From: Brijesh Singh <brijesh.singh@amd.com>
To: edk2-devel@lists.01.org
Cc: Brijesh Singh <brijesh.singh@amd.com>,
Ard Biesheuvel <ard.biesheuvel@linaro.org>,
Jordan Justen <jordan.l.justen@intel.com>,
Tom Lendacky <thomas.lendacky@amd.com>,
Laszlo Ersek <lersek@redhat.com>
Subject: [PATCH v4 3/8] OvmfPkg/VirtioNetDxe: alloc RxBuf using AllocateSharedPages()
Date: Thu, 14 Sep 2017 16:22:42 -0500 [thread overview]
Message-ID: <20170914212247.124937-4-brijesh.singh@amd.com> (raw)
In-Reply-To: <20170914212247.124937-1-brijesh.singh@amd.com>
When device is behind the IOMMU, VirtioNetDxe is required to use the
device address in bus master operations. RxBuf is allocated using
AllocatePool() which returns the system physical address.
The patch uses VIRTIO_DEVICE_PROTOCOL.AllocateSharedPages() to allocate
the RxBuf and map with VirtioMapAllBytesInSharedBuffer() so that we can
obtain the device address for RxBuf.
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: Jordan Justen <jordan.l.justen@intel.com>
Cc: Tom Lendacky <thomas.lendacky@amd.com>
Cc: Laszlo Ersek <lersek@redhat.com>
Contributed-under: TianoCore Contribution Agreement 1.1
Signed-off-by: Brijesh Singh <brijesh.singh@amd.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
---
OvmfPkg/VirtioNetDxe/VirtioNet.h | 4 +
OvmfPkg/VirtioNetDxe/SnpInitialize.c | 78 +++++++++++++++-----
OvmfPkg/VirtioNetDxe/SnpReceive.c | 5 +-
OvmfPkg/VirtioNetDxe/SnpSharedHelpers.c | 7 +-
OvmfPkg/VirtioNetDxe/TechNotes.txt | 2 +-
5 files changed, 75 insertions(+), 21 deletions(-)
diff --git a/OvmfPkg/VirtioNetDxe/VirtioNet.h b/OvmfPkg/VirtioNetDxe/VirtioNet.h
index 6762fc9d1d6e..995593f4b236 100644
--- a/OvmfPkg/VirtioNetDxe/VirtioNet.h
+++ b/OvmfPkg/VirtioNetDxe/VirtioNet.h
@@ -4,6 +4,7 @@
Protocol instances for virtio-net devices.
Copyright (C) 2013, Red Hat, Inc.
+ Copyright (c) 2017, AMD Inc, All rights reserved.<BR>
This program and the accompanying materials are licensed and made available
under the terms and conditions of the BSD License which accompanies this
@@ -86,6 +87,9 @@ typedef struct {
// VirtioNetInitRing
UINT8 *RxBuf; // VirtioNetInitRx
UINT16 RxLastUsed; // VirtioNetInitRx
+ UINTN RxBufNrPages; // VirtioNetInitRx
+ EFI_PHYSICAL_ADDRESS RxBufDeviceBase; // VirtioNetInitRx
+ VOID *RxBufMap; // VirtioNetInitRx
VRING TxRing; // VirtioNetInitRing
VOID *TxRingMap; // VirtioRingMap and
diff --git a/OvmfPkg/VirtioNetDxe/SnpInitialize.c b/OvmfPkg/VirtioNetDxe/SnpInitialize.c
index 8eabdbff6f5e..b739875d4c01 100644
--- a/OvmfPkg/VirtioNetDxe/SnpInitialize.c
+++ b/OvmfPkg/VirtioNetDxe/SnpInitialize.c
@@ -242,8 +242,9 @@ VirtioNetInitTx (
@param[in,out] Dev The VNET_DEV driver instance about to enter the
EfiSimpleNetworkInitialized state.
- @retval EFI_OUT_OF_RESOURCES Failed to allocate RX destination area.
- @return Status codes from VIRTIO_CFG_WRITE().
+ @return Status codes from VIRTIO_CFG_WRITE() or
+ VIRTIO_DEVICE_PROTOCOL.AllocateSharedPages or
+ VirtioMapAllBytesInSharedBuffer().
@retval EFI_SUCCESS RX setup successful. The device is live and may
already be writing to the receive area.
*/
@@ -255,13 +256,15 @@ VirtioNetInitRx (
IN OUT VNET_DEV *Dev
)
{
- EFI_STATUS Status;
- UINTN VirtioNetReqSize;
- UINTN RxBufSize;
- UINT16 RxAlwaysPending;
- UINTN PktIdx;
- UINT16 DescIdx;
- UINT8 *RxPtr;
+ EFI_STATUS Status;
+ UINTN VirtioNetReqSize;
+ UINTN RxBufSize;
+ UINT16 RxAlwaysPending;
+ UINTN PktIdx;
+ UINT16 DescIdx;
+ UINTN NumBytes;
+ EFI_PHYSICAL_ADDRESS RxBufDeviceAddress;
+ VOID *RxBuffer;
//
// In VirtIo 1.0, the NumBuffers field is mandatory. In 0.9.5, it depends on
@@ -286,11 +289,39 @@ VirtioNetInitRx (
//
RxAlwaysPending = (UINT16) MIN (Dev->RxRing.QueueSize / 2, VNET_MAX_PENDING);
- Dev->RxBuf = AllocatePool (RxAlwaysPending * RxBufSize);
- if (Dev->RxBuf == NULL) {
- return EFI_OUT_OF_RESOURCES;
+ //
+ // The RxBuf is shared between guest and hypervisor, use
+ // AllocateSharedPages() to allocate this memory region and map it with
+ // BusMasterCommonBuffer so that it can be accessed by both guest and
+ // hypervisor.
+ //
+ NumBytes = RxAlwaysPending * RxBufSize;
+ Dev->RxBufNrPages = EFI_SIZE_TO_PAGES (NumBytes);
+ Status = Dev->VirtIo->AllocateSharedPages (
+ Dev->VirtIo,
+ Dev->RxBufNrPages,
+ &RxBuffer
+ );
+ if (EFI_ERROR (Status)) {
+ return Status;
}
+ ZeroMem (RxBuffer, NumBytes);
+
+ Status = VirtioMapAllBytesInSharedBuffer (
+ Dev->VirtIo,
+ VirtioOperationBusMasterCommonBuffer,
+ RxBuffer,
+ NumBytes,
+ &Dev->RxBufDeviceBase,
+ &Dev->RxBufMap
+ );
+ if (EFI_ERROR (Status)) {
+ goto FreeSharedBuffer;
+ }
+
+ Dev->RxBuf = RxBuffer;
+
//
// virtio-0.9.5, 2.4.2 Receiving Used Buffers From the Device
//
@@ -310,7 +341,7 @@ VirtioNetInitRx (
// link each chain into (from) the available ring as well
//
DescIdx = 0;
- RxPtr = Dev->RxBuf;
+ RxBufDeviceAddress = Dev->RxBufDeviceBase;
for (PktIdx = 0; PktIdx < RxAlwaysPending; ++PktIdx) {
//
// virtio-0.9.5, 2.4.1.2 Updating the Available Ring
@@ -321,16 +352,16 @@ VirtioNetInitRx (
//
// virtio-0.9.5, 2.4.1.1 Placing Buffers into the Descriptor Table
//
- Dev->RxRing.Desc[DescIdx].Addr = (UINTN) RxPtr;
+ Dev->RxRing.Desc[DescIdx].Addr = RxBufDeviceAddress;
Dev->RxRing.Desc[DescIdx].Len = (UINT32) VirtioNetReqSize;
Dev->RxRing.Desc[DescIdx].Flags = VRING_DESC_F_WRITE | VRING_DESC_F_NEXT;
Dev->RxRing.Desc[DescIdx].Next = (UINT16) (DescIdx + 1);
- RxPtr += Dev->RxRing.Desc[DescIdx++].Len;
+ RxBufDeviceAddress += Dev->RxRing.Desc[DescIdx++].Len;
- Dev->RxRing.Desc[DescIdx].Addr = (UINTN) RxPtr;
+ Dev->RxRing.Desc[DescIdx].Addr = RxBufDeviceAddress;
Dev->RxRing.Desc[DescIdx].Len = (UINT32) (RxBufSize - VirtioNetReqSize);
Dev->RxRing.Desc[DescIdx].Flags = VRING_DESC_F_WRITE;
- RxPtr += Dev->RxRing.Desc[DescIdx++].Len;
+ RxBufDeviceAddress += Dev->RxRing.Desc[DescIdx++].Len;
}
//
@@ -351,10 +382,21 @@ VirtioNetInitRx (
Status = Dev->VirtIo->SetQueueNotify (Dev->VirtIo, VIRTIO_NET_Q_RX);
if (EFI_ERROR (Status)) {
Dev->VirtIo->SetDeviceStatus (Dev->VirtIo, 0);
- FreePool (Dev->RxBuf);
+ goto UnmapSharedBuffer;
}
return Status;
+
+UnmapSharedBuffer:
+ Dev->VirtIo->UnmapSharedBuffer (Dev->VirtIo, Dev->RxBufMap);
+
+FreeSharedBuffer:
+ Dev->VirtIo->FreeSharedPages (
+ Dev->VirtIo,
+ Dev->RxBufNrPages,
+ RxBuffer
+ );
+ return Status;
}
diff --git a/OvmfPkg/VirtioNetDxe/SnpReceive.c b/OvmfPkg/VirtioNetDxe/SnpReceive.c
index 99abd7ebe454..c42489636ea0 100644
--- a/OvmfPkg/VirtioNetDxe/SnpReceive.c
+++ b/OvmfPkg/VirtioNetDxe/SnpReceive.c
@@ -82,6 +82,7 @@ VirtioNetReceive (
UINT8 *RxPtr;
UINT16 AvailIdx;
EFI_STATUS NotifyStatus;
+ UINTN RxBufOffset;
if (This == NULL || BufferSize == NULL || Buffer == NULL) {
return EFI_INVALID_PARAMETER;
@@ -143,7 +144,9 @@ VirtioNetReceive (
*HeaderSize = Dev->Snm.MediaHeaderSize;
}
- RxPtr = (UINT8 *)(UINTN) Dev->RxRing.Desc[DescIdx + 1].Addr;
+ RxBufOffset = (UINTN)(Dev->RxRing.Desc[DescIdx + 1].Addr -
+ Dev->RxBufDeviceBase);
+ RxPtr = Dev->RxBuf + RxBufOffset;
CopyMem (Buffer, RxPtr, RxLen);
if (DestAddr != NULL) {
diff --git a/OvmfPkg/VirtioNetDxe/SnpSharedHelpers.c b/OvmfPkg/VirtioNetDxe/SnpSharedHelpers.c
index 57c7395848bd..ee4f9ed36ecd 100644
--- a/OvmfPkg/VirtioNetDxe/SnpSharedHelpers.c
+++ b/OvmfPkg/VirtioNetDxe/SnpSharedHelpers.c
@@ -39,7 +39,12 @@ VirtioNetShutdownRx (
IN OUT VNET_DEV *Dev
)
{
- FreePool (Dev->RxBuf);
+ Dev->VirtIo->UnmapSharedBuffer (Dev->VirtIo, Dev->RxBufMap);
+ Dev->VirtIo->FreeSharedPages (
+ Dev->VirtIo,
+ Dev->RxBufNrPages,
+ Dev->RxBuf
+ );
}
diff --git a/OvmfPkg/VirtioNetDxe/TechNotes.txt b/OvmfPkg/VirtioNetDxe/TechNotes.txt
index 37250b14a98c..fedbaee07ec7 100644
--- a/OvmfPkg/VirtioNetDxe/TechNotes.txt
+++ b/OvmfPkg/VirtioNetDxe/TechNotes.txt
@@ -247,7 +247,7 @@ In VirtioNetInitRx, the guest allocates the fixed size Receive Destination
Area, which accommodates all packets delivered asynchronously by the host. To
each packet, a slice of this area is dedicated; each slice is further
subdivided into virtio-net request header and network packet data. The
-(guest-physical) addresses of these sub-slices are denoted with A2, A3, A4 and
+(device-physical) addresses of these sub-slices are denoted with A2, A3, A4 and
so on. Importantly, an even-subscript "A" always belongs to a virtio-net
request header, while an odd-subscript "A" always belongs to a packet
sub-slice.
--
2.9.5
next prev parent reply other threads:[~2017-09-14 21:20 UTC|newest]
Thread overview: 12+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-09-14 21:22 [PATCH v4 0/8] OvmfPkg/VirtioNetDxe: map host address to device address Brijesh Singh
2017-09-14 21:22 ` [PATCH v4 1/8] OvmfPkg/VirtioNetDxe: add helper VirtioNetUninitRing() Brijesh Singh
2017-09-14 21:22 ` [PATCH v4 2/8] OvmfPkg/VirtioNetDxe: map VRINGs using VirtioRingMap() Brijesh Singh
2017-09-14 21:22 ` Brijesh Singh [this message]
2017-09-14 21:22 ` [PATCH v4 4/8] OvmfPkg/VirtioNetDxe: dynamically alloc transmit header Brijesh Singh
2017-09-14 21:22 ` [PATCH v4 5/8] OvmfPkg/VirtioNetDxe: update TechNotes Brijesh Singh
2017-09-14 21:22 ` [PATCH v4 6/8] OvmfPkg/VirtioNetDxe: add Tx packet map/unmap helper functions Brijesh Singh
2017-09-14 21:22 ` [PATCH v4 7/8] OvmfPkg/VirtioNetDxe: map caller-supplied Tx packet to device-address Brijesh Singh
2017-09-14 21:49 ` Laszlo Ersek
2017-09-14 21:22 ` [PATCH v4 8/8] OvmfPkg/VirtioNetDxe: negotiate VIRTIO_F_IOMMU_PLATFORM Brijesh Singh
2017-09-15 0:10 ` [PATCH v4 0/8] OvmfPkg/VirtioNetDxe: map host address to device address Laszlo Ersek
2017-09-15 0:14 ` Laszlo Ersek
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-list from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20170914212247.124937-4-brijesh.singh@amd.com \
--to=devel@edk2.groups.io \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox