]> git.karo-electronics.de Git - karo-tx-linux.git/commit
ARM: KVM: Fix 64-bit coprocessor handling
authorChristoffer Dall <christoffer.dall@linaro.org>
Tue, 6 Aug 2013 01:08:41 +0000 (18:08 -0700)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Fri, 27 Sep 2013 00:18:03 +0000 (17:18 -0700)
commit008fe511c11fa20d1060c9e2ddb9c829ac44e0a7
tree1fd93ca25ec7f8a141f802f049d62887f1f7cb9c
parent464de2dc1539e6355fad1342ad106f08f7642edd
ARM: KVM: Fix 64-bit coprocessor handling

commit 240e99cbd00aa541b572480e3ea7ecb0d480bc79 upstream.

The PAR was exported as CRn == 7 and CRm == 0, but in fact the primary
coprocessor register number was determined by CRm for 64-bit coprocessor
registers as the user space API was modeled after the coprocessor
access instructions (see the ARM ARM rev. C - B3-1445).

However, just changing the CRn to CRm breaks the sorting check when
booting the kernel, because the internal kernel logic always treats CRn
as the primary register number, and it makes the table sorting
impossible to understand for humans.

Alternatively we could change the logic to always have CRn == CRm, but
that becomes unclear in the number of ways we do look up of a coprocessor
register.  We could also have a separate 64-bit table but that feels
somewhat over-engineered.  Instead, keep CRn the primary representation
of the primary coproc. register number in-kernel and always export the
primary number as CRm as per the existing user space ABI.

Note: The TTBR registers just magically worked because they happened to
follow the CRn(0) regs and were considered CRn(0) in the in-kernel
representation.

Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>
Signed-off-by: Kim Phillips <kim.phillips@linaro.org>
Cc: Gleb Natapov <gleb@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Jonghwan Choi <jhbird.choi@samsung.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
arch/arm/kvm/coproc.c
arch/arm/kvm/coproc.h
arch/arm/kvm/coproc_a15.c