efi: libstub: Always pass screen_info via config table
Ard Biesheuvel
maintainerPATCH 1/1 efi: libstub: Always pass screen_info via config table
In some cases, we expose the kernel's struct screen_info to the EFI stub directly, so it gets populated before even entering the kernel. This means the early console is available as soon as the early param parsing happens, which is nice. It also means we need two different ways to pass this information, as this trick only works if the EFI stub is baked into the core kernel image, which is not always the case.
Huacai reports that the preparatory refactoring that was needed to implement this alternative method for zboot resulted in a non-functional efifb earlycon for other cases as well, due to the reordering of the kernel image relocation with the population of the screen_info struct, and the latter now takes place after copying the image to its new location, which means we copy the old, uninitialized state.
To fix this and simplify things at the same time, let's just always use the config table method.
--- arch/arm64/kernel/image-vars.h | 1 - arch/loongarch/kernel/image-vars.h | 1 - arch/riscv/kernel/image-vars.h | 1 - drivers/firmware/efi/libstub/efi-stub.c | 9 --------- drivers/firmware/efi/libstub/screen_info.c | 19 ------------------- 5 files changed, 31 deletions(-)
arch/arm64/kernel/image-vars.h +0 -1
@@ -27,7 +27,6 @@ PROVIDE(__efistub__text = _text);PROVIDE (__efistub__end = _end);PROVIDE (__efistub___inittext_end = __inittext_end);PROVIDE (__efistub__edata = _edata); -PROVIDE (__efistub_screen_info = screen_info);PROVIDE (__efistub__ctype = _ctype);PROVIDE (__pi___memcpy = __pi_memcpy);
arch/loongarch/kernel/image-vars.h +0 -1
@@ -12,7 +12,6 @@ __efistub_kernel_entry = kernel_entry; __efistub_kernel_asize = kernel_asize; __efistub_kernel_fsize = kernel_fsize; __efistub_kernel_offset = kernel_offset; -__efistub_screen_info = screen_info;#endif
arch/riscv/kernel/image-vars.h +0 -1
@@ -29,7 +29,6 @@ __efistub__start = _start; __efistub__start_kernel = _start_kernel; __efistub__end = _end; __efistub__edata = _edata; -__efistub_screen_info = screen_info;#endif
drivers/firmware/efi/libstub/efi-stub.c +0 -9
@@ -47,15 +47,6 @@static u64 virtmap_base =EFI_RT_VIRTUAL_BASE ;static bool flat_va_mapping = (EFI_RT_VIRTUAL_OFFSET !=0 ); -struct screen_info *__weak alloc_screen_info (void ) -{ -return &screen_info; -} - -void __weak free_screen_info (struct screen_info *si) -{ -} -static struct screen_info *setup_graphics (void ) {efi_guid_t gop_proto =EFI_GRAPHICS_OUTPUT_PROTOCOL_GUID ;
drivers/firmware/efi/libstub/screen_info.c +0 -19
@@ -5,25 +5,6 @@#include "efistub.h" -/* -* There are two ways of populating the core kernel's struct screen_info via the stub: -* - using a configuration table, like below, which relies on the EFI init code -* to locate the table and copy the contents; -* - by linking directly to the core kernel's copy of the global symbol. -* -* The latter is preferred because it makes the EFIFB earlycon available very -* early, but it only works if the EFI stub is part of the core kernel image -* itself. The zboot decompressor can only use the configuration table -* approach. -* -* In order to support both methods from the same build of the EFI stub -* library, provide this dummy global definition of struct screen_info. If it -* is required to satisfy a link dependency, it means we need to override the -* __weak alloc and free methods with the ones below, and those will be pulled -* in as well. -*/ -struct screen_info screen_info; -static efi_guid_t screen_info_guid =LINUX_EFI_SCREEN_INFO_TABLE_GUID ;struct screen_info *alloc_screen_info (void )
Signature · 1 line
2.39.2
Ard Biesheuvel
maintainerQuoted 18 lines · Ard Biesheuvel
In some cases, we expose the kernel's struct
screen_infoto the EFI stub directly, so it gets populated before even entering the kernel. This means the early console is available as soon as the early param parsing happens, which is nice. It also means we need two different ways to pass this information, as this trick only works if the EFI stub is baked into the core kernel image, which is not always the case.Huacai reports that the preparatory refactoring that was needed to implement this alternative method for zboot resulted in a non-functional efifb earlycon for other cases as well, due to the reordering of the kernel image relocation with the population of the
screen_infostruct, and the latter now takes place after copying the image to its new location, which means we copy the old, uninitialized state.To fix this and simplify things at the same time, let's just always use the config table method.
Quoted 8 lines
Cc Xuefeng Li <[email protected]>Cc Xuerui Wang <[email protected]>Reported-by Huacai Chen <[email protected]>Fixes 42c8ea3dca094ab8 ("efi: libstub: Factor out EFI stub entrypoint into separate file")Signed-off-by Ard Biesheuvel <[email protected]>
Please disregard - I think I can simplify this without losing the very early earlycon.
Quoted 97 lines
--- arch/arm64/kernel/image-vars.h | 1 - arch/loongarch/kernel/image-vars.h | 1 - arch/riscv/kernel/image-vars.h | 1 - drivers/firmware/efi/libstub/efi-stub.c | 9 --------- drivers/firmware/efi/libstub/screen_info.c | 19 ------------------- 5 files changed, 31 deletions(-)arch/arm64/kernel/image-vars.h +0
-1@@ -27,7 +27,6 @@ PROVIDE(__efistub__text = _text);PROVIDE (__efistub__end = _end);PROVIDE (__efistub___inittext_end = __inittext_end);PROVIDE (__efistub__edata = _edata); -PROVIDE (__efistub_screen_info = screen_info);PROVIDE (__efistub__ctype = _ctype);PROVIDE (__pi___memcpy = __pi_memcpy);arch/loongarch/kernel/image-vars.h +0
-1@@ -12,7 +12,6 @@ __efistub_kernel_entry = kernel_entry; __efistub_kernel_asize = kernel_asize; __efistub_kernel_fsize = kernel_fsize; __efistub_kernel_offset = kernel_offset; -__efistub_screen_info = screen_info;#endif arch/riscv/kernel/image-vars.h +0
-1@@ -29,7 +29,6 @@ __efistub__start = _start; __efistub__start_kernel = _start_kernel; __efistub__end = _end; __efistub__edata = _edata; -__efistub_screen_info = screen_info;#endif drivers/firmware/efi/libstub/efi-stub.c +0
-9@@ -47,15 +47,6 @@static u64 virtmap_base =EFI_RT_VIRTUAL_BASE ;static bool flat_va_mapping = (EFI_RT_VIRTUAL_OFFSET !=0 ); -struct screen_info *__weak alloc_screen_info (void ) -{ -return &screen_info; -} - -void __weak free_screen_info (struct screen_info *si) -{ -} -static struct screen_info *setup_graphics (void ) {efi_guid_t gop_proto =EFI_GRAPHICS_OUTPUT_PROTOCOL_GUID ;drivers/firmware/efi/libstub/screen_info.c +0
-20@@ -5,25 +5,6 @@#include "efistub.h" -/* -* There are two ways of populating the core kernel's struct screen_info via the stub: -* - using a configuration table, like below, which relies on the EFI init code -* to locate the table and copy the contents; -* - by linking directly to the core kernel's copy of the global symbol. -* -* The latter is preferred because it makes the EFIFB earlycon available very -* early, but it only works if the EFI stub is part of the core kernel image -* itself. The zboot decompressor can only use the configuration table -* approach. -* -* In order to support both methods from the same build of the EFI stub -* library, provide this dummy global definition of struct screen_info. If it -* is required to satisfy a link dependency, it means we need to override the -* __weak alloc and free methods with the ones below, and those will be pulled -* in as well. -*/ -struct screen_info screen_info; -static efi_guid_t screen_info_guid =LINUX_EFI_SCREEN_INFO_TABLE_GUID ;struct screen_info *alloc_screen_info (void ) --2.39.2
Huacai Chen
maintainerHi, Ard,
I'm a bit confused, It seems this patch cannot solve the problem reported in [1] for LoongArch.
[1] https://lore.kernel.org/linux-efi/[email protected]/
Huacai
Quoted 131 lines · Ard Biesheuvel
Quoted 18 lines · Ard Biesheuvel
In some cases, we expose the kernel's struct
screen_infoto the EFI stub directly, so it gets populated before even entering the kernel. This means the early console is available as soon as the early param parsing happens, which is nice. It also means we need two different ways to pass this information, as this trick only works if the EFI stub is baked into the core kernel image, which is not always the case.Huacai reports that the preparatory refactoring that was needed to implement this alternative method for zboot resulted in a non-functional efifb earlycon for other cases as well, due to the reordering of the kernel image relocation with the population of the
screen_infostruct, and the latter now takes place after copying the image to its new location, which means we copy the old, uninitialized state.To fix this and simplify things at the same time, let's just always use the config table method.
Quoted 8 lines
Cc Xuefeng Li <[email protected]>Cc Xuerui Wang <[email protected]>Reported-by Huacai Chen <[email protected]>Fixes 42c8ea3dca094ab8 ("efi: libstub: Factor out EFI stub entrypoint into separate file")Signed-off-by Ard Biesheuvel <[email protected]>Please disregard - I think I can simplify this without losing the very early earlycon.
Quoted 97 lines
--- arch/arm64/kernel/image-vars.h | 1 - arch/loongarch/kernel/image-vars.h | 1 - arch/riscv/kernel/image-vars.h | 1 - drivers/firmware/efi/libstub/efi-stub.c | 9 --------- drivers/firmware/efi/libstub/screen_info.c | 19 ------------------- 5 files changed, 31 deletions(-)arch/arm64/kernel/image-vars.h +0
-1@@ -27,7 +27,6 @@ PROVIDE(__efistub__text = _text);PROVIDE (__efistub__end = _end);PROVIDE (__efistub___inittext_end = __inittext_end);PROVIDE (__efistub__edata = _edata); -PROVIDE (__efistub_screen_info = screen_info);PROVIDE (__efistub__ctype = _ctype);PROVIDE (__pi___memcpy = __pi_memcpy);arch/loongarch/kernel/image-vars.h +0
-1@@ -12,7 +12,6 @@ __efistub_kernel_entry = kernel_entry; __efistub_kernel_asize = kernel_asize; __efistub_kernel_fsize = kernel_fsize; __efistub_kernel_offset = kernel_offset; -__efistub_screen_info = screen_info;#endif arch/riscv/kernel/image-vars.h +0
-1@@ -29,7 +29,6 @@ __efistub__start = _start; __efistub__start_kernel = _start_kernel; __efistub__end = _end; __efistub__edata = _edata; -__efistub_screen_info = screen_info;#endif drivers/firmware/efi/libstub/efi-stub.c +0
-9@@ -47,15 +47,6 @@static u64 virtmap_base =EFI_RT_VIRTUAL_BASE ;static bool flat_va_mapping = (EFI_RT_VIRTUAL_OFFSET !=0 ); -struct screen_info *__weak alloc_screen_info (void ) -{ -return &screen_info; -} - -void __weak free_screen_info (struct screen_info *si) -{ -} -static struct screen_info *setup_graphics (void ) {efi_guid_t gop_proto =EFI_GRAPHICS_OUTPUT_PROTOCOL_GUID ;drivers/firmware/efi/libstub/screen_info.c +0
-20@@ -5,25 +5,6 @@#include "efistub.h" -/* -* There are two ways of populating the core kernel's struct screen_info via the stub: -* - using a configuration table, like below, which relies on the EFI init code -* to locate the table and copy the contents; -* - by linking directly to the core kernel's copy of the global symbol. -* -* The latter is preferred because it makes the EFIFB earlycon available very -* early, but it only works if the EFI stub is part of the core kernel image -* itself. The zboot decompressor can only use the configuration table -* approach. -* -* In order to support both methods from the same build of the EFI stub -* library, provide this dummy global definition of struct screen_info. If it -* is required to satisfy a link dependency, it means we need to override the -* __weak alloc and free methods with the ones below, and those will be pulled -* in as well. -*/ -struct screen_info screen_info; -static efi_guid_t screen_info_guid =LINUX_EFI_SCREEN_INFO_TABLE_GUID ;struct screen_info *alloc_screen_info (void ) --2.39.2
Ard Biesheuvel
maintainerQuoted 7 lines · Huacai Chen
Hi, Ard,
I'm a bit confused, It seems this patch cannot solve the problem reported in [1] for LoongArch.
[1] https://lore.kernel.org/linux-efi/[email protected]/
The intent was to always pass the screen_info contents via the configuration table, so I am surprised that this does not work.
However, I am preparing a v2 that preserves the old behavior - I'll send it out in a minute.
Quoted 133 lines · Ard Biesheuvel
Quoted 131 lines · Ard Biesheuvel
Quoted 18 lines
In some cases, we expose the kernel's struct
screen_infoto the EFI stub directly, so it gets populated before even entering the kernel. This means the early console is available as soon as the early param parsing happens, which is nice. It also means we need two different ways to pass this information, as this trick only works if the EFI stub is baked into the core kernel image, which is not always the case.Huacai reports that the preparatory refactoring that was needed to implement this alternative method for zboot resulted in a non-functional efifb earlycon for other cases as well, due to the reordering of the kernel image relocation with the population of the
screen_infostruct, and the latter now takes place after copying the image to its new location, which means we copy the old, uninitialized state.To fix this and simplify things at the same time, let's just always use the config table method.
Quoted 8 lines
Cc Xuefeng Li <[email protected]>Cc Xuerui Wang <[email protected]>Reported-by Huacai Chen <[email protected]>Fixes 42c8ea3dca094ab8 ("efi: libstub: Factor out EFI stub entrypoint into separate file")Signed-off-by Ard Biesheuvel <[email protected]>Please disregard - I think I can simplify this without losing the very early earlycon.
Quoted 97 lines
--- arch/arm64/kernel/image-vars.h | 1 - arch/loongarch/kernel/image-vars.h | 1 - arch/riscv/kernel/image-vars.h | 1 - drivers/firmware/efi/libstub/efi-stub.c | 9 --------- drivers/firmware/efi/libstub/screen_info.c | 19 ------------------- 5 files changed, 31 deletions(-)arch/arm64/kernel/image-vars.h +0
-1@@ -27,7 +27,6 @@ PROVIDE(__efistub__text = _text);PROVIDE (__efistub__end = _end);PROVIDE (__efistub___inittext_end = __inittext_end);PROVIDE (__efistub__edata = _edata); -PROVIDE (__efistub_screen_info = screen_info);PROVIDE (__efistub__ctype = _ctype);PROVIDE (__pi___memcpy = __pi_memcpy);arch/loongarch/kernel/image-vars.h +0
-1@@ -12,7 +12,6 @@ __efistub_kernel_entry = kernel_entry; __efistub_kernel_asize = kernel_asize; __efistub_kernel_fsize = kernel_fsize; __efistub_kernel_offset = kernel_offset; -__efistub_screen_info = screen_info;#endif arch/riscv/kernel/image-vars.h +0
-1@@ -29,7 +29,6 @@ __efistub__start = _start; __efistub__start_kernel = _start_kernel; __efistub__end = _end; __efistub__edata = _edata; -__efistub_screen_info = screen_info;#endif drivers/firmware/efi/libstub/efi-stub.c +0
-9@@ -47,15 +47,6 @@static u64 virtmap_base =EFI_RT_VIRTUAL_BASE ;static bool flat_va_mapping = (EFI_RT_VIRTUAL_OFFSET !=0 ); -struct screen_info *__weak alloc_screen_info (void ) -{ -return &screen_info; -} - -void __weak free_screen_info (struct screen_info *si) -{ -} -static struct screen_info *setup_graphics (void ) {efi_guid_t gop_proto =EFI_GRAPHICS_OUTPUT_PROTOCOL_GUID ;drivers/firmware/efi/libstub/screen_info.c +0
-20@@ -5,25 +5,6 @@#include "efistub.h" -/* -* There are two ways of populating the core kernel's struct screen_info via the stub: -* - using a configuration table, like below, which relies on the EFI init code -* to locate the table and copy the contents; -* - by linking directly to the core kernel's copy of the global symbol. -* -* The latter is preferred because it makes the EFIFB earlycon available very -* early, but it only works if the EFI stub is part of the core kernel image -* itself. The zboot decompressor can only use the configuration table -* approach. -* -* In order to support both methods from the same build of the EFI stub -* library, provide this dummy global definition of struct screen_info. If it -* is required to satisfy a link dependency, it means we need to override the -* __weak alloc and free methods with the ones below, and those will be pulled -* in as well. -*/ -struct screen_info screen_info; -static efi_guid_t screen_info_guid =LINUX_EFI_SCREEN_INFO_TABLE_GUID ;struct screen_info *alloc_screen_info (void ) --2.39.2
WANG Xuerui
reviewerHi,
Quoted 25 lines · Ard Biesheuvel
In some cases, we expose the kernel's struct
screen_infoto the EFI stub directly, so it gets populated before even entering the kernel. This means the early console is available as soon as the early param parsing happens, which is nice. It also means we need two different ways to pass this information, as this trick only works if the EFI stub is baked into the core kernel image, which is not always the case.Huacai reports that the preparatory refactoring that was needed to implement this alternative method for zboot resulted in a non-functional efifb earlycon for other cases as well, due to the reordering of the kernel image relocation with the population of the
screen_infostruct, and the latter now takes place after copying the image to its new location, which means we copy the old, uninitialized state.To fix this and simplify things at the same time, let's just always use the config table method.
Cc Xuefeng Li <[email protected]>Cc Xuerui Wang <[email protected]>Reported-by Huacai Chen <[email protected]>Fixes 42c8ea3dca094ab8 ("efi: libstub: Factor out EFI stub entrypoint into separate file")Signed-off-by Ard Biesheuvel <[email protected]>
Quoted 8 lines
--- arch/arm64/kernel/image-vars.h | 1 - arch/loongarch/kernel/image-vars.h | 1 - arch/riscv/kernel/image-vars.h | 1 - drivers/firmware/efi/libstub/efi-stub.c | 9 --------- drivers/firmware/efi/libstub/screen_info.c | 19 ------------------- 5 files changed, 31 deletions(-)
The cleanup looks good to me, thanks!
I'm currently unable to verify though, because my loongarch box currently has an RX 6400 plugged in which won't get initialized by the firmware (likely related to newer x86 emulation capability required by the card's VBIOS) so I can't see the efifb. I'll swap that card out maybe this week to test this (and the other "Reprobe after parsing config tables" patch) though!
Signature · 3 lines
WANG "xen0n" Xuerui Linux/LoongArch mailing list: https://lore.kernel.org/loongarch/