kernel/git/stable/linux-stable.git/drivers/firmware, branch linux-6.7.y

x86/efistub: Reinstate soft limit for initrd loading

2024-04-03T13:11:47+00:00

commit decd347c2a75d32984beb8807d470b763a53b542 upstream. Commit 8117961d98fb2 ("x86/efi: Disregard setup header of loaded image") dropped the memcopy of the image's setup header into the boot_params struct provided to the core kernel, on the basis that EFI boot does not need it and should rely only on a single protocol to interface with the boot chain. It is also a prerequisite for being able to increase the section alignment to 4k, which is needed to enable memory protections when running in the boot services. So only the setup_header fields that matter to the core kernel are populated explicitly, and everything else is ignored. One thing was overlooked, though: the initrd_addr_max field in the setup_header is not used by the core kernel, but it is used by the EFI stub itself when it loads the initrd, where its default value of INT_MAX is used as the soft limit for memory allocation. This means that, in the old situation, the initrd was virtually always loaded in the lower 2G of memory, but now, due to initrd_addr_max being 0x0, the initrd may end up anywhere in memory. This should not be an issue principle, as most systems can deal with this fine. However, it does appear to tickle some problems in older UEFI implementations, where the memory ends up being corrupted, resulting in errors when unpacking the initramfs. So set the initrd_addr_max field to INT_MAX like it was before. Fixes: 8117961d98fb2 ("x86/efi: Disregard setup header of loaded image") Reported-by: Radek Podgorny Closes: https://lore.kernel.org/all/a99a831a-8ad5-4cb0-bff9-be637311f771@podgorny.cz Signed-off-by: Ard Biesheuvel Signed-off-by: Greg Kroah-Hartman

efi/libstub: Cast away type warning in use of max()

2024-04-03T13:11:47+00:00

commit 61d130f261a3c15ae2c4b6f3ac3517d5d5b78855 upstream. Avoid a type mismatch warning in max() by switching to max_t() and providing the type explicitly. Fixes: 3cb4a4827596abc82e ("efi/libstub: fix efi_random_alloc() ...") Signed-off-by: Ard Biesheuvel Signed-off-by: Greg Kroah-Hartman

efi: fix panic in kdump kernel

2024-04-03T13:11:45+00:00

[ Upstream commit 62b71cd73d41ddac6b1760402bbe8c4932e23531 ] 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 Signed-off-by: Sasha Levin

efi/libstub: fix efi_random_alloc() to allocate memory at alloc_min or higher address

2024-04-03T13:11:44+00:00

[ Upstream commit 3cb4a4827596abc82e55b80364f509d0fefc3051 ] 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 Signed-off-by: Sasha Levin

x86/efistub: Don't clear BSS twice in mixed mode

2024-03-26T22:19:00+00:00

[ Upstream commit df7ecce842b846a04d087ba85fdb79a90e26a1b0 ] 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 Signed-off-by: Sasha Levin

x86/efistub: Clear decompressor BSS in native EFI entrypoint

2024-03-26T22:19:00+00:00

[ Upstream commit b3810c5a2cc4a6665f7a65bed5393c75ce3f3aa2 ] The EFI stub on x86 no longer invokes the decompressor as a subsequent boot stage, but calls into the decompression code directly while running in the context of the EFI boot services. This means that when using the native EFI entrypoint (as opposed to the EFI handover protocol, which clears BSS explicitly), the firmware PE image loader is being relied upon to ensure that BSS is zeroed before the EFI stub is entered from the firmware. As Radek's report proves, this is a bad idea. Not all loaders do this correctly, which means some global variables that should be statically initialized to 0x0 may have junk in them. So clear BSS explicitly when entering via efi_pe_entry(). Note that zeroing BSS from C code is not generally safe, but in this case, the following assignment and dereference of a global pointer variable ensures that the memset() cannot be deferred or reordered. Cc: # v6.1+ Reported-by: Radek Podgorny Closes: https://lore.kernel.org/all/a99a831a-8ad5-4cb0-bff9-be637311f771@podgorny.cz Signed-off-by: Ard Biesheuvel Signed-off-by: Sasha Levin

firmware: arm_scmi: Fix double free in SMC transport cleanup path

2024-03-26T22:18:16+00:00

[ Upstream commit f1d71576d2c9ec8fdb822173fa7f3de79475e9bd ] When the generic SCMI code tears down a channel, it calls the chan_free callback function, defined by each transport. Since multiple protocols might share the same transport_info member, chan_free() might want to clean up the same member multiple times within the given SCMI transport implementation. In this case, it is SMC transport. This will lead to a NULL pointer dereference at the second time: | scmi_protocol scmi_dev.1: Enabled polling mode TX channel - prot_id:16 | arm-scmi firmware:scmi: SCMI Notifications - Core Enabled. | arm-scmi firmware:scmi: unable to communicate with SCMI | Unable to handle kernel NULL pointer dereference at virtual address 0000000000000000 | Mem abort info: | ESR = 0x0000000096000004 | EC = 0x25: DABT (current EL), IL = 32 bits | SET = 0, FnV = 0 | EA = 0, S1PTW = 0 | FSC = 0x04: level 0 translation fault | Data abort info: | ISV = 0, ISS = 0x00000004, ISS2 = 0x00000000 | CM = 0, WnR = 0, TnD = 0, TagAccess = 0 | GCS = 0, Overlay = 0, DirtyBit = 0, Xs = 0 | user pgtable: 4k pages, 48-bit VAs, pgdp=0000000881ef8000 | [0000000000000000] pgd=0000000000000000, p4d=0000000000000000 | Internal error: Oops: 0000000096000004 [#1] PREEMPT SMP | Modules linked in: | CPU: 4 PID: 1 Comm: swapper/0 Not tainted 6.7.0-rc2-00124-g455ef3d016c9-dirty #793 | Hardware name: FVP Base RevC (DT) | pstate: 61400009 (nZCv daif +PAN -UAO -TCO +DIT -SSBS BTYPE=--) | pc : smc_chan_free+0x3c/0x6c | lr : smc_chan_free+0x3c/0x6c | Call trace: | smc_chan_free+0x3c/0x6c | idr_for_each+0x68/0xf8 | scmi_cleanup_channels.isra.0+0x2c/0x58 | scmi_probe+0x434/0x734 | platform_probe+0x68/0xd8 | really_probe+0x110/0x27c | __driver_probe_device+0x78/0x12c | driver_probe_device+0x3c/0x118 | __driver_attach+0x74/0x128 | bus_for_each_dev+0x78/0xe0 | driver_attach+0x24/0x30 | bus_add_driver+0xe4/0x1e8 | driver_register+0x60/0x128 | __platform_driver_register+0x28/0x34 | scmi_driver_init+0x84/0xc0 | do_one_initcall+0x78/0x33c | kernel_init_freeable+0x2b8/0x51c | kernel_init+0x24/0x130 | ret_from_fork+0x10/0x20 | Code: f0004701 910a0021 aa1403e5 97b91c70 (b9400280) | ---[ end trace 0000000000000000 ]--- Simply check for the struct pointer being NULL before trying to access its members, to avoid this situation. This was found when a transport doesn't really work (for instance no SMC service), the probe routines then tries to clean up, and triggers a crash. Signed-off-by: Andre Przywara Fixes: 1dc6558062da ("firmware: arm_scmi: Add smc/hvc transport") Reviewed-by: Cristian Marussi Link: https://lore.kernel.org/r/20240126122325.2039669-1-andre.przywara@arm.com Signed-off-by: Sudeep Holla Signed-off-by: Sasha Levin

efi/capsule-loader: fix incorrect allocation size

2024-03-06T14:53:52+00:00

[ Upstream commit fccfa646ef3628097d59f7d9c1a3e84d4b6bb45e ] gcc-14 notices that the allocation with sizeof(void) on 32-bit architectures is not enough for a 64-bit phys_addr_t: drivers/firmware/efi/capsule-loader.c: In function 'efi_capsule_open': drivers/firmware/efi/capsule-loader.c:295:24: error: allocation of insufficient size '4' for type 'phys_addr_t' {aka 'long long unsigned int'} with size '8' [-Werror=alloc-size] 295 | cap_info->phys = kzalloc(sizeof(void *), GFP_KERNEL); | ^ Use the correct type instead here. Fixes: f24c4d478013 ("efi/capsule-loader: Reinstate virtual capsule mapping") Signed-off-by: Arnd Bergmann Signed-off-by: Ard Biesheuvel Signed-off-by: Sasha Levin

efi: Don't add memblocks for soft-reserved memory

2024-03-01T12:41:41+00:00

[ Upstream commit 0bcff59ef7a652fcdc6d535554b63278c2406c8f ] Adding memblocks for soft-reserved regions prevents them from later being hotplugged in by dax_kmem. Signed-off-by: Andrew Bresticker Signed-off-by: Ard Biesheuvel Signed-off-by: Sasha Levin

efi: runtime: Fix potential overflow of soft-reserved region size

2024-03-01T12:41:41+00:00

[ Upstream commit de1034b38a346ef6be25fe8792f5d1e0684d5ff4 ] md_size will have been narrowed if we have >= 4GB worth of pages in a soft-reserved region. Signed-off-by: Andrew Bresticker Signed-off-by: Ard Biesheuvel Signed-off-by: Sasha Levin