From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mail-pf1-f172.google.com (mail-pf1-f172.google.com [209.85.210.172]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 6ABCD3D0C1D for ; Tue, 18 Aug 2026 05:12:47 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.210.172 ARC-Seal:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1787029969; cv=none; b=BdxdRda/mhXfIl2/M/uVwLPAtLfmML1U2OeY6tclEbn77vac97Lm7cImm6CIHLJyn+rsOSparQ/6Y6vjzUWZn/42Sqd/0dyBV/nJd0xcSnIp0TaN0+kj7c1QaNIknur+Ea9tvO0XpKN7glvqbQ6K/Uy48VzSdas+49yjK8WS46o= ARC-Message-Signature:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1787029969; c=relaxed/simple; bh=bMPw0FwBRsEunQYzzG6BOXyVL5rIEfi5TeDivlNmsxo=; h=From:Date:Subject:MIME-Version:Content-Type:Message-Id:References: In-Reply-To:To:Cc; b=dbstPHqs/prJNqgL/KT50WU4QciwXU7+4Gv/x1UWIbjs/bugqW+HqIwIQ4ayvrfik7KB1ssx0S1KNb/ecNL6QxCq3oDMRKReKIXshC8GnX6Y8ayrlRe143KqtjaqZvAtXk3x2wiRRqN4wIiiNQyL6phnGQjyZxLqYPDKj2NQr7A= ARC-Authentication-Results:i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=gmail.com; spf=pass smtp.mailfrom=gmail.com; dkim=pass (2048-bit key) header.d=gmail.com header.i=@gmail.com header.b=YnOcY5Ko; arc=none smtp.client-ip=209.85.210.172 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=gmail.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=gmail.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=gmail.com header.i=@gmail.com header.b="YnOcY5Ko" Received: by mail-pf1-f172.google.com with SMTP id d2e1a72fcca58-84f38f3b36eso2726429b3a.1 for ; Mon, 17 Aug 2026 22:12:47 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20251104; t=1787029967; x=1787634767; darn=lists.linux.dev; h=cc:to:in-reply-to:references:message-id:content-transfer-encoding :content-type:mime-version:subject:date:from:from:to:cc:subject:date :message-id:reply-to:content-type; bh=Q4geQjhEDcsmapDV/ZLiGXecFcqlx1SUv1jdKP/WSd8=; b=YnOcY5KoWVhiPavsFHITZX6Lh5OuR00A7OTxJeSGOMPRvQ0UyUS0PFNGSRRQl17/9a aStyeSnU3oHSOznm/hQmzhW89WDNDv6h5GW7lNIbEBZFrl4lCUs28EsH1PblPpvSRrJe RAPJSUge9XGAczSHZNt9C3NEK667kxLlhfz+y+8R8mOykCsJreKWWB/WebO1whTZItOz yF/fgRPQP+4pWsFblkuueEjnhiR/lXzoPCkpSygpe7x+Bsz54UJcujScRo2GOn/DG4X+ IxRrcZIXB0KNSqtM31ByuwXMdTVAmS1BMQzbj/ukkcw/y99OQRdN8b9DNwJx+JSrPspE 3owQ== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20251104; t=1787029967; x=1787634767; h=cc:to:in-reply-to:references:message-id:content-transfer-encoding :content-type:mime-version:subject:date:from:x-gm-gg :x-gm-message-state:from:to:cc:subject:date:message-id:reply-to :content-type; bh=Q4geQjhEDcsmapDV/ZLiGXecFcqlx1SUv1jdKP/WSd8=; b=UM1CExJJhQbzVEfiz1SDv0mQn+lcwAyuAylCT7qM/iuif12y6ZXhdJzARVDkQ9u8x9 Ti3RpiUjIYixhFpMuvXWrdeaC9tn05xsC2X0wq8EssTwiSrsev4s1MbkimwCD8sLpdIR ceAzXhB0eZtC1qJ6u7a59Bmsxvho/U6L1vi10vT67fXmbNwvRZ3u/4jsb8wZ4yPhC01F x48UIjxEGhqxWiajDlp8Jm+I96mflih3dxgzh9AHlaAgrMRYmeqy6vEYMPVVJDsFJMcP Zu6Oq/A+UqGKBckYArRA7hDaQBboUZ+6rjxs7s8CpJcZRdwH0kkN8kQ6bZdEs3e5KExa 4dGg== X-Forwarded-Encrypted: i=1; AHgh+Rqs+nnKUTNYwmPCS+pbf+9kMFu4b8skt4QPPSfAU6cVPrHymDd4xcgFQHG1uGqyfaH1fdwfcyxBjPk=@lists.linux.dev X-Gm-Message-State: AOJu0YyW4l5PxTNfsdGyEBwt4ap5C6VVabYyjlx6rsRweKF8qnwGTH3V zmi80FNyIwd/HcBGRMhDouH3HuDwwo7bWkVOpBAQRyuL2DiMqceXV+7+ X-Gm-Gg: AR+sD13GwxJ0nESRZnzjyiuIDZZClwWdBhsk9ApPaGaBYLe0ayuOYO8mxplhRMeNn/X h9fW0wnFCbGpTsHn2jQF8Li5N3Blpjp3EflE5hxk1XOSgBKk4pBwwwrp8lAMt+2UtVqXHVlU17Y rdwajQgXi0sLcOjnE0WK1ckgnYGj7jMykMeWb204t1EBxu/bSv0e8flCOTErZnXRDCMHnKQYU6A tZtzPXXrngUJHSOldJeDRhJ602qYrq8+z2mPiMzDrBigJa5LTRFK5lYdnw/MbcDwLHT9Wd7Zimq aJ+GwOWnQjq4ah1oLZl+q+Gnjo7AcsXUs1lECi7Z+dAcqA/1hN3jQXI9462saIoP7CZKWEVxRI2 BQM2wJJacUU+xv1/B+dMbqmMILCljbgjoszRXh30ayMJakKL2aJmwjePz30s/VxInhDG75lOHXh VnbYddThJ92yIqpukMT30CqfIeBpL1PUuKHMjvm9EOLVuWiD1l5Du0ofYV7y3sSvvoZQ== X-Received: by 2002:a05:6a00:1256:b0:845:e04b:565f with SMTP id d2e1a72fcca58-84fde122876mr28993614b3a.14.1787029966608; Mon, 17 Aug 2026 22:12:46 -0700 (PDT) Received: from [127.0.1.1] ([2600:1700:e140:14d0:d043:a40d:25c1:432c]) by smtp.gmail.com with ESMTPSA id d2e1a72fcca58-851b6fc0e84sm1031160b3a.43.2026.08.17.22.12.44 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Mon, 17 Aug 2026 22:12:46 -0700 (PDT) From: Connor Kite Date: Mon, 17 Aug 2026 22:12:25 -0700 Subject: [PATCH RFC v2 10/13] hw/virtio/vhost-user: handle data movement with shadow vqs Precedence: bulk X-Mailing-List: virtio-fs@lists.linux.dev List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: 7bit Message-Id: <20260817-vhost-user-isolated-memory-v2-10-948aae960abb@gmail.com> References: <20260817-vhost-user-isolated-memory-v2-0-948aae960abb@gmail.com> In-Reply-To: <20260817-vhost-user-isolated-memory-v2-0-948aae960abb@gmail.com> To: qemu-devel@nongnu.org Cc: "Michael S. Tsirkin" , Stefano Garzarella , =?utf-8?q?Alex_Benn=C3=A9e?= , Viresh Kumar , Gerd Hoffmann , Mathieu Poirier , Manos Pitsidianakis , Raphael Norwitz , Kevin Wolf , Hanna Reitz , =?utf-8?q?Marc-Andr=C3=A9_Lureau?= , Paolo Bonzini , Fam Zheng , Stefan Hajnoczi , Milan Zamazal , Akihiko Odaki , Dmitry Osipenko , qemu-block@nongnu.org, virtio-fs@lists.linux.dev, "Gonglei (Arei)" , zhenwei pi , =?utf-8?q?Daniel_P=2E_Berrang=C3=A9?= , Eric Blake , Markus Armbruster , Jason Wang , Peter Xu , =?utf-8?q?Eugenio_P=C3=A9rez?= , Alyssa Ross , Demi Marie Obenour , Connor Kite , 20260817233147.2867623-1-connorkite@gmail.com X-Mailer: b4 0.13.0 X-Developer-Signature: v=1; a=ed25519-sha256; t=1787029936; l=14307; i=connorkite@gmail.com; s=20260723; h=from:subject:message-id; bh=bMPw0FwBRsEunQYzzG6BOXyVL5rIEfi5TeDivlNmsxo=; b=l+69CvZv18/AgSDA22GB8a6pNf2J8er2Y8bngQir+ldlLYyWyDZABBr7A+7tpX8JE+L3eSUAL kmnpQ+dTI9/DssO6GBTKNuQcJZwVTjSAiT/L4vaKujaE7mwosTy1fcb X-Developer-Key: i=connorkite@gmail.com; a=ed25519; pk=xg/3N8AntFCYogaIgN5NC/KkT7UlZB6ktPIRliJ3hv4= - Add start logic for shadow virtqueues, which sets vring addresses. - Update logic for sending vring addresses to backend to point to the shadow vrings when isolation mode is active. - Implement handlers for intercepted avail and used descriptors. These handlers copy buffer contents between bounce buffers in the isolation region and the buffers made available by the guest - Implement logic to stop svqs Signed-off-by: Connor Kite --- hw/virtio/vhost-shadow-virtqueue.c | 2 +- hw/virtio/vhost-shadow-virtqueue.h | 1 + hw/virtio/vhost-user.c | 292 ++++++++++++++++++++++++++++++++++++- 3 files changed, 286 insertions(+), 9 deletions(-) diff --git a/hw/virtio/vhost-shadow-virtqueue.c b/hw/virtio/vhost-shadow-virtqueue.c index f54a61439a..80b0828d7c 100644 --- a/hw/virtio/vhost-shadow-virtqueue.c +++ b/hw/virtio/vhost-shadow-virtqueue.c @@ -350,7 +350,7 @@ int vhost_svq_add(VhostShadowVirtqueue *svq, const struct iovec *out_sg, } /* Convenience wrapper to add a guest's element to SVQ */ -static int vhost_svq_add_element(VhostShadowVirtqueue *svq, +int vhost_svq_add_element(VhostShadowVirtqueue *svq, VirtQueueElement *elem) { return vhost_svq_add(svq, elem->out_sg, elem->out_num, elem->out_addr, diff --git a/hw/virtio/vhost-shadow-virtqueue.h b/hw/virtio/vhost-shadow-virtqueue.h index 1e0cc9e5e4..10a8db6c39 100644 --- a/hw/virtio/vhost-shadow-virtqueue.h +++ b/hw/virtio/vhost-shadow-virtqueue.h @@ -183,6 +183,7 @@ VhostShadowVirtqueue *vhost_svq_new(const VhostShadowVirtqueueOps *ops, void vhost_svq_free(gpointer vq); void vhost_svq_set_base_addr(VhostShadowVirtqueue *svq, void *addr); +int vhost_svq_add_element(VhostShadowVirtqueue *svq, VirtQueueElement *elem); G_DEFINE_AUTOPTR_CLEANUP_FUNC(VhostShadowVirtqueue, vhost_svq_free); diff --git a/hw/virtio/vhost-user.c b/hw/virtio/vhost-user.c index ace328f5eb..ca3aaf7a07 100644 --- a/hw/virtio/vhost-user.c +++ b/hw/virtio/vhost-user.c @@ -1364,6 +1364,231 @@ static int init_isolation_regions(struct vhost_dev *dev, return 0; } +static inline int vhost_user_get_reg_idx(struct vhost_dev *dev, hwaddr gpa) +{ + int i; + struct vhost_memory_region *reg; + + for (i = 0; i < dev->mem->nregions; i++) { + reg = dev->mem->regions + i; + + if (gpa >= reg->guest_phys_addr && + reg->guest_phys_addr + reg->memory_size > gpa) { + return i; + } + } + + return -1; +} + +static int vhost_user_svq_handle_used(VhostShadowVirtqueue *svq, + VirtQueueElement *elem, + void *opaque) +{ + uint64_t offset; + void *dst; + void *src; + size_t len, rem; + void *reg_last_addr; + void *reg_first_addr; + struct vhost_dev *dev = opaque; + struct vhost_user *u = dev->opaque; + void *shmem_addr = u->iso_mem_ctx.shared_mem_addr; + size_t shmem_size = u->iso_mem_ctx.size; + const DMAMap *map; + DMAMap needle; + int reg_idx; + + for (int i = 0; i < elem->in_num; i++) { + needle.translated_addr = elem->in_addr[i]; + needle.size = elem->in_sg[i].iov_len - 1; + map = vhost_iova_tree_find_gpa(svq->iova_tree, &needle); + + if (!map) { + return -EFAULT; + } + + offset = needle.translated_addr - map->translated_addr; + src = (void *)int128_get64(int128_add(int128_make64(map->iova + + offset), u->iso_mem_ctx.iso_iova_offset)); + dst = elem->in_sg[i].iov_base; + len = elem->in_sg[i].iov_len; + + /* Confirm that buffer range is fully within iso region */ + if ((uint64_t)src + len - 1 > (uint64_t)shmem_addr + shmem_size - 1 || + (uint64_t)src + len <= (uint64_t)src || + (uint64_t)src < (uint64_t)shmem_addr) { + + return -EFAULT; + } + + reg_idx = vhost_user_get_reg_idx(dev, needle.translated_addr); + if (reg_idx < 0) { + return -EFAULT; + } + + reg_first_addr = (void *)dev->mem->regions[reg_idx].userspace_addr; + reg_last_addr = (void *)(dev->mem->regions[reg_idx].userspace_addr + + dev->mem->regions[reg_idx].memory_size - 1); + + /* Confirm that hva from elem matches expected vhost memory region */ + if (dst < reg_first_addr || dst > reg_last_addr) { + return -EFAULT; + } + + /* + * copy buffer contents from shared memory into guest memory. If + * the buffer extends across region boundaries, it must be split and + * copied to the correct regions. + */ + while ((uint64_t)reg_last_addr - (uint64_t)dst + 1 < len) { + rem = ((uint64_t)dst + len - 1) - (uint64_t)reg_last_addr; + len -= rem; + memcpy(dst, src, len); + + reg_idx++; + if (reg_idx >= dev->mem->nregions) { + return -EFAULT; + } + + src = (void *)((uint64_t)src + len); + dst = (void *)dev->mem->regions[reg_idx].userspace_addr; + len = rem; + reg_first_addr = (void *)dev->mem->regions[reg_idx].userspace_addr; + reg_last_addr = (void *)(dev->mem->regions[reg_idx].userspace_addr + + dev->mem->regions[reg_idx].memory_size - 1); + } + + memcpy(dst, src, len); + } + + return 0; +} + +static int vhost_user_svq_handle_avail(VhostShadowVirtqueue *svq, + VirtQueueElement *elem, + void *opaque) +{ + hwaddr offset; + const DMAMap *map; + DMAMap needle; + void *dst; + void *src; + void *reg_first_addr; + void *reg_last_addr; + size_t len, rem; + struct vhost_dev *dev = opaque; + struct vhost_user *u = dev->opaque; + void *shmem_addr = u->iso_mem_ctx.shared_mem_addr; + size_t shmem_size = u->iso_mem_ctx.size; + int reg_idx; + + for (int i = 0; i < elem->out_num; i++) { + needle.translated_addr = elem->out_addr[i]; + needle.size = elem->out_sg[i].iov_len - 1; + map = vhost_iova_tree_find_gpa(svq->iova_tree, &needle); + + if (!map) { + return -EFAULT; + } + + offset = needle.translated_addr - map->translated_addr; + dst = (void *)int128_get64(int128_add(int128_make64(map->iova + + offset), u->iso_mem_ctx.iso_iova_offset)); + src = elem->out_sg[i].iov_base; + len = elem->out_sg[i].iov_len; + + /* Confirm that buffer range is fully within iso region */ + if ((uint64_t)dst + len - 1 > (uint64_t)shmem_addr + shmem_size - 1 || + (uint64_t)dst + len <= (uint64_t)dst || + (uint64_t)dst < (uint64_t)shmem_addr) { + + return -EFAULT; + } + + reg_idx = vhost_user_get_reg_idx(dev, needle.translated_addr); + if (reg_idx < 0) { + return -EFAULT; + } + + reg_first_addr = (void *)dev->mem->regions[reg_idx].userspace_addr; + reg_last_addr = (void *)(dev->mem->regions[reg_idx].userspace_addr + + dev->mem->regions[reg_idx].memory_size - 1); + + /* Confirm that hva from elem matches expected vhost memory region */ + if (src < reg_first_addr || src > reg_last_addr) { + return -EFAULT; + } + + /* + * copy buffer contents from guest memory into shared memory. If + * the buffer extends across region boundaries, it must be split and + * copied from the correct regions. + */ + while ((uint64_t)reg_last_addr - (uint64_t)src + 1 < len) { + rem = ((uint64_t)src + len - 1) - (uint64_t)reg_last_addr; + len -= rem; + memcpy(dst, src, len); + + reg_idx++; + if (reg_idx >= dev->mem->nregions) { + return -EFAULT; + } + + dst = (void *)((uint64_t)dst + len); + src = (void *)dev->mem->regions[reg_idx].userspace_addr; + len = rem; + + reg_first_addr = (void *)dev->mem->regions[reg_idx].userspace_addr; + reg_last_addr = (void *)(dev->mem->regions[reg_idx].userspace_addr + + dev->mem->regions[reg_idx].memory_size - 1); + } + + memcpy(dst, src, len); + } + + vhost_svq_add_element(svq, elem); + + return 0; +} + +static int vhost_user_get_vq_index(struct vhost_dev *dev, int idx) +{ + assert(idx >= dev->vq_index && idx < dev->vq_index + dev->nvqs); + + return idx; +} + +static int vhost_user_svqs_vring_map(struct vhost_dev *dev) +{ + int ret; + struct vhost_user *u = dev->opaque; + void *vring_base = u->iso_mem_ctx.vring_hva_addr; + uint64_t vring_last = (uint64_t)vring_base + + u->iso_mem_ctx.vring_region_size - 1; + + for (int i = 0; i < u->iso_mem_ctx.shadow_vqs->len; i++) { + vhost_user_get_vq_index(dev, dev->vq_index + i); /* bounds checking */ + + VirtQueue *vq = virtio_get_queue(dev->vdev, dev->vq_index + i); + VhostShadowVirtqueue *svq = + g_ptr_array_index(u->iso_mem_ctx.shadow_vqs, i); + + assert((uint64_t)vring_base + + vhost_svq_vring_total_size(dev->vdev, vq) - 1 <= vring_last); + + vhost_svq_set_base_addr(svq, vring_base); + ret = vhost_svq_start(svq, dev->vdev, vq, u->iso_mem_ctx.tree); + if (ret < 0) { + return ret; + } + vring_base = (void *)((uint64_t)vring_base + + vhost_svq_vring_total_size(dev->vdev, vq)); + } + + return 0; +} + static int vhost_user_set_mem_table(struct vhost_dev *dev, struct vhost_memory *mem) { @@ -1407,6 +1632,11 @@ static int vhost_user_set_mem_table(struct vhost_dev *dev, return ret; } + ret = vhost_user_svqs_vring_map(dev); + if (ret < 0) { + return ret; + } + ret = vhost_user_write(dev, &msg, fds, fd_num); if (ret < 0) { return ret; @@ -1763,13 +1993,6 @@ static int vhost_set_vring_file(struct vhost_dev *dev, return 0; } -static int vhost_user_get_vq_index(struct vhost_dev *dev, int idx) -{ - assert(idx >= dev->vq_index && idx < dev->vq_index + dev->nvqs); - - return idx; -} - static int vhost_user_set_vring_kick(struct vhost_dev *dev, struct vhost_vring_file *file) { @@ -1875,6 +2098,8 @@ static int vhost_user_set_vring_err(struct vhost_dev *dev, static int vhost_user_set_vring_addr(struct vhost_dev *dev, struct vhost_vring_addr *addr) { + struct vhost_user *u = dev->opaque; + ptrdiff_t offset = u->iso_mem_ctx.iso_iova_offset; VhostUserMsg msg = { .hdr.request = VHOST_USER_SET_VRING_ADDR, .hdr.flags = VHOST_USER_VERSION, @@ -1882,6 +2107,25 @@ static int vhost_user_set_vring_addr(struct vhost_dev *dev, .hdr.size = sizeof(msg.payload.addr), }; + if (u->user->memory_isolation) { + int svq_idx; + VhostShadowVirtqueue *svq; + struct vhost_vring_addr svq_addr; + + vhost_user_get_vq_index(dev, addr->index); /* bounds checking */ + svq_idx = addr->index - dev->vq_index; + svq = g_ptr_array_index(u->iso_mem_ctx.shadow_vqs, + svq_idx); + + svq_addr.avail_user_addr = (uint64_t)(uintptr_t)svq->vring.avail - + offset; + svq_addr.desc_user_addr = (uint64_t)(uintptr_t)svq->vring.desc - offset; + svq_addr.used_user_addr = (uint64_t)(uintptr_t)svq->vring.used - offset; + svq_addr.index = addr->index; + + msg.payload.addr = svq_addr; + } + /* * wait for a reply if logging is enabled to make sure * backend is actually logging changes @@ -2857,13 +3101,18 @@ static int vhost_user_postcopy_notifier(NotifierWithReturn *notifier, return 0; } +static const VhostShadowVirtqueueOps vhost_user_svq_ops = { + .avail_handler = vhost_user_svq_handle_avail, + .used_callback = vhost_user_svq_handle_used +}; + static void vhost_user_init_svq(struct vhost_dev *dev, struct vhost_user *u) { /*Modified from vhost-vdpa*/ u->iso_mem_ctx.shadow_vqs = g_ptr_array_new_full(dev->nvqs, vhost_svq_free); for (int i = 0; i < dev->nvqs; i++) { VhostShadowVirtqueue *svq; - svq = vhost_svq_new(NULL, NULL); + svq = vhost_svq_new(&vhost_user_svq_ops, dev); g_ptr_array_add(u->iso_mem_ctx.shadow_vqs, svq); } } @@ -3562,8 +3811,35 @@ void vhost_user_async_close(DeviceState *d, } } +static void vhost_user_svqs_stop(struct vhost_dev *dev) +{ + struct vhost_user *u = dev->opaque; + for (int i = 0; i < u->iso_mem_ctx.shadow_vqs->len; i++) { + vhost_svq_stop(g_ptr_array_index(u->iso_mem_ctx.shadow_vqs, i)); + } +} + static int vhost_user_dev_start(struct vhost_dev *dev, bool started) { + struct vhost_user *u = dev->opaque; + if (u->user->memory_isolation) { + if (vhost_dev_has_iommu(dev)) { + error_report("Memory isolation is not supported with IOMMU enabled." + "Please disable one and try again"); + return -1; + } + + if (virtio_vdev_has_feature(dev->vdev, VIRTIO_F_RING_PACKED)) { + error_report("Memory isolation is not supported with packed" + "vrings. Please use split vrings"); + return -1; + } + + if (!started) { + vhost_user_svqs_stop(dev); + } + } + if (!vhost_user_has_protocol_feature(dev, VHOST_USER_PROTOCOL_F_STATUS)) { return 0; } -- 2.43.0