Signed-off-by: Akihiko Odaki <akihiko.odaki@gmail.com>
---
 ui/cocoa.m | 49 +++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 49 insertions(+)

diff --git a/ui/cocoa.m b/ui/cocoa.m
index 995301502be..8b83f91723a 100644
--- a/ui/cocoa.m
+++ b/ui/cocoa.m
@@ -540,6 +540,43 @@ - (void) setContentDimensions
     }
 }
 
+- (void) updateUIInfo
+{
+    NSSize frameSize;
+    QemuUIInfo info;
+
+    if (!qemu_console_is_graphic(dcl.con)) {
+        return;
+    }
+
+    if ([self window]) {
+        NSDictionary *description = [[[self window] screen] deviceDescription];
+        CGDirectDisplayID display = [[description objectForKey:@"NSScreenNumber"] unsignedIntValue];
+        NSSize screenSize = [[[self window] screen] frame].size;
+        CGSize screenPhysicalSize = CGDisplayScreenSize(display);
+
+        frameSize = isFullscreen ? screenSize : [self frame].size;
+        info.width_mm = frameSize.width / screenSize.width * screenPhysicalSize.width;
+        info.height_mm = frameSize.height / screenSize.height * screenPhysicalSize.height;
+    } else {
+        frameSize = [self frame].size;
+        info.width_mm = 0;
+        info.height_mm = 0;
+    }
+
+    info.xoff = 0;
+    info.yoff = 0;
+    info.width = frameSize.width;
+    info.height = frameSize.height;
+
+    dpy_set_ui_info(dcl.con, &info);
+}
+
+- (void)viewDidMoveToWindow
+{
+    [self updateUIInfo];
+}
+
 - (void) switchSurface:(pixman_image_t *)image
 {
     COCOA_DEBUG("QemuCocoaView: switchSurface\n");
@@ -1258,6 +1295,16 @@ - (NSApplicationTerminateReply)applicationShouldTerminate:
     return [self verifyQuit];
 }
 
+- (void)windowDidChangeScreen:(NSNotification *)notification
+{
+    [cocoaView updateUIInfo];
+}
+
+- (void)windowDidResize:(NSNotification *)notification
+{
+    [cocoaView updateUIInfo];
+}
+
 /* Called when the user clicks on a window's close button */
 - (BOOL)windowShouldClose:(id)sender
 {
@@ -1936,6 +1983,8 @@ static void cocoa_switch(DisplayChangeListener *dcl,
 
     COCOA_DEBUG("qemu_cocoa: cocoa_switch\n");
 
+    [cocoaView updateUIInfo];
+
     // The DisplaySurface will be freed as soon as this callback returns.
     // We take a reference to the underlying pixman image here so it does
     // not disappear from under our feet; the switchSurface method will
-- 
2.30.1 (Apple Git-130)

On Wed, Jun 16, 2021 at 11:19:10PM +0900, Akihiko Odaki wrote:
> Signed-off-by: Akihiko Odaki <akihiko.odaki@gmail.com>
> ---
>  ui/cocoa.m | 49 +++++++++++++++++++++++++++++++++++++++++++++++++
>  1 file changed, 49 insertions(+)
> 
> diff --git a/ui/cocoa.m b/ui/cocoa.m
> index 995301502be..8b83f91723a 100644
> --- a/ui/cocoa.m
> +++ b/ui/cocoa.m
> @@ -540,6 +540,43 @@ - (void) setContentDimensions
>      }
>  }

Added to UI queue.

thanks,
  Gerd

On Wed, 16 Jun 2021 at 15:19, Akihiko Odaki <akihiko.odaki@gmail.com> wrote:
>
> Signed-off-by: Akihiko Odaki <akihiko.odaki@gmail.com>
> ---
>  ui/cocoa.m | 49 +++++++++++++++++++++++++++++++++++++++++++++++++
>  1 file changed, 49 insertions(+)

Hi; I was looking at the cocoa.m code to see about maybe deleting the
unnecessary autorelease pools, and I ran into some code I was a bit
unsure about that was added in this patch.

In particular I'm wondering about the interactions across threads here.

> +- (void) updateUIInfo
> +{
> +    NSSize frameSize;
> +    QemuUIInfo info;
> +
> +    if (!qemu_console_is_graphic(dcl.con)) {
> +        return;
> +    }
> +
> +    if ([self window]) {
> +        NSDictionary *description = [[[self window] screen] deviceDescription];
> +        CGDirectDisplayID display = [[description objectForKey:@"NSScreenNumber"] unsignedIntValue];
> +        NSSize screenSize = [[[self window] screen] frame].size;
> +        CGSize screenPhysicalSize = CGDisplayScreenSize(display);
> +
> +        frameSize = isFullscreen ? screenSize : [self frame].size;
> +        info.width_mm = frameSize.width / screenSize.width * screenPhysicalSize.width;
> +        info.height_mm = frameSize.height / screenSize.height * screenPhysicalSize.height;
> +    } else {
> +        frameSize = [self frame].size;
> +        info.width_mm = 0;
> +        info.height_mm = 0;
> +    }
> +
> +    info.xoff = 0;
> +    info.yoff = 0;
> +    info.width = frameSize.width;
> +    info.height = frameSize.height;
> +
> +    dpy_set_ui_info(dcl.con, &info);

This function makes some cocoa calls, and it also calls a QEMU
UI layer function, dpy_set_ui_info().

> +- (void)windowDidChangeScreen:(NSNotification *)notification
> +{
> +    [cocoaView updateUIInfo];

We call it from the cocoa UI thread in callbacks like this.

>  /* Called when the user clicks on a window's close button */
>  - (BOOL)windowShouldClose:(id)sender
>  {
> @@ -1936,6 +1983,8 @@ static void cocoa_switch(DisplayChangeListener *dcl,
>
>      COCOA_DEBUG("qemu_cocoa: cocoa_switch\n");
>
> +    [cocoaView updateUIInfo];

We also call it from the QEMU thread, when the QEMU thread calls
this cocoa_switch callback function.

(1) A question for Akihiko:
Are all the cocoa calls we make in updateUIInfo safe to
make from a non-UI thread? Would it be preferable for this
call in cocoa_switch() to be moved inside the dispatch_async block?
(Moving it would mean that I wouldn't have to think about whether
any of the code in it needs to have an autorelease pool :-))

(2) A question for Gerd:
Is it safe to call dpy_set_ui_info() from a non-QEMU-main-thread?
It doesn't appear to do any locking that would protect against
multiple threads calling it simultaneously.

thanks
-- PMM

On Sat, Feb 5, 2022 at 1:19 AM Peter Maydell <peter.maydell@linaro.org> wrote:
>
> On Wed, 16 Jun 2021 at 15:19, Akihiko Odaki <akihiko.odaki@gmail.com> wrote:
> >
> > Signed-off-by: Akihiko Odaki <akihiko.odaki@gmail.com>
> > ---
> >  ui/cocoa.m | 49 +++++++++++++++++++++++++++++++++++++++++++++++++
> >  1 file changed, 49 insertions(+)
>
> Hi; I was looking at the cocoa.m code to see about maybe deleting the
> unnecessary autorelease pools, and I ran into some code I was a bit
> unsure about that was added in this patch.
>
> In particular I'm wondering about the interactions across threads here.
>
> > +- (void) updateUIInfo
> > +{
> > +    NSSize frameSize;
> > +    QemuUIInfo info;
> > +
> > +    if (!qemu_console_is_graphic(dcl.con)) {
> > +        return;
> > +    }
> > +
> > +    if ([self window]) {
> > +        NSDictionary *description = [[[self window] screen] deviceDescription];
> > +        CGDirectDisplayID display = [[description objectForKey:@"NSScreenNumber"] unsignedIntValue];
> > +        NSSize screenSize = [[[self window] screen] frame].size;
> > +        CGSize screenPhysicalSize = CGDisplayScreenSize(display);
> > +
> > +        frameSize = isFullscreen ? screenSize : [self frame].size;
> > +        info.width_mm = frameSize.width / screenSize.width * screenPhysicalSize.width;
> > +        info.height_mm = frameSize.height / screenSize.height * screenPhysicalSize.height;
> > +    } else {
> > +        frameSize = [self frame].size;
> > +        info.width_mm = 0;
> > +        info.height_mm = 0;
> > +    }
> > +
> > +    info.xoff = 0;
> > +    info.yoff = 0;
> > +    info.width = frameSize.width;
> > +    info.height = frameSize.height;
> > +
> > +    dpy_set_ui_info(dcl.con, &info);
>
> This function makes some cocoa calls, and it also calls a QEMU
> UI layer function, dpy_set_ui_info().
>
> > +- (void)windowDidChangeScreen:(NSNotification *)notification
> > +{
> > +    [cocoaView updateUIInfo];
>
> We call it from the cocoa UI thread in callbacks like this.
>
> >  /* Called when the user clicks on a window's close button */
> >  - (BOOL)windowShouldClose:(id)sender
> >  {
> > @@ -1936,6 +1983,8 @@ static void cocoa_switch(DisplayChangeListener *dcl,
> >
> >      COCOA_DEBUG("qemu_cocoa: cocoa_switch\n");
> >
> > +    [cocoaView updateUIInfo];
>
> We also call it from the QEMU thread, when the QEMU thread calls
> this cocoa_switch callback function.
>
> (1) A question for Akihiko:
> Are all the cocoa calls we make in updateUIInfo safe to
> make from a non-UI thread? Would it be preferable for this
> call in cocoa_switch() to be moved inside the dispatch_async block?
> (Moving it would mean that I wouldn't have to think about whether
> any of the code in it needs to have an autorelease pool :-))

It is not safe. Apparently I totally forgot about threads when I wrote this.

It should be moved in the dispatch_async block as you suggest. Should
I write a patch, or will you write one before you delete autorelease
pools?

Regards,
Akihiko Odaki

>
> (2) A question for Gerd:
> Is it safe to call dpy_set_ui_info() from a non-QEMU-main-thread?
> It doesn't appear to do any locking that would protect against
> multiple threads calling it simultaneously.
>
> thanks
> -- PMM

On Sat, 5 Feb 2022 at 02:06, Akihiko Odaki <akihiko.odaki@gmail.com> wrote:
>
> On Sat, Feb 5, 2022 at 1:19 AM Peter Maydell <peter.maydell@linaro.org> wrote

> > (1) A question for Akihiko:
> > Are all the cocoa calls we make in updateUIInfo safe to
> > make from a non-UI thread? Would it be preferable for this
> > call in cocoa_switch() to be moved inside the dispatch_async block?
> > (Moving it would mean that I wouldn't have to think about whether
> > any of the code in it needs to have an autorelease pool :-))
>
> It is not safe. Apparently I totally forgot about threads when I wrote this.
>
> It should be moved in the dispatch_async block as you suggest. Should
> I write a patch, or will you write one before you delete autorelease
> pools?

I'll write a patchset. If you have time to test it when I send it out
that would be great.

Incidentally, I think the answer to my other question
> > Is it safe to call dpy_set_ui_info() from a non-QEMU-main-thread?

is "no, and so the body of updateUIInfo should be enclosed in a
with_iothread_lock block".

-- PMM

On Wed, Feb 9, 2022 at 3:07 AM Peter Maydell <peter.maydell@linaro.org> wrote:
>
> On Sat, 5 Feb 2022 at 02:06, Akihiko Odaki <akihiko.odaki@gmail.com> wrote:
> >
> > On Sat, Feb 5, 2022 at 1:19 AM Peter Maydell <peter.maydell@linaro.org> wrote
>
> > > (1) A question for Akihiko:
> > > Are all the cocoa calls we make in updateUIInfo safe to
> > > make from a non-UI thread? Would it be preferable for this
> > > call in cocoa_switch() to be moved inside the dispatch_async block?
> > > (Moving it would mean that I wouldn't have to think about whether
> > > any of the code in it needs to have an autorelease pool :-))
> >
> > It is not safe. Apparently I totally forgot about threads when I wrote this.
> >
> > It should be moved in the dispatch_async block as you suggest. Should
> > I write a patch, or will you write one before you delete autorelease
> > pools?
>
> I'll write a patchset. If you have time to test it when I send it out
> that would be great.

Thanks, I will test the patchset soon after I receive it.

>
> Incidentally, I think the answer to my other question
> > > Is it safe to call dpy_set_ui_info() from a non-QEMU-main-thread?
>
> is "no, and so the body of updateUIInfo should be enclosed in a
> with_iothread_lock block".
>
> -- PMM

> (2) A question for Gerd:
> Is it safe to call dpy_set_ui_info() from a non-QEMU-main-thread?

No.

take care,
  Gerd

On Mon, 14 Feb 2022 at 10:27, Gerd Hoffmann <kraxel@redhat.com> wrote:
>
> > (2) A question for Gerd:
> > Is it safe to call dpy_set_ui_info() from a non-QEMU-main-thread?
>
> No.

Is it OK to do so if the other thread is holding the iothread lock?
(This is how we do a lot of the other "need to call a QEMU function"
work from the cocoa UI thread.)

thanks
-- PMM

On Mon, Feb 14, 2022 at 10:43:32AM +0000, Peter Maydell wrote:
> On Mon, 14 Feb 2022 at 10:27, Gerd Hoffmann <kraxel@redhat.com> wrote:
> >
> > > (2) A question for Gerd:
> > > Is it safe to call dpy_set_ui_info() from a non-QEMU-main-thread?
> >
> > No.
> 
> Is it OK to do so if the other thread is holding the iothread lock?
> (This is how we do a lot of the other "need to call a QEMU function"
> work from the cocoa UI thread.)

That should work.

take care,
  Gerd