On some Adreno 505 and 506 devices we are encountering driver crashes during
glLinkProgram(). The only circumstance in which we have been able to reproduce
locally is when the show-overdraw debug option is enabled. The reason appears to
be that, due to shader optimisation, the debug overdraw variants of many shaders
have identical source code. The crash seems to occur when linking a shader which
has identical source code to a previously linked shader.
This does not, however, explain the non-insignificant numbers of crashes in the
wild because a) it's unlikely many users are enabling overdraw debugging, and b)
some crash reports predate the commit which enabled shader
optimisation. However, it is possible that for a different reason we are
compiling multiple shaders with identical source code.
To attempt to work around this crash this change adds a random comment to the
end of each shader source string, on the affected devices.
Differential Revision: https://phabricator.services.mozilla.com/D83571
It's not immediately obvious to me why we limit this check based on target;
I guess once upon a time there was no ELF or Mach-O support, so we wanted to
ensure people didn't shoot themselves in the foot. If that's the case,
testing indicates that Mach-O support isn't quite ready for prime-time and
we haven't got all the bits straightened out for our normal Linux builds.
So we're just going to enable it for Android here.
Differential Revision: https://phabricator.services.mozilla.com/D83560
I also added a jit-test for the IsPackedArray intrinsic. Now that this doesn't
depend on TI anymore, the behavior should be more predictable and this way we
can make sure we don't regress/break the packed-array optimization.
Differential Revision: https://phabricator.services.mozilla.com/D83611
The failure suggests that an additional unexpected device has been found. Added some logs to list the devices when that happens. The logs will be removed once the problem has been found.
Differential Revision: https://phabricator.services.mozilla.com/D83298
It's not immediately obvious to me why we limit this check based on target;
I guess once upon a time there was no ELF or Mach-O support, so we wanted to
ensure people didn't shoot themselves in the foot. If that's the case,
testing indicates that Mach-O support isn't quite ready for prime-time and
we haven't got all the bits straightened out for our normal Linux builds.
So we're just going to enable it for Android here.
Differential Revision: https://phabricator.services.mozilla.com/D83560
As said in llvm coding standard else should not be used after interrupts statements.
It is my first time contribution for open source.
Differential Revision: https://phabricator.services.mozilla.com/D82743
This is going to be useful for the new print preview UI, which is in a
doorhanger and thus much more likely to be less than the page size.
We (ab)use the existing print preview scaling mechanism. We only need it
after reflowing all pages, so this works.
This whole scaling mechanism is all-in-all not amazing, but the patch is
less gross than I initially thought. It's nice, actually.
We could put the new behavior behind a pref trivially, if that's wanted,
but I honestly thing this behavior is better even without the doorhanger
ui.
Differential Revision: https://phabricator.services.mozilla.com/D83309
Also removes some mozilla:: where its no longer needed
Note that this patch was created by running
sed -e -i '.bak' 's/typedef ([a-zA-Z0-9:]+) ([a-zA-Z0-9:]+)/using \2 = \1/'
and then fixing up a few cases that didn't work.
Differential Revision: https://phabricator.services.mozilla.com/D83294
Now, these classes are used only by `TextFragmentData` and they can be not
exposed. Therefore, we should hide them with making them private nested
classes of `TextFragmentData`.
Differential Revision: https://phabricator.services.mozilla.com/D82714
Although the new name is long, but I have no better idea. The class's purpose
is to keep white-space visibility around modifying DOM position. Therefore,
I use "keeper" for the name.
Differential Revision: https://phabricator.services.mozilla.com/D82713
This is a workaround, rather than an ideal fix. The patch contains
a comment describing why it's required and what the proper fix is.
Differential Revision: https://phabricator.services.mozilla.com/D83590
This adds a goal of using less then 1% of total time collecthing the nursery, which has the effect of increasing the nursery size where a lot of nursery garbage is created but little is tenured.
In local testing this is a win on ARES6 and octane, but perfherder paints a more ambiguous picture. I'd like to land this anyway and we can back it out if it causes problems.
Differential Revision: https://phabricator.services.mozilla.com/D80003
