From 3cb4a4827596abc82e55b80364f509d0fefc3051 Mon Sep 17 00:00:00 2001 From: "KONDO KAZUMA(近藤　和真)" Date: Fri, 22 Mar 2024 10:47:02 +0000 Subject: efi/libstub: fix efi_random_alloc() to allocate memory at alloc_min or higher address Following warning is sometimes observed while booting my servers: [ 3.594838] DMA: preallocated 4096 KiB GFP_KERNEL pool for atomic allocations [ 3.602918] swapper/0: page allocation failure: order:10, mode:0xcc1(GFP_KERNEL|GFP_DMA), nodemask=(null),cpuset=/,mems_allowed=0-1 ... [ 3.851862] DMA: preallocated 1024 KiB GFP_KERNEL|GFP_DMA pool for atomic allocation If 'nokaslr' boot option is set, the warning always happens. On x86, ZONE_DMA is small zone at the first 16MB of physical address space. When this problem happens, most of that space seems to be used by decompressed kernel. Thereby, there is not enough space at DMA_ZONE to meet the request of DMA pool allocation. The commit 2f77465b05b1 ("x86/efistub: Avoid placing the kernel below LOAD_PHYSICAL_ADDR") tried to fix this problem by introducing lower bound of allocation. But the fix is not complete. efi_random_alloc() allocates pages by following steps. 1. Count total available slots ('total_slots') 2. Select a slot ('target_slot') to allocate randomly 3. Calculate a starting address ('target') to be included target_slot 4. Allocate pages, which starting address is 'target' In step 1, 'alloc_min' is used to offset the starting address of memory chunk. But in step 3 'alloc_min' is not considered at all. As the result, 'target' can be miscalculated and become lower than 'alloc_min'. When KASLR is disabled, 'target_slot' is always 0 and the problem happens everytime if the EFI memory map of the system meets the condition. Fix this problem by calculating 'target' considering 'alloc_min'. Cc: linux-efi@vger.kernel.org Cc: Tom Englund Cc: linux-kernel@vger.kernel.org Fixes: 2f77465b05b1 ("x86/efistub: Avoid placing the kernel below LOAD_PHYSICAL_ADDR") Signed-off-by: Kazuma Kondo Signed-off-by: Ard Biesheuvel --- drivers/firmware/efi/libstub/randomalloc.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/firmware/efi/libstub/randomalloc.c b/drivers/firmware/efi/libstub/randomalloc.c index 4e96a855fdf47b..7e185285955021 100644 --- a/drivers/firmware/efi/libstub/randomalloc.c +++ b/drivers/firmware/efi/libstub/randomalloc.c @@ -120,7 +120,7 @@ efi_status_t efi_random_alloc(unsigned long size, continue; } - target = round_up(md->phys_addr, align) + target_slot * align; + target = round_up(max(md->phys_addr, alloc_min), align) + target_slot * align; pages = size / EFI_PAGE_SIZE; status = efi_bs_call(allocate_pages, EFI_ALLOCATE_ADDRESS, -- cgit 1.2.3-korg From cefcd4fe2e3aaf792c14c9e56dab89e3d7a65d02 Mon Sep 17 00:00:00 2001 From: Ard Biesheuvel Date: Fri, 22 Mar 2024 17:03:58 +0200 Subject: x86/efistub: Call mixed mode boot services on the firmware's stack Normally, the EFI stub calls into the EFI boot services using the stack that was live when the stub was entered. According to the UEFI spec, this stack needs to be at least 128k in size - this might seem large but all asynchronous processing and event handling in EFI runs from the same stack and so quite a lot of space may be used in practice. In mixed mode, the situation is a bit different: the bootloader calls the 32-bit EFI stub entry point, which calls the decompressor's 32-bit entry point, where the boot stack is set up, using a fixed allocation of 16k. This stack is still in use when the EFI stub is started in 64-bit mode, and so all calls back into the EFI firmware will be using the decompressor's limited boot stack. Due to the placement of the boot stack right after the boot heap, any stack overruns have gone unnoticed. However, commit 5c4feadb0011983b ("x86/decompressor: Move global symbol references to C code") moved the definition of the boot heap into C code, and now the boot stack is placed right at the base of BSS, where any overruns will corrupt the end of the .data section. While it would be possible to work around this by increasing the size of the boot stack, doing so would affect all x86 systems, and mixed mode systems are a tiny (and shrinking) fraction of the x86 installed base. So instead, record the firmware stack pointer value when entering from the 32-bit firmware, and switch to this stack every time a EFI boot service call is made. Cc: # v6.1+ Signed-off-by: Ard Biesheuvel --- arch/x86/boot/compressed/efi_mixed.S | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/arch/x86/boot/compressed/efi_mixed.S b/arch/x86/boot/compressed/efi_mixed.S index f4e22ef774ab6b..719e939050cbfa 100644 --- a/arch/x86/boot/compressed/efi_mixed.S +++ b/arch/x86/boot/compressed/efi_mixed.S @@ -49,6 +49,11 @@ SYM_FUNC_START(startup_64_mixed_mode) lea efi32_boot_args(%rip), %rdx mov 0(%rdx), %edi mov 4(%rdx), %esi + + /* Switch to the firmware's stack */ + movl efi32_boot_sp(%rip), %esp + andl $~7, %esp + #ifdef CONFIG_EFI_HANDOVER_PROTOCOL mov 8(%rdx), %edx // saved bootparams pointer test %edx, %edx @@ -254,6 +259,9 @@ SYM_FUNC_START_LOCAL(efi32_entry) /* Store firmware IDT descriptor */ sidtl (efi32_boot_idt - 1b)(%ebx) + /* Store firmware stack pointer */ + movl %esp, (efi32_boot_sp - 1b)(%ebx) + /* Store boot arguments */ leal (efi32_boot_args - 1b)(%ebx), %ebx movl %ecx, 0(%ebx) @@ -318,5 +326,6 @@ SYM_DATA_END(efi32_boot_idt) SYM_DATA_LOCAL(efi32_boot_cs, .word 0) SYM_DATA_LOCAL(efi32_boot_ds, .word 0) +SYM_DATA_LOCAL(efi32_boot_sp, .long 0) SYM_DATA_LOCAL(efi32_boot_args, .long 0, 0, 0) SYM_DATA(efi_is64, .byte 1) -- cgit 1.2.3-korg From df7ecce842b846a04d087ba85fdb79a90e26a1b0 Mon Sep 17 00:00:00 2001 From: Ard Biesheuvel Date: Fri, 22 Mar 2024 17:01:45 +0100 Subject: x86/efistub: Don't clear BSS twice in mixed mode Clearing BSS should only be done once, at the very beginning. efi_pe_entry() is the entrypoint from the firmware, which may not clear BSS and so it is done explicitly. However, efi_pe_entry() is also used as an entrypoint by the mixed mode startup code, in which case BSS will already have been cleared, and doing it again at this point will corrupt global variables holding the firmware's GDT/IDT and segment selectors. So make the memset() conditional on whether the EFI stub is running in native mode. Fixes: b3810c5a2cc4a666 ("x86/efistub: Clear decompressor BSS in native EFI entrypoint") Signed-off-by: Ard Biesheuvel --- drivers/firmware/efi/libstub/x86-stub.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/firmware/efi/libstub/x86-stub.c b/drivers/firmware/efi/libstub/x86-stub.c index 2096ae09438e47..1edf93e638975c 100644 --- a/drivers/firmware/efi/libstub/x86-stub.c +++ b/drivers/firmware/efi/libstub/x86-stub.c @@ -476,7 +476,8 @@ efi_status_t __efiapi efi_pe_entry(efi_handle_t handle, efi_status_t status; char *cmdline_ptr; - memset(_bss, 0, _ebss - _bss); + if (efi_is_native()) + memset(_bss, 0, _ebss - _bss); efi_system_table = sys_table_arg; -- cgit 1.2.3-korg From 62b71cd73d41ddac6b1760402bbe8c4932e23531 Mon Sep 17 00:00:00 2001 From: Oleksandr Tymoshenko Date: Sat, 23 Mar 2024 06:33:33 +0000 Subject: efi: fix panic in kdump kernel Check if get_next_variable() is actually valid pointer before calling it. In kdump kernel this method is set to NULL that causes panic during the kexec-ed kernel boot. Tested with QEMU and OVMF firmware. Fixes: bad267f9e18f ("efi: verify that variable services are supported") Signed-off-by: Oleksandr Tymoshenko Signed-off-by: Ard Biesheuvel --- drivers/firmware/efi/efi.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/drivers/firmware/efi/efi.c b/drivers/firmware/efi/efi.c index 8859fb0b006d3b..fdf07dd6f4591d 100644 --- a/drivers/firmware/efi/efi.c +++ b/drivers/firmware/efi/efi.c @@ -203,6 +203,8 @@ static bool generic_ops_supported(void) name_size = sizeof(name); + if (!efi.get_next_variable) + return false; status = efi.get_next_variable(&name_size, &name, &guid); if (status == EFI_UNSUPPORTED) return false; -- cgit 1.2.3-korg