From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:51998) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1emfsz-0001hl-GM for qemu-devel@nongnu.org; Fri, 16 Feb 2018 08:17:10 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1emfsy-0006D2-7M for qemu-devel@nongnu.org; Fri, 16 Feb 2018 08:17:09 -0500 Received: from mx3-rdu2.redhat.com ([66.187.233.73]:44952 helo=mx1.redhat.com) by eggs.gnu.org with esmtps (TLS1.0:DHE_RSA_AES_256_CBC_SHA1:32) (Exim 4.71) (envelope-from ) id 1emfsy-0006CM-31 for qemu-devel@nongnu.org; Fri, 16 Feb 2018 08:17:08 -0500 Received: from smtp.corp.redhat.com (int-mx06.intmail.prod.int.rdu2.redhat.com [10.11.54.6]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by mx1.redhat.com (Postfix) with ESMTPS id B3E3C404085B for ; Fri, 16 Feb 2018 13:17:07 +0000 (UTC) From: "Dr. David Alan Gilbert (git)" Date: Fri, 16 Feb 2018 13:16:22 +0000 Message-Id: <20180216131625.9639-27-dgilbert@redhat.com> In-Reply-To: <20180216131625.9639-1-dgilbert@redhat.com> References: <20180216131625.9639-1-dgilbert@redhat.com> Subject: [Qemu-devel] [PATCH v3 26/29] vhost: Huge page align and merge List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: qemu-devel@nongnu.org, maxime.coquelin@redhat.com, marcandre.lureau@redhat.com, peterx@redhat.com, imammedo@redhat.com, mst@redhat.com Cc: quintela@redhat.com, aarcange@redhat.com From: "Dr. David Alan Gilbert" Align RAMBlocks to page size alignment, and adjust the merging code to deal with partial overlap due to that alignment. This is needed for postcopy so that we can place/fetch whole hugepages when under userfault. Signed-off-by: Dr. David Alan Gilbert --- hw/virtio/trace-events | 3 ++- hw/virtio/vhost.c | 66 ++++++++++++++++++++++++++++++++++++++++++-------- 2 files changed, 58 insertions(+), 11 deletions(-) diff --git a/hw/virtio/trace-events b/hw/virtio/trace-events index 857c495e65..1422ff03ab 100644 --- a/hw/virtio/trace-events +++ b/hw/virtio/trace-events @@ -3,7 +3,8 @@ # hw/virtio/vhost.c vhost_commit(bool started, bool changed) "Started: %d Changed: %d" vhost_region_add_section(const char *name, uint64_t gpa, uint64_t size, uint64_t host) "%s: 0x%"PRIx64"+0x%"PRIx64" @ 0x%"PRIx64 -vhost_region_add_section_abut(const char *name, uint64_t new_size) "%s: 0x%"PRIx64 +vhost_region_add_section_merge(const char *name, uint64_t new_size, uint64_t gpa, uint64_t owr) "%s: size: 0x%"PRIx64 " gpa: 0x%"PRIx64 " owr: 0x%"PRIx64 +vhost_region_add_section_aligned(const char *name, uint64_t gpa, uint64_t size, uint64_t host) "%s: 0x%"PRIx64"+0x%"PRIx64" @ 0x%"PRIx64 vhost_section(const char *name, int r) "%s:%d" # hw/virtio/vhost-user.c diff --git a/hw/virtio/vhost.c b/hw/virtio/vhost.c index 4a44e6e6bf..f39450d39e 100644 --- a/hw/virtio/vhost.c +++ b/hw/virtio/vhost.c @@ -518,10 +518,28 @@ static void vhost_region_add_section(struct vhost_dev *dev, uint64_t mrs_gpa = section->offset_within_address_space; uintptr_t mrs_host = (uintptr_t)memory_region_get_ram_ptr(section->mr) + section->offset_within_region; + RAMBlock *mrs_rb = section->mr->ram_block; + size_t mrs_page = qemu_ram_pagesize(mrs_rb); trace_vhost_region_add_section(section->mr->name, mrs_gpa, mrs_size, mrs_host); + /* Round the section to it's page size */ + /* First align the start down to a page boundary */ + uint64_t alignage = mrs_host & (mrs_page - 1); + if (alignage) { + mrs_host -= alignage; + mrs_size += alignage; + mrs_gpa -= alignage; + } + /* Now align the size up to a page boundary */ + alignage = mrs_size & (mrs_page - 1); + if (alignage) { + mrs_size += mrs_page - alignage; + } + trace_vhost_region_add_section_aligned(section->mr->name, mrs_gpa, mrs_size, + mrs_host); + if (dev->n_tmp_sections) { /* Since we already have at least one section, lets see if * this extends it; since we're scanning in order, we only @@ -538,18 +556,46 @@ static void vhost_region_add_section(struct vhost_dev *dev, prev_sec->offset_within_region; uint64_t prev_host_end = range_get_last(prev_host_start, prev_size); - if (prev_gpa_end + 1 == mrs_gpa && - prev_host_end + 1 == mrs_host && - section->mr == prev_sec->mr && - (!dev->vhost_ops->vhost_backend_can_merge || - dev->vhost_ops->vhost_backend_can_merge(dev, + if (mrs_gpa <= (prev_gpa_end + 1)) { + /* OK, looks like overlapping/intersecting - it's possible that + * the rounding to page sizes has made them overlap, but they should + * match up in the same RAMBlock if they do. + */ + if (mrs_gpa < prev_gpa_start) { + error_report("%s:Section rounded to %"PRIx64 + " prior to previous %"PRIx64, + __func__, mrs_gpa, prev_gpa_start); + /* A way to cleanly fail here would be better */ + return; + } + /* Offset from the start of the previous GPA to this GPA */ + size_t offset = mrs_gpa - prev_gpa_start; + + if (prev_host_start + offset == mrs_host && + section->mr == prev_sec->mr && + (!dev->vhost_ops->vhost_backend_can_merge || + dev->vhost_ops->vhost_backend_can_merge(dev, mrs_host, mrs_size, prev_host_start, prev_size))) { - /* The two sections abut */ - need_add = false; - prev_sec->size = int128_add(prev_sec->size, section->size); - trace_vhost_region_add_section_abut(section->mr->name, - mrs_size + prev_size); + uint64_t max_end = MAX(prev_host_end, mrs_host + mrs_size); + need_add = false; + prev_sec->offset_within_address_space = + MIN(prev_gpa_start, mrs_gpa); + prev_sec->offset_within_region = + MIN(prev_host_start, mrs_host) - + (uintptr_t)memory_region_get_ram_ptr(prev_sec->mr); + prev_sec->size = int128_make64(max_end - MIN(prev_host_start, + mrs_host)); + trace_vhost_region_add_section_merge(section->mr->name, + int128_get64(prev_sec->size), + prev_sec->offset_within_address_space, + prev_sec->offset_within_region); + } else { + error_report("%s: Overlapping but not coherent sections " + "at %"PRIx64, + __func__, mrs_gpa); + return; + } } } -- 2.14.3