There is a race condition where the sysfs-vsync thread accesses a value
(in particular a function ptr) that may not have been set yet as this
is done externally.
Change-Id: If1afb1533344e994398b9b6f5bd3dfb125313416
The APEX macros always rounded to int, losing too much precision. Ensure
that we always keep at least two decimal digits.
The more precise F-number comes from the value recorded by the camera
closed source driver.
Added a new macro APEX_SHUTTER_TO_EXPOSURE(x) (credit to tryptophane).
Change-Id: Ief43c15c6517d8dc48d4d4bc9e93df9eb01e6b91
This libsecion is reverse engineered from the libsecion.so blob and then
written based on the Exynos 5 libion available in AOSP.
Change-Id: Ibe69047c78491e3c7f3173b18fcabdf327fcfbd0
Signed-off-by: angelsl <hidingfromhidden@gmail.com>
Stock gralloc doesn't have these alignment hacks. In fact
they currently do nothing and won't until we fix HWC.
Change-Id: I3427cf3fed84a023ba6ed69b0e0fe0b63ea00886
This caused the ION pointer to get clobbered with a UMP
pointer when FIMC1 memory was requested.
After a while, camera would be unable to allocate buffers
due to ION memory not being freed properly.
Change-Id: I1ba1d27da924a0bb81631548e83c7f3e5789f319
Two reverts squashed, due to Mali-derping being fixed
in frameworks/native:
Revert "gralloc: avoid using framebuffer"
This reverts commit 5125ccc5b6.
Revert "gralloc: Send first 2 FB buffers to carveout and next to UMP"
This reverts commit d288deb284.
Change-Id: I530dc255b7543e03d0e8277d3ad12ed11d6649ab
Mobile's blobs behave this way, except they don't do the
4k alignment done by the FIMC1 allocs.
The 4k alignment should fix issues such as NHK World Live TV,
but fails to do so.
Change-Id: I1ee4cd13e23f9f2f6c6616ee9eed1096b571a82c
Mobile's blobs don't appear to have been compiled with this
present in their headers, and none of the Insignal sources
actually use this.
Change-Id: Ib8611f6da90ea78aa8eec451c76f56ddb32b45a4
This is the story of poor }. It is a painful saga.
Samsung left } floating inside the galaxy of framebuffer_alloc,
unable to join its brethren on the planet of STANDARD_LINUX_SCREEN
when they were shifted out of the #ifdef dimension.
codeworkx brutally eliminated poor } from the galaxy, it was
nowhere to be seen. codeworkx then took pity on the } and
placed it back on the planet of STANDARD_LINUX_SCREEN, but
placed it on a continent where it was all alone, unable to
ever see its mate {
Take pity on poor }, let it rejoin its people on the correct
continent of the planet STANDARD_LINUX_SCREEN, so they may
all shift out of the #ifdef dimension together.
Change-Id: I7a700b73ccd2c91f91d9af3704930759f46978c7
Taken from QCOM gralloc:
commit d80a52cbf961d737f6f3beef47b868a581b98591
Author: Saurabh Shah <saurshah@codeaurora.org>
Date: Tue Oct 2 14:32:40 2012 -0700
gralloc: Send first 2 FB buffers to carveout and next to IOMMU
Gralloc uses the GRALLOC_USAGE_HW_FB flag to allocate memory from carveout.
This is fine for primary framebuffer, but for external framebuffer we need IOMMU
heap buffers, since the ioctls for external work with those.
Ideally, the primary buffers also need to come from IOMMU heap, which is ongoing
work. This patch allocates the first 2 FB buffer requests (assuming they are for
primary) from carveout and allocates all subsequent buffers from IOMMU heap.
Bug: 7124159
Change-Id: Ie53b2e1e7fcac136acd2d251f39cefe367fc7125
Signed-off-by: Iliyan Malchev <malchev@google.com>
Window 2 is used to query global info about the LCD.
Kanged from patch for aries by Greg Hackmann <ghackmann@google.com>
Change-Id: Idf754d4536337d6c06652c1d0c744dc7c0936b15
Howard Su