[PATCH 3/4] arm64: fix hyp mode mismatch detection
ard.biesheuvel at linaro.org
Mon Mar 16 04:05:42 PDT 2015
On 16 March 2015 at 12:03, Marc Zyngier <marc.zyngier at arm.com> wrote:
> On 16/03/15 10:56, Mark Rutland wrote:
>> On Fri, Mar 13, 2015 at 08:21:19PM +0000, Ard Biesheuvel wrote:
>>> On 13 March 2015 at 17:14, Mark Rutland <mark.rutland at arm.com> wrote:
>>>> Commit 828e9834e9a5b7e6 ("arm64: head: create a new function for setting
>>>> the boot_cpu_mode flag") added BOOT_CPU_MODE_EL1, a nonzero value
>>>> replacing uses of zero. However it failed to update __boot_cpu_mode
>>>> A CPU booted at EL2 writes BOOT_CPU_MODE_EL2 to __boot_cpu_mode, and
>>>> a CPU booted at EL1 writes BOOT_CPU_MODE_EL1 to __boot_cpu_mode.
>>>> Later is_hyp_mode_mismatched() determines there to be a mismatch if
>>>> __boot_cpu_mode != __boot_cpu_mode.
>>>> If all CPUs are booted at EL1, __boot_cpu_mode will be set to
>>>> BOOT_CPU_MODE_EL1, but __boot_cpu_mode will retain its initial value
>>>> of zero, and is_hyp_mode_mismatched will erroneously determine that the
>>>> boot modes are mismatched. This hasn't been a problem so far, but later
>>>> patches which will make use of is_hyp_mode_mismatched() expect it to
>>>> work correctly.
>>>> This patch initialises __boot_cpu_mode to BOOT_CPU_MODE_EL1, fixing
>>>> the erroneous mismatch detection when all CPUs are booted at EL1.
>>> Maybe it's just me, but isn't it *much* easier to understand to
>>> initialise both values to 0, and use 'both are non-zero' as the error
>>> 'HYP mode available' would then be '__boot_cpu_mode ==
>>> BOOT_CPU_MODE_EL2 && __boot_cpu_mode == 0'
>> I agree that change would make this easier to follow.
>> Marc, are you happy with Ard's proposed change?
> Absolutely. If that makes it more obvious, please go for it.
Actually, simply checking for __boot_cpu_mode == 0 (i.e., no CPUs
booted at EL1) would be sufficient to implement 'is HYP mode
More information about the linux-arm-kernel