When trying to kexec into a new kernel on a platform where multiple CPU
cores are present, but no SMP bringup code is available yet, the
kexec_load system call fails with:
kexec_load failed: Invalid argument
The SMP test added to machine_kexec_prepare() in commit 2103f6cba61a8b8b
("ARM: 7807/1: kexec: validate CPU hotplug support") wants to prohibit
kexec on SMP platforms where it cannot disable secondary CPUs.
However, this test is too strict: if the secondary CPUs couldn't be
enabled in the first place, there's no need to disable them later at
kexec time. Hence skip the test in the absence of SMP bringup code.
This allows to add all CPU cores to the DTS from the beginning, without
having to implement SMP bringup first, improving DT compatibility.
Signed-off-by: Geert Uytterhoeven <[email protected]>
---
arch/arm/include/asm/smp_plat.h | 1 +
arch/arm/kernel/machine_kexec.c | 3 ++-
arch/arm/kernel/smp.c | 5 +++++
3 files changed, 8 insertions(+), 1 deletion(-)
diff --git a/arch/arm/include/asm/smp_plat.h b/arch/arm/include/asm/smp_plat.h
index 0ad7d490ee6f2657..993e5224d8f7eeeb 100644
--- a/arch/arm/include/asm/smp_plat.h
+++ b/arch/arm/include/asm/smp_plat.h
@@ -104,6 +104,7 @@ static inline u32 mpidr_hash_size(void)
return 1 << mpidr_hash.bits;
}
+extern int platform_can_secondary_boot(void);
extern int platform_can_cpu_hotplug(void);
#endif
diff --git a/arch/arm/kernel/machine_kexec.c b/arch/arm/kernel/machine_kexec.c
index de2b085ad7535da7..8bf3b7c098881b95 100644
--- a/arch/arm/kernel/machine_kexec.c
+++ b/arch/arm/kernel/machine_kexec.c
@@ -46,7 +46,8 @@ int machine_kexec_prepare(struct kimage *image)
* and implements CPU hotplug for the current HW. If not, we won't be
* able to kexec reliably, so fail the prepare operation.
*/
- if (num_possible_cpus() > 1 && !platform_can_cpu_hotplug())
+ if (num_possible_cpus() > 1 && platform_can_secondary_boot() &&
+ !platform_can_cpu_hotplug())
return -EINVAL;
/*
diff --git a/arch/arm/kernel/smp.c b/arch/arm/kernel/smp.c
index 86ef244c5a24b4fa..cca5b87581855244 100644
--- a/arch/arm/kernel/smp.c
+++ b/arch/arm/kernel/smp.c
@@ -145,6 +145,11 @@ void __init smp_init_cpus(void)
smp_ops.smp_init_cpus();
}
+int platform_can_secondary_boot(void)
+{
+ return !!smp_ops.smp_boot_secondary;
+}
+
int platform_can_cpu_hotplug(void)
{
#ifdef CONFIG_HOTPLUG_CPU
--
1.9.1
On 02/26/2015 03:37 AM, Geert Uytterhoeven wrote:
> When trying to kexec into a new kernel on a platform where multiple CPU
> cores are present, but no SMP bringup code is available yet, the
> kexec_load system call fails with:
>
> kexec_load failed: Invalid argument
>
> The SMP test added to machine_kexec_prepare() in commit 2103f6cba61a8b8b
> ("ARM: 7807/1: kexec: validate CPU hotplug support") wants to prohibit
> kexec on SMP platforms where it cannot disable secondary CPUs.
> However, this test is too strict: if the secondary CPUs couldn't be
> enabled in the first place, there's no need to disable them later at
> kexec time. Hence skip the test in the absence of SMP bringup code.
>
> This allows to add all CPU cores to the DTS from the beginning, without
> having to implement SMP bringup first, improving DT compatibility.
Acked-by: Stephen Warren <[email protected]>
On Thu, Feb 26, 2015 at 11:37:08AM +0100, Geert Uytterhoeven wrote:
> When trying to kexec into a new kernel on a platform where multiple CPU
> cores are present, but no SMP bringup code is available yet, the
> kexec_load system call fails with:
>
> kexec_load failed: Invalid argument
>
> The SMP test added to machine_kexec_prepare() in commit 2103f6cba61a8b8b
> ("ARM: 7807/1: kexec: validate CPU hotplug support") wants to prohibit
> kexec on SMP platforms where it cannot disable secondary CPUs.
> However, this test is too strict: if the secondary CPUs couldn't be
> enabled in the first place, there's no need to disable them later at
> kexec time. Hence skip the test in the absence of SMP bringup code.
Hmm. I don't think we should relax it in this manner - I think there's
an easier solution to this.
> diff --git a/arch/arm/kernel/machine_kexec.c b/arch/arm/kernel/machine_kexec.c
> index de2b085ad7535da7..8bf3b7c098881b95 100644
> --- a/arch/arm/kernel/machine_kexec.c
> +++ b/arch/arm/kernel/machine_kexec.c
> @@ -46,7 +46,8 @@ int machine_kexec_prepare(struct kimage *image)
> * and implements CPU hotplug for the current HW. If not, we won't be
> * able to kexec reliably, so fail the prepare operation.
> */
> - if (num_possible_cpus() > 1 && !platform_can_cpu_hotplug())
> + if (num_possible_cpus() > 1 && platform_can_secondary_boot() &&
> + !platform_can_cpu_hotplug())
if (num_online_cpus() > 1 && !platform_can_cpu_hotplug())
> return -EINVAL;
Neither test is actually accurate though: when we have implementations
where the secondary CPUs spin inside the kernel when they're "unplugged"
that is not sufficient to be able to kexec.
We should probably fix that, and make platform_can_cpu_hotplug() report
whether it really is possible to hotplug all secondary CPUs into such
a state that kexec can work.
--
FTTC broadband for 0.8mile line: currently at 10.5Mbps down 400kbps up
according to speedtest.net.
On 02/26/2015 10:42 AM, Russell King - ARM Linux wrote:
> On Thu, Feb 26, 2015 at 11:37:08AM +0100, Geert Uytterhoeven wrote:
>> When trying to kexec into a new kernel on a platform where multiple CPU
>> cores are present, but no SMP bringup code is available yet, the
>> kexec_load system call fails with:
>>
>> kexec_load failed: Invalid argument
>>
>> The SMP test added to machine_kexec_prepare() in commit 2103f6cba61a8b8b
>> ("ARM: 7807/1: kexec: validate CPU hotplug support") wants to prohibit
>> kexec on SMP platforms where it cannot disable secondary CPUs.
>> However, this test is too strict: if the secondary CPUs couldn't be
>> enabled in the first place, there's no need to disable them later at
>> kexec time. Hence skip the test in the absence of SMP bringup code.
>
> Hmm. I don't think we should relax it in this manner - I think there's
> an easier solution to this.
>
>> diff --git a/arch/arm/kernel/machine_kexec.c b/arch/arm/kernel/machine_kexec.c
>> index de2b085ad7535da7..8bf3b7c098881b95 100644
>> --- a/arch/arm/kernel/machine_kexec.c
>> +++ b/arch/arm/kernel/machine_kexec.c
>> @@ -46,7 +46,8 @@ int machine_kexec_prepare(struct kimage *image)
>> * and implements CPU hotplug for the current HW. If not, we won't be
>> * able to kexec reliably, so fail the prepare operation.
>> */
>> - if (num_possible_cpus() > 1 && !platform_can_cpu_hotplug())
>> + if (num_possible_cpus() > 1 && platform_can_secondary_boot() &&
>> + !platform_can_cpu_hotplug())
>
> if (num_online_cpus() > 1 && !platform_can_cpu_hotplug())
I can't remember the call stack here. Is num_online_cpus() guaranteed
not to change from this point through to when the kexec actually happens?
>
>> return -EINVAL;
>
> Neither test is actually accurate though: when we have implementations
> where the secondary CPUs spin inside the kernel when they're "unplugged"
> that is not sufficient to be able to kexec.
>
> We should probably fix that, and make platform_can_cpu_hotplug() report
> whether it really is possible to hotplug all secondary CPUs into such
> a state that kexec can work.
>
On Thu, Feb 26, 2015 at 8:00 PM, Stephen Warren <[email protected]> wrote:
> On 02/26/2015 10:42 AM, Russell King - ARM Linux wrote:
>>
>> On Thu, Feb 26, 2015 at 11:37:08AM +0100, Geert Uytterhoeven wrote:
>>>
>>> When trying to kexec into a new kernel on a platform where multiple CPU
>>> cores are present, but no SMP bringup code is available yet, the
>>> kexec_load system call fails with:
>>>
>>> kexec_load failed: Invalid argument
>>>
>>> The SMP test added to machine_kexec_prepare() in commit 2103f6cba61a8b8b
>>> ("ARM: 7807/1: kexec: validate CPU hotplug support") wants to prohibit
>>> kexec on SMP platforms where it cannot disable secondary CPUs.
>>> However, this test is too strict: if the secondary CPUs couldn't be
>>> enabled in the first place, there's no need to disable them later at
>>> kexec time. Hence skip the test in the absence of SMP bringup code.
>>
>>
>> Hmm. I don't think we should relax it in this manner - I think there's
>> an easier solution to this.
>>
>>> diff --git a/arch/arm/kernel/machine_kexec.c
>>> b/arch/arm/kernel/machine_kexec.c
>>> index de2b085ad7535da7..8bf3b7c098881b95 100644
>>> --- a/arch/arm/kernel/machine_kexec.c
>>> +++ b/arch/arm/kernel/machine_kexec.c
>>> @@ -46,7 +46,8 @@ int machine_kexec_prepare(struct kimage *image)
>>> * and implements CPU hotplug for the current HW. If not, we
>>> won't be
>>> * able to kexec reliably, so fail the prepare operation.
>>> */
>>> - if (num_possible_cpus() > 1 && !platform_can_cpu_hotplug())
>>> + if (num_possible_cpus() > 1 && platform_can_secondary_boot() &&
>>> + !platform_can_cpu_hotplug())
>>
>>
>> if (num_online_cpus() > 1 && !platform_can_cpu_hotplug())
>
> I can't remember the call stack here. Is num_online_cpus() guaranteed not to
> change from this point through to when the kexec actually happens?
Yeah, I had similar thoughts when I added the new test.
include/linux/cpumask.h:
* cpu_possible_mask- has bit 'cpu' set iff cpu is populatable
* cpu_present_mask - has bit 'cpu' set iff cpu is populated
* cpu_online_mask - has bit 'cpu' set iff cpu available to scheduler
* cpu_active_mask - has bit 'cpu' set iff cpu available to migration
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds