In the old implementation, function vm_get_suspend_mode is used
as the interface to check whether the UOS is in the S3, That is
not saft way.
This patch use Port Link State (PLS) to substitute the old logic.
According to xHCI spec 5.4.8, the PLS should be U3 during the
UOS is in the S3 state.
Tracked-On: #1893
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Liang Yang <liang3.yang@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
According to xHCI 5.4.8, the Port Link State Change (PLC) bit
should not be set to 1 during suspend process. This patch is used
to fix it.
Tracked-On: #1893
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Liang Yang <liang3.yang@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
Under current implementation, when USB control transfer failed
with Endpoint Stall error, there is no response reporting to UOS.
This logic will result of timeout in UOS kernel and hence a longer
enumeration process.
This patch reports the Stall Error to UOS by xHCI Completion Event,
which will fix this issue.
Tracked-On: #1895
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Liang Yang <liang3.yang@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
Device trap has great impact on latency of real time (RT) tasks.
This patch provide a virtio poll mode to avoid trap.
According to the virtio spec, backend devices can declare the
notification is not needed so that frontend will never trap.
This means the backends make commitment to the frontends they have a
poll mechanism which don’t need any frontends notification.
This patch uses a periodic timer to give backends pseudo notifications
so that drive them processing data in their virtqueues. People should
choose a appropriate notification peroid interval to use this poll
mode. Too big interval may cause virtqueue processing latency while
too small interval may cause high SOS CPU usage. The suggested interval
is between 100us to 1ms.
The poll mode is not enabled by default and traditional trap
notification mode will be used. To use poll mode for RT with interval
1ms. You can add following acrn-dm parameter.
--virtio_poll 1000000
Tracked-On: #1956
Signed-off-by: Jie Deng <jie.deng@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
It is preferred to state the absence of a return value explicitly in the
doxygen-stile comments. Currently there are different styles of doing this,
including:
@return None
@return NULL
@return void
@return N/A
This patch unifies the above with `@return None`.
Tracked-On: #1595
Signed-off-by: Junjie Mao <junjie.mao@intel.com>
v4 -> v5
Replace strcpy with strncpy
Refine resource free if error happens
Remove meaningless *in_progress* label for pthread wait/signal
Rename the coreu thread routine name (virtio_coreu_thread)
v3 -> v4
Move setsocketopt to connect_to_daemon
Cleanup coreu file descriptor after close
Reconnect CoreU daemon if the socket descriptor is invalid
v2 -> v3
Move the daemon connection to vdev_init
Diagram the CoreU virtualization architecture
Create a seperate thread for sending and receving the CoreU message
Change the socket name
v1 -> v2
No change
v1
Initial CoreU back-end driver
Signed-off-by: Yu Shiqiang <shiqiang.yu@intel.com>
Simple atomic add/dec do no guarantee reference count full
synchronization without a lock. Compare and swap operations
are required for correct implementation.
Tracked-On: #1875
Signed-off-by: Shuo A Liu <shuo.a.liu@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Acked-by: Yin Fengwei <fengwei.yin@intel.com>
Fix the prefix comparison, we need to compare on prefix length, not
on devpath len, otherwise we always fail.
Tracked-On: #1848
Signed-off-by: Alexander Usyskin <alexander.usyskin@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Acked-by: Acked-by: Yu Wang <yu1.wang@intel.com>
Reset directory name to receive clean prints.
Tracked-On: #1847
Signed-off-by: Alexander Usyskin <alexander.usyskin@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Acked-by: Acked-by: Yu Wang <yu1.wang@intel.com>
Edge triggered mevent on sysfs file is triggered
immediately after mevent_add(), hence this has to be ignored.
The issue that is resolved that the 'first' flag was
global and wasn't reset on each mevent_add().
Tracked-On: #1846
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Acked-by: Acked-by: Yu Wang <yu1.wang@intel.com>
Prevent intercepting reset callback if reset state
transition is already in progress.
Tracked-On: #1846
Signed-off-by: Alexander Usyskin <alexander.usyskin@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Acked-by: Acked-by: Yu Wang <yu1.wang@intel.com>
(EV_MSC, MSC_TIMESTAMP) is added to each frame just before the
SYN event since kernel 4.15. EV_MSC is configured as
INPUT_PASS_TO_ALL. In the use case of virtio-input, there is
a loop as follows:
- A mt frame with (EV_MSC, MSC_TIMESTAMP) is passed to FE.
- FE will call virtinput_status to pass (EV_MSC, MSC_TIMESTAMP)
back to BE.
- BE writes this event to evdev. Because (EV_MSC, MSC_TIMESTAMP)
is configured as INPUT_PASS_TO_ALL, it will be written into
the event buffer of evdev then be read out by BE without
SYN followed.
- Each mt frame will introduce one (EV_MSC, MSC_TIMESTAMP).
Later the frame becomes larger and larger...
This patch fixed above issue by ignoring MSC_TIMESTAMP from guest.
Besides that timestamp is added for every status event from guest
before writing to evdev.
Tracked-On: #1670
Signed-off-by: Jian Jun Chen <jian.jun.chen@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
The FW reset is currently detected from two points upon
a read failure from native read and from the reset
handler.
The fix removes the detection from the mevent rx callback,
leaving a single detection point.
To prevent reset hiccup, hw_ready is not set if a full rescan is
performed, it will be set only when virtio FW will request the FW reset.
Tracked-On: #1632
Signed-off-by: Aviad Nissel <aviad.nissel@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
vmei_host_client_to_vmei() may return NULL we need to check
for the return value.
Tracked-On: #1630
Signed-off-by: Alexander Usyskin <alexander.usyskin@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Simplify the flow by adding mutex_type variable
and call pthread_mutexattr_destroy() on the error path.
Tracked-On: #1630
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Alexander Usyskin <alexander.usyskin@intel.com>
MEI HBM the disconnect replay doesn't have the address set,
breaking the protocol, fix the issue.
Tracked-On: #1570
Signed-off-by: Alexander Usyskin <alexander.usyskin@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Acked-by: Wang, Yu1 <yu1.wang@intel.com>
- remove ASSERT & DEASSET IRQ line IOCTLs
- remove PULSE IRQ line IOCTLs, use set/clear
IRQ line instead.
- Use IC_SET_IRQLINE to set or clear IRQ line
Tracked-On: #861
Signed-off-by: Yonghua Huang <yonghua.huang@intel.com>
Reviewed-by: Jason Chen CJ <jason.cj.chen@intel.com>
Reviewed-by: Eddie Dong <eddie.dong@intel.com>
Acked-by: Anthony Xu <anthony.xu@intel.com>
Virtio-console device specific configuration is readonly, callback
for the cfgwrite is not required. This patch removed the unused
virtio_console_cfgwrite.
Tracked-On: #1364
Signed-off-by: Jian Jun Chen <jian.jun.chen@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
Remove the changing operation for 'xdev->native_ports[].state' in
pci_xhci_cmd_disable_slot. The operation should not be done in this
funciton.
Tracked-On: #1589
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Liang Yang <liang3.yang@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
On certain SOC, during the system suspend and resume process, connecting
and disconnecting events may happen. In previous implementation, DM didn't
clear PED bit in the xHCI PORTSC register, this will induce many invalid
polling operations in UOS and fail to enumerate one or more USB devices.
Tracked-On: #1589
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Liang Yang <liang3.yang@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
The function for port unassigning: pci_xhci_clr_native_port_assigned
should reset the 'info' member to all zero when it is called.
Tracked-On: #1434
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Liang Yang <liang3.yang@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
This issue is result from reporting interrupt when the interrupt ability
of virtual xHCI is disabled.
As Event Ring State Machine showed in Figure 20 of xHCI spec, following
rules should be followed:
1 when RS bit of USBCMD register is zero, Event Ring should not be accessed;
2 when INTE bit of USBCMD register is zero, Interrupt should not be sent.
Tracked-On: #1566
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Liang Yang <liang3.yang@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
When usb device numbers reach up to XHCI_MAX_SLOTS. The slot_allocated
array will get out of range. This patch is used to fix this issue.
Tracked-On: #1479
Signed-off-by: Liang Yang <liang3.yang@intel.com>
Reviewed-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
After excap pointer is assigned, it should be checked whether it's
possible to get assignment for NULL pointer or not. This patch
fixes this issue.
Tracked-On: #1479
Signed-off-by: Liang Yang <liang3.yang@intel.com>
Reviewed-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Liu Shuo <shuo.a.liu@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
sscanf and strcpy are banned according to the security requirements.
So replace them with their safe alternative.
Tracked-on: #1496
Signed-off-by: Jie Deng <jie.deng@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
physical RPMB is accessed if "phisycal_rpmb" is specified
in launch.sh.
Also it reserves some RPMB area with a fixed size(32KB) for
AttKB and future usage, which is RO for UOS.
Tracked-On: #1544
Signed-off-by: Huang, Yang <yang.huang@intel.com>
Acked-by: Zhu Bing <bing.zhu@intel.com>
In vmei_start() the HBM host client is created,
it handles HBM protocol.
The HBM client is a management object and
is not exposed by the mei native driver.
The communication between TX and RX threads
is handed via internal pipe(2).
Second, we connect all fixed address clients as the mei
protocol doesn't provide connection mechanism for them,
they appear always connected.
Last, the hw_ready is set.
Define virtio_mei_ops, ass all the handlers are now available.
Tracked-On: #1536
Signed-off-by: Liu Shuo <shuo.a.liu@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Acked-by: Wang, Yu1 <yu1.wang@intel.com>
Rx is triggered from epoll event (mevent), a host client
is retrieved from associated event data. P proper mei message header is
attached to the packet and sent to the virtio FE device.
Tracked-On: #1536
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Acked-by: Wang, Yu1 <yu1.wang@intel.com>
TX from virtio driver is handled firt via
virio tx notify handler vmei_notify_tx(),
placed into host clinet tx circular buffer.
TX thread will then write the data to the device.
Tracked-On: #1536
Signed-off-by: Vitaly Lubart <vitaly.lubart@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Acked-by: Wang, Yu1 <yu1.wang@intel.com>
Implement the FW part of the HBM protocol.
Currently the support version is 2.0.
The HBM protocol handles client management, such
initialization handshake, connection, power management,
and the flow control.
Tracked-On: #1536
Signed-off-by: Vitaly Lubart <vitaly.lubart@intel.com>
Signed-off-by: Alexander Usyskin <alexander.usyskin@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Acked-by: Wang, Yu1 <yu1.wang@intel.com>
ME client enumeration is received from sysfs attributes
of the native device.
V3: Use bounded functions.
Tracked-On: #1536
Signed-off-by: Vitaly Lubart <vitaly.lubart@intel.com>
Signed-off-by: Alexander Usyskin <alexander.usyskin@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Acked-by: Wang, Yu1 <yu1.wang@intel.com>
mei mediator perform reset handshake via
the cfgread/write handlers and provide access
to fw status registers.
Tracked-On: #1536
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Acked-by: Wang, Yu1 <yu1.wang@intel.com>
There are multiple types of resets that need to be handled
by the mei device.
The reset may be initiated from both sides host or fw.
The host requests reset on probe and remove,
power state transaction, and errors,
while ME FW may request reset upon error.
If the native device undergo reset the host application
has to be notified.
Tracked-On: #1536
Signed-off-by: Aviad Nissel <aviad.nissel@intel.com>
Signed-off-by: Alexander Usyskin <alexander.usyskin@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Acked-by: Wang, Yu1 <yu1.wang@intel.com>
1. virtio_mei structure represents an instance of mei device.
2. vmei_me_client represents an ME application in the MEI FW.
3. vmei_host_client represent a host application talking to the
ME application, ME application can support multiple connections.
4. Add debug helpers
Tracked-On: #1536
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Acked-by: Wang, Yu1 <yu1.wang@intel.com>
The virtio has two virtio queues, TX and RX.
TX have two segment buffers one for header and one for data.
The virtio_mei declares host buffer of size 128 slots, each
slot is a 4 bytes value.
For synchronization hw_ready and host_reset configuration flags
are used that emulates mei reset flow.
And last fw status registers must be also readable via virtio
mei device.
Tracked-On: #1536
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Acked-by: Wang, Yu1 <yu1.wang@intel.com>
mei handles objects on the list, hence reference counting
infrastructure is required for easier multithreading.
Tracked-On: #1536
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Acked-by: Wang, Yu1 <yu1.wang@intel.com>
libuuid sports only uuid (big endian encoding),
though mei requires guids (little endian encoding).
The base types are based on <linux/uuid.h> header.
Tracked-On: #1536
Signed-off-by: Alexander Usyskin <alexander.usyskin@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Acked-by: Wang, Yu1 <yu1.wang@intel.com>
Remove virtio_heci to be replaced with virtio_mei
Tracked-On: #1536
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Acked-by: Wang, Yu1 <yu1.wang@intel.com>
This patch is one workaround to resolve a crash issue for certain brand
touch screen (eGalaxTouch EXC7200-7368v1.01). The formal fix should
follow the xHCI spec to stop xfer and generate transfer completed event
trb prior to Stop Endpoint Command complete event trb. It should be a
big change and for short term, do nothing for the stop endpoint command
which is no other side effect be observed so far.
Tracked-On: #1413
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Liang Yang <liang3.yang@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
The function pci_xhci_get_native_port_index_by_path didn't
compare the bus number during the process of native USB
device searching. This patch is used to fix it.
Tracked-On: #1434
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Liang Yang <liang3.yang@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
This patch is used to enable multiple hubs in single layer under
Flat Mode.
Tracked-On: #1434
Signed-off-by: Liang Yang <liang3.yang@intel.com>
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
To support the multiple hubs and multiple layers of hub, the
port_map_tbl should be replaced by native_ports to record all
native devices' state, including assignment state, emulation
state etc.
Tracked-On: #1434
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Signed-off-by: Liang Yang <liang3.yang@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
This patch introduce struct usb_devpath to indentify uniquely an USB
device, which is basic element for multiple hub support.
Tracked-On: #1434
Signed-off-by: Liang Yang <liang3.yang@intel.com>
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
Previous design use bus and port(root hub port) for the identification
to the native USB device. It cannot work properly under the multi-hub
situation due to external USB hub is introduced.
This patch removes old hub implementation code, and subsequent patches
will be add support for mutiple hubs.
Tracked-On: #1434
Signed-off-by: Liang Yang <liang3.yang@intel.com>
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
As RPMB mux kernel module is going to be created,
there are two corresponding changes required for DM:
1. The name has been changed to /dev/rpmbmux.
2. DM does NOT check MAC of RPMB result returned by kernel
module because DM doesn't own the real key.
Tracked-On: #1508
Signed-off-by: Huang, Yang <yang.huang@intel.com>
Acked-by: Zhu Bing <bing.zhu@intel.com>
Per i6300esb spec, when WDT_INT_TYPE(bit 0 and 1 of WDT config register)
are set to 00, IRQ feature should be supported.
The WDT_INT_ACTIVE bit is set when the first stage of the 35-bit
down-counter reaches zero. An interrupt will be generated if WDT_INT_TYPE
is configured to do so (See WDT Configuration Register). This is a sticky
bit and is only cleared by writing a 1.
SMI feature(WDT_INT_TYPE are set to 0x10) is not supported.
Tracked-On: #1498
Signed-off-by: Victor Sun <victor.sun@intel.com>
Acked-by: Yin Fengwei <fengwei.yin@intel.com>
In the current implementation sigev_notify is configured as
SIGEV_THREAD. When wdt expires an async thread is created and
the registered timer callback is called in the context of this
thread, then the watchdog interrupt emulation would require the
thread to assert intr on this pci dev.
There would be a race condition that when the wdt pci device is
freed in pci device deinit and then a timer expires. In this case
the wdt expired thread will access a freed buffer which would cause
problem such as heap corruption and segment fault.
In this patch we replace timer API with acrn_timer which is based
on timerfd/epoll mechanism to avoid the race condition.
Tracked-On: #1489
Signed-off-by: Victor Sun <victor.sun@intel.com>
Reviewed-by: Jian Jun Chen <jian.jun.chen@intel.com>
Reviewed-by: Minggui Cao <minggui.cao@intel.com>
Acked-by: Yin Fengwei <fengwei.yin@intel.com>
VirtIO v1.0 spec 04 5.2.5:
- Protocol unit size is always 512 bytes.
- blk_size (logical block size) and physical_block_exp (physical block
size) do not affect the units in the protocol, only performance.
VirtIO v1.0 spec 04 5.2.6.1:
- A driver MUST NOT submit a request which would cause a read or write
beyond capacity.
Reject the requests that violate these terms.
v1 -> v2:
- add more comments for clarity
Tracked-On: #1422
Signed-off-by: Peter Fang <peter.fang@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
NHLT table contains the settings some audio drivers need.
An ACPI method is used to get NHLT table address & length.
In current DM code, the NHLT talbe length in the ACPI method
is hardcoded, which will cause troubles when the length of the
table changed.
This patch replaces the hardcoded NHLT table length according to
the table length read from SOS.
Tracked-On: #1461
Signed-off-by: Binbin Wu <binbin.wu@intel.com>
Acked-by: Yin Fengwei <fengwei.yin@intel.com>
The write_dsdt_xhci function is use for describe the xdci dsdt table.
Correct its name.
Tracked-On: #1444
Signed-off-by: Yu Wang <yu1.wang@intel.com>
Reviewed-by: Binbin Wu <binbin.wu@intel.com>
movb is used for registers STATUS and CFGGENERATION whose size is 1
byte. Previously hv cannot report the correct MMIO trap size for
movb and virtio hard coded their size to 4 as a workaround. hv fixed
movb instruction emulation and MMIO size can be reported correctly.
This patch removes those workaround.
commit 9df8790ffc ("hv: Fix two minor issues in instruction emulation code")
Tracked-On: #1449
Signed-off-by: Jian Jun Chen <jian.jun.chen@intel.com>
Acked-by: Yin Fengwei <fengwei.yin@intel.com>
Add a function to do msix table deinit.
1. call api to reset msix entry in hypervisor
2. free virtual msix table memory
3. unmap pba page if any
4. unmap the pages passhtru to uos in MSIX BAR if any
Tracked-On: #1222
Signed-off-by: Binbin Wu <binbin.wu@intel.com>
Acked-by: Yin Fengwei <fengwei.yin@intel.com>
Normally, for devices support MSI-X, PBA is passed-through to guest.
However, PBA and MSI-X table share the same bar, and part of PBA and
MSI-X table may share a same page for some devices.
If that is the case, the part of PBA within the page should be emulated
rather than passed-through.
This patch adds PBA emulation support for MSI-X.
Tracked-On: #1222
Signed-off-by: Binbin Wu <binbin.wu@intel.com>
Acked-by: Yin Fengwei <fengwei.yin@intel.com>
This is an issue result from incomplete process logic of commit:
"ba68bd4 DM USB: xHCI: fix enumeration error after rebooting".
This patch is used to fix it.
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Liang Yang <liang3.yang@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
Tracked-On: #1425
Use strnlen instead of strlen to avoid potential security issue.
Tracked-On: #1364
Signed-off-by: Jian Jun Chen <jian.jun.chen@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
According to PCI spec 3.0 section 6.2.2 "Device Control", guest
could write the command register to control device response to
io/mem access.
The origial code register/unregister the memory range which is
not suitable because it can't handle the sequence:
1. disble the device response to specific memory range
2. reboot guest (DM will try to free the memory range which
was freed in step 1 already)
Tracked-On: #1277
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
Remove the workaround and will submit the fixing patch.
Tracked-On: #1277
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
When the physical USB device is disconnected before DM's emulation
is ready, the virtual connection state is not cleared properly. This
will cause the DM refuse to do emulation for future physical connection.
This patch clears those states mentioned above and hence fix the issue.
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Liang Yang <liang3.yang@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
Tracked-On: #1367
When USB hub is disconnected, its connection status is not cleared
in the DM, this defect will cause the related assigned port could
not be used any more.
This patch is used fix this issue.
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Liang Yang <liang3.yang@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
Tracked-On: #1365
Previous implementation binds libusb (in other words: usbfs) with
native device when DM receives the Enable Slot command. But according
to xHCI spec 4.6.5, the binding relationship is decided when the
Address Device command is received, so this implementation is not
consistent with hardware behaviors.
And this incompatible could induce following issue. When two or more
USB devices are connected at the same time, eg, connecting two devices
before Guest OS is booted, the virtual slot id may bind to wrong root
hub port.
This patch will do the binding when Address Device command is received
and related issues will be fixed.
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Liang Yang <liang3.yang@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
Tracked-On: #1366
According to xHCI spec, the Stop Endpoint command should execute
no matter the endpoint is in halted state or not, but current
implementation just refuse do the command and return error to
Guest OS. It is wrong and this patch is used to fix it.
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Liang Yang <liang3.yang@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
Tracked-On: #1366
One additional command parameter is added for virtio-net to support
vhost net. The command line for vhost net is as follows:
-s n,virtio-net,tap_xxx,vhost
Tracked-On: #1329
Signed-off-by: Jian Jun Chen <jian.jun.chen@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
vhost proxy interacts with vhost kernel thru vhost char dev. Internal
interfaces are implemented based on ioctls of vhost char dev in this
patch.
Tracked-On: #1329
Signed-off-by: Jian Jun Chen <jian.jun.chen@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
There are 2 eventfds for one virtqueue, one is for kick and the other
is for notify. eventfd used for kick is associated with a PIO/MMIO
region. eventfd used for notify is associated with a MSIx/INTx. The
eventfd pair is registered to VHM thru VHM char dev.
VHM irqfd currently only support MSIx. If INTx is used, vhost proxy
uses mevent to poll the call fd from vhost then inject interrupt to
guest.
Tracked-On: #1329
Signed-off-by: Jian Jun Chen <jian.jun.chen@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
vhost kernel driver needs the information of memory mapping between
GPA and the virtual addresses in device model process. This is
required for virtqueue related operations. This patch gets memory
mapping information from vmctx then conveys to vhost.
Tracked-On: #1329
Signed-off-by: Jian Jun Chen <jian.jun.chen@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
vhost_vq related interfaces are implemented in this patch. They are
vhost_vq_init/vhost_vq_deinit/vhost_vq_start/vhost_vq_stop.
Tracked-On: #1329
Signed-off-by: Jian Jun Chen <jian.jun.chen@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
This patch adds the vhost support to the device model virtio. A vhost
proxy is implemented based on the virtio framework and vhost char dev.
Key data structures and external interfaces are implemented in this
patch.
Tracked-On: #1329
Signed-off-by: Jian Jun Chen <jian.jun.chen@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
Some virtio ring structures and virtio feature bits are using the
same name/definition as those in kernel header files(linux/
virtio_ring.h, linux/virtio_config.h). Kernel header files must
be included to perform ioctls to support vhost. There are
compiling errors due to duplicated definitions. In this patch
the following renamings are done:
VRING_DESC_F_NEXT -> ACRN_VRING_DESC_F_NEXT
VRING_DESC_F_WRITE -> ACRN_VRING_DESC_F_WRITE
VRING_DESC_F_INDIRECT -> ACRN_VRING_DESC_F_INDIRECT
VRING_AVAIL_F_NO_INTERRUPT -> ACRN_VRING_AVAIL_F_NO_INTERRUPT
VRING_USED_F_NO_NOTIFY -> ACRN_VRING_USED_F_NO_NOTIFY
VIRTIO_F_NOTIFY_ON_EMPTY -> ACRN_VIRTIO_F_NOTIFY_ON_EMPTY
VIRTIO_RING_F_INDIRECT_DESC -> ACRN_VIRTIO_RING_F_INDIRECT_DESC
VIRTIO_RING_F_EVENT_IDX -> ACRN_VIRTIO_RING_F_EVENT_IDX
VIRTIO_F_VERSION_1 -> ACRN_VIRTIO_F_VERSION_1
vring_avail -> virtio_vring_avail
vring_used -> virtio_vring_used
vring_size -> virtio_vring_size
Tracked-On: #1329
Signed-off-by: Jian Jun Chen <jian.jun.chen@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
Netmap/vale attempts to speed up network communication by bypassing the
TCP/IP network stack, which requires patching the physical NIC driver so
that applications developed based on netmap can interact directly with
the physical NIC driver. It may make sense for some specific scenarios
which requres very high bandwith (10Gb/s or 100Gb/s), we can even put up
with the complexity and compatibility introduced by this techology.
However for ACRN, a virtualization solution for IoT, there is no need to
support this backend. For 1Gb NICs or below, the VBS-U/tap solution
can already achieve near-native bandwidth. To keep simplicity and
improve compatibility, remove the netmap/vale support in dm.
Tracked-On: #1313
Signed-off-by: Jie Deng <jie.deng@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
Acked-by: Anthony Xu <anthony.xu@intel.com>
CAVS audio driver depends on the NHLT table to get topology info.
Enable NHLT table in DM for audio passthrough.
Also increase the size reserved for NHLT in ACPI table from 2048B to 2560B.
Tracked-On: #1284
Signed-off-by: Binbin Wu <binbin.wu@intel.com>
Reviewed-by: Geoffroy Van Cutsem <geoffroy.vancutsem@intel.com>
Acked-by: Yin Fengwei <fengwei.yin@intel.com>
This is temperory workaround for DM crash when doing fastboot
reboot. In fastboot, it will disable USB host functionality by
disable device respsone to one PCI bar. While DM code just release
the bar in this case. Which break the reboot functionality.
The workaround is to remove the assert to avoid DM abort. This is
safe because reboot will remove all memory range registered.
We will have offiical fixing later.
Tracked-On: #1277
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
USB device is connected before UOS is booted up, this scenario is
called 'cold plug' for easy to refer.
Under 'cold plug' situation, the libusb will not report 'connect'
event to device model, hence UOS will not discover 'cold plugged'
device.
This patch add support to fix this issue.
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Liang Yang <liang3.yang@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
Tracked-On: #1242
Flat Mode for hub emulation means DM emulates USB devices under hub but
hide hub itself. Under this design the Guest OS cannot see any emulated
hub. So in the perspective of Guest OS, all the emulated devices are
under root hub.
This patch is used to enable feature as mentioned above. And please NOTE,
it is the initial version of hub flat Mode hub emulation, there are one
limitation: only one physical hub is supported. If second physical hub is
connected, the connect and disconnect behavior in second hub may affect
the function of first emulated hub.
The USB HUB device model should be the final long term solution, but it is
very complex. Use flat mode HUB emulation as the short term solution first
to support some USB touch devices which integrated internal HUB.
Signed-off-by: Liang Yang <liang3.yang@intel.com>
Reviewed-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
Tracked-On: #1243
In current code, when read/write msix table, it first handle the case
the offset is in pba range when msix talbe and pba share the same bar.
But the code didn't add the condition whether pba bar equals msix table
bar. It will cause problems for the passthrugh devices,whose pba and msix
table don't share the same bar.
Tracked-On: #1209
Signed-off-by: Binbin Wu <binbin.wu@intel.com>
Acked-by: Yin Fengwei <fengwei.yin@intel.com>
if some pci devices behind gvt got failures when initlizing at init_pci(), gvt
instance have no chance to be destroied even acrn-dm exits.
NOTE: this patch can not work standalone, a following patch to kernel
side(643d40961cf: "vhm: init client->kthread_exit true") is requied as well,
otherwise it will stucks during destroying gvt instance.
Tracked-On: #1141
Signed-off-by: Li Zhijian <zhijianli88@163.com>
Unmap ptdev BAR when deinit to comply with native system, who zap out
all pre-allocated BARs.
Tracked-On: #1146
Signed-off-by: Edwin Zhai <edwin.zhai@intel.com>
Acked-by: Yin Fengwei <fengwei.yin@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
1. In default prescaler, the wdt clock is 1 KHz for a 20-bit counter,
which means approximate 1 second for 10 bits;
2. the default reset timer in seconds need to left shift 10 bits to
represent the value that set to i6300esb register;
Tracked-On: #1142
Signed-off-by: Victor Sun <victor.sun@intel.com>
Reviewed-by: Minggui Cao <minggui.cao@intel.com>
Acked-by: Yin Fengwei <fengwei.yin@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
To align with the configuration of the HOST tool, the SW masters
are organized in slices of 8 masters each.
The slice is also the minimal unit to allocate the SW masters for
each UOS.
The patch is to update the parameter checking function.
Tracked-On: #1138
Signed-off-by: Zhi Jin <zhi.jin@intel.com>
Reviewed-by: Chen Gang <gang.c.chen@intel.com>
This patch is used to fix a potential issue resulted from typo.
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Liang Yang <liang3.yang@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
When reset VBS-K status should be set to VIRTIO_DEV_INIT_SUCCESS
because at the time the char dev of VBS-K is still opened and
when set_status callback is called later, it depends on
VIRTIO_DEV_INIT_SUCCESS to resume.
Signed-off-by: Jian Jun Chen <jian.jun.chen@intel.com>
Reviewed-by: Shuo Liu <shuo.a.liu@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
A new ioctl is introduced in VBS-K to issue reset command to kernel
VBS-K driver. This is used to support VBS-K S3. When FE enters S3
reset command is sent to device model. Backend driver in device model
should use this ioctl to inform the VBS-K drvier in kernel.
Signed-off-by: Jian Jun Chen <jian.jun.chen@intel.com>
Reviewed-by: Shuo Liu <shuo.a.liu@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
RPMB frontend driver in UOS kernel has fixed unstable issue,
which requires BE for update as well. E.g. structure adjustment,
definition modification and so on.
Signed-off-by: Deng Wei <wei.a.deng@intel.com>
Signed-off-by: Huang Yang <yang.huang@intel.com>
Acked-by: Zhu Bing <bing.zhu@intel.com>
This patch enable the support for SOS S3 from the perspective
of USB xHCI.
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Liang Yang <liang3.yang@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
The xHCI emulation greatly depends on the user space library libusb
which is based on the usbfs module in Linux kernel. The libusb will
bind usbfs to physical USB device which makes hardware control over
libusb in user space possible.
The pci_xhci_dev_create is called in pci_xhci_native_usb_dev_conn_cb
which is a callback function triggered by physical USB device plugging.
This function will bind the physical USB device to usbfs in SOS, which
we depend to create the communication between UOS xHCI driver with
physical USB device.
This design will fail if the reconnection happened in the SOS, which
will bind class driver to the physical USB device instead of usbfs,
hence the libusb device handle in DM is invalid.
Currently, the native S3 will disable the vbus for all xHCI ports and
re-drive during S3 resume. This behavior cause native USB driver unbind
the usbfs and bind to related class driver, then made the DM lost
control and failed to continue emulation.
To fix this issue, place the pci_xhci_dev_create in the function
pci_xhci_cmd_enable_slot. According to the xHCI spec 4.5.3 Figure 10,
the UOS always send Enable Slot command when a device is attached or
recovered from errors (by Disable Slot command). So every time the SOS
can't resuming normally or some unexpected disconnection happens, this
desigen will always survive by Disable Slot and Enable Slot command
series from UOS xHCI driver.
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Liang Yang <liang3.yang@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
Replace 'native_assign_ports' with 'port_map_tbl' to be more accurate
for the role of this variable plays.
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Liang Yang <liang3.yang@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
The variable native_assign_ports in struct pci_xhci_vdev is used
to record wether certain root hub port in SOS is assigned to UOS.
The logic uses zero to express 'not assigned' and nonzero to express
'assigned'. In this patch, use macro to replace number to express
better.
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Liang Yang <liang3.yang@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
Currently the maximum number of bus and port for xHCI are
both set to 255, it is theoretically possible but in fact
not neccessary. This patch changes those two values to be
more proper: 4 buses and 20 ports.
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Liang Yang <liang3.yang@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
Current design cannot get physical USB device information without
the creation of pci_xhci_dev_emu. This brings some difficulties in
certain situations, hence struct usb_native_devinfo is introduced
to describe neccessary information to solve this trouble.
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Liang Yang <liang3.yang@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
PORTSC (Port Status and Control Register) register play a very
important role in USB sub-system. This patch is used to refine
related manipulation functions.
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Liang Yang <liang3.yang@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
Current DM design use two variables to do the indexing of xHCI
Event Ring: er_enq_idx and er_events_cnt. They are members of
the struct pci_xhci_rtsregs.
In UOS, during the process of xHCI resuming, the xHCI driver
will restore the ERSTBA (Event Ring Segment Table Base Address)
register to be the value before suspending. And at this point,
the old DM implementation will set both er_enq_idx and
er_events_cnt to be zero, so the DM will access the Event Ring
from the start position in the buffer. But at the same time the
UOS xHCI driver still wants to access the old position in the
Event Ring before suspending, which will result of unexpected
errors.
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Liang Yang <liang3.yang@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
This device model is to configure the virtual IPU PCI device.
In order to execute this DM the lauch script needs to add
virtio-ipu as parameter to acrn-dm
For e.g. -s 21,virtio-ipu
Signed-off-by: Bandi,Kushal <kushal.bandi@intel.com>
Randomness sourced from /dev/random which does not block
once it has been seeded at bootup and you will always get
something when you read from that file. This is true on
Freebsd but unfortunately things are not the same on Linux.
Most cases, you can't read anything from /dev/random especially
on current acrn platform which lacking random events.
virtio_rnd inherted from freebsd doesn't work anymore.
This patch makes virtio_rnd working on Linux based SOS. It uses
blocking IO to sevice the front-end random driver and delays the
read operation into a new thread to avoid blocking the main
notify thread.
Signed-off-by: Jie Deng <jie.deng@intel.com>
Reviewed-by: Shuo Liu <shuo.a.liu@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
1. support "writeback" and "writethru" mode toggling for virtio-blk
conditionally. When starting DM with "writethru" parameter in
virtio-blk, guest OS could not toggle cache mode. When starting DM
with "writeback" parameter in virtio-blk, guest OS could toggle
cache mode.
------------------------------
DM cmdline | toggle support
------------+-----------------
writeback | yes
writethru | no
------------------------------
2. To toggle cache mode, run below command in guest OS:
echo "write back" > /sys/devices/xxx/vdx/cache_type
OR
echo "write through" > /sys/devices/xxx/vdx/cache_type
Signed-off-by: Conghui Chen <conghui.chen@intel.com>
Reviewed-by: Shuo Liu <shuo.a.liu@intel.com>
Acked-by: Yin Fengwei <fengwei.yin@intel.com>
currently, each virtio device has their own virtio_ops implementation.
Take virtio-blk for example:
static struct virtio_ops virtio_blk_ops = {
"virtio_blk",
1,
sizeof(struct virtio_blk_config),
virtio_blk_reset,
virtio_blk_notify,
virtio_blk_cfgread,
virtio_blk_cfgwrite,
NULL,
NULL,
VIRTIO_BLK_S_HOSTCAPS,
};
If start DM with two virtio-blk, this global variable will be
assigined to two virtio-blk instances. Changing hv_caps for one
instance will affect others. But different instances may need
different capabilities.
To support this requirement, we suggest to move hv_caps to
virtio_base structure, and each instance can return their own
capabilities.
Signed-off-by: Conghui Chen <conghui.chen@intel.com>
Reviewed-by: Yin Fengwei <fengwei.yin@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
With latest FW, ethernet/wifi BDF changes from 3:0.0/4:0.0 to
2:0.0/3:0.0.
Signed-off-by: Edwin Zhai <edwin.zhai@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
When guest enter/exit S3, we need to do
1. stop watchdog timer when guest enter S3 to avoid watchdog
timer reset guest when guest is in S3 state.
2. reset watchdog timer when guest exit from S3.
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Add microframe index register support, which is an important timing
component for isochronous transport.
Change-Id: I615664275b539cfb713d7795edd3f213b0302b92
Tracked-On:
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Liang Yang <liang3.yang@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
Current ring buffer processing logic assumes every transaction
will be submited to physical device before next transaction
coming. So it use two states 0 (free) and 1 (used) to represent
the state of every data block in the ring buffer. With the help
of the two state, the ring buffer could accept and process data
normally.
But this logic is not proper for ISOC transfer, which generally
submits many transactions even none of them arrive the physical
device. So this patch uses three values to represent the state
of data block in the ring buffer:
USB_XFER_BLK_FREE: this block could be filled with new data;
USB_XFER_BLK_HANDLING: this block is submited to physical device
but response from device is still not received;
USB_XFER_BLK_HANDLED: this block has been processed by physical
device.
The new logic will do different things for each state, which will
make the ISOC transfer work successfully.
Change-Id: I5559cae24c739633289742d64dd51751797b81a7
Tracked-On:
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Liang Yang <liang3.yang@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
The xHCI speed emulation is not right, which will cause failure
during enumeration of certain USB device. This patch is used to
fix it.
Change-Id: I2d996298983882ed6921a75a10dec9e8684a393e
Tracked-On:
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Liang Yang <liang3.yang@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
This patch is used to add support for USB 3.0 devices. Currently
USB 3.0 disk is supported and tested successfully.
Change-Id: I3fbfbe9c28bc4b14af0417104f8fa822f9758908
Tracked-On:
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Liang Yang <liang3.yang@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
Original code will reset the whole USB device when xHCI Reset
Endpoint command is received, this behavior is not right. This
patch is used to fix it.
And according to xhci spec 4.6.8, if the endpoint is not in the
halted state, xHC should reject to execute this command and the
Context State Error should be returned. This patch also add this
logic.
Change-Id: I55a5918148d82d103fb3eb27d582f9676f9f61d3
Tracked-On:
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Liang Yang <liang3.yang@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
Mainly change the logs for transfer submission and completion,
which are very important parts in USB emulation code.
Change-Id: I4e04f1426e164ca3693e70946ed51380201e49ee
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Liang Yang <liang3.yang@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
This function add high level reset_vdev function. Which is
implemented to call deinit/init pairing to emulate the virtual
device reset operation.
This patch also add the system reset which keep the UOS RAM
content functionality to DM.
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Acked-by: Anthony Xu <anthony.xu@intel.com>
Guest has erquirement to support system/full reboot and S3. Which could
trigger different reset path in guest
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Acked-by: Anthony Xu <anthony.xu@intel.com>
We should keep variable name in function declaration. It makes
things clearer and easier to be understood.
Signed-off-by: Jie Deng <jie.deng@intel.com>
Acked-by: Yu Wang <yu1.wang@intel.com>
There has one new DRD driver followed usb role framework which is just
upstreamed to Linux community. This patch updates the xHCI DM to be
compatible with it. DM DRD code follows DRD spec to implement and make
it more reasonable.
Signed-off-by: Liang Yang <liang3.yang@intel.com>
Reviewed-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Yu Wang <yu1.wang@intel.com>
Acked-by: Anthony Xu <anthony.xu@intel.com>
For dedicated xHCI extended capability, it need set corresponding PCI VID/PID.
This patch sets the Intel Apollo Lake platform PCI VID/PID for DRD
capability which will be checked for enabling DRD fucntion in new DRD
driver. Besides, this patch refines the PCI VID/PID related code.
Signed-off-by: Liang Yang <liang3.yang@intel.com>
Reviewed-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Yu Wang <yu1.wang@intel.com>
Acked-by: Anthony Xu <anthony.xu@intel.com>
6300esb has bit in its register to show whether the watchdog
timeout is hit.
This patch adds this bit support. So the guest could query
whether last reset is triggered by watchdog reset.
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Reviewed-by: Cao Minggui <minggui.cao@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
Tested-by: Binbin Wu <binbin.wu@intel.com>
The lpc dm causes that UOS can't boot if the parameters are set incorrectly,
it is not friendly to users.
This patch optimizes the lpc error handle flow. UOS always can boot successfully
whatever the lpc settings are.
Signed-off-by: Yuan Liu <yuan1.liu@intel.com>
Reviewed-by: Yu Wang <yu1.wang@intel.com>
WIFI dev has no FLR, so 'reset' in sysfs calls secondary bus reset,
which cause PCI configuration mess(all FF) then passthrough failure.
To fix it, this patch makes no reset before passthrough by default,
until append this option.
Signed-off-by: Edwin Zhai <edwin.zhai@intel.com>
Acked-by: Anthony Xu <anthony.xu@intel.com>
Current option of removing vGSI capability is global, which exposes
vIOAPIC link for all ptdev even only one need this. This patch makes
it as ptdev local option to lower the system level impact. To keep
vGSI for MSI capable ptdev, just explicitly append ",keep_gsi" in
option list, like "-s 14,passthru,0/e/0,keep_gsi"
Signed-off-by: Edwin Zhai <edwin.zhai@intel.com>
Acked-by: Anthony Xu <anthony.xu@intel.com>
Previously, either rnd->fd or vbs_k->fd isn't be closed in some cases.
this patch will close them in time.
V2: fix vbs_k->fd leak as well
Signed-off-by: Li Zhijian <lizhijian@cn.fujitsu.com>
- hardcode the devices' GSI info based on the platform
- reject the passthrough if the following requirement is not met
all the PCI devices that are sharing the same GSI should be assigned
to same VM to avoid physical GSI sharing between multiple VMs.
v4 -> v5
* Move the gsi_dev_mapping_tables definition in a separate file
* Add the GSI info that might be used by GPIO
* Update the HW name
v3 - > v4
* Refine the format of raw data to improve the readability
* Remove the redundant code when adding the new dev into the gsi
sharing group
v2 -> v3
* Add the MSI/MSI-x capability check
Do not add the device which supports MSI/MSI-x to the GSI sharing
group.
v1 -> v2
* Update the GSI raw data based on SBL
* Free the resources when gsi sharing violation occurs
* Move the MACRO PCI_BDF(b, d, f) to pci_core.h since passthrough.c
and gsi_sharing.c are both using it
Signed-off-by: Shiqing Gao <shiqing.gao@intel.com>
Reviewed-by: Edwin Zhai <edwin.zhai@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
The device model is a userspace application on SOS to config the
PCI devices for the UOS. Audio mediator device model is to config
the virtual audio PCI device.
Signed-off-by: Zhu Yingjiang <yingjiang.zhu@linux.intel.com>
When enabling GPU passthru for guest, stolen memory needs to be
disabled. This change disables stolen memory in passthru mode.
Signed-off-by: Anitha Chrisanthus <anitha.chrisanthus@intel.com>
Reviewed-by: Gong Zhipeng <zhipeng.gong@intel.com>
To enable GPU passthru, BAR2 allocation had problems when the allocation
is above the 4GB memory.This change is to accomodate the 256MB aperture
allocation in BAR2 to a more managable address.
v2: changed the MEMBASE64 address instead of increasing the mem_size.
Signed-off-by: Anitha Chrisanthus <anitha.chrisanthus@intel.com>
Reviewed-by: Gong Zhipeng <zhipeng.gong@intel.com>
acrn-dm runs to segmentation fault when failed to create VMs with
improper parameters.
If vdevs failed to be created, they are still be freed in deinit(),
and dereference the null pointers leads to segfault.
Signed-off-by: Xinyun Liu <xinyun.liu@intel.com>
Reviewed-by: Yin Fengwei <fengwei.yin@intel.com>
Acked-by: Anthony Xu <anthony.xu@intel.com>
This patch fix some potential crash issues, like wild
pointers access, buffer overflow and etc.
Change-Id: Iddd8e1820da426adc6b9b4d9da9e44017d9f365c
Signed-off-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Yu Wang <yu1.wang@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
With current implementation:
vm_init_vdevs only handles the negative error code, while passthru_init
returns positive error code when error occurs.
This causes unexpected dm crash since the real error is not being
handled properly.
What this patch does:
Change the error code to be negative value in passthru_init because it
is common in Linux kernel to return negative value when error occurs.
v2 -> v3
* add more comments about the reason to convert the return value
v1 -> v2:
* add a wrapper API to convert the error returned from pci_system_init
to the ERROR we defined in DM
* use the defined errno as the return value rather than -1
Signed-off-by: Shiqing Gao <shiqing.gao@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
The Intel Trace Hub (aka. North Peak, NPK) is a trace aggregator for
Software, Firmware, and Hardware. On the virtualization platform, it
can be used to output the traces from SOS/UOS/Hypervisor/FW together
with unified timestamps.
There are 2 software visible MMIO space in the npk pci device. One is
the CSR which maps the configuration registers, and the other is the
STMR which is organized as many Masters, and used to send the traces.
Each Master has a fixed number of Channels, which is 128 on GP. Each
channel occupies 64B, so the offset of each Master is 8K (64B*128).
Here is the detailed layout of STMR:
M=NPK_SW_MSTR_STP (1024 on GP)
+-------------------+
| m[M],c[C-1] |
Base(M,C-1) +-------------------+
| ... |
+-------------------+
| m[M],c[0] |
Base(M,0) +-------------------+
| ... |
+-------------------+
| m[i+1],c[1] |
Base(i+1,1) +-------------------+
| m[i+1],c[0] |
Base(i+1,0) +-------------------+
| ... |
+-------------------+
| m[i],c[1] |
Base(i,1)=SW_BAR+0x40 +-------------------+
| m[i],c[0] | 64B
Base(i,0)=SW_BAR +-------------------+
i=NPK_SW_MSTR_STRT (256 on GP)
CSR and STMR are treated differently in npk virtualization because:
1. CSR configuration should come from just one OS, instead of each OS.
In our case, it should come from SOS.
2. For performance and timing concern, the traces from each OS should
be written to STMR directly.
Based on these, the npk virtualization is implemented in this way:
1. The physical CSR is owned by SOS, and dm/npk emulates a software
one for the UOS, to keep the npk driver on UOS unchanged. Some CSR
initial values are configured to make the UOS npk driver think it
is working on a real npk. The CSR configuration from UOS is ignored
by dm, and it will not bring any side-effect. Because traces are the
only things needed from UOS, the location to send traces to and the
trace format are not affected by the CSR configuration.
2. Part of the physical STMR will be reserved for the SOS, and the
others will be passed through to the UOS, so that the UOS can write
the traces to the MMIO space directly.
A parameter is needed to indicate the offset and size of the Masters
to pass through to the UOS. For example, "-s 0:2,npk,512/256", there
are 256 Masters from #768 (256+512, #256 is the starting Master for
software tracing) passed through to the UOS.
CSR STMR
SOS: +--------------+ +----------------------------------+
| physical CSR | | Reserved for SOS | |
+--------------+ +----------------------------------+
UOS: +--------------+ +---------------+
| sw CSR by dm | | mapped to UOS |
+--------------+ +---------------+
Here is an overall flow about how it works.
1. System boots up, and the npk driver on SOS is loaded.
2. The dm is launched with parameters to enable npk virtualization.
3. The dm/npk sets up a bar for CSR, and some values are initialized
based on the parameters, for example, the total number of Masters for
the UOS.
4. The dm/npk sets up a bar for STMR, and maps part of the physical
STMR to it with an offset, according to the parameters.
5. The UOS boots up, and the native npk driver on the UOS is loaded.
6. Enable the traces from UOS, and the traces are written directly to
STMR, but not output by npk for now.
7. Enable the npk output on SOS, and now the traces are output by npk
to the selected target.
8. If the memory is the selected target, the traces can be retrieved
from memory on SOS, after stopping the traces.
Signed-off-by: Zhi Jin <zhi.jin@intel.com>
Reviewed-by: Zhang Di <di.zhang@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
In virtio_heci struct there have deiniting/pending_reset/resetting
variables. All these variables represent the status of virtio heci devices.
Change them into one enum type variable for vheci status.
Signed-off-by: Long Liu <long.liu@intel.com>
Reviewed-by: Yu Wang <yu1.wang@intel.com>
Reviewed-by: Shuo Liu <shuo.a.liu@intel.com>
Acked-by: Eddie Dong <eddie.dong@intel.com>
NHLT table is used by some audio driver for topology data, but current
default audio driver doens't depend on it. Disable it by default due
to possible boot failure with different SOS firmware.
Signed-off-by: Edwin Zhai <edwin.zhai@intel.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
This patch implements scalable xHCI parameter for extended capabilities.
For future supported platform, user can be easy to specify their
platform to emulate corresponding xHCI capabilities.
The new usage:
-s <n>,xhci,[bus1-port1,bus2-port2]:[tablet]:[log=x]:[cap=x]
The old usage:
-s <n>,xhci,[bus1-port1,bus2-port2]:[tablet]:[log=x]
Change-Id: Ie8ba056d57cac9446bcf3f39b342c7ac22245c61
Signed-off-by: Liang Yang <liang3.yang@intel.com>
Reviewed-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Yu Wang <yu1.wang@intel.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
This patch implements Intel ApolloLake xHCI extended capabilities. It
includes two dual role switch registers for switching shared
USB2&USB3 phys between xHCI and xDCI.
Change-Id: I2533537d8a4224da3cf9b2e7475aab9f65347a4a
Signed-off-by: Liang Yang <liang3.yang@intel.com>
Reviewed-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Yu Wang <yu1.wang@intel.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Some xHCI extended capabilities are writable. This patch adds the
writing function for excap.
Change-Id: Ie8b144b47ffa261f97d0461bf97b0c4d312a9333
Signed-off-by: Liang Yang <liang3.yang@intel.com>
Reviewed-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Yu Wang <yu1.wang@intel.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Every platform should have their own xHCI specific extended
capabilities, but the current xHCI DM is not scalable for them. This
patch refines related logic to make it scalable.
Current code only support 4 registers(4*32) as basic extended
capabilites. Base on this new implementation, the mmio range from
excapoff to regsend will cover real excap size according to the cap
parameter.
Change-Id: Ic55a4494e090ec255939cdb8f32950e3c8a66082
Signed-off-by: Liang Yang <liang3.yang@intel.com>
Reviewed-by: Xiaoguang Wu <xiaoguang.wu@intel.com>
Reviewed-by: Yu Wang <yu1.wang@intel.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
The VIRTIO_PCI_CAP_PCI_CFG capability creates an alternative access
method to the common configuration, notification, ISR and device-
specific configuration regions.
To access a device region, the driver writes into the capability
structure (ie. within the PCI configuration space) as follows:
- The driver sets the BAR to access by writing to cap.bar
- The driver sets the size of the access by writing 1, 2 or 4 to
cap.length
- The driver sets the offset within the BAR by writing to cap.offset
At that point, pci_cfg_data will provide a window of size cap.length
into the given cap.bar at offset cap.offset.
Signed-off-by: Jian Jun Chen <jian.jun.chen@intel.com>
Reviewed-by: Hao Li <hao.l.li@intel.com>
Reviewed-by: Zhao Yakui <yakui.zhao@intel.com>
Acked-by: Kevin Tian <kevin.tian@intel.com>
Virtio 1.0 introduced several PCIY_VENDOR capabilities. When trying to
write to these capabilities no action is taken so the registers in
the capability of VIRTIO_PCI_CAP_PCI_CFG such as bar, offset and length
remain the default value 0. Later a read or write of pci_cfg_data needs
these information to perform the indirect read or write to the bar
region.
Signed-off-by: Jian Jun Chen <jian.jun.chen@intel.com>
Reviewed-by: Hao Li <hao.l.li@intel.com>
Reviewed-by: Zhao Yakui <yakui.zhao@intel.com>
Acked-by: Kevin Tian <kevin.tian@intel.com>
pci_emul_find_capability can be used to get the offset of a PCI
capability in PCI configuration space.
Signed-off-by: Jian Jun Chen <jian.jun.chen@intel.com>
Reviewed-by: Hao Li <hao.l.li@intel.com>
Reviewed-by: Zhao Yakui <yakui.zhao@intel.com>
Acked-by: Kevin Tian <kevin.tian@intel.com>
Move rpmb_sim.c and rpmb_backend.c to hw/platform/rpmb/
Signed-off-by: Huang Yang <yang.huang@intel.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
A simulated rpmbfile requires to enable 4MB access by writing
the last byte.
Otherwise, the read operation should be failed if no write
was operated on the address greater than the read address.
Writing the last byte during file creating ensures the whole
4MB address is readable.
Signed-off-by: Huang Yang <yang.huang@intel.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Xiaoguang Wu