From: "Gary Lin" <glin@suse.com>
To: devel@edk2.groups.io
Cc: Jordan Justen <jordan.l.justen@intel.com>,
Laszlo Ersek <lersek@redhat.com>,
Ard Biesheuvel <ard.biesheuvel@arm.com>
Subject: [PATCH v2 08/12] OvmfPkg/LsiScsiDxe: Map DMA buffer
Date: Thu, 16 Jul 2020 15:46:03 +0800 [thread overview]
Message-ID: <20200716074607.18048-9-glin@suse.com> (raw)
In-Reply-To: <20200716074607.18048-1-glin@suse.com>
Map DMA buffer and perpare for the implementation of LsiScsiPassThru().
v2:
- Replace 0x10000 with SIZE_64KB macro for the DMA buffer data array
- Remove DUAL_ADDRESS_CYCLE from PciIo since we don't really need
64-bit DMA address
Cc: Jordan Justen <jordan.l.justen@intel.com>
Cc: Laszlo Ersek <lersek@redhat.com>
Cc: Ard Biesheuvel <ard.biesheuvel@arm.com>
Signed-off-by: Gary Lin <glin@suse.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
---
OvmfPkg/LsiScsiDxe/LsiScsi.c | 62 +++++++++++++++++++++++++++++++++++-
OvmfPkg/LsiScsiDxe/LsiScsi.h | 14 ++++++++
2 files changed, 75 insertions(+), 1 deletion(-)
diff --git a/OvmfPkg/LsiScsiDxe/LsiScsi.c b/OvmfPkg/LsiScsiDxe/LsiScsi.c
index b21387db6e5a..4e84afa40085 100644
--- a/OvmfPkg/LsiScsiDxe/LsiScsi.c
+++ b/OvmfPkg/LsiScsiDxe/LsiScsi.c
@@ -356,6 +356,8 @@ LsiScsiControllerStart (
{
EFI_STATUS Status;
LSI_SCSI_DEV *Dev;
+ UINTN Pages;
+ UINTN BytesMapped;
Dev = AllocateZeroPool (sizeof (*Dev));
if (Dev == NULL) {
@@ -411,11 +413,45 @@ LsiScsiControllerStart (
goto CloseProtocol;
}
- Status = LsiScsiReset (Dev);
+ //
+ // Create buffers for data transfer
+ //
+ Pages = EFI_SIZE_TO_PAGES (sizeof (*Dev->Dma));
+ Status = Dev->PciIo->AllocateBuffer (
+ Dev->PciIo,
+ AllocateAnyPages,
+ EfiBootServicesData,
+ Pages,
+ (VOID **)&Dev->Dma,
+ EFI_PCI_ATTRIBUTE_MEMORY_CACHED
+ );
if (EFI_ERROR (Status)) {
goto RestoreAttributes;
}
+ BytesMapped = EFI_PAGES_TO_SIZE (Pages);
+ Status = Dev->PciIo->Map (
+ Dev->PciIo,
+ EfiPciIoOperationBusMasterCommonBuffer,
+ Dev->Dma,
+ &BytesMapped,
+ &Dev->DmaPhysical,
+ &Dev->DmaMapping
+ );
+ if (EFI_ERROR (Status)) {
+ goto FreeBuffer;
+ }
+
+ if (BytesMapped != EFI_PAGES_TO_SIZE (Pages)) {
+ Status = EFI_OUT_OF_RESOURCES;
+ goto Unmap;
+ }
+
+ Status = LsiScsiReset (Dev);
+ if (EFI_ERROR (Status)) {
+ goto Unmap;
+ }
+
Status = gBS->CreateEvent (
EVT_SIGNAL_EXIT_BOOT_SERVICES,
TPL_CALLBACK,
@@ -462,6 +498,19 @@ CloseExitBoot:
UninitDev:
LsiScsiReset (Dev);
+Unmap:
+ Dev->PciIo->Unmap (
+ Dev->PciIo,
+ Dev->DmaMapping
+ );
+
+FreeBuffer:
+ Dev->PciIo->FreeBuffer (
+ Dev->PciIo,
+ Pages,
+ Dev->Dma
+ );
+
RestoreAttributes:
Dev->PciIo->Attributes (
Dev->PciIo,
@@ -524,6 +573,17 @@ LsiScsiControllerStop (
LsiScsiReset (Dev);
+ Dev->PciIo->Unmap (
+ Dev->PciIo,
+ Dev->DmaMapping
+ );
+
+ Dev->PciIo->FreeBuffer (
+ Dev->PciIo,
+ EFI_SIZE_TO_PAGES (sizeof (*Dev->Dma)),
+ Dev->Dma
+ );
+
Dev->PciIo->Attributes (
Dev->PciIo,
EfiPciIoAttributeOperationSet,
diff --git a/OvmfPkg/LsiScsiDxe/LsiScsi.h b/OvmfPkg/LsiScsiDxe/LsiScsi.h
index 8c2acff6e86f..9f9e5c7fed00 100644
--- a/OvmfPkg/LsiScsiDxe/LsiScsi.h
+++ b/OvmfPkg/LsiScsiDxe/LsiScsi.h
@@ -12,6 +12,17 @@
#ifndef _LSI_SCSI_DXE_H_
#define _LSI_SCSI_DXE_H_
+typedef struct {
+ //
+ // Allocate 64KB for read/write buffer. It seems sufficient for the common
+ // boot scenarios.
+ //
+ // NOTE: The number of bytes for data transmission is bounded by DMA Byte
+ // Count (DBC), a 24-bit register, so the maximum is 0xFFFFFF (16MB-1).
+ //
+ UINT8 Data[SIZE_64KB];
+} LSI_SCSI_DMA_BUFFER;
+
typedef struct {
UINT32 Signature;
UINT64 OrigPciAttrs;
@@ -19,6 +30,9 @@ typedef struct {
EFI_PCI_IO_PROTOCOL *PciIo;
UINT8 MaxTarget;
UINT8 MaxLun;
+ LSI_SCSI_DMA_BUFFER *Dma;
+ EFI_PHYSICAL_ADDRESS DmaPhysical;
+ VOID *DmaMapping;
EFI_EXT_SCSI_PASS_THRU_MODE PassThruMode;
EFI_EXT_SCSI_PASS_THRU_PROTOCOL PassThru;
} LSI_SCSI_DEV;
--
2.25.1
next prev parent reply other threads:[~2020-07-16 7:46 UTC|newest]
Thread overview: 21+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-07-16 7:45 [PATCH v2 00/12] Introduce LsiScsi driver to OvmfPkg Gary Lin
2020-07-16 7:45 ` [PATCH v2 01/12] OvmfPkg/LsiScsiDxe: Create the empty driver Gary Lin
2020-07-16 7:45 ` [PATCH v2 02/12] OvmfPkg/LsiScsiDxe: Install the skeleton of driver binding Gary Lin
2020-07-16 7:45 ` [PATCH v2 03/12] OvmfPkg/LsiScsiDxe: Report the name of the driver Gary Lin
2020-07-16 7:45 ` [PATCH v2 04/12] OvmfPkg/LsiScsiDxe: Probe PCI devices and look for LsiScsi Gary Lin
2020-07-16 7:46 ` [PATCH v2 05/12] OvmfPkg/LsiScsiDxe: Install stubbed EXT_SCSI_PASS_THRU Gary Lin
2020-07-16 7:46 ` [PATCH v2 06/12] OvmfPkg/LsiScsiDxe: Report Targets and LUNs Gary Lin
2020-07-16 15:04 ` Laszlo Ersek
2020-07-16 16:19 ` Laszlo Ersek
2020-07-16 7:46 ` [PATCH v2 07/12] OvmfPkg/LsiScsiDxe: Open PciIo protocol and initialize the device Gary Lin
2020-07-16 16:07 ` Laszlo Ersek
2020-07-16 7:46 ` Gary Lin [this message]
2020-07-16 16:11 ` [PATCH v2 08/12] OvmfPkg/LsiScsiDxe: Map DMA buffer Laszlo Ersek
2020-07-16 7:46 ` [PATCH v2 09/12] OvmfPkg/LsiScsiDxe: Examine the incoming SCSI Request Packet Gary Lin
2020-07-16 7:46 ` [PATCH v2 10/12] OvmfPkg/LsiScsiDxe: Process the " Gary Lin
2020-07-16 17:37 ` Laszlo Ersek
2020-07-17 2:28 ` Gary Lin
2020-07-16 7:46 ` [PATCH v2 11/12] OvmfPkg/LsiScsiDxe: Calculate the transferred bytes Gary Lin
2020-07-16 18:21 ` Laszlo Ersek
2020-07-17 3:15 ` Gary Lin
2020-07-16 7:46 ` [PATCH v2 12/12] Maintainers.txt: Add Gary Lin as the reviewer for LsiScsi driver Gary Lin
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=20200716074607.18048-9-glin@suse.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