From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:43480) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1YZyvX-00045P-EZ for qemu-devel@nongnu.org; Mon, 23 Mar 2015 05:45:45 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1YZyvT-0005nb-Un for qemu-devel@nongnu.org; Mon, 23 Mar 2015 05:45:43 -0400 Received: from [59.151.112.132] (port=5678 helo=heian.cn.fujitsu.com) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1YZyvT-0005n4-2B for qemu-devel@nongnu.org; Mon, 23 Mar 2015 05:45:39 -0400 Message-ID: <550FDF1D.9090100@cn.fujitsu.com> Date: Mon, 23 Mar 2015 17:38:37 +0800 From: Chen Fan MIME-Version: 1.0 References: <20150323092319.7da8cf84@nial.brq.redhat.com> <550FD7D1.2050307@cn.fujitsu.com> <20150323104328.62d2910b@nial.brq.redhat.com> In-Reply-To: <20150323104328.62d2910b@nial.brq.redhat.com> Content-Type: text/plain; charset="windows-1252"; format=flowed Content-Transfer-Encoding: 7bit Subject: Re: [Qemu-devel] [PATCH v2 1/2] cpu/apic: drop icc bus/bridge/ List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Igor Mammedov Cc: zhugh.fnst@cn.fujitsu.com, qemu-devel@nongnu.org, tangchen@cn.fujitsu.com, isimatu.yasuaki@jp.fujitsu.com, guz.fnst@cn.fujitsu.com, afaerber@suse.de On 03/23/2015 05:43 PM, Igor Mammedov wrote: > On Mon, 23 Mar 2015 17:07:29 +0800 > Chen Fan wrote: > >> On 03/23/2015 04:23 PM, Igor Mammedov wrote: >>> On Mon, 23 Mar 2015 13:54:23 +0800 >>> Chen Fan wrote: >>> >>>> ICC bus was invented only to provide hotplug capability to >>>> CPU and APIC because at the time being hotplug was available only for >>>> BUS attached devices. >>>> >>>> Now this patch is to drop ICC bus impl, and switch to bus-less >>>> CPU+APIC hotplug, handling them in the same manner as pc-dimm. >>>> >>>> Signed-off-by: Chen Fan >>>> --- >>>> hw/i386/pc.c | 29 +++++++++++------------------ >>>> hw/i386/pc_piix.c | 9 +-------- >>>> hw/i386/pc_q35.c | 9 +-------- >>>> hw/intc/apic.c | 6 +++--- >>>> hw/intc/apic_common.c | 11 ++--------- >>>> include/hw/i386/apic_internal.h | 5 ++--- >>>> include/hw/i386/pc.h | 2 +- >>>> target-i386/cpu.c | 2 -- >>>> 8 files changed, 21 insertions(+), 52 deletions(-) >>>> >>>> diff --git a/hw/i386/pc.c b/hw/i386/pc.c >>>> index 4b46c29..5d15473 100644 >>>> --- a/hw/i386/pc.c >>>> +++ b/hw/i386/pc.c >>> [...] >>> >>>> @@ -1093,8 +1083,11 @@ void pc_cpus_init(const char *cpu_model, DeviceState *icc_bridge) >>>> /* map APIC MMIO area if CPU has APIC */ >>>> if (cpu && cpu->apic_state) { >>>> /* XXX: what if the base changes? */ >>>> - sysbus_mmio_map_overlap(SYS_BUS_DEVICE(icc_bridge), 0, >>>> - APIC_DEFAULT_ADDRESS, 0x1000); >>>> + apic = APIC_COMMON(cpu->apic_state); >>>> + memory_region_add_subregion_overlap(CPU(cpu)->as->root, >>>> + APIC_DEFAULT_ADDRESS, >>>> + &apic->io_memory, >>>> + 0x1000); >>> Why is it here? >>> Shouldn't it be mapped not once but for each CPU since we are using >>> per CPU address spaces? >>> >>> Split this change out into a separate patch please, with commit message >>> describing what it does. >> Hi Igor, >> >> in your previous mail said, "It might be that kvm_irqchip don't >> need it at all." >> I don't know why kvm_irqchip don't need it ? > That's why it's MIGHT, I'm not sure since I've not look at that code for a while. > >> because I have test that for kernel_irqchip=on, qemu emulator the >> kvm-apic object, >> and sent the MSI to kernel irqchip for pcie devices. it also need map >> the region. > Can we have this test as a patch to qemu/tests? so it would be easier to > discuss it. no problem. Thanks, Chen > >> Thanks, >> Chen >> >> >>> PS: >>> It should be part of APIC code or at worst case part of CPU's realize. >>> >>> PS2: >>> new cpu tests don't test actual CPU execution, so they can't validate >>> this change. To test it you need to run test in TCG (at least) or >>> TCG + KVM mode, with some guest code that programs and checks APIC >>> of each CPU. >>> >>> PS3: >>> the rest of the patch I'd suggest to merge with 2/2 patch that >>> removes unused icc_bridge code, there isn't point in splitting >>> that from removing icc_bridge from other files. >>> >>> [...] >>>> >>>> diff --git a/target-i386/cpu.c b/target-i386/cpu.c >>>> index f01690b..2385e6b 100644 >>>> --- a/target-i386/cpu.c >>>> +++ b/target-i386/cpu.c >>>> @@ -42,7 +42,6 @@ >>>> >>>> #include "sysemu/sysemu.h" >>>> #include "hw/qdev-properties.h" >>>> -#include "hw/cpu/icc_bus.h" >>>> #ifndef CONFIG_USER_ONLY >>>> #include "hw/xen/xen.h" >>>> #include "hw/i386/apic_internal.h" >>>> @@ -2941,7 +2940,6 @@ static void x86_cpu_common_class_init(ObjectClass *oc, void *data) >>>> >>>> xcc->parent_realize = dc->realize; >>>> dc->realize = x86_cpu_realizefn; >>>> - dc->bus_type = TYPE_ICC_BUS; >>> that isn't the only place in this file that should be changed. >>> >>> See x86_cpu_apic_create(): >>> cpu->apic_state = qdev_try_create(qdev_get_parent_bus(dev), apic_type); >>> >>> probably it's not right to try get parent bus from bus-less device, >>> qdev_try_create() call should be replaced by object_new()/object_unref() pair. >>> >>>> dc->props = x86_cpu_properties; >>>> >>>> xcc->parent_reset = cc->reset; >>> . >>> > . >