kernel/linux.git/drivers/firmware/efi, branch v6.19.11

x86/efi: defer freeing of boot services memory

2026-03-12T11:09:30+00:00

commit a4b0bf6a40f3c107c67a24fbc614510ef5719980 upstream. efi_free_boot_services() frees memory occupied by EFI_BOOT_SERVICES_CODE and EFI_BOOT_SERVICES_DATA using memblock_free_late(). There are two issue with that: memblock_free_late() should be used for memory allocated with memblock_alloc() while the memory reserved with memblock_reserve() should be freed with free_reserved_area(). More acutely, with CONFIG_DEFERRED_STRUCT_PAGE_INIT=y efi_free_boot_services() is called before deferred initialization of the memory map is complete. Benjamin Herrenschmidt reports that this causes a leak of ~140MB of RAM on EC2 t3a.nano instances which only have 512MB or RAM. If the freed memory resides in the areas that memory map for them is still uninitialized, they won't be actually freed because memblock_free_late() calls memblock_free_pages() and the latter skips uninitialized pages. Using free_reserved_area() at this point is also problematic because __free_page() accesses the buddy of the freed page and that again might end up in uninitialized part of the memory map. Delaying the entire efi_free_boot_services() could be problematic because in addition to freeing boot services memory it updates efi.memmap without any synchronization and that's undesirable late in boot when there is concurrency. More robust approach is to only defer freeing of the EFI boot services memory. Split efi_free_boot_services() in two. First efi_unmap_boot_services() collects ranges that should be freed into an array then efi_free_boot_services() later frees them after deferred init is complete. Link: https://lore.kernel.org/all/ec2aaef14783869b3be6e3c253b2dcbf67dbc12a.camel@kernel.crashing.org Fixes: 916f676f8dc0 ("x86, efi: Retain boot service code until after switching to virtual mode") Cc: Signed-off-by: Mike Rapoport (Microsoft) Reviewed-by: Benjamin Herrenschmidt Signed-off-by: Ard Biesheuvel Signed-off-by: Greg Kroah-Hartman

EFI/CPER: don't go past the ARM processor CPER record buffer

2026-03-04T12:19:37+00:00

[ Upstream commit eae21beecb95a3b69ee5c38a659f774e171d730e ] There's a logic inside GHES/CPER to detect if the section_length is too small, but it doesn't detect if it is too big. Currently, if the firmware receives an ARM processor CPER record stating that a section length is big, kernel will blindly trust section_length, producing a very long dump. For instance, a 67 bytes record with ERR_INFO_NUM set 46198 and section length set to 854918320 would dump a lot of data going a way past the firmware memory-mapped area. Fix it by adding a logic to prevent it to go past the buffer if ERR_INFO_NUM is too big, making it report instead: [Hardware Error]: Hardware error from APEI Generic Hardware Error Source: 1 [Hardware Error]: event severity: recoverable [Hardware Error]: Error 0, type: recoverable [Hardware Error]: section_type: ARM processor error [Hardware Error]: MIDR: 0xff304b2f8476870a [Hardware Error]: section length: 854918320, CPER size: 67 [Hardware Error]: section length is too big [Hardware Error]: firmware-generated error record is incorrect [Hardware Error]: ERR_INFO_NUM is 46198 Signed-off-by: Mauro Carvalho Chehab Reviewed-by: Jonathan Cameron Acked-by: Ard Biesheuvel Reviewed-by: Hanjun Guo [ rjw: Subject and changelog tweaks ] Link: https://patch.msgid.link/41cd9f6b3ace3cdff7a5e864890849e4b1c58b63.1767871950.git.mchehab+huawei@kernel.org Signed-off-by: Rafael J. Wysocki Signed-off-by: Sasha Levin

EFI/CPER: don't dump the entire memory region

2026-03-04T12:19:37+00:00

[ Upstream commit 55cc6fe5716f678f06bcb95140882dfa684464ec ] The current logic at cper_print_fw_err() doesn't check if the error record length is big enough to handle offset. On a bad firmware, if the ofset is above the actual record, length -= offset will underflow, making it dump the entire memory. The end result can be: - the logic taking a lot of time dumping large regions of memory; - data disclosure due to the memory dumps; - an OOPS, if it tries to dump an unmapped memory region. Fix it by checking if the section length is too small before doing a hex dump. Signed-off-by: Mauro Carvalho Chehab Reviewed-by: Jonathan Cameron Acked-by: Ard Biesheuvel Reviewed-by: Hanjun Guo [ rjw: Subject tweaks ] Link: https://patch.msgid.link/1752b5ba63a3e2f148ddee813b36c996cc617e86.1767871950.git.mchehab+huawei@kernel.org Signed-off-by: Rafael J. Wysocki Signed-off-by: Sasha Levin

efi: Fix reservation of unaccepted memory table

2026-02-26T23:01:37+00:00

[ Upstream commit 0862438c90487e79822d5647f854977d50381505 ] The reserve_unaccepted() function incorrectly calculates the size of the memblock reservation for the unaccepted memory table. It aligns the size of the table, but fails to account for cases where the table's starting physical address (efi.unaccepted) is not page-aligned. If the table starts at an offset within a page and its end crosses into a subsequent page that the aligned size does not cover, the end of the table will not be reserved. This can lead to the table being overwritten or inaccessible, causing a kernel panic in accept_memory(). This issue was observed when starting Intel TDX VMs with specific memory sizes (e.g., > 64GB). Fix this by calculating the end address first (including the unaligned start) and then aligning it up, ensuring the entire range is covered by the reservation. Fixes: 8dbe33956d96 ("efi/unaccepted: Make sure unaccepted table is mapped") Reported-by: Moritz Sanft Signed-off-by: Kiryl Shutsemau (Meta) Reviewed-by: Tom Lendacky Acked-by: Mike Rapoport (Microsoft) Signed-off-by: Ard Biesheuvel Signed-off-by: Sasha Levin

mm: rename cpu_bitmap field to flexible_array

2026-01-19T20:30:00+00:00

The cpu_bitmap flexible array now contains more than just the cpu_bitmap. In preparation for changing the static mm_struct definitions to cover for the additional space required, change the cpu_bitmap type from "unsigned long" to "char", require an unsigned long alignment of the flexible array, and rename the field from "cpu_bitmap" to "flexible_array". Introduce the MM_STRUCT_FLEXIBLE_ARRAY_INIT macro to statically initialize the flexible array. This covers the init_mm and efi_mm static definitions. This is a preparation step for fixing the missing mm_cid size for static mm_struct definitions. Link: https://lkml.kernel.org/r/20251224173358.647691-3-mathieu.desnoyers@efficios.com Fixes: af7f588d8f73 ("sched: Introduce per-memory-map concurrency ID") Signed-off-by: Mathieu Desnoyers Reviewed-by: Thomas Gleixner Cc: Mark Brown Cc: Aboorva Devarajan Cc: Al Viro Cc: Baolin Wang Cc: Christan König Cc: Christian Brauner Cc: Christoph Lameter Cc: David Hildenbrand Cc: David Rientjes Cc: Dennis Zhou Cc: Johannes Weiner Cc: "Liam R . Howlett" Cc: Lorenzo Stoakes Cc: Martin Liu Cc: Masami Hiramatsu Cc: Mateusz Guzik Cc: Matthew Wilcox Cc: Miaohe Lin Cc: Michal Hocko Cc: Mike Rapoport Cc: "Paul E. McKenney" Cc: Roman Gushchin Cc: SeongJae Park Cc: Shakeel Butt Cc: Steven Rostedt Cc: Suren Baghdasaryan Cc: Sweet Tea Dorminy Cc: Tejun Heo Cc: Vlastimil Babka Cc: Wei Yang Cc: Yu Zhao Cc: Peter Zijlstra (Intel) Cc: Signed-off-by: Andrew Morton

efi/cper: Fix cper_bits_to_str buffer handling and return value

2026-01-14T10:34:42+00:00

The return value calculation was incorrect: `return len - buf_size;` Initially `len = buf_size`, then `len` decreases with each operation. This results in a negative return value on success. Fix by returning `buf_size - len` which correctly calculates the actual number of bytes written. Fixes: a976d790f494 ("efi/cper: Add a new helper function to print bitmasks") Signed-off-by: Morduan Zang Signed-off-by: Ard Biesheuvel

efi: Wipe INITRD config table from memory after consumption

2026-01-08T11:41:38+00:00

When the EFI stub itself loads the initrd and puts it in memory (rather than simply passing on a struct boot_params or device tree that already carries initrd information), it exposes this information to the core kernel via a INITRD configuration table. Given that config tables are preserved across kexec, this means that subsequent kexec boots will observe the same information, even though it most likely has become stale by that point. On x86, this information is usually superseded by the initrd info passed via bootparams, in which case this stale information is simply ignored. However, when performing a kexec boot without passing an initrd, the loader falls back to this stale information and explodes. So wipe the base and size from the INITRD config table as soon as it has been consumed. This fixes the issue for kexec on all EFI architectures. Reported-by: James Le Cuirot Tested-by: James Le Cuirot Acked-by: H. Peter Anvin (Intel) Link: https://lore.kernel.org/all/20251126173209.374755-2-chewi@gentoo.org Signed-off-by: Ard Biesheuvel

arm64: efi: Fix NULL pointer dereference by initializing user_ns

2025-12-24T20:32:57+00:00

Linux 6.19-rc2 (9448598b22c5 ("Linux 6.19-rc2")) is crashing with a NULL pointer dereference on arm64 hosts: Unable to handle kernel NULL pointer dereference at virtual address 00000000000000c8 pc : cap_capable (security/commoncap.c:82 security/commoncap.c:128) Call trace: cap_capable (security/commoncap.c:82 security/commoncap.c:128) (P) security_capable (security/security.c:?) ns_capable_noaudit (kernel/capability.c:342 kernel/capability.c:381) __ptrace_may_access (./include/linux/rcupdate.h:895 kernel/ptrace.c:326) ptrace_may_access (kernel/ptrace.c:353) do_task_stat (fs/proc/array.c:467) proc_tgid_stat (fs/proc/array.c:673) proc_single_show (fs/proc/base.c:803) I've bissected the problem to commit a5baf582f4c0 ("arm64/efi: Call EFI runtime services without disabling preemption"). >From my analyzes, the crash occurs because efi_mm lacks a user_ns field initialization. This was previously harmless, but commit a5baf582f4c0 ("arm64/efi: Call EFI runtime services without disabling preemption") changed the EFI runtime call path to use kthread_use_mm(&efi_mm), which temporarily adopts efi_mm as the current mm for the calling kthread. When a thread has an active mm, LSM hooks like cap_capable() expect mm->user_ns to be valid for credential checks. With efi_mm.user_ns being NULL, capability checks during possible /proc access dereference the NULL pointer and crash. Fix by initializing efi_mm.user_ns to &init_user_ns. Fixes: a5baf582f4c0 ("arm64/efi: Call EFI runtime services without disabling preemption") Signed-off-by: Breno Leitao Acked-by: Rik van Riel Signed-off-by: Ard Biesheuvel

efi/libstub: gop: Fix EDID support in mixed-mode

2025-12-24T20:32:31+00:00

The efi_edid_discovered_protocol and efi_edid_active_protocol have mixed mode fields. So all their attributes should be accessed through the efi_table_attr() helper. Doing so fixes the upper 32 bits of the 64 bit gop_edid pointer getting set to random values (followed by a crash at boot) when booting a x86_64 kernel on a machine with 32 bit UEFI like the Asus T100TA. Fixes: 17029cdd8f9d ("efi/libstub: gop: Add support for reading EDID") Cc: Thomas Zimmermann Cc: Javier Martinez Canillas Signed-off-by: Hans de Goede Signed-off-by: Ard Biesheuvel

Merge tag 'loongarch-6.19' of git://git.kernel.org/pub/scm/linux/kernel/git/chenhuacai/linux-loongson

2025-12-12T17:44:03+00:00

Pull LoongArch updates from Huacai Chen: - Add basic LoongArch32 support Note: Build infrastructures of LoongArch32 are not enabled yet, because we need to adjust irqchip drivers and wait for GNU toolchain be upstream first. - Select HAVE_ARCH_BITREVERSE in Kconfig - Fix build and boot for CONFIG_RANDSTRUCT - Correct the calculation logic of thread_count - Some bug fixes and other small changes * tag 'loongarch-6.19' of git://git.kernel.org/pub/scm/linux/kernel/git/chenhuacai/linux-loongson: (22 commits) LoongArch: Adjust default config files for 32BIT/64BIT LoongArch: Adjust VDSO/VSYSCALL for 32BIT/64BIT LoongArch: Adjust misc routines for 32BIT/64BIT LoongArch: Adjust user accessors for 32BIT/64BIT LoongArch: Adjust system call for 32BIT/64BIT LoongArch: Adjust module loader for 32BIT/64BIT LoongArch: Adjust time routines for 32BIT/64BIT LoongArch: Adjust process management for 32BIT/64BIT LoongArch: Adjust memory management for 32BIT/64BIT LoongArch: Adjust boot & setup for 32BIT/64BIT LoongArch: Adjust common macro definitions for 32BIT/64BIT LoongArch: Add adaptive CSR accessors for 32BIT/64BIT LoongArch: Add atomic operations for 32BIT/64BIT LoongArch: Add new PCI ID for pci_fixup_vgadev() LoongArch: Add and use some macros for AVEC LoongArch: Correct the calculation logic of thread_count LoongArch: Use unsigned long for _end and _text LoongArch: Use __pmd()/__pte() for swap entry conversions LoongArch: Fix arch_dup_task_struct() for CONFIG_RANDSTRUCT LoongArch: Fix build errors for CONFIG_RANDSTRUCT ...