From patchwork Tue Jun 19 06:39:48 2012 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Benjamin Herrenschmidt X-Patchwork-Id: 165676 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 6BB6EB6EEC for ; Tue, 19 Jun 2012 16:59:12 +1000 (EST) Received: from localhost ([::1]:57199 helo=lists.gnu.org) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1Sgs7f-0005SM-B6 for incoming@patchwork.ozlabs.org; Tue, 19 Jun 2012 02:41:07 -0400 Received: from eggs.gnu.org ([208.118.235.92]:55733) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1Sgs7O-00053U-M7 for qemu-devel@nongnu.org; Tue, 19 Jun 2012 02:40:55 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1Sgs7I-00063H-J2 for qemu-devel@nongnu.org; Tue, 19 Jun 2012 02:40:50 -0400 Received: from gate.crashing.org ([63.228.1.57]:47799) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1Sgs7I-00062w-5m for qemu-devel@nongnu.org; Tue, 19 Jun 2012 02:40:44 -0400 Received: from pasglop.ozlabs.ibm.com (localhost.localdomain [127.0.0.1]) by gate.crashing.org (8.14.1/8.13.8) with ESMTP id q5J6e0iU025357; Tue, 19 Jun 2012 01:40:32 -0500 From: Benjamin Herrenschmidt To: qemu-devel@nongnu.org Date: Tue, 19 Jun 2012 16:39:48 +1000 Message-Id: <1340087992-2399-10-git-send-email-benh@kernel.crashing.org> X-Mailer: git-send-email 1.7.9.5 In-Reply-To: <1340087992-2399-1-git-send-email-benh@kernel.crashing.org> References: <1340087992-2399-1-git-send-email-benh@kernel.crashing.org> X-detected-operating-system: by eggs.gnu.org: GNU/Linux 2.6 (newer, 2) X-Received-From: 63.228.1.57 Cc: anthony@codemonkey.ws, David Gibson Subject: [Qemu-devel] [PATCH 09/13] iommu: Add facility to cancel in-use dma memory maps 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 From: David Gibson One new complication raised by IOMMU support over only handling DMA directly to physical addresses is handling dma_memory_map() case (replacing cpu_physical_memory_map()) when the IOMMU translation the IOVAs covered by such a map are invalidated or changed while the map is active. This should never happen with correct guest software, but we do need to handle buggy guests. This case might also occur during handovers between different guest software stages if the handover protocols aren't fully seamless. The iommu implementation will have to wait for maps to be removed before it can "complete" an invalidation of a translation, which can take a long time. In order to make it possible to speed that process up, we add a "Cancel" callback to the map function which the clients can optionally provide. The core makes no use of that, but the iommu backend implementation may choose to keep track of maps and call the respective cancel callback whenever a translation within a map is removed, allowing the driver to do things like cancel async IOs etc. Signed-off-by: David Gibson Signed-off-by: Benjamin Herrenschmidt --- dma-helpers.c | 49 ++++++++++++++++++++++++++++--------------------- dma.h | 23 +++++++++++++++++++---- 2 files changed, 47 insertions(+), 25 deletions(-) diff --git a/dma-helpers.c b/dma-helpers.c index b4ee827..6e6c7b3 100644 --- a/dma-helpers.c +++ b/dma-helpers.c @@ -107,6 +107,28 @@ static void dma_complete(DMAAIOCB *dbs, int ret) } } +static void dma_aio_cancel(BlockDriverAIOCB *acb) +{ + DMAAIOCB *dbs = container_of(acb, DMAAIOCB, common); + + trace_dma_aio_cancel(dbs); + + if (dbs->acb) { + BlockDriverAIOCB *acb = dbs->acb; + dbs->acb = NULL; + dbs->in_cancel = true; + bdrv_aio_cancel(acb); + dbs->in_cancel = false; + } + dbs->common.cb = NULL; + dma_complete(dbs, 0); +} + +static void dma_bdrv_cancel_cb(void *opaque) +{ + dma_aio_cancel(&((DMAAIOCB *)opaque)->common); +} + static void dma_bdrv_cb(void *opaque, int ret) { DMAAIOCB *dbs = (DMAAIOCB *)opaque; @@ -127,7 +149,8 @@ static void dma_bdrv_cb(void *opaque, int ret) while (dbs->sg_cur_index < dbs->sg->nsg) { cur_addr = dbs->sg->sg[dbs->sg_cur_index].base + dbs->sg_cur_byte; cur_len = dbs->sg->sg[dbs->sg_cur_index].len - dbs->sg_cur_byte; - mem = dma_memory_map(dbs->sg->dma, cur_addr, &cur_len, dbs->dir); + mem = dma_memory_map_with_cancel(dbs->sg->dma, dma_bdrv_cancel_cb, dbs, + cur_addr, &cur_len, dbs->dir); if (!mem) break; qemu_iovec_add(&dbs->iov, mem, cur_len); @@ -149,23 +172,6 @@ static void dma_bdrv_cb(void *opaque, int ret) assert(dbs->acb); } -static void dma_aio_cancel(BlockDriverAIOCB *acb) -{ - DMAAIOCB *dbs = container_of(acb, DMAAIOCB, common); - - trace_dma_aio_cancel(dbs); - - if (dbs->acb) { - BlockDriverAIOCB *acb = dbs->acb; - dbs->acb = NULL; - dbs->in_cancel = true; - bdrv_aio_cancel(acb); - dbs->in_cancel = false; - } - dbs->common.cb = NULL; - dma_complete(dbs, 0); -} - static AIOPool dma_aio_pool = { .aiocb_size = sizeof(DMAAIOCB), .cancel = dma_aio_cancel, @@ -353,7 +359,9 @@ void dma_context_init(DMAContext *dma, DMATranslateFunc translate, dma->unmap = unmap; } -void *iommu_dma_memory_map(DMAContext *dma, dma_addr_t addr, dma_addr_t *len, +void *iommu_dma_memory_map(DMAContext *dma, + DMACancelMapFunc cb, void *cb_opaque, + dma_addr_t addr, dma_addr_t *len, DMADirection dir) { int err; @@ -361,7 +369,7 @@ void *iommu_dma_memory_map(DMAContext *dma, dma_addr_t addr, dma_addr_t *len, void *buf; if (dma->map) { - return dma->map(dma, addr, len, dir); + return dma->map(dma, cb, cb_opaque, addr, len, dir); } plen = *len; @@ -397,5 +405,4 @@ void iommu_dma_memory_unmap(DMAContext *dma, void *buffer, dma_addr_t len, cpu_physical_memory_unmap(buffer, len, dir == DMA_DIRECTION_FROM_DEVICE, access_len); - } diff --git a/dma.h b/dma.h index 14fe17d..f1fcb71 100644 --- a/dma.h +++ b/dma.h @@ -49,10 +49,15 @@ typedef int DMATranslateFunc(DMAContext *dma, target_phys_addr_t *paddr, target_phys_addr_t *len, DMADirection dir); + +typedef void DMACancelMapFunc(void *); typedef void* DMAMapFunc(DMAContext *dma, + DMACancelMapFunc cb, + void *cb_opaque, dma_addr_t addr, dma_addr_t *len, DMADirection dir); + typedef void DMAUnmapFunc(DMAContext *dma, void *buffer, dma_addr_t len, @@ -129,11 +134,15 @@ static inline int dma_memory_set(DMAContext *dma, dma_addr_t addr, } void *iommu_dma_memory_map(DMAContext *dma, + DMACancelMapFunc *cb, void *opaque, dma_addr_t addr, dma_addr_t *len, DMADirection dir); -static inline void *dma_memory_map(DMAContext *dma, - dma_addr_t addr, dma_addr_t *len, - DMADirection dir) +static inline void *dma_memory_map_with_cancel(DMAContext *dma, + DMACancelMapFunc *cb, + void *opaque, + dma_addr_t addr, + dma_addr_t *len, + DMADirection dir) { if (!dma_has_iommu(dma)) { target_phys_addr_t xlen = *len; @@ -144,9 +153,15 @@ static inline void *dma_memory_map(DMAContext *dma, *len = xlen; return p; } else { - return iommu_dma_memory_map(dma, addr, len, dir); + return iommu_dma_memory_map(dma, cb, opaque, addr, len, dir); } } +static inline void *dma_memory_map(DMAContext *dma, + dma_addr_t addr, dma_addr_t *len, + DMADirection dir) +{ + return dma_memory_map_with_cancel(dma, NULL, NULL, addr, len, dir); +} void iommu_dma_memory_unmap(DMAContext *dma, void *buffer, dma_addr_t len,