From patchwork Tue Nov 8 19:16:41 2022 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: John Cabaj X-Patchwork-Id: 1701466 Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@legolas.ozlabs.org Authentication-Results: legolas.ozlabs.org; spf=pass (sender SPF authorized) smtp.mailfrom=lists.ubuntu.com (client-ip=91.189.94.19; helo=huckleberry.canonical.com; envelope-from=kernel-team-bounces@lists.ubuntu.com; receiver=) Authentication-Results: legolas.ozlabs.org; dkim=fail reason="signature verification failed" (2048-bit key; unprotected) header.d=canonical.com header.i=@canonical.com header.a=rsa-sha256 header.s=20210705 header.b=ilTsbaqH; dkim-atps=neutral Received: from huckleberry.canonical.com (huckleberry.canonical.com [91.189.94.19]) (using TLSv1.2 with cipher ECDHE-ECDSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by legolas.ozlabs.org (Postfix) with ESMTPS id 4N6HsZ31CXz23lg for ; Wed, 9 Nov 2022 06:17:05 +1100 (AEDT) Received: from localhost ([127.0.0.1] helo=huckleberry.canonical.com) by huckleberry.canonical.com with esmtp (Exim 4.86_2) (envelope-from ) id 1osU5b-0002Gx-0d; Tue, 08 Nov 2022 19:16:51 +0000 Received: from smtp-relay-internal-0.internal ([10.131.114.225] helo=smtp-relay-internal-0.canonical.com) by huckleberry.canonical.com with esmtps (TLS1.2:ECDHE_RSA_AES_128_GCM_SHA256:128) (Exim 4.86_2) (envelope-from ) id 1osU5Y-0002Gp-T5 for kernel-team@lists.ubuntu.com; Tue, 08 Nov 2022 19:16:48 +0000 Received: from mail-il1-f197.google.com (mail-il1-f197.google.com [209.85.166.197]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature RSA-PSS (2048 bits) server-digest SHA256) (No client certificate requested) by smtp-relay-internal-0.canonical.com (Postfix) with ESMTPS id C0EA03F129 for ; Tue, 8 Nov 2022 19:16:47 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=canonical.com; s=20210705; t=1667935007; bh=9NdXdsCZ74MLqkrc6zJyg4CL6LlK1qcATwhRgFWniTo=; h=From:To:Subject:Date:Message-Id:MIME-Version; b=ilTsbaqHEaCsjOYBLwRldl7sroEqAiOF2By771sKXicAYD0eK4NEWXUNWxnYRBDZX qJdhUz4Qj12WvNUD18kzYKVJZxY6NoqatlDN2qvSrI65AbE5oqHDUec8OWEa0JppRA PWW9awzUb3g+KOPJBkZg9jkAtUE4z+YFJUz+1RxtOu2DRmZ+5zG2JSo5KdVMgEQKeK uwtBKVxpQCsYf1Tcy3nTWjD/nKttrWxHas6YQYPefWNIgCMzky1dJTAvmf/UqAoqsN VPnXECiU6INiCuSNnvtWVFbygG9SodVRNWOPxYVDpO8pxFXTE+gMNVIBHQX3Jp1wSD xu8o74604wcEA== Received: by mail-il1-f197.google.com with SMTP id h6-20020a92c266000000b00300624bf414so11856621ild.14 for ; Tue, 08 Nov 2022 11:16:47 -0800 (PST) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20210112; h=content-transfer-encoding:mime-version:message-id:date:subject:to :from:x-gm-message-state:from:to:cc:subject:date:message-id:reply-to; bh=9NdXdsCZ74MLqkrc6zJyg4CL6LlK1qcATwhRgFWniTo=; b=kqmfGuu1pZyFjVJQvROkiyGG+EIDeNk1WxSBFAGqwh1fMfpmt8j3nspab7aGpbBpX3 07JvmiCz5e14imULulz4DRqa6nVLkBGEggnDvk0N4giqNfh9Zv2V5orfflYiPpaq/3Iz jY67gu1BSAcu6kbyNJZrVXCjqImEsDREDcThVb5Eqr0SqRMElXd5a26nEwzLHa+90La4 Xx6ffrtgl52220secdyhNg8MT3MOuYY3XJWeeeskz+54YdN821PSlzB5FQPbpmzgQY/c xbEWLKVtU5Rom9j9PwucBLYIN5o6p2Y4z0mJpXoJho6YpQ603uMKPd/G9wXI8sbzivux x6Hw== X-Gm-Message-State: ACrzQf2OGNZX8N8qxxaRquapcug2NbRfCw9bFbs3p6KbgIPu/AjKQ79I VsdfrJcHFRc6DHTW6rkT3HyGKseZmkZjr5El2UB3OKwDBQv8V3eDRNlbo6AocuVaaLjOomWa2xo Luedd3ZxPBejLVbmDgHkNfKgQEYLWJUVQcs9AXqcFMA== X-Received: by 2002:a5d:89c6:0:b0:6d7:a3a6:c27e with SMTP id a6-20020a5d89c6000000b006d7a3a6c27emr736880iot.96.1667935003111; Tue, 08 Nov 2022 11:16:43 -0800 (PST) X-Google-Smtp-Source: AMsMyM41UUfHE4iPIUnOmCZi8kvSWh8zuaf9pUzloUUrCV1un90MMYJk3WdB6IVqSIumWk7bjc5hPA== X-Received: by 2002:a5d:89c6:0:b0:6d7:a3a6:c27e with SMTP id a6-20020a5d89c6000000b006d7a3a6c27emr736874iot.96.1667935002647; Tue, 08 Nov 2022 11:16:42 -0800 (PST) Received: from smtp.gmail.com (h69-130-246-116.mdtnwi.broadband.dynamic.tds.net. [69.130.246.116]) by smtp.gmail.com with ESMTPSA id b10-20020a920b0a000000b003001147e7f3sm4088249ilf.30.2022.11.08.11.16.42 for (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Tue, 08 Nov 2022 11:16:42 -0800 (PST) From: John Cabaj To: kernel-team@lists.ubuntu.com Subject: [SRU][kinetic:linux-azure][PATCH v2] UBUNTU: SAUCE: PCI: hv: Only reuse existing IRTE allocation for Multi-MSI Date: Tue, 8 Nov 2022 13:16:41 -0600 Message-Id: <20221108191641.47619-1-john.cabaj@canonical.com> X-Mailer: git-send-email 2.34.1 MIME-Version: 1.0 X-BeenThere: kernel-team@lists.ubuntu.com X-Mailman-Version: 2.1.20 Precedence: list List-Id: Kernel team discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: kernel-team-bounces@lists.ubuntu.com Sender: "kernel-team" From: Dexuan Cui BugLink: https://bugs.launchpad.net/bugs/1995408 Jeffrey added Multi-MSI support to the pci-hyperv driver by the 4 patches: 08e61e861a0e ("PCI: hv: Fix multi-MSI to allow more than one MSI vector") 455880dfe292 ("PCI: hv: Fix hv_arch_irq_unmask() for multi-MSI") b4b77778ecc5 ("PCI: hv: Reuse existing IRTE allocation in compose_msi_msg()") a2bad844a67b ("PCI: hv: Fix interrupt mapping for multi-MSI") It turns out that the third patch (b4b77778ecc5) causes a performance regression because all the interrupts now happen on 1 physical CPU (or two pCPUs, if one pCPU doesn't have enough vectors). When a guest has many PCI devices, it may suffer from soft lockups if the workload is heavy, e.g., see https://lwn.net/ml/linux-kernel/20220804025104.15673-1-decui@microsoft.com/ Commit b4b77778ecc5 itself is good. The real issue is that the hypercall in hv_irq_unmask() -> hv_arch_irq_unmask() -> hv_do_hypercall(HVCALL_RETARGET_INTERRUPT...) only changes the target virtual CPU rather than physical CPU; with b4b77778ecc5, the pCPU is determined only once in hv_compose_msi_msg() where only vCPU0 is specified; consequently the hypervisor only uses 1 target pCPU for all the interrupts. Note: before b4b77778ecc5, the pCPU is determined twice, and when the pCPU is determinted the second time, the vCPU in the effective affinity mask is used (i.e., it isn't always vCPU0), so the hypervisor chooses a different pCPU for each interrupt. The hypercall will be fixed in future to update the pCPU as well, but that will take quite a while, so let's restore the old behavior in hv_compose_msi_msg(), i.e., don't reuse the existing IRTE allocation for single-MSI and MSI-X; for multi-MSI, we choose the vCPU in a round-robin manner for each PCI device, so the interrupts of different devices can happen on different pCPUs, though the interrupts of each device happen on some single pCPU. The hypercall fix may not be backported to all old versions of Hyper-V, so we want to have this guest side change for ever (or at least till we're sure the old affected versions of Hyper-V are no longer supported). Fixes: b4b77778ecc5 ("PCI: hv: Reuse existing IRTE allocation in compose_msi_msg()") Co-developed-by: Jeffrey Hugo Signed-off-by: Jeffrey Hugo Co-developed-by: Carl Vanderlip Signed-off-by: Carl Vanderlip Signed-off-by: Dexuan Cui Signed-off-by: John Cabaj Acked-by: Tim Gardner Acked-by: Stefan Bader --- drivers/pci/controller/pci-hyperv.c | 61 ++++++++++++++++++++++------- 1 file changed, 46 insertions(+), 15 deletions(-) diff --git a/drivers/pci/controller/pci-hyperv.c b/drivers/pci/controller/pci-hyperv.c index db814f7b93ba..c10548243d65 100644 --- a/drivers/pci/controller/pci-hyperv.c +++ b/drivers/pci/controller/pci-hyperv.c @@ -1613,7 +1613,7 @@ static void hv_pci_compose_compl(void *context, struct pci_response *resp, } static u32 hv_compose_msi_req_v1( - struct pci_create_interrupt *int_pkt, struct cpumask *affinity, + struct pci_create_interrupt *int_pkt, u32 slot, u8 vector, u8 vector_count) { int_pkt->message_type.type = PCI_CREATE_INTERRUPT_MESSAGE; @@ -1640,18 +1640,39 @@ static int hv_compose_msi_req_get_cpu(struct cpumask *affinity) return cpumask_first_and(affinity, cpu_online_mask); } -static u32 hv_compose_msi_req_v2( - struct pci_create_interrupt2 *int_pkt, struct cpumask *affinity, - u32 slot, u8 vector, u8 vector_count) +/* + * Make sure the dummy vCPU values for multi-MSI don't all point to vCPU0. + */ +static int hv_compose_multi_msi_req_get_cpu(void) { + static DEFINE_SPINLOCK(multi_msi_cpu_lock); + + /* -1 means starting with CPU 0 */ + static int cpu_next = -1; + + unsigned long flags; int cpu; + spin_lock_irqsave(&multi_msi_cpu_lock, flags); + + cpu_next = cpumask_next_wrap(cpu_next, cpu_online_mask, nr_cpu_ids, + false); + cpu = cpu_next; + + spin_unlock_irqrestore(&multi_msi_cpu_lock, flags); + + return cpu; +} + +static u32 hv_compose_msi_req_v2( + struct pci_create_interrupt2 *int_pkt, int cpu, + u32 slot, u8 vector, u8 vector_count) +{ int_pkt->message_type.type = PCI_CREATE_INTERRUPT_MESSAGE2; int_pkt->wslot.slot = slot; int_pkt->int_desc.vector = vector; int_pkt->int_desc.vector_count = vector_count; int_pkt->int_desc.delivery_mode = DELIVERY_MODE; - cpu = hv_compose_msi_req_get_cpu(affinity); int_pkt->int_desc.processor_array[0] = hv_cpu_number_to_vp_number(cpu); int_pkt->int_desc.processor_count = 1; @@ -1660,18 +1681,15 @@ static u32 hv_compose_msi_req_v2( } static u32 hv_compose_msi_req_v3( - struct pci_create_interrupt3 *int_pkt, struct cpumask *affinity, + struct pci_create_interrupt3 *int_pkt, int cpu, u32 slot, u32 vector, u8 vector_count) { - int cpu; - int_pkt->message_type.type = PCI_CREATE_INTERRUPT_MESSAGE3; int_pkt->wslot.slot = slot; int_pkt->int_desc.vector = vector; int_pkt->int_desc.reserved = 0; int_pkt->int_desc.vector_count = vector_count; int_pkt->int_desc.delivery_mode = DELIVERY_MODE; - cpu = hv_compose_msi_req_get_cpu(affinity); int_pkt->int_desc.processor_array[0] = hv_cpu_number_to_vp_number(cpu); int_pkt->int_desc.processor_count = 1; @@ -1710,12 +1728,18 @@ static void hv_compose_msi_msg(struct irq_data *data, struct msi_msg *msg) struct pci_create_interrupt3 v3; } int_pkts; } __packed ctxt; + bool multi_msi; u64 trans_id; u32 size; int ret; + int cpu; + + msi_desc = irq_data_get_msi_desc(data); + multi_msi = !msi_desc->pci.msi_attrib.is_msix && + msi_desc->nvec_used > 1; /* Reuse the previous allocation */ - if (data->chip_data) { + if (data->chip_data && multi_msi) { int_desc = data->chip_data; msg->address_hi = int_desc->address >> 32; msg->address_lo = int_desc->address & 0xffffffff; @@ -1723,7 +1747,6 @@ static void hv_compose_msi_msg(struct irq_data *data, struct msi_msg *msg) return; } - msi_desc = irq_data_get_msi_desc(data); pdev = msi_desc_to_pci_dev(msi_desc); dest = irq_data_get_effective_affinity_mask(data); pbus = pdev->bus; @@ -1733,11 +1756,18 @@ static void hv_compose_msi_msg(struct irq_data *data, struct msi_msg *msg) if (!hpdev) goto return_null_message; + /* Free any previous message that might have already been composed. */ + if (data->chip_data && !multi_msi) { + int_desc = data->chip_data; + data->chip_data = NULL; + hv_int_desc_free(hpdev, int_desc); + } + int_desc = kzalloc(sizeof(*int_desc), GFP_ATOMIC); if (!int_desc) goto drop_reference; - if (!msi_desc->pci.msi_attrib.is_msix && msi_desc->nvec_used > 1) { + if (multi_msi) { /* * If this is not the first MSI of Multi MSI, we already have * a mapping. Can exit early. @@ -1762,9 +1792,11 @@ static void hv_compose_msi_msg(struct irq_data *data, struct msi_msg *msg) */ vector = 32; vector_count = msi_desc->nvec_used; + cpu = hv_compose_multi_msi_req_get_cpu(); } else { vector = hv_msi_get_int_vector(data); vector_count = 1; + cpu = hv_compose_msi_req_get_cpu(dest); } memset(&ctxt, 0, sizeof(ctxt)); @@ -1775,7 +1807,6 @@ static void hv_compose_msi_msg(struct irq_data *data, struct msi_msg *msg) switch (hbus->protocol_version) { case PCI_PROTOCOL_VERSION_1_1: size = hv_compose_msi_req_v1(&ctxt.int_pkts.v1, - dest, hpdev->desc.win_slot.slot, vector, vector_count); @@ -1784,7 +1815,7 @@ static void hv_compose_msi_msg(struct irq_data *data, struct msi_msg *msg) case PCI_PROTOCOL_VERSION_1_2: case PCI_PROTOCOL_VERSION_1_3: size = hv_compose_msi_req_v2(&ctxt.int_pkts.v2, - dest, + cpu, hpdev->desc.win_slot.slot, vector, vector_count); @@ -1792,7 +1823,7 @@ static void hv_compose_msi_msg(struct irq_data *data, struct msi_msg *msg) case PCI_PROTOCOL_VERSION_1_4: size = hv_compose_msi_req_v3(&ctxt.int_pkts.v3, - dest, + cpu, hpdev->desc.win_slot.slot, vector, vector_count);