In the VS2019 project template for C++/CX D3D12, the DisplayInformation::DisplayContentsInvalidated event (documented here) is subscribed and when this event fires, the code in the project template attempts to validate the D3D12 device. The criteria being used to validate is whether the default adapter has changed:
// This method is called in the event handler for the DisplayContentsInvalidated event.
void DX::DeviceResources::ValidateDevice()
{
// The D3D Device is no longer valid if the default adapter changed since the device
// was created or if the device has been removed.
// First, get the LUID for the default adapter from when the device was created.
DXGI_ADAPTER_DESC previousDesc;
{
ComPtr<IDXGIAdapter1> previousDefaultAdapter;
DX::ThrowIfFailed(m_dxgiFactory->EnumAdapters1(0, &previousDefaultAdapter));
DX::ThrowIfFailed(previousDefaultAdapter->GetDesc(&previousDesc));
}
// Next, get the information for the current default adapter.
DXGI_ADAPTER_DESC currentDesc;
{
ComPtr<IDXGIFactory4> currentDxgiFactory;
DX::ThrowIfFailed(CreateDXGIFactory1(IID_PPV_ARGS(¤tDxgiFactory)));
ComPtr<IDXGIAdapter1> currentDefaultAdapter;
DX::ThrowIfFailed(currentDxgiFactory->EnumAdapters1(0, ¤tDefaultAdapter));
DX::ThrowIfFailed(currentDefaultAdapter->GetDesc(¤tDesc));
}
// If the adapter LUIDs don't match, or if the device reports that it has been removed,
// a new D3D device must be created.
if (previousDesc.AdapterLuid.LowPart != currentDesc.AdapterLuid.LowPart ||
previousDesc.AdapterLuid.HighPart != currentDesc.AdapterLuid.HighPart ||
FAILED(m_d3dDevice->GetDeviceRemovedReason()))
{
m_deviceRemoved = true;
}
}
However, it seems to me that the default adapter may not be the elected adapter, as you can see in this code, because it ensures the adapter can create a D3D12 device before electing it, and this same check does not exist in the validation code above:
// This method acquires the first available hardware adapter that supports Direct3D 12.
// If no such adapter can be found, *ppAdapter will be set to nullptr.
void DX::DeviceResources::GetHardwareAdapter(IDXGIAdapter1** ppAdapter)
{
ComPtr<IDXGIAdapter1> adapter;
*ppAdapter = nullptr;
for (UINT adapterIndex = 0; DXGI_ERROR_NOT_FOUND != m_dxgiFactory->EnumAdapters1(adapterIndex, &adapter); adapterIndex++)
{
DXGI_ADAPTER_DESC1 desc;
adapter->GetDesc1(&desc);
if (desc.Flags & DXGI_ADAPTER_FLAG_SOFTWARE)
{
// Don't select the Basic Render Driver adapter.
continue;
}
// Check to see if the adapter supports Direct3D 12, but don't create the
// actual device yet.
if (SUCCEEDED(D3D12CreateDevice(adapter.Get(), D3D_FEATURE_LEVEL_11_0, _uuidof(ID3D12Device), nullptr)))
{
break;
}
}
*ppAdapter = adapter.Detach();
}
Is this unintentionally relying on the default adapter being the elected adapter, or are there wider assumptions we can make based on the default adapter having changed?
What is the real meaning if DisplayInformation::DisplayContentsInvalidated? What is this telling me? Why does this suggest the device is possibly invalid? What are the cases where this event fires andthe device would or would not be invalid?
And since we are also checking for a DeviceRemovedReason here, are there cases where the default adapter doesn't change but the device was removed because of DisplayInformation::DisplayContentsInvalidated?
Do we even really need to handle this at all? What would happen if I just completely ignored this event and kept trying to render? Would I not end up handling whatever problem this event firing is indicating in the documented "device lost scenarios" per this document?
We are still investigating your question and get some information.
With regards to when the DisplayContentInvalidated Event will be triggered, it is typically invoked upon various activities including changing of the primary display or if the display configurations have changed in any other way. This would also include if there were certain types of issues occurring with the graphics, which this event should be able to be utilized to handle such situations.
With regards to the template, we ran some tests on our end and we indeed did see that the project uses the first adapter that it finds that can handle DirectX12 rather than an elected adapter. Upon further tests, we removed that particular check and found that the application does indeed continue to run as expected. The adapter may not be the one that was used to create the device at application startup. So indeed, this check does seem like it can potentially be ignored as indicated. However, having this event in the code might not hurt as it can provide the extra functionality should it be required for any situation.
Any other detailed information would likely be found in our documentation regarding this. However, it seems that the template is meant for demonstrating certain functionalities of DirextX12 and its various API’s so ignoring the DisplayContentInvalidated Event should typically not cause any issues.
Related
i am overriding the dnsEnd() function in EventListener:
#Override
public void dnsEnd(Call call, String domainName, List<InetAddress> inetAddressList) {
inetAddressList.forEach(address -> {
logger.debug("checking if url ({}) is in allowlist", address.toString());
if (!allowlist.contains(address)) {
call.cancel();
}
});
}
i know, in the documentation it says not to alter call parameters etc:
"All event methods must execute fast, without external locking, cannot throw exceptions, attempt to mutate the event parameters, or be re-entrant back into the client. Any IO - writing to files or network should be done asynchronously."
but, as i don't care about the call if it is trying to get to an address outside the allowlist, i fail to see the issue with this implementation.
I want to know if anyone has experience with this, and why it may be an issue?
I tested this and it seems to work fine.
This is fine and safe. Probably the strangest consequence of this is the canceled event will be triggered by the thread already processing the DNS event.
But cancelling is not the best way to constrain permitted IP addresses to a list. You can instead implement the Dns interface. Your implementation should delegate to Dns.SYSTEM and them filter its results to your allowlist. That way you don't have to worry about races on cancelation.
I am writing an application for my own purposes that aims to get play pause events no matter what is going on in the system. I have gotten this much working
let commandCenter = MPRemoteCommandCenter.shared()
commandCenter.togglePlayPauseCommand.isEnabled = true
commandCenter.togglePlayPauseCommand.addTarget { (MPRemoteCommandEvent) -> MPRemoteCommandHandlerStatus in
print("Play Pause Command")
return .success
}
commandCenter.nextTrackCommand.isEnabled = true
commandCenter.nextTrackCommand.addTarget { (MPRemoteCommandEvent) -> MPRemoteCommandHandlerStatus in
print("NextTrackCommand")
return .success
}
commandCenter.previousTrackCommand.isEnabled = true
commandCenter.previousTrackCommand.addTarget { (MPRemoteCommandEvent) -> MPRemoteCommandHandlerStatus in
print("previousTrackCommand")
return .success
}
commandCenter.playCommand.isEnabled = true
commandCenter.playCommand.addTarget { (MPRemoteCommandEvent) -> MPRemoteCommandHandlerStatus in
print("playCommand")
return .success
}
MPNowPlayingInfoCenter.default().playbackState = .playing
Most of those methods are there because apparently you will not get any notifications without having nextTrackCommand or previousTrackCommand or playCommand implemented.
Anyways my one issue is that as soon as you open another application that uses audio these event handlers stop getting called and I cant find a way to detect and fix this.
I would normally try doing AVAudioSession things to state this as a background application however that does not seem to work. Any ideas on how I can get playpause events no matter what state the system is in?
I would like to be able to always listen for these events OR get an indication of when someone else has taken control of the audio? Perhaps even be able to re-subscribe to these play pause events.
There's an internal queue in the system which contains all the audio event subscribers. Other applications get on top of it when you start using them.
I would like to be able to always listen for these events
There's no API for that but there's a dirty workaround. If I understand your issue correctly, this snippet:
MPNowPlayingInfoCenter.default().playbackState = .paused
MPNowPlayingInfoCenter.default().playbackState = .playing
must do the trick for you if you run it in a loop somewhere in your application.
Note that this is not 100% reliable because:
If an event is generated before two subsequent playbackState state changes right after you've switched to a different application, it would still be catched by the application in the active window;
If another application is doing the same thing, there would be a constant race condition in the queue, with unpredictable outcome.
References:
Documentation for playbackState is here;
See also a similar question;
See also a bug report for mpv with a similar
issue (a pre-MPRemoteCommandCenter one, but still very valuable)
OR get an indication of when someone else has taken control of the audio
As far as I know there's no public API for this in macOS.
I'm building a taskbar for an X11 desktop and so far I've been successful in detecting new and removed windows and changes to window titles and icons.
However, despite setting every event mask I can think of on the client windows, I've been unable to get any events when a test application adds the _NET_WM_STATE_DEMANDS_ATTENTION atom to its _NET_WM_STATE property.
I'm using Qt5 and capturing the incoming X11 events using installNativeEventFilter. However, I've also tried using xprop -spy and I see the same issue there: Even though polling the _NET_WM_STATE property shows the atom being added and removed, no property change event is ever received. Fluxbox also doesn't seem to pick up on it until something else causes it to re-query the window.
My event filter code resembles this:
xcb_generic_event_t* ev = static_cast<xcb_generic_event_t*>(message);
uint32_t type = ev->response_type;
switch (type) {
case XCB_PROPERTY_NOTIFY: {
xcb_property_notify_event_t* pev =
reinterpret_cast<xcb_property_notify_event_t*>(ev);
qDebug() << "property" << pev->window << pev->atom << (int)pev->state;
break;
/* snip */
default:
qDebug() << "unrecognized event" << type;
};
My test application uses QApplication::alert() on a timer to assert the attention flag.
Is there some special handling necessary for atom list properties? Am I doomed to poll for changes? I've tried looking at the source code for other window managers but I haven't been able to identify any specific differences.
It turns out that Qt5's native event filter on X11 doesn't pass everything through consistently. I haven't isolated that bug yet, but I wrote my own minimalist xcb event loop on a separate connection to handle window-management activities and it works fine.
We are using a custom board running version 3.2 of the kernel, but we've come across some issues when testing the robustness of the MMC layer.
First of all, our MMC slot does not have a card detection pin. The issue in question consists of the following:
Load the module (omap_hsmmc). The card is detected on power-up,
and is mounted appropriately.
Read something from the SD card (i.e. cat foo.txt)
While the file is being read, remove the card.
After many failed attempts, the system hangs.
Now, I've tracked the issue to the following code section, in drivers/mmc/card/queue.c:
static int mmc_queue_thread(void *d)
{
struct mmc_queue *mq = d;
struct request_queue *q = mq->queue;
current->flags |= PF_MEMALLOC;
down(&mq->thread_sem);
do {
struct request *req = NULL;
struct mmc_queue_req *tmp;
spin_lock_irq(q->queue_lock);
set_current_state(TASK_INTERRUPTIBLE);
req = blk_fetch_request(q);
mq->mqrq_cur->req = req;
spin_unlock_irq(q->queue_lock);
if (req || mq->mqrq_prev->req) {
set_current_state(TASK_RUNNING);
mq->issue_fn(mq, req);
} else {
if (kthread_should_stop()) {
set_current_state(TASK_RUNNING);
break;
}
up(&mq->thread_sem);
schedule();
down(&mq->thread_sem);
}
/* Current request becomes previous request and vice versa. */
mq->mqrq_prev->brq.mrq.data = NULL;
mq->mqrq_prev->req = NULL;
tmp = mq->mqrq_prev;
mq->mqrq_prev = mq->mqrq_cur;
mq->mqrq_cur = tmp;
} while (1);
up(&mq->thread_sem);
return 0;
}
Investigating this code, I've found that it hangs inside the mq->issue_fn(mq, req) call. This function prepares and issues the appropriate commands to fufill the request passed into it, and it is aware of any errors that happens when it can't communicate with the card. The error is handled in an awkward (in my opinion) manner, and does not 'bubble up' to the mmc_queue_thread. However, I've checked my version's code against that of the most recent kernel release (4.9), and I did not see any difference in the treatment of such errors, apart from a better separation of each error case (the treatment is very similar).
I suspect that the issue is caused by the inability ofthe upper layers to stop making new requests to read from the MMC card.
What I've Tried:
Rewrote the code so that the error is passed up so that I can do ret = mq->issue_fn(mq, req).
Being able to identify the specific errors, I tried to cancel the thread in different ways: calling kthread_stop, calling mmc_queue_suspend, __blk_end_request, and others. With these, the most I've been able to accomplish was keep the thread in a "harmless" state, where it still existed, but did not consume any resources. However, the user space program that triggered the call does not return, locking up in an uninterruptible state.
My questions:
Is the best way to solve this by stopping the upper layers from making new requests? Or should the thread itself be 'killed off'?
In a case such as mine, should it be assumed that because there is no card detection pin, the card should not be removed?
Update: I've found out that you can tell the driver to use polling to detect card insertion/removal. It can be done by adding the MMC_CAP_NEEDS_POLL flag to the mmc's .caps on driver initialization (on newer kernels, you can use the broken-cd property on your DT). However, the problem still happens after this modification.
Figured out the solution!
The issue was, as I suspected, that the block requests kept being issued even after the card was removed. The error was fixed in commit a8ad82cc1b22d04916d9cdb1dc75052e80ac803c from Texas' kernel repository:
commit a8ad82cc1b22d04916d9cdb1dc75052e80ac803c
Author: Sujit Reddy Thumma <sthumma#codeaurora.org>
Date: Thu Dec 8 14:05:50 2011 +0530
mmc: card: Kill block requests if card is removed
Kill block requests when the host realizes that the card is
removed from the slot and is sure that subsequent requests
are bound to fail. Do this silently so that the block
layer doesn't output unnecessary error messages.
Signed-off-by: Sujit Reddy Thumma <sthumma#codeaurora.org>
Acked-by: Adrian Hunter <adrian.hunter#intel.com>
Signed-off-by: Chris Ball <cjb#laptop.org>
The key in the commit is that a return BLKPREP_KILL was added to the mmc_prep_request(), besides some minor tweaks that added some 'granularity' to the error treatment, adding a specific code for the cases where the card was removed.
I'm trying to enhance an existing Firefox extension which relies on nsIContentPolicy to detect and abort certain network loads (in order to block the resulting UI action, i.e. tab navigation). Then handle loading that resource internally. Under rare circumstances, only after handling the load, it turns out we shouldn't have interrupted the load at all, so we flag it to be ignored and re-start it.
Under e10s/multi-process, that means the parent (where the content policy is running) must send a message to the child (handling the UI for the content) to restart the load. Today, that's done by:
function findMessageManager(aContext) {
// With e10s off, context is a <browser> with a direct reference to
// the docshell loaded therein.
var docShell = aContext && aContext.docShell;
if (!docShell) {
// But with e10s on, context is a content window and we have to work hard
// to find the docshell, from which we can find the message manager.
docShell = aContext
.QueryInterface(Ci.nsIInterfaceRequestor)
.getInterface(Ci.nsIWebNavigation)
.QueryInterface(Ci.nsIDocShellTreeItem).rootTreeItem;
}
try {
return docShell
.QueryInterface(Ci.nsIInterfaceRequestor)
.getInterface(Ci.nsIContentFrameMessageManager);
} catch (e) {
return null;
}
};
Which is crazy complex, because e10s is crazy complex. But it works; it generates some object in the parent, upon which I can call .sendAsyncMessage(), and then the addMessageListener() handler in my frame/child script receives it, and does what it needs to do.
I'd like to switch from nsIContentPolicy to http-on-modify-request as it presents more information for making a better determination (block and handle this load?) earlier. Inside that observer I can do:
var browser = httpChannel
.notificationCallbacks.getInterface(Ci.nsILoadContext)
.topFrameElement;
Which gives me an object which has a .messageManager which is some kind of message manager, and which has a .sendAsyncMessage() method. But when I use that .sendAsyncMessage(), the message disappears, never to be observed by the child.
Context: https://github.com/greasemonkey/greasemonkey/issues/2280
This should work in principle, although the docshell tree traversal may do different things in e10s and non-e10s, so you have to be careful there. In e10s rootTreeItem -> nsIContentFrameMessageManager should give you the MM equivalent to a frame script and topFrameElement.frameLoader.messageManager should give you the <browser>'s MM, which pretty much is the parent side counterpart to it.
Potential sources of confusion:
e10s on vs. off
process MM vs. frame MM hierarchy
listening in the wrong frame for the message (registering in all frames might help for debugging purposes)
This is the function I use to find the content message manager:
function contentMMFromContentWindow_Method2(aContentWindow) {
if (!gCFMM) {
gCFMM = aContentWindow.QueryInterface(Ci.nsIInterfaceRequestor)
.getInterface(Ci.nsIDocShell)
.QueryInterface(Ci.nsIInterfaceRequestor)
.getInterface(Ci.nsIContentFrameMessageManager);
}
return gCFMM;
}
So maybe get the content window that triggered that request, and then use this function.