[PATCH] x86/Xen: don't report ancient LAPIC version

[PATCH] x86/Xen: don't report ancient LAPIC version

[Jan Beulich]

Unconditionally reporting a value seen on the P4 or older invokes
functionality like io_apic_get_unique_id() on 32-bit builds, resulting
in a panic() with sufficiently many CPUs and/or IO-APICs. Doing what
that function does would be the hypervisor's responsibility anyway, so
makes no sense to be used when running on Xen. Uniformly report a more
modern version; this shouldn't matter much as both LAPIC and IO-APIC are
being managed entirely / mostly by the hypervisor.

Signed-off-by: Jan Beulich <jbeulich@suse.com>

--- a/arch/x86/xen/apic.c
+++ b/arch/x86/xen/apic.c
@@ -57,7 +57,7 @@ static u32 xen_apic_read(u32 reg)
 		return 0;
 
 	if (reg == APIC_LVR)
-		return 0x10;
+		return 0x14;
 #ifdef CONFIG_X86_32
 	if (reg == APIC_LDR)
 		return SET_APIC_LOGICAL_ID(1UL << smp_processor_id());




_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel

Re: [PATCH] x86/Xen: don't report ancient LAPIC version

[Boris Ostrovsky]

On 12/08/2017 06:17 AM, Jan Beulich wrote:
> Unconditionally reporting a value seen on the P4 or older invokes
> functionality like io_apic_get_unique_id() on 32-bit builds, resulting
> in a panic() with sufficiently many CPUs and/or IO-APICs. Doing what
> that function does would be the hypervisor's responsibility anyway, so
> makes no sense to be used when running on Xen. Uniformly report a more
> modern version; this shouldn't matter much as both LAPIC and IO-APIC are
> being managed entirely / mostly by the hypervisor.
>
> Signed-off-by: Jan Beulich <jbeulich@suse.com>

Committed to for-linus-4.15.

-boris

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel

Re: [PATCH] x86/Xen: don't report ancient LAPIC version

[Jan Beulich]

>>> On 08.12.17 at 14:22, <jgross@suse.com> wrote:
> On 08/12/17 12:17, Jan Beulich wrote:
>> Unconditionally reporting a value seen on the P4 or older invokes
>> functionality like io_apic_get_unique_id() on 32-bit builds, resulting
>> in a panic() with sufficiently many CPUs and/or IO-APICs. Doing what
>> that function does would be the hypervisor's responsibility anyway, so
>> makes no sense to be used when running on Xen. Uniformly report a more
>> modern version; this shouldn't matter much as both LAPIC and IO-APIC are
>> being managed entirely / mostly by the hypervisor.
>> 
>> Signed-off-by: Jan Beulich <jbeulich@suse.com>
> 
> Reviewed-by: Juergen Gross <jgross@suse.com>

Thanks.

> BTW: Cc: stable?

Maybe - the issue not having bothered anybody for years made
me uncertain whether anyone other than me actually cared.

Jan


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel

Re: [PATCH] x86/Xen: don't report ancient LAPIC version

[Juergen Gross]

On 08/12/17 12:17, Jan Beulich wrote:
> Unconditionally reporting a value seen on the P4 or older invokes
> functionality like io_apic_get_unique_id() on 32-bit builds, resulting
> in a panic() with sufficiently many CPUs and/or IO-APICs. Doing what
> that function does would be the hypervisor's responsibility anyway, so
> makes no sense to be used when running on Xen. Uniformly report a more
> modern version; this shouldn't matter much as both LAPIC and IO-APIC are
> being managed entirely / mostly by the hypervisor.
> 
> Signed-off-by: Jan Beulich <jbeulich@suse.com>

Reviewed-by: Juergen Gross <jgross@suse.com>

BTW: Cc: stable?


Juergen

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel