arm64: head: Ignore bogus KASLR displacement on non-relocatable kernels

Even non-KASLR kernels can be built as relocatable, to work around
broken bootloaders that violate the rules regarding physical placement
of the kernel image - in this case, the physical offset modulo 2 MiB is
used as the KASLR offset, and all absolute symbol references are fixed
up in the usual way. This workaround is enabled by default.

CONFIG_RELOCATABLE can also be disabled entirely, in which case the
relocation code and the code that captures the offset are omitted from
the build. However, since commit aacd149b6238 ("arm64: head: avoid
relocating the kernel twice for KASLR"), this code got out of sync, and
we still add the offset to the kernel virtual address before populating
the page tables even though we never capture it. This means we add a
bogus value instead, breaking the boot entirely.

Fixes: aacd149b6238 ("arm64: head: avoid relocating the kernel twice for KASLR")
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Tested-by: Mikulas Patocka <mpatocka@redhat.com>
Link: https://lore.kernel.org/r/20220827070904.2216989-1-ardb@kernel.org
Signed-off-by: Will Deacon <will@kernel.org>

diff --git a/arch/arm64/kernel/head.S b/arch/arm64/kernel/head.S
index cefe6a73ee546c26e8971bce4abf4b01f6e5f96d..814b6587ccb78fc2fcc4dc0d1e6d4d8098236bca 100644 (file)
--- a/arch/arm64/kernel/head.S
+++ b/arch/arm64/kernel/head.S
@@ -371,7 +371,9 @@ SYM_FUNC_END(create_idmap)
 SYM_FUNC_START_LOCAL(create_kernel_mapping)
        adrp    x0, init_pg_dir
        mov_q   x5, KIMAGE_VADDR                // compile time __va(_text)
+#ifdef CONFIG_RELOCATABLE
        add     x5, x5, x23                     // add KASLR displacement
+#endif
        adrp    x6, _end                        // runtime __pa(_end)
        adrp    x3, _text                       // runtime __pa(_text)
        sub     x6, x6, x3                      // _end - _text