Add the VirtioFsFuseReadFileOrDir() function, for sending the FUSE_READ or
FUSE_READDIRPLUS command to the Virtio Filesystem device.
Parsing the structured FUSE_READDIRPLUS output is complex, and cannot be
integrated into the wrapper function. Given that fact, FUSE_READ and
FUSE_READDIRPLUS turn out to need identical low-level handling, except for
the opcode. Hence the shared wrapper function.
(It's prudent to verify whether the FUSE server supports FUSE_READDIRPLUS,
so update the session init code accordingly.)
This is the first FUSE request wrapper function that deals with a variable
size tail buffer.
Cc: Ard Biesheuvel <ard.biesheuvel@arm.com>
Cc: Jordan Justen <jordan.l.justen@intel.com>
Cc: Philippe Mathieu-Daudé <philmd@redhat.com>
Ref: https://bugzilla.tianocore.org/show_bug.cgi?id=3097
Signed-off-by: Laszlo Ersek <lersek@redhat.com>
Message-Id: <20201216211125.19496-33-lersek@redhat.com>
Acked-by: Ard Biesheuvel <ard.biesheuvel@arm.com>
Introduce the VirtioFsFuseAttrToEfiFileInfo() function, for converting
FUSE inode attributes to EFI_FILE_INFO.
The EpochToEfiTime() function from EmbeddedPkg's TimeBaseLib proves
invaluable for converting the file access times.
This is the first time we consume TimeBaseLib in OvmfPkg, so add the
necessary lib class resolution. We need not modify any ArmVirtPkg DSC
files: see commit af5fed90bf ("ArmPlatformPkg,ArmVirtPkg: delete
redundant PL031 functions", 2017-05-10).
Cc: Ard Biesheuvel <ard.biesheuvel@arm.com>
Cc: Jordan Justen <jordan.l.justen@intel.com>
Cc: Philippe Mathieu-Daudé <philmd@redhat.com>
Ref: https://bugzilla.tianocore.org/show_bug.cgi?id=3097
Signed-off-by: Laszlo Ersek <lersek@redhat.com>
Message-Id: <20201216211125.19496-22-lersek@redhat.com>
Acked-by: Ard Biesheuvel <ard.biesheuvel@arm.com>
With the help of the VirtioFsFuseOpenDir() and
VirtioFsFuseReleaseFileOrDir() functions introduced previously, we can now
open and close the root directory. So let's implement
EFI_SIMPLE_FILE_SYSTEM_PROTOCOL.OpenVolume().
OpenVolume() creates a new EFI_FILE_PROTOCOL object -- a reference to the
root directory of the filesystem. Thus, we have to start tracking
references to EFI_SIMPLE_FILE_SYSTEM_PROTOCOL, lest we unbind the
virtio-fs device while files are open.
There are two methods that release an EFI_FILE_PROTOCOL object: the
Close() and the Delete() member functions. In particular, they are not
allowed to fail with regard to resource management -- they must release
resources unconditionally. Thus, for rolling back the resource accounting
that we do in EFI_SIMPLE_FILE_SYSTEM_PROTOCOL.OpenVolume(), we have to
implement the first versions of EFI_FILE_PROTOCOL.Close() and
EFI_FILE_PROTOCOL.Delete() in this patch as well.
With this patch applied, the UEFI shell can enter the root directory of
the Virtio Filesystem (such as with the "FS3:" shell command), and the
"DIR" shell command exercises FUSE_OPENDIR and FUSE_RELEASEDIR, according
to the virtiofsd log. The "DIR" command reports the root directory as if
it were empty; probably because at this time, we only allow the shell to
open and to close the root directory, but not to read it.
Cc: Ard Biesheuvel <ard.biesheuvel@arm.com>
Cc: Jordan Justen <jordan.l.justen@intel.com>
Cc: Philippe Mathieu-Daudé <philmd@redhat.com>
Ref: https://bugzilla.tianocore.org/show_bug.cgi?id=3097
Signed-off-by: Laszlo Ersek <lersek@redhat.com>
Message-Id: <20201216211125.19496-12-lersek@redhat.com>
Acked-by: Ard Biesheuvel <ard.biesheuvel@arm.com>
Introduce the VIRTIO_FS_FUSE_REQUEST and VIRTIO_FS_FUSE_RESPONSE
structures, which are the common headers for the various FUSE
request/response structures.
Introduce the VirtioFsFuseNewRequest() helper function for populating
VIRTIO_FS_FUSE_REQUEST, from parameters and from a VIRTIO_FS-level request
counter.
Introduce the VirtioFsFuseCheckResponse() helper function for verifying
most FUSE response types that begin with the VIRTIO_FS_FUSE_RESPONSE
header.
Cc: Ard Biesheuvel <ard.biesheuvel@arm.com>
Cc: Jordan Justen <jordan.l.justen@intel.com>
Cc: Philippe Mathieu-Daudé <philmd@redhat.com>
Ref: https://bugzilla.tianocore.org/show_bug.cgi?id=3097
Signed-off-by: Laszlo Ersek <lersek@redhat.com>
Message-Id: <20201216211125.19496-7-lersek@redhat.com>
Acked-by: Ard Biesheuvel <ard.biesheuvel@arm.com>
Add the VirtioFsInit(), VirtioFsUninit(), and VirtioFsExitBoot()
functions.
In VirtioFsInit():
- Verify the host-side config of the virtio-fs device.
- Save the filesystem label ("tag") for later, from the configuration area
of the virtio-fs device.
- Save the virtio queue size for later as well.
- Set up the virtio ring for sending requests.
In VirtioFsUninit():
- Reset the device.
- Tear down the virtio ring.
In VirtioFsExitBoot():
- Reset the device.
With this patch, the UEFI connect / disconnect controller operations
involve virtio setup / teardown; they are visible in the virtio-fs
daemon's log file. The virtiofsd log also confirms the device reset in
VirtioFsExitBoot(), when an OS is booted while the virtio-fs device is
bound.
Cc: Ard Biesheuvel <ard.biesheuvel@arm.com>
Cc: Jordan Justen <jordan.l.justen@intel.com>
Cc: Philippe Mathieu-Daudé <philmd@redhat.com>
Ref: https://bugzilla.tianocore.org/show_bug.cgi?id=3097
Signed-off-by: Laszlo Ersek <lersek@redhat.com>
Message-Id: <20201216211125.19496-5-lersek@redhat.com>
Acked-by: Ard Biesheuvel <ard.biesheuvel@arm.com>
Complete the Supported, Start, and Stop member functions of
EFI_DRIVER_BINDING_PROTOCOL sufficiently for exercising the UEFI driver
model:
- bind virtio-fs devices,
- produce placeholder EFI_SIMPLE_FILE_SYSTEM_PROTOCOL instances on them.
On the "TO_START" (= Virtio) side, the VirtioFsBindingSupported() function
verifies the Virtio subsystem ID for the virtio-fs device (decimal 26 --
see
<https://github.com/oasis-tcs/virtio-spec/blob/87fa6b5d8155/virtio-fs.tex>).
Beyond that, no actual Virtio setup is performed for now. Those bits are
going to be implemented later in this series.
On the "BY_START" (= UEFI filesystem) side, the VirtioFsOpenVolume()
function -- which is the sole EFI_SIMPLE_FILE_SYSTEM_PROTOCOL member
function -- is a stub; it always returns EFI_NO_MEDIA, for now.
The "CONNECT", "DISCONNECT", and "MAP -R" UEFI Shell commands can be used
to test this patch.
Cc: Ard Biesheuvel <ard.biesheuvel@arm.com>
Cc: Jordan Justen <jordan.l.justen@intel.com>
Cc: Philippe Mathieu-Daudé <philmd@redhat.com>
Ref: https://bugzilla.tianocore.org/show_bug.cgi?id=3097
Signed-off-by: Laszlo Ersek <lersek@redhat.com>
Message-Id: <20201216211125.19496-4-lersek@redhat.com>
Acked-by: Ard Biesheuvel <ard.biesheuvel@arm.com>
It is anticipated that this part of the code will work for both Intel
TDX and AMD SEV, so remove the SEV specific naming and change to
ConfidentialComputing as a more architecture neutral prefix. Apart
from the symbol rename, there are no code changes.
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Message-Id: <20201216014146.2229-3-jejb@linux.ibm.com>
Reviewed-by: Jiewen Yao <Jiewen.yao@intel.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
Although the SEV secret location must always be below 4GB, the same is
not necessarily true for Intel TDX, so change the configuration table
to contain a pair of UINT64 parameters instead of UINT32 so that any X64
location can be represented.
Signed-off-by: James Bottomley <jejb@linux.ibm.com>
Message-Id: <20201216014146.2229-2-jejb@linux.ibm.com>
Reviewed-by: Jiewen Yao <Jiewen.yao@intel.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
BZ: https://bugzilla.tianocore.org/show_bug.cgi?id=2198
Under SEV-ES, a IOIO_PROT intercept generates a #VC exception. VMGEXIT
must be used to allow the hypervisor to handle this intercept.
Add support to construct the required GHCB values to support a IOIO_PROT
NAE event. Parse the instruction that generated the #VC exception,
setting the required register values in the GHCB and creating the proper
SW_EXITINFO1 value in the GHCB.
Cc: Jordan Justen <jordan.l.justen@intel.com>
Cc: Laszlo Ersek <lersek@redhat.com>
Cc: Ard Biesheuvel <ard.biesheuvel@arm.com>
Acked-by: Laszlo Ersek <lersek@redhat.com>
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Regression-tested-by: Laszlo Ersek <lersek@redhat.com>
Sean reports that having two DEC files under OvmfPkg violates the DEC
spec:
> An EDK II Package (directory) is a directory that contains an EDK II
> package declaration (DEC) file. Only one DEC file is permitted per
> directory. EDK II Packages cannot be nested within other EDK II
> Packages.
This issue originates from commit 656419f922 ("Add BhyvePkg, to support
the bhyve hypervisor", 2020-07-31).
Remedy the problem as follows. (Note that these steps are not split to
multiple patches in order to keep Bhyve buildable across the transition.)
(1) Delete "OvmfPkg/Bhyve/BhyvePkg.dec".
(2) Point the [Packages] sections of the Bhyve-specific AcpiPlatformDxe,
BhyveRfbDxe, and BhyveFwCtlLib INF files to "OvmfPkg.dec".
(3) Migrate the artifacts that "BhyvePkg.dec" used to have on top of
"OvmfPkg.dec" as follows:
(3a) Merge the copyright notices from Rebecca Cran and Pluribus Networks
into "OvmfPkg.dec".
(3b) Merge the "BhyveFwCtlLib" class header definition into "OvmfPkg.dec".
(3c) Merge value 0x2F8 for the fixed PcdDebugIoPort into
"BhyvePkgX64.dsc".
(4) Unnest the the Include/Library/ and Library/ subtrees from under
OvmfPkg/Bhyve to the corresponding, preexistent subtrees in OvmfPkg.
The goal is to keep the [Includes] section in the "OvmfPkg.dec" file
unchanged, plus simplify references in "BhyvePkgX64.dsc". Non-library
modules remain under "OvmfPkg/Bhyve/".
(4a) The BhyveFwCtlLib class header, and sole instance, are already
uniquely named, so their movements need not involve file renames.
(4b) Rename the Bhyve-specific PlatformBootManagerLib instance to
PlatformBootManagerLibBhyve, in additon to moving it, for
distinguishing it from OvmfPkg's preexistent lib instance. Apply the
name change to all three of the lib instance directory name, the INF
file, and the BASE_NAME define in the INF file.
(4c) Update lib class resolutions in "BhyvePkgX64.dsc" accordingly.
(5) Replace the "ACPI table storage" FILE_GUID in
"OvmfPkg/Bhyve/AcpiTables/AcpiTables.inf" with a new GUID, and
open-code the "ACPI table storage" GUID in the "ACPITABLE" FDF rule
instead, replacing $(NAMED_GUID). This step is necessary because CI
requires unique FILE_GUIDs over all INF files, and OVMF's original
"AcpiTables.inf" already uses the "ACPI table storage" GUID as
FILE_GUID.
Cc: Ard Biesheuvel <ard.biesheuvel@arm.com>
Cc: Jordan Justen <jordan.l.justen@intel.com>
Cc: Philippe Mathieu-Daudé <philmd@redhat.com>
Cc: Rebecca Cran <rebecca@bsdio.com>
Cc: Sean Brogan <spbrogan@outlook.com>
Fixes: 656419f922
Reported-by: Sean Brogan <spbrogan@outlook.com>
Signed-off-by: Laszlo Ersek <lersek@redhat.com>
Message-Id: <20200801155024.16439-1-lersek@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Rebecca Cran <rebecca@bsdio.com>
This is the second part of LsiScsiPassThru(). LsiScsiProcessRequest() is
added to translate the SCSI Request Packet into the LSI 53C895A
commands. This function utilizes the so-called Script buffer to transmit
a series of commands to the chip and then polls the DMA Status (DSTAT)
register until the Scripts Interrupt Instruction Received (SIR) bit
sets. Once the script is done, the SCSI Request Packet will be modified
to reflect the result of the script. The Cumulative SCSI Byte Count
(CSBC) register is fetched before and after the script to calculate the
transferred bytes and update InTransferLength/OutTransferLength if
necessary.
v3:
- Set DStat, SIst0, and SIst1 to 0 before using them
- Amend the if statements for the DMA data instruction and add the
assertions for the data direction
- Also set SenseDataLength to 0 on the error path
- Fix typos and amend comments
- Amend the error handling of the calculation of transferred bytes
v2:
- Use the BITx macros for the most of LSI_* constants
- Fix a typo: contorller => controller
- Add SeaBIOS lsi-scsi driver as one of the references of the script
- Cast the result of sizeof to UINT32 for the instructions of the
script
- Drop the backslashes
- Replace LSI_SCSI_DMA_ADDR_LOW with LSI_SCSI_DMA_ADDR since we
already removed DUAL_ADDRESS_CYCLE
- Add more comments for the script
- Fix the check of the script size at the end of the script
- Always set SenseDataLength to 0 to avoid the caller to access
SenseData
- Improve the error handling in LsiScsiProcessRequest()
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>
Message-Id: <20200717061130.8881-11-glin@suse.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
Open PciIo protocol and use it to initialize the device. The
initialization of LSI 53C895A is simple: just set the SRST bit in
Interrupt Status Zero register to reset the device.
v2:
- Use the BITx macros for the bit constants
- Add the closing of PciIo protocol in LsiScsiControllerStop()
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>
Message-Id: <20200717061130.8881-8-glin@suse.com>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
In preparation of moving the legacy x86 loading to an implementation
of the QEMU load image library class, introduce a protocol header
and GUID that we will use to identify legacy loaded x86 Linux kernels
in the protocol database.
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
In an upcoming patch, we will introduce a separate DXE driver that
exposes the virtual SimpleFileSystem implementation that carries the
kernel and initrd passed via the QEMU command line, and a separate
library that consumes it, to be incorporated into the boot manager.
Since the GUID used for the SimpleFileSystem implementation's device
path will no longer be for internal use only, create a well defined
GUID to identify the media device path.
Ref: https://bugzilla.tianocore.org/show_bug.cgi?id=2566
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Reviewed-by: Laszlo Ersek <lersek@redhat.com>
In Intel datasheet 316966-002 (the "q35 spec"), Table 5-1 "DRAM Controller
Register Address Map (D0:F0)" leaves the byte register at config space
offset 0x9C unused.
On QEMU's Q35 board, for detecting the "SMRAM at default SMBASE" feature,
firmware is expected to write MCH_DEFAULT_SMBASE_QUERY (0xFF) to offset
MCH_DEFAULT_SMBASE_CTL (0x9C), and read back the register. If the value is
MCH_DEFAULT_SMBASE_IN_RAM (0x01), then the feature is available, and the
range mentioned below is open (accessible to code running outside of SMM).
Then, once firmware writes MCH_DEFAULT_SMBASE_LCK (0x02) to the register,
the MCH_DEFAULT_SMBASE_SIZE (128KB) range at 0x3_0000 (SMM_DEFAULT_SMBASE)
gets closed and locked down, and the register becomes read-only. The area
is reopened, and the register becomes read/write, at platform reset.
Add the above-listed macros to "Q35MchIch9.h".
(There are some other unused offsets in Table 5-1; for example we had
scavenged 0x50 for implementing the extended TSEG feature. 0x9C is the
first byte-wide register standing in isolation after 0x50.)
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: Jordan Justen <jordan.l.justen@intel.com>
Ref: https://bugzilla.tianocore.org/show_bug.cgi?id=1512
Signed-off-by: Laszlo Ersek <lersek@redhat.com>
Reviewed-by: Jiewen Yao <jiewen.yao@intel.com>
Message-Id: <20200129214412.2361-4-lersek@redhat.com>
Reviewed-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
