From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from lists.gnu.org (lists.gnu.org [209.51.188.17]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.lore.kernel.org (Postfix) with ESMTPS id B8590CCFA04 for ; Mon, 3 Nov 2025 09:53:20 +0000 (UTC) Received: from localhost ([::1] helo=lists1p.gnu.org) by lists.gnu.org with esmtp (Exim 4.90_1) (envelope-from ) id 1vFrEk-0004FD-BA; Mon, 03 Nov 2025 04:52:31 -0500 Received: from eggs.gnu.org ([2001:470:142:3::10]) by lists.gnu.org with esmtps (TLS1.2:ECDHE_RSA_AES_256_GCM_SHA384:256) (Exim 4.90_1) (envelope-from ) id 1vFrEj-0004F2-4I for qemu-devel@nongnu.org; Mon, 03 Nov 2025 04:52:29 -0500 Received: from us-smtp-delivery-124.mimecast.com ([170.10.133.124]) by eggs.gnu.org with esmtps (TLS1.2:ECDHE_RSA_AES_256_GCM_SHA384:256) (Exim 4.90_1) (envelope-from ) id 1vFrEf-0007OR-Q9 for qemu-devel@nongnu.org; Mon, 03 Nov 2025 04:52:28 -0500 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=redhat.com; s=mimecast20190719; t=1762163541; h=from:from:reply-to:subject:subject:date:date:message-id:message-id: to:to:cc:cc:mime-version:mime-version:content-type:content-type: content-transfer-encoding:content-transfer-encoding: in-reply-to:in-reply-to:references:references; bh=VcAIOZZ3d5XNJNg3P8cQJlolVFcqvq0I8RonPpmyS4Y=; b=G85SdFPj1ugJsQR7O99BNCvsUFC2cad+Q8BxaVYSnitWL982QVDzNMZuSf+7m/GR842DU3 RQpAYlH+cQ+8BKKHxKCiZFu5iYBbQc9LpBsRWgs4Px2Pm2QqHWCbSupBxXL0TCP0SRE2/I CPU0c3Bc2WOwHrDDd7RG59CVqOnE6BM= Received: from mail-wm1-f70.google.com (mail-wm1-f70.google.com [209.85.128.70]) by relay.mimecast.com with ESMTP with STARTTLS (version=TLSv1.3, cipher=TLS_AES_256_GCM_SHA384) id us-mta-526-UEqGaw64OyG-tBeRuNojfw-1; Mon, 03 Nov 2025 04:52:20 -0500 X-MC-Unique: UEqGaw64OyG-tBeRuNojfw-1 X-Mimecast-MFC-AGG-ID: UEqGaw64OyG-tBeRuNojfw_1762163539 Received: by mail-wm1-f70.google.com with SMTP id 5b1f17b1804b1-47113dfdd20so18098475e9.1 for ; Mon, 03 Nov 2025 01:52:20 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=redhat.com; s=google; t=1762163539; x=1762768339; darn=nongnu.org; h=content-transfer-encoding:mime-version:references:in-reply-to :message-id:subject:cc:to:from:date:from:to:cc:subject:date :message-id:reply-to; bh=VcAIOZZ3d5XNJNg3P8cQJlolVFcqvq0I8RonPpmyS4Y=; b=RpigcgH4M+EE8EOml7DEvOXCEtqd7dmddQctrum8ghORn0fyZoJLmpHZQuP3+/qbEr iFv6gyqrMvnQ8PSRAeZ4ChbuSrT1Q7kXD7EJPciToC9d+4/WnLUcskLrr+A1uRXMWlvg hmf9UMb69EGM9536yMp4nzyU5hvP9xJH4nDiRrXGoX3Yn7M3Lh8vU4Tia+aqu5xDMq+w hjOko9dtss3wKeDQgsetgmfQIM2WHE2v9g/pqnHF+N6dk0jWU3DpyBzP7Pzpv0p0gXGr k0JCOmb6IxnQ5TzRB5vTSaNXZQKtOEM0iabG38G3ujwhkOu5biixhzs51yUi07d/8jm0 E0VQ== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1762163539; x=1762768339; h=content-transfer-encoding:mime-version:references:in-reply-to :message-id:subject:cc:to:from:date:x-gm-message-state:from:to:cc :subject:date:message-id:reply-to; bh=VcAIOZZ3d5XNJNg3P8cQJlolVFcqvq0I8RonPpmyS4Y=; b=t0CGroVLlzENsuRKTzHm9klEGA+k8x7lqr/GrR0S3r07BdjwQPa/QsNaI+wNvuYbha 4GoCdQQjq7JAj/0Yp4qgAdqYIyHsDgpKjQyGe2t/wgTjv75ylM9VEzR84KjgCMzoOjFU j07g46AigkWLbcWZ3/tAu0T02hQi90Y4UJvLKfoXt/QI0qEH3agOP+VpFNzDo8uDfKbR ftxoajI1GkfzK9q3JC+EEoqLnxK0twANPT/0wWSBnTRjVdBH3liM5iMfLGJ6RXOvnVjI //xl1DlXwdpYWvQ0LCkoErBOlDR000Q+2Tt7eVf2z8C/0xelWx3QnFRLCNDGWTC7mYug KUsQ== X-Forwarded-Encrypted: i=1; AJvYcCVdTgA/E2VYj08Ze2pAbQOxGt/hOfEiXVtGwBYSPpE/cYQg+/2yg9SOlZmjt55Wwulk7rebM5VWJI1S@nongnu.org X-Gm-Message-State: AOJu0Yz6piF50UlD+/J1bYiIWN7sO6DAL+DzNKCbs2vzOuA/87z9kGeu 9b7nHAlkQYf7mSw76IUqeUHPPOwHrmXyKcbA5+iD1+0E+OArYXdW5ZEgjTb2QcFj+LaPocsPn33 sasBOiS/ByP89ITbP5gVeBOBdrz/Pe9y1mP9jjHOFu79eE19drL18yQUy X-Gm-Gg: ASbGncvAQTRnPGes6yYwSXTug2+bfEIQTLaf7wm4eqLJbNynWICKDh2w1e3pi3GKNXX TbsbR0w9y/zv7gw6r/Ah8XzpUUNqU8tUyu2Mpgo6oRiVbjtY6PC8PhgBnykzXMynP7LlA93YM1T n3wpWud+c8XHbm44iDIKfx9Go4XouJ/ngxUN2ZXn2l4Lw6hjrE7aNe8HvGHXsDOe80AMrGUEAvs d+IcBvIpAYS0gn2WGcS7T/zFlAQXQMMZ96kEBtNX69mBbLBIHaeVkrmY2218O8qrNnmylCMm3ze QX2Ra8kn0t57tbh0xV7NowvbBVOeI4pgroyQcMlyezJp6gbKEDCvz3E2EXqdwgI9tA== X-Received: by 2002:a05:600d:8314:b0:477:4345:7c59 with SMTP id 5b1f17b1804b1-47743458444mr34153175e9.40.1762163538866; Mon, 03 Nov 2025 01:52:18 -0800 (PST) X-Google-Smtp-Source: AGHT+IEHrpKvuEcpdjJgG2drC9CG2GulQfK0IBVYPY0JGIQAUg5kL10oY2PxuO7fAavRV3IeUgPX/g== X-Received: by 2002:a05:600d:8314:b0:477:4345:7c59 with SMTP id 5b1f17b1804b1-47743458444mr34152935e9.40.1762163538430; Mon, 03 Nov 2025 01:52:18 -0800 (PST) Received: from fedora ([85.93.96.130]) by smtp.gmail.com with ESMTPSA id 5b1f17b1804b1-4773c374f84sm144441465e9.0.2025.11.03.01.52.17 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Mon, 03 Nov 2025 01:52:18 -0800 (PST) Date: Mon, 3 Nov 2025 10:52:16 +0100 From: Igor Mammedov To: Gavin Shan Cc: qemu-arm@nongnu.org, qemu-devel@nongnu.org, mst@redhat.com, anisinha@redhat.com, gengdongjiu1@gmail.com, peter.maydell@linaro.org, pbonzini@redhat.com, mchehab+huawei@kernel.org, Jonathan.Cameron@huawei.com, shan.gavin@gmail.com Subject: Re: [PATCH RESEND v2 3/3] target/arm/kvm: Support multiple memory CPERs injection Message-ID: <20251103105216.1f4241d7@fedora> In-Reply-To: <88a41137-d5fb-4b61-a3f2-dd73133c17ec@redhat.com> References: <20251007060810.258536-1-gshan@redhat.com> <20251007060810.258536-4-gshan@redhat.com> <20251017162746.2a99015b@fedora> <20251031145539.3551b0a5@fedora> <88a41137-d5fb-4b61-a3f2-dd73133c17ec@redhat.com> X-Mailer: Claws Mail 4.3.1 (GTK 3.24.49; x86_64-redhat-linux-gnu) MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Received-SPF: pass client-ip=170.10.133.124; envelope-from=imammedo@redhat.com; helo=us-smtp-delivery-124.mimecast.com X-Spam_score_int: -20 X-Spam_score: -2.1 X-Spam_bar: -- X-Spam_report: (-2.1 / 5.0 requ) BAYES_00=-1.9, DKIMWL_WL_HIGH=-0.001, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, DKIM_VALID_EF=-0.1, RCVD_IN_DNSWL_NONE=-0.0001, RCVD_IN_MSPIKE_H3=-0.01, RCVD_IN_MSPIKE_WL=-0.01, RCVD_IN_VALIDITY_CERTIFIED_BLOCKED=0.001, RCVD_IN_VALIDITY_RPBL_BLOCKED=0.001, SPF_HELO_PASS=-0.001, SPF_PASS=-0.001 autolearn=ham autolearn_force=no X-Spam_action: no action X-BeenThere: qemu-devel@nongnu.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: qemu-devel-bounces+qemu-devel=archiver.kernel.org@nongnu.org Sender: qemu-devel-bounces+qemu-devel=archiver.kernel.org@nongnu.org On Mon, 3 Nov 2025 09:02:54 +1000 Gavin Shan wrote: > On 10/31/25 11:55 PM, Igor Mammedov wrote: > > On Sun, 19 Oct 2025 10:36:16 +1000 > > Gavin Shan wrote: > >> On 10/18/25 12:27 AM, Igor Mammedov wrote: > >>> On Tue, 7 Oct 2025 16:08:10 +1000 > >>> Gavin Shan wrote: > >>> > >>>> In the combination of 64KB host and 4KB guest, a problematic host page > >>>> affects 16x guest pages. In this specific case, it's reasonable to > >>>> push 16 consecutive memory CPERs. Otherwise, QEMU can run into core > >>>> dump due to the current error can't be delivered as the previous error > >>>> isn't acknoledges. It's caused by the nature the host page can be > >>>> accessed in parallel due to the mismatched host and guest page sizes. > >>> > >>> can you explain a bit more what goes wrong? > >>> > >>> I'm especially interested in parallel access you've mentioned > >>> and why batch adding error records is needed > >>> as opposed to adding records every time invalid access happens? > >>> > >>> PS: > >>> Assume I don't remember details on how HEST works, > >>> Answering it in this format also should improve commit message > >>> making it more digestible for uninitiated. > >>> > >> > >> Thanks for your review and I'm trying to answer your question below. Please let > >> me know if there are more questions. > >> > >> There are two signals (BUS_MCEERR_AR and BUS_MCEERR_AO) and BUS_MCEERR_AR is > >> concerned here. This signal BUS_MCEERR_AR is sent by host's stage2 page fault > >> handler when the resolved host page has been marked as marked as poisoned. > >> The stage2 page fault handler is invoked on every access to the host page. > >> > >> In the combination where host and guest has 64KB and 4KB separately, A 64KB > >> host page corresponds to 16x consecutive 4KB guest pages. It means we're > >> accessing the 64KB host page when any of those 16x consecutive 4KB guest pages > >> is accessed. In other words, a problematic 64KB host page affects the accesses > >> on 16x 4KB guest pages. Those 16x 4KB guest pages can be owned by different > >> threads on the guest and they run in parallel, potentially to access those > >> 16x 4KB guest pages in parallel. It potentially leading to 16x BUS_MCEERR_AR > >> signals at one point. > >> > >> In current implementation, the error record is built as the following calltrace > >> indicates. There are 16 error records in the extreme case (parallel accesses on > >> 16x 4KB guest pages, mapped to one 64KB host page). However, we can't handle > >> multiple error records at once due to the acknowledgement mechanism in > >> ghes_record_cper_errors(). For example, the first error record has been sent, > >> but not consumed by the guest yet. We fail to send the second error record. > >> > >> kvm_arch_on_sigbus_vcpu > >> acpi_ghes_memory_errors > >> ghes_gen_err_data_uncorrectable_recoverable // Generic Error Data Entry > >> acpi_ghes_build_append_mem_cper // Memory Error > >> ghes_record_cper_errors > >> > >> So this series improves this situation by simply sending 16x error records in > >> one shot for the combination of 64KB host + 4KB guest. > > > > 1) What I'm concerned about is that it target one specific case only. > > Imagine if 1st cpu get error on page1 and another on page2=(page1+host_page_size) > > and so on for other CPUs. Then we are back where we were before this series. > > > > Also in abstract future when ARM gets 1Gb pages, that won't scale well. > > > > Can we instead of making up CPERs to cover whole host page, > > create 1/vcpu GHES source? > > That way when vcpu trips over bad page, it would have its own > > error status block to put errors in. > > That would address [1] and deterministically scale > > (well assuming that multiple SEA error sources are possible in theory) > > > > I think it's a good idea to have individual error source for each vCPU. In this > way, the read_ack_reg won't be a limitation. I hope Jonathan is ok to this scheme. > > Currently, we have two (fixed) error sources like below. I assume the index of > the error source per vCPU will starts from (ACPI_HEST_SRC_ID_QMP + 1) based on > CPUState::cpu_index. I'd suggest ditch cpu index and use arch_id instead. > > enum AcpiGhesSourceID { > ACPI_HEST_SRC_ID_SYNC, > ACPI_HEST_SRC_ID_QMP, /* Use it only for QMP injected errors */ > }; > > > > PS: > > I also wonder what real HW does when it gets in similar situation > > (i.e. error status block is not yet acknowledged but another async > > error arrived for the same error source)? > > > > I believe real HW also have this specific issue. As to ARM64, I ever did some > google search and was told the error follows the firmware-first policy and > handled by a component of trustfirmware-a. However, I was unable to get the > source code. So it's hard for me to know how this specific issue is handled > there. Perhaps Jonathan can help with finding how real hw works around it? My idea using per cpu source is just a speculation based on spec on how workaround the problem, I don't really know if guest OS will be able to handle it (aka, need to be tested is it's viable). That also probably was a reason in previous review, why should've waited for multiple sources support be be merged first before this series. > Thanks, > Gavin > > >>>> Imporve push_ghes_memory_errors() to push 16x consecutive memory CPERs > >>>> for this specific case. The maximal error block size is bumped to 4KB, > >>>> providing enough storage space for those 16x memory CPERs. > >>>> > >>>> Signed-off-by: Gavin Shan > >>>> --- > >>>> hw/acpi/ghes.c | 2 +- > >>>> target/arm/kvm.c | 46 +++++++++++++++++++++++++++++++++++++++++++++- > >>>> 2 files changed, 46 insertions(+), 2 deletions(-) > >>>> > >>>> diff --git a/hw/acpi/ghes.c b/hw/acpi/ghes.c > >>>> index 045b77715f..5c87b3a027 100644 > >>>> --- a/hw/acpi/ghes.c > >>>> +++ b/hw/acpi/ghes.c > >>>> @@ -33,7 +33,7 @@ > >>>> #define ACPI_HEST_ADDR_FW_CFG_FILE "etc/acpi_table_hest_addr" > >>>> > >>>> /* The max size in bytes for one error block */ > >>>> -#define ACPI_GHES_MAX_RAW_DATA_LENGTH (1 * KiB) > >>>> +#define ACPI_GHES_MAX_RAW_DATA_LENGTH (4 * KiB) > >>>> > >>>> /* Generic Hardware Error Source version 2 */ > >>>> #define ACPI_GHES_SOURCE_GENERIC_ERROR_V2 10 > >>>> diff --git a/target/arm/kvm.c b/target/arm/kvm.c > >>>> index c5d5b3b16e..3ecb85e4b7 100644 > >>>> --- a/target/arm/kvm.c > >>>> +++ b/target/arm/kvm.c > >>>> @@ -11,6 +11,7 @@ > >>>> */ > >>>> > >>>> #include "qemu/osdep.h" > >>>> +#include "qemu/units.h" > >>>> #include > >>>> > >>>> #include > >>>> @@ -2433,10 +2434,53 @@ static void push_ghes_memory_errors(CPUState *c, AcpiGhesState *ags, > >>>> uint64_t paddr) > >>>> { > >>>> GArray *addresses = g_array_new(false, false, sizeof(paddr)); > >>>> + uint64_t val, start, end, guest_pgsz, host_pgsz; > >>>> int ret; > >>>> > >>>> kvm_cpu_synchronize_state(c); > >>>> - g_array_append_vals(addresses, &paddr, 1); > >>>> + > >>>> + /* > >>>> + * Sort out the guest page size from TCR_EL1, which can be modified > >>>> + * by the guest from time to time. So we have to sort it out dynamically. > >>>> + */ > >>>> + ret = read_sys_reg64(c->kvm_fd, &val, ARM64_SYS_REG(3, 0, 2, 0, 2)); > >>>> + if (ret) { > >>>> + goto error; > >>>> + } > >>>> + > >>>> + switch (extract64(val, 14, 2)) { > >>>> + case 0: > >>>> + guest_pgsz = 4 * KiB; > >>>> + break; > >>>> + case 1: > >>>> + guest_pgsz = 64 * KiB; > >>>> + break; > >>>> + case 2: > >>>> + guest_pgsz = 16 * KiB; > >>>> + break; > >>>> + default: > >>>> + error_report("unknown page size from TCR_EL1 (0x%" PRIx64 ")", val); > >>>> + goto error; > >>>> + } > >>>> + > >>>> + host_pgsz = qemu_real_host_page_size(); > >>>> + start = paddr & ~(host_pgsz - 1); > >>>> + end = start + host_pgsz; > >>>> + while (start < end) { > >>>> + /* > >>>> + * The precise physical address is provided for the affected > >>>> + * guest page that contains @paddr. Otherwise, the starting > >>>> + * address of the guest page is provided. > >>>> + */ > >>>> + if (paddr >= start && paddr < (start + guest_pgsz)) { > >>>> + g_array_append_vals(addresses, &paddr, 1); > >>>> + } else { > >>>> + g_array_append_vals(addresses, &start, 1); > >>>> + } > >>>> + > >>>> + start += guest_pgsz; > >>>> + } > >>>> + > >>>> ret = acpi_ghes_memory_errors(ags, ACPI_HEST_SRC_ID_SYNC, addresses); > >>>> if (ret) { > >>>> goto error; > >>> > >> > > >