also need to be able to provide a device tree to the kernel. This is done with the "dtb=" command line option, and is process in the same manner as the "initrd=" option that is described above.
Signature · 1 line
1.8.3.1
Re: [PATCH 2/3] doc: arm64: add description of EFI stub support
On Fri, 29 Nov 2013 17:05:11 -0500, Mark Salter <[email protected]> wrote:
Even documentation updates should have a mildly sane commit text. State why these changes came about. Give a future reader at least some clues as to why these changes were made now.
Otherwise:
g.
Quoted 73 lines
Signed-off-by Mark Salter <[email protected]>CC Catalin Marinas <[email protected]>CC Will Deacon <[email protected]>CC linux-arm-kernel at lists.infradead.orgCC matt.fleming at intel.comCC linux-efi at vger.kernel.orgCC linux-doc at vger.kernel.orgCC Rob Landley <[email protected]>CC Leif Lindholm <[email protected]>CC roy.franz at linaro.org--- Documentation/arm64/booting.txt | 4 ++++ Documentation/efi-stub.txt | 10 ++++++++-- 2 files changed, 12 insertions(+), 2 deletions(-)Documentation/arm64/booting.txt +4
-0@@ -85,6 +85,10 @@ The decompressed kernel image contains a 64-byte header as follows: Header notes: - code0/code1 are responsible for branching to stext. +- when booting through EFI, code0/code1 are initially skipped. + res5 is an offset to the PE header and the PE header has the EFI + entry point (efi_stub_entry). When the stub has done its work, it + jumps to code0 to resume the normal boot process. The image must be placed at the specified offset (currently 0x80000) from the start of the system RAM and called there. The start of theDocumentation/efi-stub.txt +8
-2@@ -12,6 +12,11 @@ arch/arm/boot/compressed/efi-header.S and arch/arm/boot/compressed/efi-stub.c. EFI stub code that is shared between architectures is in drivers/firmware/efi/efi-stub-helper.c. +For arm64, there is no compressed kernel support, so the Image itself +masquerades as a PE/COFF image and the EFI stub is linked into the +kernel. The arm64 EFI stub lives in arch/arm64/kernel/efi-entry.S +and arch/arm64/kernel/efi-stub.c. + By using the EFI boot stub it's possible to boot a Linux kernel without the use of a conventional EFI boot loader, such as grub or elilo. Since the EFI boot stub performs the jobs of a boot loader, in@@ -28,7 +33,8 @@ the extension the EFI firmware loader will refuse to execute it. It's not possible to execute bzImage.efi from the usual Linux file systems because EFI firmware doesn't have support for them. For ARM the arch/arm/boot/zImage should be copied to the system partition, and it -may not need to be renamed. +may not need to be renamed. Similarly for arm64, arch/arm64/boot/Image +should be copied but not necessarily renamed. **** Passing kernel parameters from the EFI shell@@ -72,7 +78,7 @@ is passed to bzImage.efi. **** The "dtb=" option -For the ARM architecture, we also need to be able to provide a device +For ARM and arm64 architecture, we also need to be able to provide a device tree to the kernel. This is done with the "dtb=" command line option, and is process in the same manner as the "initrd=" option that is described above.Signature · 7 lines
1.8.3.1 _______________________________________________ linux-arm-kernel mailing list linux-arm-kernel at lists.infradead.org http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
PATCH 3/3 arm64: add EFI runtime services
In reply to Mark SalterThis patch adds EFI runtime support for arm64. The runtime support allows the kernel to access various EFI runtime services provided by EFI firmware. Things like reboot, real time clock, EFI boot variables, and others.
--- arch/arm64/Kconfig | 15 ++ arch/arm64/include/asm/efi.h | 18 ++ arch/arm64/kernel/Makefile | 1 + arch/arm64/kernel/efi.c | 507 +++++++++++++++++++++++++++++++++++++++++++ arch/arm64/kernel/setup.c | 6 + include/linux/efi.h | 2 +- 6 files changed, 548 insertions(+), 1 deletion(-) create mode 100644 arch/arm64/include/asm/efi.h create mode 100644 arch/arm64/kernel/efi.c
arch/arm64/Kconfig +15 -0
@@ -250,6 +250,19 @@ config CMDLINE_FORCE This is useful if you cannot or don't want to change the command-line options your boot loader passes to the kernel. +config EFI + bool "EFI runtime service support" + depends on !CPU_BIG_ENDIAN && !ARM64_64K_PAGES && OF + select UCS2_STRING + select LIBFDT + help + This enables the kernel to use UEFI runtime services that are + available (such as the UEFI variable services). + + This option is only useful on systems that have UEFI firmware. + However, even with this option, the resultant kernel should + continue to boot on existing non-UEFI platforms. + config EFI_STUB bool "EFI stub support" depends on !CPU_BIG_ENDIAN && !ARM64_64K_PAGES && OF
@@ -291,6 +304,8 @@ source "net/Kconfig" source "drivers/Kconfig" +source "drivers/firmware/Kconfig" + source "fs/Kconfig" source "arch/arm64/kvm/Kconfig"
arch/arm64/include/asm/efi.h new file +18 -0
@@ -0,0 +1,18 @@ +#ifndef _ASM_ARM64_EFI_H +#define _ASM_ARM64_EFI_H + +#include <asm/io.h> + +#ifdef CONFIG_EFI +extern void efi_init (void ); +#else +#define efi_init () +#endif + +#define efi_remap (cookie, size) __ioremap((cookie), (size), PAGE_KERNEL_EXEC) +#define efi_ioremap (cookie, size) __ioremap((cookie), (size), \ + __pgprot(PROT_DEVICE_nGnRE)) +#define efi_unmap (cookie) __iounmap((cookie)) +#define efi_iounmap (cookie) __iounmap((cookie)) + +#endif /* _ASM_ARM64_EFI_H */
arch/arm64/kernel/Makefile +1 -0
@@ -21,6 +21,7 @@ arm64-obj-$(CONFIG_HW_PERF_EVENTS) += perf_event.o arm64-obj-$(CONFIG_HAVE_HW_BREAKPOINT)+= hw_breakpoint.o arm64-obj-$(CONFIG_EARLY_PRINTK) += early_printk.o arm64-obj-$(CONFIG_EFI_STUB) += efi-stub.o efi-entry.o +arm64-obj-$(CONFIG_EFI) += efi.o obj-y += $(arm64-obj-y) vdso/ obj-m += $(arm64-obj-m)
arch/arm64/kernel/efi.c new file +507 -0
507 lines not on this page · raw message
arch/arm64/kernel/setup.c +6 -0
@@ -41,6 +41,7 @@#include <linux/memblock.h> #include <linux/of_fdt.h> #include <linux/of_platform.h> +#include <linux/efi.h> #include <asm/fixmap.h> #include <asm/cputype.h>
@@ -55,6 +56,7 @@#include <asm/traps.h> #include <asm/memblock.h> #include <asm/psci.h> +#include <asm/efi.h> unsigned int processor_id;EXPORT_SYMBOL (processor_id);
@@ -222,9 +224,13 @@ void __init setup_arch(char **cmdline_p)arm64_memblock_init (); +efi_init ();paging_init ();request_standard_resources (); +if (efi_enabled (EFI_BOOT )) +efi_enter_virtual_mode (); +unflatten_device_tree ();psci_init ();
include/linux/efi.h +1 -1
@@ -655,7 +655,7 @@ extern int __init efi_setup_pcdp_console(char *);#define EFI_64BIT 5/* Is the firmware 64-bit? */ #ifdef CONFIG_EFI -# ifdef CONFIG_X86 +# if defined(CONFIG_X86 ) || defined(CONFIG_ARM64 )extern int efi_enabled (int facility);# else static inline int efi_enabled (int facility)
Signature · 1 line
1.8.3.1
Catalin Marinas
maintainerRe: [PATCH 3/3] arm64: add EFI runtime services
Quoted 12 lines · Mark Salter
arch/arm64/include/asm/efi.h new file +6
-0@@ -0,0 +1,18 @@ +#ifndef _ASM_ARM64_EFI_H +#define _ASM_ARM64_EFI_H + +#include <asm/io.h> + +#ifdef CONFIG_EFI
arch/arm64/include/asm/efi.h line 16
Quoted 10 lines
arch/arm64/include/asm/efi.h new file +10
-0+extern void efi_init (void ); +#else +#define efi_init () +#endif + +#define efi_remap (cookie, size) __ioremap((cookie), (size), PAGE_KERNEL_EXEC) +#define efi_ioremap (cookie, size) __ioremap((cookie), (size), \ + __pgprot(PROT_DEVICE_nGnRE)) +#define efi_unmap (cookie) __iounmap((cookie)) +#define efi_iounmap (cookie) __iounmap((cookie))
I prefer to use the ioremap_*() functions rather than the lower-level __ioremap(). The ioremap_cache() should give us executable memory.
Looking at the code, I think we have a bug with ioremap_cache() using MT_NORMAL since it doesn't have the shareability bit (Device memory does not require this on AArch64). PROT_DEFAULT should change to pgprot_default | PTE_DIRTY.
Quoted 24 lines
arch/arm64/kernel/efi.c new file +18
-0@@ -0,0 +1,507 @@ +/* +* Extensible Firmware Interface +* +* Based on Extensible Firmware Interface Specification version 2.3.1 +* +* Copyright (C) 2013 Linaro Ltd. +* +* Adapted for arm64 from arch/arm/kernel/efi.c code +*/ + +#include <linux/efi.h> +#include <linux/export.h> +#include <linux/memblock.h> +#include <linux/of.h> +#include <linux/of_fdt.h> +#include <linux/sched.h> +#include <linux/slab.h> +#include <linux/bootmem.h>
arch/arm64/kernel/efi.c line 28
Quoted 10 lines
arch/arm64/kernel/efi.c new file +10
-0+ +#include <asm/cacheflush.h> +#include <asm/efi.h> +#include <asm/tlbflush.h> +#include <asm/mmu_context.h> + +#define efi_early_remap (a, b) \ + ((__force void *)early_ioremap((a), (b))) +#define efi_early_unmap (a, b) \ + early_iounmap((void __iomem *)(a), (b))
I lost track of the early_ioremap status for arm/arm64? Was there more progress since October (I think)?
Quoted 55 lines
arch/arm64/kernel/efi.c new file +55
-0+static int __init fdt_find_efi_params (unsigned long node,const char *uname, +int depth,void *data) +{ +unsigned long len, size; +__be32 *prop; + +if (depth !=1 || + (strcmp (uname,"chosen" ) !=0 &&strcmp (uname,"chosen at 0" ) !=0 )) +return 0 ; + +pr_info ("Getting EFI parameters from FDT.\n" ); + + prop =of_get_flat_dt_prop (node,"linux,uefi-system-table" , &len); +if (!prop) { +pr_err ("No EFI system table in FDT\n" ); +return 0 ; + } + efi_system_table =of_read_ulong (prop, len/4 ); + + prop =of_get_flat_dt_prop (node,"linux,uefi-mmap-start" , &len); +if (!prop) { +pr_err ("No EFI memmap in FDT\n" ); +return 0 ; + } + memmap.phys_map = (void *)of_read_ulong (prop, len/4 ); + + prop =of_get_flat_dt_prop (node,"linux,uefi-mmap-size" , &len); +if (!prop) { +pr_err ("No EFI memmap size in FDT\n" ); +return 0 ; + } + size =of_read_ulong (prop, len/4 ); + memmap.map_end = memmap.phys_map + size; + +/* reserve memmap */ +memblock_reserve ((u64 )memmap.phys_map, size); + + prop =of_get_flat_dt_prop (node,"linux,uefi-mmap-desc-size" , &len); +if (!prop) { +pr_err ("No EFI memmap descriptor size in FDT\n" ); +return 0 ; + } + memmap.desc_size =of_read_ulong (prop, len/4 ); + + memmap.nr_map = size / memmap.desc_size; + + prop =of_get_flat_dt_prop (node,"linux,uefi-mmap-desc-ver" , &len); +if (!prop) { +pr_err ("No EFI memmap descriptor version in FDT\n" ); +return 0 ; + } + memmap.desc_version =of_read_ulong (prop, len/4 ); + +if (uefi_debug) { +pr_info (" EFI system table @ %p\n" , (void *)efi_system_table);
arch/arm64/kernel/efi.c line 120
Quoted 10 lines
arch/arm64/kernel/efi.c new file +10
-0+pr_info (" EFI mmap @ %p-%p\n" , memmap.phys_map, + memmap.map_end); +pr_info (" EFI mmap descriptor size = 0x%lx\n" , + memmap.desc_size); +pr_info (" EFI mmap descriptor version = 0x%lx\n" , + memmap.desc_version); + } + +return 1 ; +}
Are these checks generic to other architectures? We may do with some helpers to avoid duplication.
arch/arm64/kernel/efi.c line 95
Quoted 2 lines
arch/arm64/kernel/efi.c new file +2
-0+ +#define PGD_END (&idmap_pg_dir[sizeof(idmap_pg_dir)/sizeof(pgd_t)])
Just &idmap_pg_dir[PTRS_PER_PGD] would do (or idmap_pg_dir + ARRAY_SIZE(idmap_pg_dir)).
arch/arm64/kernel/efi.c line 130
Quoted 8 lines
arch/arm64/kernel/efi.c new file +8
-0+#ifndef CONFIG_SMP +#define PMD_FLAGS (PMD_ATTRINDX(MT_NORMAL) | PMD_TYPE_SECT | PMD_SECT_AF) +#else +#define PMD_FLAGS (PMD_ATTRINDX(MT_NORMAL) | PMD_TYPE_SECT | PMD_SECT_AF | \ + PMD_SECT_S) +#endif + +static void __init create_idmap (unsigned long addr,unsigned long len)
I would change this to use the existing create_mapping() function which takes care of allocating pud/pmd/ptes. We shouldn't duplicate this functionality in two places. create_mapping() may need a slight change since it assumes swapper_pg_dir but it's not much. It also uses memblock_alloc() for early allocations.
arch/arm64/kernel/efi.c line 245
Quoted 3 lines
arch/arm64/kernel/efi.c new file +3
-0+static __init int memmap_walk (struct efi_memory_map *memmap, +int (*func)(efi_memory_desc_t *,void *), +void *private_data,int early)
Is this generic enough as a common helper between arm and arm64 (and maybe x86)?
Quoted 1 line
arch/arm64/kernel/efi.c new file +1
-0+static __init int reserve_region (efi_memory_desc_t *md,void *priv)
[...]
Quoted 1 line
+static
__initvoidreserve_regions(void)
[...]
Quoted 1 line
+static int
__initremap_region(efi_memory_desc_t *md, void *priv)
[...]
Quoted 1 line
+static int
__initremap_regions(efi_memory_desc_t *map)
Same here (I haven't looked at the other implementations).
Quoted 14 lines
arch/arm64/kernel/efi.c new file +14
-0+/* +* Called from setup_arch with interrupts disabled. +*/ +void __init efi_enter_virtual_mode (void ) +{ +void *newmap; +efi_status_t status; +u64 mapsize, total_freed =0 ; +int count; + +if (!efi_enabled (EFI_BOOT )) { +pr_info ("EFI services will not be available.\n" ); +return ; + }
arch/arm64/kernel/efi.c line 434
Quoted 10 lines
arch/arm64/kernel/efi.c new file +10
-0+ +pr_info ("Remapping and enabling EFI services.\n" ); + + mapsize = memmap.map_end - memmap.phys_map; + memmap.map = (__force void *)ioremap_cache ((phys_addr_t )memmap.phys_map, + mapsize); + memmap.map_end = memmap.map + mapsize; + +/* Map the regions we reserved earlier */ + newmap =alloc_bootmem (mapsize);
memblock_alloc() (also check the other bootmem uses in this patch, arm64 is using memblock).
Quoted 30 lines
arch/arm64/kernel/efi.c new file +30
-0+if (newmap ==NULL ) { +pr_err ("Failed to allocate new EFI memmap\n" ); +return ; + } + + count =remap_regions (newmap); +if (count <=0 ) { +pr_err ("Failed to remap EFI regions.\n" ); +return ; + } + + efi.memmap = &memmap; + + efi.systab = (__force void *)efi_lookup_mapped_addr (efi_system_table); +if (efi.systab) +set_bit (EFI_SYSTEM_TABLES , &arm_efi_facility); + +/* +* efi.systab->runtime is a pointer to something guaranteed by +* the UEFI specification to be 1:1 mapped. +*/ + runtime = (__force void *)efi_lookup_mapped_addr ((u64 )efi.systab->runtime); + +/* Call SetVirtualAddressMap with the physical address of the map */ + efi.set_virtual_address_map = runtime->set_virtual_address_map; + +/* boot time idmap_pg_dir is incomplete, so fill in missing parts */ +efi_setup_idmap (); + +cpu_switch_mm (idmap_pg_dir, &init_mm);
arch/arm64/kernel/efi.c line 474
Quoted 10 lines
arch/arm64/kernel/efi.c new file +10
-0+flush_tlb_all (); +flush_cache_all (); + + status = efi.set_virtual_address_map (count * memmap.desc_size, + memmap.desc_size, + memmap.desc_version, + newmap); +cpu_set_reserved_ttbr0 (); +flush_tlb_all (); +flush_cache_all ();
What is the point of cache flusing here? See my comment in the first patch about this not being guaranteed.
arch/arm64/kernel/efi.c line 476
Quoted 2 lines
arch/arm64/kernel/efi.c new file +2
-0+ +free_bootmem ((unsigned long )newmap, mapsize);
memblock.
Signature · 1 line
Catalin
Re: [PATCH 3/3] arm64: add EFI runtime services
Quoted 2 lines · Catalin Marinas
I lost track of the
early_ioremapstatus for arm/arm64? Was there more progress since October (I think)?
See the two patch series:
https://lkml.org/lkml/2013/11/25/474
and
https://lkml.org/lkml/2013/11/27/621
The latter early_ioremap depends on the first although there are no arm64 bits in the first. All of the arm64 early_ioremap stuff is in the latter.
--Mark
Catalin Marinas
maintainerRe: [PATCH 3/3] arm64: add EFI runtime services
Quoted 8 lines · Mark Salter
Quoted 2 lines · Catalin Marinas
I lost track of the
early_ioremapstatus for arm/arm64? Was there more progress since October (I think)?See the two patch series:
Quoted 7 lines
and
https://lkml.org/lkml/2013/11/27/621
The latter
early_ioremapdepends on the first although there are no arm64 bits in the first. All of the arm64early_ioremapstuff is in the latter.
Thanks. I was even cc'ed but too many emails in my inbox ;). I'll have a look.
Signature · 1 line
Catalin
Re: [PATCH 3/3] arm64: add EFI runtime services
In reply to Catalin MarinasQuoted 10 lines · Catalin Marinas
arch/arm64/include/asm/efi.h line 16
Quoted 6 lines · Mark Salter
arch/arm64/include/asm/efi.h new file +6
-0+ +#define efi_remap (cookie, size) __ioremap((cookie), (size), PAGE_KERNEL_EXEC) +#define efi_ioremap (cookie, size) __ioremap((cookie), (size), \ + __pgprot(PROT_DEVICE_nGnRE)) +#define efi_unmap (cookie) __iounmap((cookie)) +#define efi_iounmap (cookie) __iounmap((cookie))I prefer to use the ioremap_*() functions rather than the lower-level
__ioremap(). Theioremap_cache()should give us executable memory.
okay
Quoted 6 lines
Looking at the code, I think we have a bug with
ioremap_cache()usingMT_NORMALsince it doesn't have the shareability bit (Device memory does not require this on AArch64).PROT_DEFAULTshould change topgprot_default|PTE_DIRTY.
ah, yes.
Quoted 54 lines
Quoted 54 lines
arch/arm64/kernel/efi.c new file +54
-0+if (depth !=1 || + (strcmp (uname,"chosen" ) !=0 &&strcmp (uname,"chosen at 0" ) !=0 )) +return 0 ; + +pr_info ("Getting EFI parameters from FDT.\n" ); + + prop =of_get_flat_dt_prop (node,"linux,uefi-system-table" , &len); +if (!prop) { +pr_err ("No EFI system table in FDT\n" ); +return 0 ; + } + efi_system_table =of_read_ulong (prop, len/4 ); + + prop =of_get_flat_dt_prop (node,"linux,uefi-mmap-start" , &len); +if (!prop) { +pr_err ("No EFI memmap in FDT\n" ); +return 0 ; + } + memmap.phys_map = (void *)of_read_ulong (prop, len/4 ); + + prop =of_get_flat_dt_prop (node,"linux,uefi-mmap-size" , &len); +if (!prop) { +pr_err ("No EFI memmap size in FDT\n" ); +return 0 ; + } + size =of_read_ulong (prop, len/4 ); + memmap.map_end = memmap.phys_map + size; + +/* reserve memmap */ +memblock_reserve ((u64 )memmap.phys_map, size); + + prop =of_get_flat_dt_prop (node,"linux,uefi-mmap-desc-size" , &len); +if (!prop) { +pr_err ("No EFI memmap descriptor size in FDT\n" ); +return 0 ; + } + memmap.desc_size =of_read_ulong (prop, len/4 ); + + memmap.nr_map = size / memmap.desc_size; + + prop =of_get_flat_dt_prop (node,"linux,uefi-mmap-desc-ver" , &len); +if (!prop) { +pr_err ("No EFI memmap descriptor version in FDT\n" ); +return 0 ; + } + memmap.desc_version =of_read_ulong (prop, len/4 ); + +if (uefi_debug) { +pr_info (" EFI system table @ %p\n" , (void *)efi_system_table); +pr_info (" EFI mmap @ %p-%p\n" , memmap.phys_map, + memmap.map_end); +pr_info (" EFI mmap descriptor size = 0x%lx\n" , + memmap.desc_size); + pr_info(" EFI mmap descriptor version = 0x%lx\n",
Quoted 8 lines
arch/arm64/kernel/efi.c line 120
Quoted 5 lines
arch/arm64/kernel/efi.c new file +5
-0+ memmap.desc_version); + } + +return 1 ; +}Are these checks generic to other architectures? We may do with some helpers to avoid duplication.
That whole function could probably be shared.
Quoted 6 lines
Review
Quoted 3 lines
(quoted diff) +2
-0+ +#define PGD_END (&idmap_pg_dir[sizeof(idmap_pg_dir)/sizeof(pgd_t)])Just &idmap_pg_dir[PTRS_PER_PGD] would do (or
idmap_pg_dir+ARRAY_SIZE(idmap_pg_dir)).
Should have been ARRAY_SIZE. I didn't want to use PTRS_PER_PGD in case that changed.
Quoted 15 lines
arch/arm64/kernel/efi.c line 130
Quoted 9 lines
arch/arm64/kernel/efi.c new file +8
-0+#ifndef CONFIG_SMP +#define PMD_FLAGS (PMD_ATTRINDX(MT_NORMAL) | PMD_TYPE_SECT | PMD_SECT_AF) +#else +#define PMD_FLAGS (PMD_ATTRINDX(MT_NORMAL) | PMD_TYPE_SECT | PMD_SECT_AF | \ + PMD_SECT_S) +#endif + +static void __init create_idmap (unsigned long addr,unsigned long len)I would change this to use the existing
create_mapping()function which takes care of allocating pud/pmd/ptes. We shouldn't duplicate this functionality in two places.create_mapping()may need a slight change since it assumesswapper_pg_dirbut it's not much. It also usesmemblock_alloc()for early allocations.
I'll take a look at this.
Quoted 7 lines
arch/arm64/kernel/efi.c line 245
Quoted 4 lines
arch/arm64/kernel/efi.c new file +3
-0+static __init int memmap_walk (struct efi_memory_map *memmap, +int (*func)(efi_memory_desc_t *,void *), +void *private_data,int early)Is this generic enough as a common helper between arm and arm64 (and maybe x86)?
Probably not as-is, but one could be made.
Quoted 10 lines
Quoted 2 lines
+static
__initintreserve_region(efi_memory_desc_t *md, void *priv)[...]
Quoted 1 line
+static
__initvoidreserve_regions(void)[...]
Quoted 1 line
+static int
__initremap_region(efi_memory_desc_t *md, void *priv)[...]
Quoted 1 line
+static int
__initremap_regions(efi_memory_desc_t *map)Same here (I haven't looked at the other implementations).
Only in arm/arm64 patches and they are different.
Quoted 20 lines
Quoted 20 lines
arch/arm64/kernel/efi.c new file +19
-0+/* +* Called from setup_arch with interrupts disabled. +*/ +void __init efi_enter_virtual_mode (void ) +{ +void *newmap; +efi_status_t status; +u64 mapsize, total_freed =0 ; +int count; + +if (!efi_enabled (EFI_BOOT )) { +pr_info ("EFI services will not be available.\n" ); +return ; + } + +pr_info ("Remapping and enabling EFI services.\n" ); + + mapsize = memmap.map_end - memmap.phys_map; + memmap.map = (__force void *)ioremap_cache((phys_addr_t)memmap.phys_map,
Quoted 8 lines
arch/arm64/kernel/efi.c line 434
Quoted 5 lines
arch/arm64/kernel/efi.c new file +5
-0+ mapsize); + memmap.map_end = memmap.map + mapsize; + +/* Map the regions we reserved earlier */ + newmap =alloc_bootmem (mapsize);
memblock_alloc()(also check the other bootmem uses in this patch, arm64 is using memblock).
Okay, the bootmem function do use the memblock interface with some leak detection added, but I can use memblock directly.
Quoted 35 lines
Quoted 35 lines
arch/arm64/kernel/efi.c new file +34
-0+if (newmap ==NULL ) { +pr_err ("Failed to allocate new EFI memmap\n" ); +return ; + } + + count =remap_regions (newmap); +if (count <=0 ) { +pr_err ("Failed to remap EFI regions.\n" ); +return ; + } + + efi.memmap = &memmap; + + efi.systab = (__force void *)efi_lookup_mapped_addr (efi_system_table); +if (efi.systab) +set_bit (EFI_SYSTEM_TABLES , &arm_efi_facility); + +/* +* efi.systab->runtime is a pointer to something guaranteed by +* the UEFI specification to be 1:1 mapped. +*/ + runtime = (__force void *)efi_lookup_mapped_addr ((u64 )efi.systab->runtime); + +/* Call SetVirtualAddressMap with the physical address of the map */ + efi.set_virtual_address_map = runtime->set_virtual_address_map; + +/* boot time idmap_pg_dir is incomplete, so fill in missing parts */ +efi_setup_idmap (); + +cpu_switch_mm (idmap_pg_dir, &init_mm); +flush_tlb_all (); +flush_cache_all (); + + status = efi.set_virtual_address_map(count * memmap.desc_size,
Quoted 8 lines
arch/arm64/kernel/efi.c line 474
Quoted 6 lines
arch/arm64/kernel/efi.c new file +6
-0+ memmap.desc_size, + memmap.desc_version, + newmap); +cpu_set_reserved_ttbr0 (); +flush_tlb_all (); +flush_cache_all ();What is the point of cache flusing here?
Paranoia based on not knowing what the firmware will be caching. If it doesn't matter and there's nothing to worry about I can drop it.
Re: [PATCH 3/3] arm64: add EFI runtime services
In reply to Mark SalterApologies for late feedback.
arch/arm64/kernel/efi.c line 4
Quoted 10 lines · Mark Salter
arch/arm64/kernel/efi.c new file +4
-0@@ -0,0 +1,507 @@ +/* +* Extensible Firmware Interface +* +* Based on Extensible Firmware Interface Specification version 2.3.1
Actually, for arm64, we're relying on UEFI 2.4 or later.
arch/arm64/kernel/efi.c line 6
Quoted 2 lines
arch/arm64/kernel/efi.c new file +2
-0+* +* Copyright (C) 2013 Linaro Ltd.
And I'd say you've easily done enough here to motivate adding Red Hat.
Quoted 10 lines
arch/arm64/kernel/efi.c new file +10
-0+* +* Adapted for arm64 from arch/arm/kernel/efi.c code +*/ + +#include <linux/efi.h> +#include <linux/export.h> +#include <linux/memblock.h> +#include <linux/of.h> +#include <linux/of_fdt.h> +#include <linux/sched.h>
arch/arm64/kernel/efi.c line 26
Quoted 10 lines
arch/arm64/kernel/efi.c new file +10
-0+#include <linux/slab.h> +#include <linux/bootmem.h> + +#include <asm/cacheflush.h> +#include <asm/efi.h> +#include <asm/tlbflush.h> +#include <asm/mmu_context.h> + +#define efi_early_remap (a, b) \ + ((__force void *)early_ioremap((a), (b)))
Do we not want early_memremap() here, rather than early_ioremap()?
/
Leif
Re: [PATCH 3/3] arm64: add EFI runtime services
Quoted 15 lines · Leif Lindholm
Apologies for late feedback.
arch/arm64/kernel/efi.c line 4
Quoted 10 lines · Mark Salter
arch/arm64/kernel/efi.c new file +4
-0@@ -0,0 +1,507 @@ +/* +* Extensible Firmware Interface +* +* Based on Extensible Firmware Interface Specification version 2.3.1 Actually, for arm64, we're relying on UEFI 2.4 or later.
You're right.
Quoted 20 lines
arch/arm64/kernel/efi.c line 6
Quoted 3 lines
arch/arm64/kernel/efi.c new file +2
-0+* +* Copyright (C) 2013 Linaro Ltd. And I'd say you've easily done enough here to motivate adding Red Hat.
Quoted 14 lines
arch/arm64/kernel/efi.c new file +14
-0+* +* Adapted for arm64 from arch/arm/kernel/efi.c code +*/ + +#include <linux/efi.h> +#include <linux/export.h> +#include <linux/memblock.h> +#include <linux/of.h> +#include <linux/of_fdt.h> +#include <linux/sched.h> +#include <linux/slab.h> +#include <linux/bootmem.h> + +#include <asm/cacheflush.h>
Quoted 9 lines
arch/arm64/kernel/efi.c line 26
Quoted 6 lines
arch/arm64/kernel/efi.c new file +6
-0+#include <asm/efi.h> +#include <asm/tlbflush.h> +#include <asm/mmu_context.h> + +#define efi_early_remap (a, b) \ + ((__force void *)early_ioremap((a), (b)))Do we not want
early_memremap()here, rather thanearly_ioremap()?
Yes.