In addition to the above I uninstalled chrome, then re-installed and immediately tried full screen -- before signing in, loading anything, and even before it started the background process thing.  Still drawing a blank screen.

And another note: it's just the drawing (of everything except for the F11  message at the top) that is broken -- eg, with a youtube video clicking the place where the full screen on/off button is supposed to be does get it out of the (broken) full screen mode.

bug opener means Chrome F11 full screen mode, but not video full screen. This issue repro even at blank chrome tab as per bug opener. this is not a media bug.
eli@barzilay.org, please try to uninstall, reinstall Chrome to try again.

It's all kinds of full screens that results in that: F11 on any page, youtube fullscreen (ie, hitting "f"), and other video sources.

I have tried uninstalling and re-installing -- multiple times, and each time I chose the option of not keeping settings/browsing history/whatever.

Some more information points:

1. I have four monitors in a layout of left/high-middle/right and fourth one under the high-middle one.  Opening a new window in chrome sometime results in a window that stretches across all *three* monitors except for the low-middle one, and that window is black and doesn't respond to the usual keypresses (ie, Ctrl+W doesn't close it).  When this happens I close it using the Windows close button or Alt+F4, and open a new one.  It usually gets resolved after a few attempts.

2. Another relevant bit about this setup: AFAICT, the three top monitors are running on the nvidia quadro, and the low-middle one is the laptop's builtin screen, which is some Intel integrated thing.

3. Turning hardware acceleration off (in settings -> advanced) makes the full screen problem go away, but not the blank super-wide problem I described in #1.

Does the bug happen if you are using only one screen?

As I said: it happens on certain screen configurations, and there's one
screen that even if it's the only one connected, the bug happens.  (The
blank-on-fullscreen one, not the super-wide window, of course.)

I'm also using an Nvidia card with Optimus (and I suspect that the bug is related to Optimus). According to chrome://gpu, the Nvidia driver version is 10.18.13.6839, and the Intel driver version is 20.19.15.4390.

I've gathered some more information about when this happens for me:

1. The bug only happens when chrome://gpu says the Driver vendor is NVIDIA.
If I start Chrome when the internal laptop screen is the "Main display" in Windows's settings, then I get the Intel driver until I restart Chrome (even if I change the display settings), and the bug doesn't happen.
If I start Chrome when the external screen is the "Main display", I get the NVIDIA driver and the bug can happen.

Going into Nvidia's control centre and changing the GPU for chrome.exe doesn't seem to affect anything - I still get the same GPUs in the same situations according to chrome://gpu.

2. The bug only happens when the external monitor is the "Main display".
Even if I start Chrome with the NVIDIA driver and then change the display settings, I only get the black screen when the external display is either the main display in a multi-screen "extend" setup, or the only active monitor.

3. The bug only happens when going into full screen mode on the external display.
The internal laptop screen is never affected, even if the external monitor is the main display in an "extend" setup.

The two workarounds seem to be disabling "Use hardware acceleration when available" and starting Chrome with the main display as the internal laptop screen.

I have the same exact issue. See http://superuser.com/questions/1098204/chrome-html5-videos-in-fullscreen-are-black-when-intel-graphics-are-enabled-or-h 

I have the exact same issue on brand new fully updated Windows 10 Thinkpad with an external Dell monitor. In both Chrome and Opera full-screen videos of any kind went black (with sound on), so I had to disable hardware acceleration in both; it worked. No issue with FF or Edge, not sure why: because they had acceleration turned off by default or because they don't have a bug.

It is worrysome that this issue happens in Canary: Canary also only works properly when HW acceleration is off.

please note that my laptop has NVIDIA QUADRA M1000M discrete graphics along with Intel HD 530 built in graphics.

PS: I am also having  issue 627714 .

Issue does not occur when the internal GPU (in my case Intel HD Graphics 530) is disabled in the BIOS (display HW is set to discrete only, instead of hybrid).

This suggests that the issue is related to the Optimus software OR the Intel driver. Unfortunately one cannot test the issue using only the Intel GPU, as you cannot connect an external display when only the Intel GPU is on (the HDMI/miniDisplay ports are hardwired to the discrete GPU on my laptop)

Still having this issue in Version 53.0.2785.34 beta-m (64-bit), but it no longer happens in Canary.

GPU output (disappears in 7 days):
https://securesha.re?#u=8699c139678c75345b7848450db8f631.bin
pword: 30KJb8XOv5YEv

Issue fixed in 53.0.2785.70 beta-m (64-bit)

Also fixed in 52.0.2743.116 m (64-bit)

I am running 52.0.2743.116 m (64-bit) on a Lenovo W540 which has a Quadro K1100M.  The issue remains: On external displays, if I press F11 on any page, it is black.  F11 does work as expected when Chrome is on the built-in display.  I have the newest nVidia and Intel drivers.

Yes, the issue persists for me: both the black full-screen and the occasional insistence of chrome to open a useless blackened window that spans all of my monitors.

The issue still persists for me on 52.0.2743.116m (64-bit).

I am using a MSI GT70 laptop connected to a BenQ XL2430T monitor via a MiniDP to DisplayPort cable.

Temp fixes involve disabling hardware acceleration in Chrome, switching the main display back to the laptop display, or using Firefox. But all of them are a hassle especially considering I was happily fullscreening until late June when an update caused this issue.

I have this issue as well. MSI GS70 Laptop. GPU: GTX765M CPU: Intel 4700HQ.
Very annoying issue.

F11 and youtube full screen both black when Hrdwr accl on.

Dragging chrome window between external and laptop monitor occasionally yields black chrome window as well.

Just commenting to say the issue is still here with the new Chrome version, 53.0.2785.89m.

still happens.
Version 53.0.2785.116 m (64-bit)

jbauman@ could you take a look?

Could people seeing this attach the contents of about:gpu to this bug?

Also, it would be great if you someone could try to bisect and see exactly when this started happening. You can follow the instructions from https://www.chromium.org/developers/bisect-builds-py - I think a command-line that works is

python tools/bisect-builds.py -a win -g 369907 -b 403382 --verify-range

I've attached the contents of about:gpu. I'll see if I can do some bisecting...

Deleted: gpu.html 152 KB

gpu.html 152 KB View Download

I've bisected it, and found that the problem actually changed slightly after it first occurred.

When it first occurred, going into full screen mode would still take over interaction with the whole screen (i.e. I could see the cursor change over text and links etc. and not interact with other windows), but it would only produce a black area in the top-left corner of the screen that was the size of the chrome window - you could still see the remains of other windows from before entering full screen mode.
When it changed, the black area became the size of the whole screen.

Bisecting for the first problem gives this:
You are probably looking for a change made after 370302 (known good), but no later than 370327 (first known bad).
CHANGELOG URL:
  https://chromium.googlesource.com/chromium/src/+log/72c13ed8293f8bd61ecb91a4ccc23a975da30fb2..3e32a8e504116ed017db403ee899449eb7fbec32

Bisecting for the second problem (where the black area changed from just being in the top left corner to covering the whole screen) gives this:
You are probably looking for a change made after 374846 (known good), but no later than 374851 (first known bad).
CHANGELOG URL:
  https://chromium.googlesource.com/chromium/src/+log/63a79760afb4a25023368311e6906e7462b87c28..c8107b9d3cac87c58560596d66d7de8627d486b4

Thanks, that's interesting. If you want a workaround for the problem you could try launching chrome with the --disable-direct-composition command-line flag. I'll see if I can find a way to try and repro this issue.

Thanks! I can confirm that --disable-direct-composition works as a workaround.

Added the --disable-direct-composition quick fix to super user,
Thanks very much!

http://superuser.com/questions/1098204/chrome-html5-videos-in-fullscreen-are-black-when-intel-graphics-are-enabled-or-h/1119807#1119807

confirming --disable-direct-composition works for me as well.

I can also confirm that --disable-direct-composition works!

Just wanted to say that the --disable-direct-composition flag worked for me as well. Not honestly sure what we are giving up by disabling this feature, but hopefully it is fixed soon!

I had this issue as well and after finding this discussion I tried the "--disable-direct-composition" and now my full screen works (MSI GS70 & Samsung external monitor (HDMI) )

Finally found this through superuser.com and used the "--disable-direct-composition" workaround.
(980M and external monitor configured as "main")

That was a long one!
(=PA=)

Any word on when this might be fixed?

Worked for me as well the workaround "--disable-direct-composition" 
(980M and external monitor configured via HDMI)

Chrome 54.0.2840.87 m (64-bit) in Windows 10. Alienware 17 R3

I've been having this issue for a while. Worth noting it doesn't only affect chrome, but all browsers.
I tend to use my laptop at my desk with the lid closed and external monitor only.

--disable-direct-composition workaround works for me.

Monitor in use: Dell UH2175H (as main display) - via displayport.

Deleted: gpu.html 153 KB

gpu.html 153 KB View Download

affects me as well, Razer Blade Stealth (7gen-i7) hooked up to external GPU enclosure via Thunderbolt, going out to an external monitor via DP.  When not connected to external GPU everything works fine.

Deleted: gpu.html 152 KB

gpu.html 152 KB View Download

Unfortunately, the parameter "--disable-direct-composition" doesn't help me.

I'm on Chrome 55 with Windows 10 with my Yoga 260 (Intel GPU only) using the Thinkplus USB3-Dock (I have two screens, each connected using Displayport to the USB3-dock). As soon as I start playback of a video (e.g. Amazon Prime Video), the external displays go black.

Deleted: gpu.html 206 KB

gpu.html 206 KB View Download

Can confirm this issue on my New Razer Blade when I'm hooked up to an external monitor.  Running chrome with the "--disable-direct-composition" flag "fixes" the issue.

Would just like to say that I'm having this issue again on a new MSI GS63VR (upgraded from a MSI GT60 in my previous post). Unfortunately, the "--disable-direct-composition" flag no longer solves the issue for me; does anybody know how to check that the command line flag is actually being used by Chrome?

Alternative (but annoying) way to solve the issue is to disable the Intel graphics via Device Manager, open Chrome, and then re-enable the Intel graphics. This will permit fullscreen operations so long as Chrome isn't closed again.

problem persists in 56.0.2924.87 (64-bit)

Any word from the owner (jbau...@chromium.org) on when this is going to be fixed? It's been about seven months since this issue arose.

This affects anyone who uses two monitors and a mix of integrated/dedicated GPU. Not necessarily a niche group....

Sorry, I'd recommend checking with NVIDIA (or Microsoft). I suspect there's a bug in their dual-GPU implementation.

Yeah, it looks to be a bug in DirectComposition when Chrome creates a swapchain that's the same size as the main display, if that main display is on the NVIDIA card.

Thank you for the response! I suspected as much considering how many multi-monitor bugs we've seen this year

jbau, why does it not happen with any other application or browser?

My guess is that no other application that's been tried creates a BGRA DirectComposition swapchain that's fullscreen. Maybe they use other mechanisms to draw (e.g. edge uses an IDCompositionSurface or a YUV swapchain, while firefox doesn't use DirectComposition).

I'm working on creating a little demo application to narrow down when this happens.

has anyone at nvidia been contacted? you guys with @chromium addresses have a better chance at getting their attention.

thank you for your help!

I can confirm this problem still exists with the latest Intel and Nvidia drivers and Chrome 56.0.2924.87 (64-bit). Full screen only works on laptop screen. The problem exists also in Edge. The workaround works for me.

Ok, I've attached a test case I made.

The window should display red. If the system has an NVIDIA and an Intel GPU and a monitor connected to the NVIDIA GPU is set as the main display then it'll display white. This is because it has a DirectComposition swapchain the same size as the main monitor, and if that monitor is on the NVIDIA GPU then the swapchain is created in a special way and dwm can't open its share handle.

Deleted: DCompTest.zip 1.2 MB

DCompTest.zip 1.2 MB Download

I've just run the test case on my machine, it works as you described:
It displays red when the primary monitor is the laptop screen (Intel GPU), and white when the primary monitor is the external screen (NVIDIA GPU), regardless of whether or not any other screens are turned on. When the screens are duplicated, it displays red.

problem persists in 57.0.2987.98 (64-bit)

I am no longer experiencing this issue in Google Chrome 57.0.2987.110 (64-bit), Nvidia driver 378.92, Intel Graphics Driver 21.20.16.4590. Unsure which update fixed the issue but based on the comments, suspect it was probably one of the graphics drivers.

I am no longer experiencing this issue in Google Chrome 57.0.2987.110 (64-bit) either!

Im still experiencing this issue in Google Chrome Version 60.0.3112.78 (Official Build) (64-bit), NVidia driver 382.05

Also experiencing this on 60.0.3112.101 with NVidia driver 377.35

reporters: Is this still occurring for anyone?

It used to occur on my computer but has not for quite a while.  My chrome has no arguments passed to the .exe and hardware acceleration is enabled.  Previous workarounds involved passing an argument and disabling hardware acceleration.

The NextAction date has arrived: 2018-03-23