summaryrefslogtreecommitdiff
AgeCommit message (Collapse)Author
2007-12-11Make relocation validate client computed values when debuggingKeith Packard
2007-12-11i915: wait for buffer idle before writing relocationsKeith Packard
When writing a relocation entry, make sure the target buffer is idle, otherwise the GPU may see inconsistent data.
2007-12-11Bump driver minor for relocation optimzationsKeith Packard
2007-12-11Allow relocation to be skipped when buffers don't move.Keith Packard
One of the costs of superioctl has been the need to perform relocations inside the kernel. The cost of mapping the buffers to the CPU and writing data is fairly high, especially if those buffers have been mapped and read by the GPU. If we assume that buffers don't move around very often, we can have the client compute the relocations itself using the previous GPU address. When that object doesn't move, the kernel can skip computing and writing the updated data. Here's a patch which adds a new field to struct drm_bo_info_req called 'presumed_offset', and a new DRM_BO_HINT_PRESUMED_OFFSET that is set when this field has been filled in by the client. There are two separate optimizations performed when the presumed_offset is correct: 1. i915_exec_reloc checks to see if all previous buffer offsets were guessed correctly. If so, there's no need for it to look at *any* of the relocations for a buffer. When this happens, it skips the whole relocation process, simply returning success. 2. i915_apply_reloc checks to see if the target buffer offset was guessed correctly. If so, it skips mapping the relocatee, computing the relocation and writing the value. If no relocations are needed, the relocatee should never be mapped to the CPU, and so the kernel shouldn't need to wait for any fences to pass.
2007-12-10drm: move agp include outside CONFIG_AGP as it isn't dependant on agp in kernelDave Airlie
2007-12-08mach64: comment bus master / ring buffer behavior and securityJosé Fonseca
2007-12-07Fix pipe<->plane mapping vs. vblank handling (again)Jesse Barnes
If drmMinor >= 6, the intel DDX driver will enable vblank events on both pipes. If drmMinor >= 10 on pre-965 chipsets, the intel DDX driver will swap the pipe<->plane mapping to allow for framebuffer compression on laptop screens. This means the secondary vblank counter (corresponding to pipe B) will be incremented when vblank interrupts occur. Now Mesa waits for vblank events on whichever plane has a greater portion of the displayed window. So it will happly ask to wait for the primary counter even though that one won't increment. So we can fix this in either the DDX driver, Mesa or the kernel (though I thought we already had several times). Since current (and previous) userspace assumes it's talking about a pipe == plane situation and now uses planes when talking to the kernel, we should probably just hide the mapping details there (indeed they already are hidden there for vblank swaps), which this patch does. So as far as userland is concerned, whether we call things planes or pipes is irrelevant, as long as kernel developers understand that userland hands them planes and they have to figure out which pipe that corresponds to (which will typically be the same on 965+ hardware and reversed on pre-965 mobile chips).
2007-12-06take down stuff after asking driver to unloadDave Airlie
2007-12-05mach64: make buffer emission macros normal functionsJosé Fonseca
2007-12-05mach64: use utf-8José Fonseca
2007-12-05Remove references to the sarea_priv perf_boxes field.Kristian Høgsberg
This field isn't touched or read by any other code in the stack so it's time to retire these last few references.
2007-12-05patch from -mm kernel to use upper_32_bitsDave Airlie
2007-12-02bsd: Replace other occurrences of msleep with mtx_sleepRobert Noland
2007-12-02bsd: Now make secondary vblank workRobert Noland
We needed to specifically check for driver support and test the correct vbl_received value. Also pulled over support for _DRM_VBLANK_NEXTONMISS from the linux code.
2007-12-01bsd: Hook secondary vblank support.Robert Noland
2007-12-01bsd: Fix typo in i915_drv.cRobert Noland
2007-12-01mtx_sleep is preferred to msleepRobert Noland
Calling semantics are the same and both were introduced in 5.0
2007-12-01Clarify order of operationsRobert Noland
2007-12-01DRM_DEBUG already prints the function name.Robert Noland
2007-12-01drm: Add _DRM_DRIVER map flag.Robert Noland
This flag indicates that the driver is responsible for the map.
2007-12-01bsd: Move counter initialization to load time.Robert Noland
2007-11-30nouveau: Properly identify NV40 and NV44 generation.Maarten Maathuis
2007-11-29Beside the emitted warning, the added cast (u64 -> unsigned) strips outJiri Slaby
part of address on 64 bit. Cast to unsigned long instead. Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
2007-11-29drm: enable udev node creationDave Airlie
2007-11-29drm: oops not a cleanup..Dave Airlie
2007-11-29Fix up drm_ati_pcigart_infoRobert C. Noland III
2007-11-29drm: more cleanupsDave Airlie
2007-11-28i965: oops force mi batchbuffer startDave Airlie
2007-11-22drm: major whitespace/coding style realignment with kernelDave Airlie
2007-11-22drm: cleanup drm_regman.c coding styleDave Airlie
2007-11-22i915: add context handle to superioctl structDave Airlie
This will be used later for lockless operation.
2007-11-21drm: don't reset to 0 irq_enabled when client open file descriptorJerome Glisse
2007-11-19Fix capitalization of __linux__ define.Eric Anholt
2007-11-19drm: fix dead lock in drm_buffer_object_transferJerome Glisse
2007-11-18Bug #13233: Fix build on FreeBSD.Robert Noland
2007-11-18Bug #13233: Chase move of agp headers on FreeBSD.Robert Noland
2007-11-18radeon: refactor out the fb/agp location read/write.Dave Airlie
Add a new get param to get the fb location into userspace. Mesa currently hits MMIO to do this, but this isn't always possible.
2007-11-16nouveau: also mention the number of succcessfully copied bios bytes.Stephane Marchesin
2007-11-15nouveau: renameJeremy Kolb
2007-11-15nouveau: flip buffer into gart.Jeremy Kolb
2007-11-15nouveau: be verbose about PPC bios for now.Stephane Marchesin
2007-11-15nouveau: revert the nv34 context size change, it was not the culprit after all.Stephane Marchesin
2007-11-15nouveau: use get_property instead of of_get_property on pre-2.6.22 kernels.Stephane Marchesin
2007-11-15mm fixups.Thomas Hellstrom
2007-11-15i915: remove excess debug outputDave Airlie
2007-11-15nouveau: Copy the PPC bios to RAMIN on init, that lets us do proper output ↵Stephane Marchesin
detection in user space.
2007-11-15intel: add flushing for i8xx chipsets.Dave Airlie
Add a nut vs hammer style chipset flush for the i8xx chipsets - reenable TTM code paths
2007-11-14nouveau: nv30: missing ramin init, does it brake other hw?Patrice Mandin
2007-11-14Add new shared header file drm_internal.h.Kristian Høgsberg
This header file is shared across linux and bsd, but is not installed for user space to access. It's the place to put prototypes and data types that aren't platform or chipset specific, but still internal to the drm.
2007-11-14nouveau: adjust the size of the NV34 context. That fixes mobile PPC cards.Stephane Marchesin