]> git.proxmox.com Git - mirror_ubuntu-kernels.git/commit
ARM: 9195/1: entry: avoid explicit literal loads
authorArd Biesheuvel <ardb@kernel.org>
Wed, 20 Apr 2022 08:41:31 +0000 (09:41 +0100)
committerRussell King (Oracle) <rmk+kernel@armlinux.org.uk>
Fri, 20 May 2022 11:32:32 +0000 (12:32 +0100)
commit508074607c7b95b24f0adf633fdf606761bb7824
tree6bb3ce001060faa2c5da3c985e219c10e557330c
parent952f03316352c606bebef56ba8f9642edbb8e348
ARM: 9195/1: entry: avoid explicit literal loads

ARMv7 has MOVW/MOVT instruction pairs to load symbol addresses into
registers without having to rely on literal loads that go via the
D-cache.  For older cores, we now support a similar arrangement, based
on PC-relative group relocations.

This means we can elide most literal loads entirely from the entry path,
by switching to the ldr_va macro to emit the appropriate sequence
depending on the target architecture revision.

While at it, switch to the bl_r macro for invoking the right PABT/DABT
helpers instead of setting the LR register explicitly, which does not
play well with cores that speculate across function returns.

Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
arch/arm/include/asm/assembler.h
arch/arm/kernel/entry-armv.S
arch/arm/kernel/entry-common.S
arch/arm/kernel/entry-header.S