From patchwork Mon May 21 19:35:01 2018 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Stefano Stabellini X-Patchwork-Id: 917824 Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@bilbo.ozlabs.org Authentication-Results: ozlabs.org; spf=pass (mailfrom) smtp.mailfrom=nongnu.org (client-ip=2001:4830:134:3::11; helo=lists.gnu.org; envelope-from=qemu-devel-bounces+incoming=patchwork.ozlabs.org@nongnu.org; receiver=) Authentication-Results: ozlabs.org; dmarc=fail (p=none dis=none) header.from=kernel.org Authentication-Results: ozlabs.org; dkim=fail reason="signature verification failed" (1024-bit key; unprotected) header.d=kernel.org header.i=@kernel.org header.b="gqmG4/8R"; dkim-atps=neutral Received: from lists.gnu.org (lists.gnu.org [IPv6:2001:4830:134:3::11]) (using TLSv1 with cipher AES256-SHA (256/256 bits)) (No client certificate requested) by ozlabs.org (Postfix) with ESMTPS id 40qTs61qz1z9s1b for ; Tue, 22 May 2018 05:50:18 +1000 (AEST) Received: from localhost ([::1]:52219 helo=lists.gnu.org) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1fKqox-0004gC-Ud for incoming@patchwork.ozlabs.org; Mon, 21 May 2018 15:50:15 -0400 Received: from eggs.gnu.org ([2001:4830:134:3::10]:41788) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1fKqaS-0000hD-ET for qemu-devel@nongnu.org; Mon, 21 May 2018 15:35:18 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1fKqaO-0003Id-VU for qemu-devel@nongnu.org; Mon, 21 May 2018 15:35:16 -0400 Received: from mail.kernel.org ([198.145.29.99]:60370) by eggs.gnu.org with esmtps (TLS1.0:DHE_RSA_AES_256_CBC_SHA1:32) (Exim 4.71) (envelope-from ) id 1fKqaO-0003H1-JB for qemu-devel@nongnu.org; Mon, 21 May 2018 15:35:12 -0400 Received: from sstabellini-ThinkPad-X260.xilinx.com (unknown [149.199.62.254]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-SHA256 (128/128 bits)) (No client certificate requested) by mail.kernel.org (Postfix) with ESMTPSA id 87FAF20879; Mon, 21 May 2018 19:35:11 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=kernel.org; s=default; t=1526931311; bh=FF9ENIth57duLx0iW5OOP4SGGcvtirinGLtLwnHJpLg=; h=From:To:Cc:Subject:Date:In-Reply-To:References:From; b=gqmG4/8RtcOQ8CMPSrHFuLrhhy8UNpzWiFyeMKKkf5Jj2zFeq5DJ8GbkP2rzA3qvQ t0VqEp4nChSfb6zJpxo1bLWhYIM1H8OhiOQTgQ1cV2xZJ5tsPaUUEtxMA77pf+FUIt IG6wrMtdhfoVtx106rNimJwIFEp+oUbqCi2jSpak= From: Stefano Stabellini To: peter.maydell@linaro.org, stefanha@gmail.com Date: Mon, 21 May 2018 12:35:01 -0700 Message-Id: <1526931304-7289-12-git-send-email-sstabellini@kernel.org> X-Mailer: git-send-email 1.9.1 In-Reply-To: References: X-detected-operating-system: by eggs.gnu.org: GNU/Linux 2.2.x-3.x [generic] [fuzzy] X-Received-From: 198.145.29.99 Subject: [Qemu-devel] [PULL 12/15] xen_disk: remove use of grant map/unmap X-BeenThere: qemu-devel@nongnu.org X-Mailman-Version: 2.1.21 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Cc: sstabellini@kernel.org, qemu-devel@nongnu.org, Paul Durrant , stefanha@redhat.com, anthony.perard@citrix.com, xen-devel@lists.xenproject.org Errors-To: qemu-devel-bounces+incoming=patchwork.ozlabs.org@nongnu.org Sender: "Qemu-devel" From: Paul Durrant Now that the (native or emulated) xen_be_copy_grant_refs() helper is always available, the xen_disk code can be significantly simplified by removing direct use of grant map and unmap operations. Signed-off-by: Paul Durrant Acked-by: Anthony Perard Signed-off-by: Stefano Stabellini --- hw/block/xen_disk.c | 352 ++++------------------------------------------------ 1 file changed, 25 insertions(+), 327 deletions(-) diff --git a/hw/block/xen_disk.c b/hw/block/xen_disk.c index d3be45a..28be8b6 100644 --- a/hw/block/xen_disk.c +++ b/hw/block/xen_disk.c @@ -36,27 +36,9 @@ /* ------------------------------------------------------------- */ -static int batch_maps = 0; - -/* ------------------------------------------------------------- */ - #define BLOCK_SIZE 512 #define IOCB_COUNT (BLKIF_MAX_SEGMENTS_PER_REQUEST + 2) -struct PersistentGrant { - void *page; - struct XenBlkDev *blkdev; -}; - -typedef struct PersistentGrant PersistentGrant; - -struct PersistentRegion { - void *addr; - int num; -}; - -typedef struct PersistentRegion PersistentRegion; - struct ioreq { blkif_request_t req; int16_t status; @@ -65,14 +47,11 @@ struct ioreq { off_t start; QEMUIOVector v; int presync; - uint8_t mapped; /* grant mapping */ uint32_t refs[BLKIF_MAX_SEGMENTS_PER_REQUEST]; - int prot; void *page[BLKIF_MAX_SEGMENTS_PER_REQUEST]; void *pages; - int num_unmap; /* aio status */ int aio_inflight; @@ -103,7 +82,6 @@ struct XenBlkDev { int protocol; blkif_back_rings_t rings; int more_work; - int cnt_map; /* request lists */ QLIST_HEAD(inflight_head, ioreq) inflight; @@ -114,13 +92,7 @@ struct XenBlkDev { int requests_finished; unsigned int max_requests; - /* Persistent grants extension */ gboolean feature_discard; - gboolean feature_persistent; - GTree *persistent_gnts; - GSList *persistent_regions; - unsigned int persistent_gnt_count; - unsigned int max_grants; /* qemu block driver */ DriveInfo *dinfo; @@ -139,10 +111,8 @@ static void ioreq_reset(struct ioreq *ioreq) ioreq->status = 0; ioreq->start = 0; ioreq->presync = 0; - ioreq->mapped = 0; memset(ioreq->refs, 0, sizeof(ioreq->refs)); - ioreq->prot = 0; memset(ioreq->page, 0, sizeof(ioreq->page)); ioreq->pages = NULL; @@ -156,37 +126,6 @@ static void ioreq_reset(struct ioreq *ioreq) qemu_iovec_reset(&ioreq->v); } -static gint int_cmp(gconstpointer a, gconstpointer b, gpointer user_data) -{ - uint ua = GPOINTER_TO_UINT(a); - uint ub = GPOINTER_TO_UINT(b); - return (ua > ub) - (ua < ub); -} - -static void destroy_grant(gpointer pgnt) -{ - PersistentGrant *grant = pgnt; - struct XenBlkDev *blkdev = grant->blkdev; - struct XenDevice *xendev = &blkdev->xendev; - - xen_be_unmap_grant_ref(xendev, grant->page); - grant->blkdev->persistent_gnt_count--; - xen_pv_printf(xendev, 3, "unmapped grant %p\n", grant->page); - g_free(grant); -} - -static void remove_persistent_region(gpointer data, gpointer dev) -{ - PersistentRegion *region = data; - struct XenBlkDev *blkdev = dev; - struct XenDevice *xendev = &blkdev->xendev; - - xen_be_unmap_grant_refs(xendev, region->addr, region->num); - xen_pv_printf(xendev, 3, "unmapped grant region %p with %d pages\n", - region->addr, region->num); - g_free(region); -} - static struct ioreq *ioreq_start(struct XenBlkDev *blkdev) { struct ioreq *ioreq = NULL; @@ -254,7 +193,6 @@ static int ioreq_parse(struct ioreq *ioreq) ioreq->req.handle, ioreq->req.id, ioreq->req.sector_number); switch (ioreq->req.operation) { case BLKIF_OP_READ: - ioreq->prot = PROT_WRITE; /* to memory */ break; case BLKIF_OP_FLUSH_DISKCACHE: ioreq->presync = 1; @@ -263,7 +201,6 @@ static int ioreq_parse(struct ioreq *ioreq) } /* fall through */ case BLKIF_OP_WRITE: - ioreq->prot = PROT_READ; /* from memory */ break; case BLKIF_OP_DISCARD: return 0; @@ -310,171 +247,6 @@ err: return -1; } -static void ioreq_unmap(struct ioreq *ioreq) -{ - struct XenBlkDev *blkdev = ioreq->blkdev; - struct XenDevice *xendev = &blkdev->xendev; - int i; - - if (ioreq->num_unmap == 0 || ioreq->mapped == 0) { - return; - } - if (batch_maps) { - if (!ioreq->pages) { - return; - } - xen_be_unmap_grant_refs(xendev, ioreq->pages, ioreq->num_unmap); - ioreq->blkdev->cnt_map -= ioreq->num_unmap; - ioreq->pages = NULL; - } else { - for (i = 0; i < ioreq->num_unmap; i++) { - if (!ioreq->page[i]) { - continue; - } - xen_be_unmap_grant_ref(xendev, ioreq->page[i]); - ioreq->blkdev->cnt_map--; - ioreq->page[i] = NULL; - } - } - ioreq->mapped = 0; -} - -static int ioreq_map(struct ioreq *ioreq) -{ - struct XenBlkDev *blkdev = ioreq->blkdev; - struct XenDevice *xendev = &blkdev->xendev; - uint32_t refs[BLKIF_MAX_SEGMENTS_PER_REQUEST]; - void *page[BLKIF_MAX_SEGMENTS_PER_REQUEST]; - int i, j, new_maps = 0; - PersistentGrant *grant; - PersistentRegion *region; - /* refs variable will contain the information necessary - * to map the grants that are needed to fulfill this request. - * - * After mapping the needed grants, the page array will contain the - * memory address of each granted page in the order specified in ioreq - * (disregarding if it's a persistent grant or not). - */ - - if (ioreq->v.niov == 0 || ioreq->mapped == 1) { - return 0; - } - if (ioreq->blkdev->feature_persistent) { - for (i = 0; i < ioreq->v.niov; i++) { - grant = g_tree_lookup(ioreq->blkdev->persistent_gnts, - GUINT_TO_POINTER(ioreq->refs[i])); - - if (grant != NULL) { - page[i] = grant->page; - xen_pv_printf(&ioreq->blkdev->xendev, 3, - "using persistent-grant %" PRIu32 "\n", - ioreq->refs[i]); - } else { - /* Add the grant to the list of grants that - * should be mapped - */ - refs[new_maps] = ioreq->refs[i]; - page[i] = NULL; - new_maps++; - } - } - /* Set the protection to RW, since grants may be reused later - * with a different protection than the one needed for this request - */ - ioreq->prot = PROT_WRITE | PROT_READ; - } else { - /* All grants in the request should be mapped */ - memcpy(refs, ioreq->refs, sizeof(refs)); - memset(page, 0, sizeof(page)); - new_maps = ioreq->v.niov; - } - - if (batch_maps && new_maps) { - ioreq->pages = xen_be_map_grant_refs(xendev, refs, new_maps, - ioreq->prot); - if (ioreq->pages == NULL) { - xen_pv_printf(&ioreq->blkdev->xendev, 0, - "can't map %d grant refs (%s, %d maps)\n", - new_maps, strerror(errno), ioreq->blkdev->cnt_map); - return -1; - } - for (i = 0, j = 0; i < ioreq->v.niov; i++) { - if (page[i] == NULL) { - page[i] = ioreq->pages + (j++) * XC_PAGE_SIZE; - } - } - ioreq->blkdev->cnt_map += new_maps; - } else if (new_maps) { - for (i = 0; i < new_maps; i++) { - ioreq->page[i] = xen_be_map_grant_ref(xendev, refs[i], - ioreq->prot); - if (ioreq->page[i] == NULL) { - xen_pv_printf(&ioreq->blkdev->xendev, 0, - "can't map grant ref %d (%s, %d maps)\n", - refs[i], strerror(errno), ioreq->blkdev->cnt_map); - ioreq->mapped = 1; - ioreq_unmap(ioreq); - return -1; - } - ioreq->blkdev->cnt_map++; - } - for (i = 0, j = 0; i < ioreq->v.niov; i++) { - if (page[i] == NULL) { - page[i] = ioreq->page[j++]; - } - } - } - if (ioreq->blkdev->feature_persistent && new_maps != 0 && - (!batch_maps || (ioreq->blkdev->persistent_gnt_count + new_maps <= - ioreq->blkdev->max_grants))) { - /* - * If we are using persistent grants and batch mappings only - * add the new maps to the list of persistent grants if the whole - * area can be persistently mapped. - */ - if (batch_maps) { - region = g_malloc0(sizeof(*region)); - region->addr = ioreq->pages; - region->num = new_maps; - ioreq->blkdev->persistent_regions = g_slist_append( - ioreq->blkdev->persistent_regions, - region); - } - while ((ioreq->blkdev->persistent_gnt_count < ioreq->blkdev->max_grants) - && new_maps) { - /* Go through the list of newly mapped grants and add as many - * as possible to the list of persistently mapped grants. - * - * Since we start at the end of ioreq->page(s), we only need - * to decrease new_maps to prevent this granted pages from - * being unmapped in ioreq_unmap. - */ - grant = g_malloc0(sizeof(*grant)); - new_maps--; - if (batch_maps) { - grant->page = ioreq->pages + (new_maps) * XC_PAGE_SIZE; - } else { - grant->page = ioreq->page[new_maps]; - } - grant->blkdev = ioreq->blkdev; - xen_pv_printf(&ioreq->blkdev->xendev, 3, - "adding grant %" PRIu32 " page: %p\n", - refs[new_maps], grant->page); - g_tree_insert(ioreq->blkdev->persistent_gnts, - GUINT_TO_POINTER(refs[new_maps]), - grant); - ioreq->blkdev->persistent_gnt_count++; - } - assert(!batch_maps || new_maps == 0); - } - for (i = 0; i < ioreq->v.niov; i++) { - ioreq->v.iov[i].iov_base += (uintptr_t)page[i]; - } - ioreq->mapped = 1; - ioreq->num_unmap = new_maps; - return 0; -} - static void ioreq_free_copy_buffers(struct ioreq *ioreq) { int i; @@ -570,32 +342,28 @@ static void qemu_aio_complete(void *opaque, int ret) goto done; } - if (xen_feature_grant_copy) { - switch (ioreq->req.operation) { - case BLKIF_OP_READ: - /* in case of failure ioreq->aio_errors is increased */ - if (ret == 0) { - ioreq_grant_copy(ioreq); - } - ioreq_free_copy_buffers(ioreq); - break; - case BLKIF_OP_WRITE: - case BLKIF_OP_FLUSH_DISKCACHE: - if (!ioreq->req.nr_segments) { - break; - } - ioreq_free_copy_buffers(ioreq); - break; - default: + switch (ioreq->req.operation) { + case BLKIF_OP_READ: + /* in case of failure ioreq->aio_errors is increased */ + if (ret == 0) { + ioreq_grant_copy(ioreq); + } + ioreq_free_copy_buffers(ioreq); + break; + case BLKIF_OP_WRITE: + case BLKIF_OP_FLUSH_DISKCACHE: + if (!ioreq->req.nr_segments) { break; } + ioreq_free_copy_buffers(ioreq); + break; + default: + break; } ioreq->status = ioreq->aio_errors ? BLKIF_RSP_ERROR : BLKIF_RSP_OKAY; - if (!xen_feature_grant_copy) { - ioreq_unmap(ioreq); - } ioreq_finish(ioreq); + switch (ioreq->req.operation) { case BLKIF_OP_WRITE: case BLKIF_OP_FLUSH_DISKCACHE: @@ -655,18 +423,13 @@ static int ioreq_runio_qemu_aio(struct ioreq *ioreq) { struct XenBlkDev *blkdev = ioreq->blkdev; - if (xen_feature_grant_copy) { - ioreq_init_copy_buffers(ioreq); - if (ioreq->req.nr_segments && (ioreq->req.operation == BLKIF_OP_WRITE || - ioreq->req.operation == BLKIF_OP_FLUSH_DISKCACHE) && - ioreq_grant_copy(ioreq)) { - ioreq_free_copy_buffers(ioreq); - goto err; - } - } else { - if (ioreq->req.nr_segments && ioreq_map(ioreq)) { - goto err; - } + ioreq_init_copy_buffers(ioreq); + if (ioreq->req.nr_segments && + (ioreq->req.operation == BLKIF_OP_WRITE || + ioreq->req.operation == BLKIF_OP_FLUSH_DISKCACHE) && + ioreq_grant_copy(ioreq)) { + ioreq_free_copy_buffers(ioreq); + goto err; } ioreq->aio_inflight++; @@ -707,9 +470,6 @@ static int ioreq_runio_qemu_aio(struct ioreq *ioreq) } default: /* unknown operation (shouldn't happen -- parse catches this) */ - if (!xen_feature_grant_copy) { - ioreq_unmap(ioreq); - } goto err; } @@ -895,10 +655,6 @@ static void blk_alloc(struct XenDevice *xendev) blkdev->ctx = iothread_get_aio_context(blkdev->iothread); blkdev->bh = aio_bh_new(blkdev->ctx, blk_bh, blkdev); - - if (xen_mode != XEN_EMULATE) { - batch_maps = 1; - } } static void blk_parse_discard(struct XenBlkDev *blkdev) @@ -981,15 +737,10 @@ static int blk_init(struct XenDevice *xendev) blkdev->file_blk = BLOCK_SIZE; - xen_pv_printf(&blkdev->xendev, 3, "grant copy operation %s\n", - xen_feature_grant_copy ? "enabled" : "disabled"); - /* fill info * blk_connect supplies sector-size and sectors */ xenstore_write_be_int(&blkdev->xendev, "feature-flush-cache", 1); - xenstore_write_be_int(&blkdev->xendev, "feature-persistent", - !xen_feature_grant_copy); xenstore_write_be_int(&blkdev->xendev, "info", info); xenstore_write_be_int(&blkdev->xendev, "max-ring-page-order", @@ -1016,19 +767,10 @@ out_error: return -1; } -/* - * We need to account for the grant allocations requiring contiguous - * chunks; the worst case number would be - * max_req * max_seg + (max_req - 1) * (max_seg - 1) + 1, - * but in order to keep things simple just use - * 2 * max_req * max_seg. - */ -#define MAX_GRANTS(max_req, max_seg) (2 * (max_req) * (max_seg)) - static int blk_connect(struct XenDevice *xendev) { struct XenBlkDev *blkdev = container_of(xendev, struct XenBlkDev, xendev); - int pers, index, qflags; + int index, qflags; bool readonly = true; bool writethrough = true; int order, ring_ref; @@ -1150,11 +892,6 @@ static int blk_connect(struct XenDevice *xendev) &blkdev->xendev.remote_port) == -1) { return -1; } - if (xenstore_read_fe_int(&blkdev->xendev, "feature-persistent", &pers)) { - blkdev->feature_persistent = FALSE; - } else { - blkdev->feature_persistent = !!pers; - } if (!blkdev->xendev.protocol) { blkdev->protocol = BLKIF_PROTOCOL_NATIVE; @@ -1189,11 +926,8 @@ static int blk_connect(struct XenDevice *xendev) return -1; } - /* Calculate the maximum number of grants needed by ioreqs */ - max_grants = MAX_GRANTS(blkdev->max_requests, - BLKIF_MAX_SEGMENTS_PER_REQUEST); /* Add on the number needed for the ring pages */ - max_grants += blkdev->nr_ring_ref; + max_grants = blkdev->nr_ring_ref; xen_be_set_max_grant_refs(xendev, max_grants); @@ -1204,8 +938,6 @@ static int blk_connect(struct XenDevice *xendev) return -1; } - blkdev->cnt_map++; - switch (blkdev->protocol) { case BLKIF_PROTOCOL_NATIVE: { @@ -1229,19 +961,6 @@ static int blk_connect(struct XenDevice *xendev) } } - if (blkdev->feature_persistent) { - /* Init persistent grants */ - blkdev->max_grants = blkdev->max_requests * - BLKIF_MAX_SEGMENTS_PER_REQUEST; - blkdev->persistent_gnts = g_tree_new_full((GCompareDataFunc)int_cmp, - NULL, NULL, - batch_maps ? - (GDestroyNotify)g_free : - (GDestroyNotify)destroy_grant); - blkdev->persistent_regions = NULL; - blkdev->persistent_gnt_count = 0; - } - blk_set_aio_context(blkdev->blk, blkdev->ctx); xen_be_bind_evtchn(&blkdev->xendev); @@ -1274,29 +993,8 @@ static void blk_disconnect(struct XenDevice *xendev) if (blkdev->sring) { xen_be_unmap_grant_refs(xendev, blkdev->sring, blkdev->nr_ring_ref); - blkdev->cnt_map--; blkdev->sring = NULL; } - - /* - * Unmap persistent grants before switching to the closed state - * so the frontend can free them. - * - * In the !batch_maps case g_tree_destroy will take care of unmapping - * the grant, but in the batch_maps case we need to iterate over every - * region in persistent_regions and unmap it. - */ - if (blkdev->feature_persistent) { - g_tree_destroy(blkdev->persistent_gnts); - assert(batch_maps || blkdev->persistent_gnt_count == 0); - if (batch_maps) { - blkdev->persistent_gnt_count = 0; - g_slist_foreach(blkdev->persistent_regions, - (GFunc)remove_persistent_region, blkdev); - g_slist_free(blkdev->persistent_regions); - } - blkdev->feature_persistent = false; - } } static int blk_free(struct XenDevice *xendev)