From patchwork Thu Feb 21 13:12:59 2013 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Cornelia Huck X-Patchwork-Id: 222284 Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@bilbo.ozlabs.org Received: from lists.gnu.org (lists.gnu.org [208.118.235.17]) (using TLSv1 with cipher AES256-SHA (256/256 bits)) (Client did not present a certificate) by ozlabs.org (Postfix) with ESMTPS id 147F02C0086 for ; Fri, 22 Feb 2013 00:13:39 +1100 (EST) Received: from localhost ([::1]:57153 helo=lists.gnu.org) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1U8Vxx-0004pa-30 for incoming@patchwork.ozlabs.org; Thu, 21 Feb 2013 08:13:37 -0500 Received: from eggs.gnu.org ([208.118.235.92]:37469) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1U8VxU-0004bf-Ft for qemu-devel@nongnu.org; Thu, 21 Feb 2013 08:13:19 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1U8VxR-0000V5-O1 for qemu-devel@nongnu.org; Thu, 21 Feb 2013 08:13:08 -0500 Received: from e06smtp14.uk.ibm.com ([195.75.94.110]:59275) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1U8VxR-0000Tm-8n for qemu-devel@nongnu.org; Thu, 21 Feb 2013 08:13:05 -0500 Received: from /spool/local by e06smtp14.uk.ibm.com with IBM ESMTP SMTP Gateway: Authorized Use Only! Violators will be prosecuted for from ; Thu, 21 Feb 2013 13:10:03 -0000 Received: from d06dlp01.portsmouth.uk.ibm.com (9.149.20.13) by e06smtp14.uk.ibm.com (192.168.101.144) with IBM ESMTP SMTP Gateway: Authorized Use Only! Violators will be prosecuted; Thu, 21 Feb 2013 13:10:03 -0000 Received: from b06cxnps4076.portsmouth.uk.ibm.com (d06relay13.portsmouth.uk.ibm.com [9.149.109.198]) by d06dlp01.portsmouth.uk.ibm.com (Postfix) with ESMTP id 3C75717D8068 for ; Thu, 21 Feb 2013 13:13:31 +0000 (GMT) Received: from d06av07.portsmouth.uk.ibm.com (d06av07.portsmouth.uk.ibm.com [9.149.37.248]) by b06cxnps4076.portsmouth.uk.ibm.com (8.13.8/8.13.8/NCO v10.0) with ESMTP id r1LDCqDi23003252 for ; Thu, 21 Feb 2013 13:12:52 GMT Received: from d06av07.portsmouth.uk.ibm.com (d06av07.portsmouth.uk.ibm.com [127.0.0.1]) by d06av07.portsmouth.uk.ibm.com (8.14.4/8.13.1/NCO v10.0 AVout) with ESMTP id r1LC9QwN016428 for ; Thu, 21 Feb 2013 07:09:26 -0500 Received: from tuxmaker.boeblingen.de.ibm.com (tuxmaker.boeblingen.de.ibm.com [9.152.85.9]) by d06av07.portsmouth.uk.ibm.com (8.14.4/8.13.1/NCO v10.0 AVin) with ESMTP id r1LC9Qo8016380; Thu, 21 Feb 2013 07:09:26 -0500 From: Cornelia Huck To: KVM Date: Thu, 21 Feb 2013 14:12:59 +0100 Message-Id: <1361452380-50398-3-git-send-email-cornelia.huck@de.ibm.com> X-Mailer: git-send-email 1.7.12.4 In-Reply-To: <1361452380-50398-1-git-send-email-cornelia.huck@de.ibm.com> References: <1361452380-50398-1-git-send-email-cornelia.huck@de.ibm.com> X-Content-Scanned: Fidelis XPS MAILER x-cbid: 13022113-1948-0000-0000-0000045DF0E3 X-detected-operating-system: by eggs.gnu.org: GNU/Linux 2.4.x-2.6.x [generic] X-Received-From: 195.75.94.110 Cc: linux-s390 , qemu-devel Subject: [Qemu-devel] [RFC PATCH 2/3] KVM: Generalize ioeventfds. X-BeenThere: qemu-devel@nongnu.org X-Mailman-Version: 2.1.14 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: qemu-devel-bounces+incoming=patchwork.ozlabs.org@nongnu.org Sender: qemu-devel-bounces+incoming=patchwork.ozlabs.org@nongnu.org Currently, ioeventfds are designed to work on architectures that can trap I/O memory writes. This won't work for architectures like s390, however; therefore provide a way for architectures to override this with an architecture-specific implementation. Signed-off-by: Cornelia Huck --- Documentation/virtual/kvm/api.txt | 5 +- include/linux/kvm_host.h | 13 +++ include/uapi/linux/kvm.h | 4 +- virt/kvm/eventfd.c | 181 ++++++++++++++++++++++++++------------ 4 files changed, 146 insertions(+), 57 deletions(-) diff --git a/Documentation/virtual/kvm/api.txt b/Documentation/virtual/kvm/api.txt index c2534c3..13c038c 100644 --- a/Documentation/virtual/kvm/api.txt +++ b/Documentation/virtual/kvm/api.txt @@ -1465,7 +1465,7 @@ struct kvm_ioeventfd { __u32 len; /* 1, 2, 4, or 8 bytes */ __s32 fd; __u32 flags; - __u8 pad[36]; + __u8 data[36]; /* for architecture-specific data */ }; The following flags are defined: @@ -1473,10 +1473,13 @@ The following flags are defined: #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch) #define KVM_IOEVENTFD_FLAG_PIO (1 << kvm_ioeventfd_flag_nr_pio) #define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << kvm_ioeventfd_flag_nr_deassign) +#define KVM_IOEVENTFD_FLAG_ARCH (1 << kvm_ioeventfd_flag_nr_arch) If datamatch flag is set, the event will be signaled only if the written value to the registered address is equal to datamatch in struct kvm_ioeventfd. +If the arch flag is set, the eventfd will use the data field. If the arch flag +is not set, the data field is not valid. 4.60 KVM_DIRTY_TLB diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h index 722cae7..d7965fb 100644 --- a/include/linux/kvm_host.h +++ b/include/linux/kvm_host.h @@ -905,8 +905,21 @@ static inline void kvm_free_irq_routing(struct kvm *kvm) {} #ifdef CONFIG_HAVE_KVM_EVENTFD +struct kvm_arch_ioeventfd; void kvm_eventfd_init(struct kvm *kvm); int kvm_ioeventfd(struct kvm *kvm, struct kvm_ioeventfd *args); +struct eventfd_ctx *kvm_ioeventfd_get_eventfd(struct kvm_arch_ioeventfd *arch); +int kvm_arch_ioeventfd_check(struct kvm_ioeventfd *args); +void kvm_arch_ioeventfd_init(struct kvm_arch_ioeventfd *arch, + struct kvm_ioeventfd *args); +int kvm_arch_ioeventfd_activate(struct kvm *kvm, + struct kvm_arch_ioeventfd *arch, + struct kvm_ioeventfd *args); +bool kvm_arch_ioeventfd_match(struct kvm_arch_ioeventfd *arch, + struct kvm_arch_ioeventfd *to_match); +bool kvm_arch_ioeventfd_match_and_release(struct kvm *kvm, + struct kvm_arch_ioeventfd *arch, + struct kvm_ioeventfd *args); #ifdef CONFIG_HAVE_KVM_IRQCHIP int kvm_irqfd(struct kvm *kvm, struct kvm_irqfd *args); diff --git a/include/uapi/linux/kvm.h b/include/uapi/linux/kvm.h index 9a2db57..dfd444b 100644 --- a/include/uapi/linux/kvm.h +++ b/include/uapi/linux/kvm.h @@ -448,12 +448,14 @@ enum { kvm_ioeventfd_flag_nr_datamatch, kvm_ioeventfd_flag_nr_pio, kvm_ioeventfd_flag_nr_deassign, + kvm_ioeventfd_flag_nr_arch, kvm_ioeventfd_flag_nr_max, }; #define KVM_IOEVENTFD_FLAG_DATAMATCH (1 << kvm_ioeventfd_flag_nr_datamatch) #define KVM_IOEVENTFD_FLAG_PIO (1 << kvm_ioeventfd_flag_nr_pio) #define KVM_IOEVENTFD_FLAG_DEASSIGN (1 << kvm_ioeventfd_flag_nr_deassign) +#define KVM_IOEVENTFD_FLAG_ARCH (1 << kvm_ioeventfd_flag_nr_arch) #define KVM_IOEVENTFD_VALID_FLAG_MASK ((1 << kvm_ioeventfd_flag_nr_max) - 1) @@ -463,7 +465,7 @@ struct kvm_ioeventfd { __u32 len; /* 1, 2, 4, or 8 bytes */ __s32 fd; __u32 flags; - __u8 pad[36]; + __u8 data[36]; /* for architecture-specific data */ }; /* for KVM_ENABLE_CAP */ diff --git a/virt/kvm/eventfd.c b/virt/kvm/eventfd.c index b6eea5c..63fe454 100644 --- a/virt/kvm/eventfd.c +++ b/virt/kvm/eventfd.c @@ -568,23 +568,37 @@ module_exit(irqfd_module_exit); * * userspace can register a PIO/MMIO address with an eventfd for receiving * notification when the memory has been touched. + * + * Architectures that use a notification mechanism different from memory + * writes may override this with architecture-specific callbacks. * -------------------------------------------------------------------- */ - -struct _ioeventfd { - struct list_head list; +#ifndef __KVM_HAVE_ARCH_IOEVENTFD +struct kvm_arch_ioeventfd { u64 addr; int length; - struct eventfd_ctx *eventfd; u64 datamatch; struct kvm_io_device dev; bool wildcard; }; +#endif + +struct _ioeventfd { + struct list_head list; + struct eventfd_ctx *eventfd; + struct kvm_arch_ioeventfd arch; +}; -static inline struct _ioeventfd * -to_ioeventfd(struct kvm_io_device *dev) +static inline struct _ioeventfd *to_ioeventfd(struct kvm_arch_ioeventfd *arch) { - return container_of(dev, struct _ioeventfd, dev); + return container_of(arch, struct _ioeventfd, arch); +} + +struct eventfd_ctx *kvm_ioeventfd_get_eventfd(struct kvm_arch_ioeventfd *arch) +{ + struct _ioeventfd *p = to_ioeventfd(arch); + + return p->eventfd; } static void @@ -595,16 +609,18 @@ ioeventfd_release(struct _ioeventfd *p) kfree(p); } +#ifndef __KVM_HAVE_ARCH_IOEVENTFD static bool -ioeventfd_in_range(struct _ioeventfd *p, gpa_t addr, int len, const void *val) +ioeventfd_in_range(struct kvm_arch_ioeventfd *arch, gpa_t addr, int len, + const void *val) { u64 _val; - if (!(addr == p->addr && len == p->length)) + if (!(addr == arch->addr && len == arch->length)) /* address-range must be precise for a hit */ return false; - if (p->wildcard) + if (arch->wildcard) /* all else equal, wildcard is always a hit */ return true; @@ -629,7 +645,13 @@ ioeventfd_in_range(struct _ioeventfd *p, gpa_t addr, int len, const void *val) return false; } - return _val == p->datamatch ? true : false; + return _val == arch->datamatch ? true : false; +} + +static inline struct kvm_arch_ioeventfd * +to_arch_ioeventfd(struct kvm_io_device *dev) +{ + return container_of(dev, struct kvm_arch_ioeventfd, dev); } /* MMIO/PIO writes trigger an event if the addr/val match */ @@ -637,12 +659,12 @@ static int ioeventfd_write(struct kvm_io_device *this, gpa_t addr, int len, const void *val) { - struct _ioeventfd *p = to_ioeventfd(this); + struct kvm_arch_ioeventfd *arch = to_arch_ioeventfd(this); - if (!ioeventfd_in_range(p, addr, len, val)) + if (!ioeventfd_in_range(arch, addr, len, val)) return -EOPNOTSUPP; - eventfd_signal(p->eventfd, 1); + eventfd_signal(kvm_ioeventfd_get_eventfd(arch), 1); return 0; } @@ -653,7 +675,7 @@ ioeventfd_write(struct kvm_io_device *this, gpa_t addr, int len, static void ioeventfd_destructor(struct kvm_io_device *this) { - struct _ioeventfd *p = to_ioeventfd(this); + struct _ioeventfd *p = to_ioeventfd(to_arch_ioeventfd(this)); ioeventfd_release(p); } @@ -663,6 +685,87 @@ static const struct kvm_io_device_ops ioeventfd_ops = { .destructor = ioeventfd_destructor, }; +int kvm_arch_ioeventfd_check(struct kvm_ioeventfd *args) +{ + if (args->flags & KVM_IOEVENTFD_FLAG_ARCH) + return -EINVAL; + + /* must be natural-word sized */ + switch (args->len) { + case 1: + case 2: + case 4: + case 8: + break; + default: + return -EINVAL; + } + + /* check for range overflow */ + if (args->addr + args->len < args->addr) + return -EINVAL; + + return 0; +} + +void kvm_arch_ioeventfd_init(struct kvm_arch_ioeventfd *arch, + struct kvm_ioeventfd *args) +{ + arch->addr = args->addr; + arch->length = args->len; + + /* The datamatch feature is optional, otherwise this is a wildcard */ + if (args->flags & KVM_IOEVENTFD_FLAG_DATAMATCH) + arch->datamatch = args->datamatch; + else + arch->wildcard = true; +} + +int kvm_arch_ioeventfd_activate(struct kvm *kvm, + struct kvm_arch_ioeventfd *arch, + struct kvm_ioeventfd *args) +{ + int pio = args->flags & KVM_IOEVENTFD_FLAG_PIO; + enum kvm_bus bus_idx = pio ? KVM_PIO_BUS : KVM_MMIO_BUS; + + kvm_iodevice_init(&arch->dev, &ioeventfd_ops); + + return kvm_io_bus_register_dev(kvm, bus_idx, arch->addr, arch->length, + &arch->dev); +} + +bool kvm_arch_ioeventfd_match(struct kvm_arch_ioeventfd *arch, + struct kvm_arch_ioeventfd *to_match) +{ + if (to_match->addr == arch->addr && + to_match->length == arch->length && + (to_match->wildcard || arch->wildcard || + to_match->datamatch == arch->datamatch)) + return true; + return false; +} + +bool kvm_arch_ioeventfd_match_and_release(struct kvm *kvm, + struct kvm_arch_ioeventfd *arch, + struct kvm_ioeventfd *args) +{ + int pio = args->flags & KVM_IOEVENTFD_FLAG_PIO; + enum kvm_bus bus_idx = pio ? KVM_PIO_BUS : KVM_MMIO_BUS; + bool wildcard = !(args->flags & KVM_IOEVENTFD_FLAG_DATAMATCH); + + if (arch->addr != args->addr || + arch->length != args->len || + arch->wildcard != wildcard) + return false; + + if (!arch->wildcard && arch->datamatch != args->datamatch) + return false; + + kvm_io_bus_unregister_dev(kvm, bus_idx, &arch->dev); + return true; +} +#endif + /* assumes kvm->slots_lock held */ static bool ioeventfd_check_collision(struct kvm *kvm, struct _ioeventfd *p) @@ -670,9 +773,7 @@ ioeventfd_check_collision(struct kvm *kvm, struct _ioeventfd *p) struct _ioeventfd *_p; list_for_each_entry(_p, &kvm->ioeventfds, list) - if (_p->addr == p->addr && _p->length == p->length && - (_p->wildcard || p->wildcard || - _p->datamatch == p->datamatch)) + if (kvm_arch_ioeventfd_match(&p->arch, &_p->arch)) return true; return false; @@ -681,26 +782,13 @@ ioeventfd_check_collision(struct kvm *kvm, struct _ioeventfd *p) static int kvm_assign_ioeventfd(struct kvm *kvm, struct kvm_ioeventfd *args) { - int pio = args->flags & KVM_IOEVENTFD_FLAG_PIO; - enum kvm_bus bus_idx = pio ? KVM_PIO_BUS : KVM_MMIO_BUS; struct _ioeventfd *p; struct eventfd_ctx *eventfd; int ret; - /* must be natural-word sized */ - switch (args->len) { - case 1: - case 2: - case 4: - case 8: - break; - default: - return -EINVAL; - } - - /* check for range overflow */ - if (args->addr + args->len < args->addr) - return -EINVAL; + ret = kvm_arch_ioeventfd_check(args); + if (ret) + return ret; /* check for extra flags that we don't understand */ if (args->flags & ~KVM_IOEVENTFD_VALID_FLAG_MASK) @@ -717,15 +805,9 @@ kvm_assign_ioeventfd(struct kvm *kvm, struct kvm_ioeventfd *args) } INIT_LIST_HEAD(&p->list); - p->addr = args->addr; - p->length = args->len; p->eventfd = eventfd; - /* The datamatch feature is optional, otherwise this is a wildcard */ - if (args->flags & KVM_IOEVENTFD_FLAG_DATAMATCH) - p->datamatch = args->datamatch; - else - p->wildcard = true; + kvm_arch_ioeventfd_init(&p->arch, args); mutex_lock(&kvm->slots_lock); @@ -735,10 +817,7 @@ kvm_assign_ioeventfd(struct kvm *kvm, struct kvm_ioeventfd *args) goto unlock_fail; } - kvm_iodevice_init(&p->dev, &ioeventfd_ops); - - ret = kvm_io_bus_register_dev(kvm, bus_idx, p->addr, p->length, - &p->dev); + ret = kvm_arch_ioeventfd_activate(kvm, &p->arch, args); if (ret < 0) goto unlock_fail; @@ -761,8 +840,6 @@ fail: static int kvm_deassign_ioeventfd(struct kvm *kvm, struct kvm_ioeventfd *args) { - int pio = args->flags & KVM_IOEVENTFD_FLAG_PIO; - enum kvm_bus bus_idx = pio ? KVM_PIO_BUS : KVM_MMIO_BUS; struct _ioeventfd *p, *tmp; struct eventfd_ctx *eventfd; int ret = -ENOENT; @@ -774,18 +851,12 @@ kvm_deassign_ioeventfd(struct kvm *kvm, struct kvm_ioeventfd *args) mutex_lock(&kvm->slots_lock); list_for_each_entry_safe(p, tmp, &kvm->ioeventfds, list) { - bool wildcard = !(args->flags & KVM_IOEVENTFD_FLAG_DATAMATCH); - - if (p->eventfd != eventfd || - p->addr != args->addr || - p->length != args->len || - p->wildcard != wildcard) + if (p->eventfd != eventfd) continue; - if (!p->wildcard && p->datamatch != args->datamatch) + if (!kvm_arch_ioeventfd_match_and_release(kvm, &p->arch, args)) continue; - kvm_io_bus_unregister_dev(kvm, bus_idx, &p->dev); ioeventfd_release(p); ret = 0; break;