https://bugs.kde.org/show_bug.cgi?id=388854
--- Comment #35 from Tobias Deiminger <haxti...@posteo.de> ---
(In reply to Albert Astals Cid from comment #34)
> That code is only used if you have QT_XCB_NATIVE_PAINTING environment
> variable set that i hope you're not since it's experimental
The variable wasn't set. So QPixmap didn't delegate to QX11PlatformPixmap, but
to QRasterPlatformPixmap.
> I can't find memory chunks large enough to hold contiguous range
> of memory for page bitmaps
Now I know why heap didn't contain expected chunks. The inner objects of
PagePrivate::PixmapObject::m_pixmap were NOT in heap, but on stack.
@Albert: As I understand, the Page QPixmap is meant to live long. Is this
correct? If yes, we have a bug here, because stack memory of course doesn't
live long.
That's what I did, maybe someone can repeat it for confirmation.
void PagePrivate::setPixmap( DocumentObserver *observer, QPixmap *pixmap, const
NormalizedRect &rect, bool isPartialPixmap )
{
if ( m_rotation == Rotation0 ) {
// ...
QMap< DocumentObserver*, PagePrivate::PixmapObject >::iterator it =
m_pixmaps.find( observer );
// ...
it.value().m_pixmap = pixmap; // break here, and observe inner objects
of QPixmap* pixmap
// ...
}
}
(gdb) b page.cpp:557
Breakpoint 1 at 0x7fffdac68109: file /home/deiminge/git/okular/core/page.cpp,
line 557.
(gdb) c
Thread 1 "okular" hit Breakpoint 1, Okular::PagePrivate::setPixmap
(this=0x555555d2a610, observer=0x555555a46ee0, pixmap=0x555555df9d60, rect=...,
isPartialPixmap=0x0) at /home/deiminge/git/okular/core/page.cpp:557
557 it.value().m_pixmap = pixmap;
(gdb) p *pixmap
$1 = {
...
members of QPixmap:
data = {
d = 0x555555d21950
}
}
(gdb) p *(QRasterPlatformPixmap*)0x555555d21950 $2 = {
...
members of QRasterPlatformPixmap:
image = {
members of QImage:
d = 0x7fffc00100d0
}
}
(gdb) p *(QImageData*)0x7fffc00100d0
$9 = {
...
width = 0x4f9,
height = 0x673,
depth = 0x20,
nbytes = 0x80476c,
devicePixelRatio = 1,
...
data = 0x7fffc6093010 // Here's the large raw pixmap
// 0x7fff... is in my stack memory range
...
}
--
You are receiving this mail because:
You are watching all bug changes.
