6a6c37af9e
[intel-GEM] ref count objects in gtt-lru.
...
If objects on the lru aren't ref counted, they'll get pulled from the gtt as
soon as they are freed. This change does cause objects to get stuck in the
gtt until they're forced out by new requests. The lru should get cleaned
when the irq occurs.
2008-05-06 21:59:06 -07:00
61253f4f67
[intel-GEM] Add memory domain support.
...
Memory domains allow the kernel to track which caches to flush and how to
move objects before buffer execution.
2008-05-06 20:00:23 -07:00
2b9ef32669
Merge commit 'anholt/drm-gem' into drm-gem
2008-05-06 14:43:56 -07:00
631e86c5c4
Start coding up memory domains
2008-05-06 14:43:49 -07:00
d2373b2a34
GEM: Use irq-based fencing rather than syncing and evicting every exec.
2008-05-06 13:28:26 -07:00
dd6976c56f
GEM: Skip relocation if presumed offset matches.
2008-05-06 11:25:53 -07:00
8551bfc6db
GEM: Save the last ioremapped page for relocations in case we need it again.
2008-05-06 11:18:57 -07:00
91cba3ae17
Dump last batch buffer when hardware lockup is detected.
2008-05-05 22:10:02 -07:00
ed6657fa8e
Monitor ACTHD register while polling for idle ring.
...
When batch buffers are executing, the ring may be stuck for a long time.
Monitor the ACTHD pointer which will show if the execution engine is
actually hung.
2008-05-05 22:09:34 -07:00
2c8f970baa
Unlock pages right after getting them.
...
pages come back from find_or_create_page locked, but must not stay locked
for long. Unlock them immediately instead of waiting until we're done with
them to avoid deadlock when applications try to touch them.
2008-05-05 17:17:19 -07:00
5b0d0fa7f8
Merge commit 'anholt/drm-gem' into drm-gem
...
Conflicts:
linux-core/i915_gem.c
2008-05-05 14:40:20 -07:00
dafe48e623
GEM: Replace drm_memrange_for_each with just evicting what we brought in.
...
I was wrong about how the data structure worked, and didn't care to fix it
to support debugging code.
2008-05-05 14:38:26 -07:00
d59a9300ec
Remove some debug messages.
2008-05-05 14:32:01 -07:00
f0bc796a02
Add object base to relocation store address.
...
The relocated value was being written to the wrong location, missing
the object base address.
2008-05-05 14:22:42 -07:00
4867780bd6
Emit clflush and chipset flush when mapping objects to gtt
2008-05-05 13:32:28 -07:00
4511e6cd80
Correct execbuffer offset. Add memory barrier and chipset flush.
2008-05-05 11:27:06 -07:00
b6f173c430
Add i915_dispatch_gem_execbuffer (broken).
...
This function submits a gem-based execbuffer to the ring.
It doesn't work yet.
2008-05-05 10:51:49 -07:00
d015219bd0
r500: add allowed range for us config/pixsize
2008-05-05 17:03:27 +10:00
166ff364fb
Don't forget to set the memrange private, and reset ring on kernel entry.
2008-05-02 17:50:46 -07:00
afe574f328
Don't include the tail guard memrange in foreach callbacking.
2008-05-02 17:49:52 -07:00
5f0614b86f
Check for do_mmap errors
2008-05-02 17:13:11 -07:00
ab3549d133
Add a bit of /proc/dri/*/gem support. Clean up some refcount/pagelock issues.
...
Track named objects in /proc/dri/0/gem_names.
Track total object count in /proc/dri/0/gem_objects.
Initialize device gem data.
return -ENODEV for gem ioctls if the driver doesn't support gem.
Call unlock_page when unbinding from gtt.
Add numerous misssing calls to drm_gem_object_unreference.
2008-05-02 16:34:16 -07:00
39e20bcd5f
Add name/open ioctls, separate handle and pointer ref counts.
...
Names are just another unique integer set (from another idr object).
Names are removed when the user refernces (handles) are all destroyed --
this required that handles for objects be counted separately from
internal kernel references (so that we can tell when the handles are all
gone).
2008-05-02 12:29:17 -07:00
49e8e3372a
Remove drm_driver argument to functions taking drm_gem_object.
...
Now that drm_gem_object has a drm_driver * in it, functions don't need both
parameters.
2008-05-02 10:36:00 -07:00
5b5b68ffd2
Fix nouveau warning when returning pointers in uint64_t objects.
2008-05-02 10:34:46 -07:00
0d547c9ed9
Add alignment to all aperture allocation requests.
...
When pinning buffers, or using execbuffer, allow the application to specify
the necessary aperture allocation alignment constraints.
2008-05-01 20:41:55 -07:00
30efad5113
Fix gem ioctls to be 32/64-bit clean.
...
mixed 32/64 bit systems need 'special' help for ioctl where the user-space
and kernel-space datatypes differ. Fixing the datatypes to be the same size,
and align the same way for both 32 and 64-bit ppc and x86 environments will
elimiante the need to have magic 32/64-bit ioctl translation code.
2008-05-01 20:31:16 -07:00
abc896638f
Use krefs for refcounting.
...
krefs are way easier than a custom-coded spinlock+int combo.
2008-05-01 20:12:39 -07:00
c10695bb7a
Unbind objects when freeing, fix some error paths, and warn in others.
2008-05-01 17:31:57 -07:00
ddc80651d5
Fix offset passed to AGP to be pages instead of bytes.
...
Fix some utterly bonged loop while we were staring at it.
2008-05-01 17:31:29 -07:00
3f641b56c7
Fix missing member settings in obj/obj_priv, and some error paths.
2008-05-01 16:48:25 -07:00
7d5f783eca
Make GEM object handles be nonzero.
2008-05-01 16:38:37 -07:00
c530011aaa
Update mm tests for GEM rename.
2008-05-01 16:38:25 -07:00
d2529d1396
Remove _args from gem ioctl argument structure tags.
2008-05-01 16:27:03 -07:00
793549116e
Add pin/unpin object ioctls for gem.
2008-05-01 15:40:02 -07:00
ccd1bae0f6
checkpoint: relocations support.
2008-05-01 15:22:21 -07:00
5af87acbc2
checkpoint: gtt binding written.
2008-05-01 14:20:44 -07:00
2140e102f9
checkpoint: rename to GEM and a few more i915 bits.
2008-05-01 11:39:20 -07:00
3ac74f3208
nv50: enable 0x400500 bit 0 after PGRAPH exception also
...
No solid idea about what these 2 bits do, but nv50 can now survive a few
PGRAPH exceptions just as nv40 does :)
2008-05-02 01:36:30 +10:00
6d8062ac1e
nouveau: guard against channels potentially not having a context, fix nv50
2008-05-02 01:36:08 +10:00
77d20928b3
nouveau: disable all card interrupts when unknown PFIFO IRQ occurs.
...
This is possibly temporary. I can trigger an unending IRQ storm on G8x
in some circumstances, and have no idea how to handle that particular PFIFO
exception correctly yet.
2008-05-02 00:53:42 +10:00
5c4c778c0d
nouveau: restore original NV_PFIFO_CACHES_REASSIGN value in fifo handler
...
Doesn't fix any issue I've seen, but is a potential issue if a FIFO IRQ
occurs during channel creation/takedown.
2008-05-02 00:52:21 +10:00
bfbe4ade32
nouveau: gather nsource in trap_info()
...
The IRQ handling stuff really is a mess.. On the TODO :)
2008-05-02 00:51:00 +10:00
e317dfdabf
nv50: PGRAPH exception handling completely different from earlier chips
2008-05-02 00:06:22 +10:00
b92efd5956
nv50: I cave... Add nv84 initial context values.
...
I swore I'd actually do this properly and not go the horrible route
we did with nv4x, but I won't get around to it just yet with so many
*actually* interesting things to do first.. One day.
Since someone already added nv86, why not!
2008-05-01 23:50:44 +10:00
1a84067950
Hacking towards hooking up execbuffer.
2008-04-30 16:03:15 -07:00
81ba8ded7e
Remove the remainder of the mmfs device.
2008-04-29 13:48:51 -07:00
3148c16364
Move mmfs tests over to be drm tests.
2008-04-29 13:45:43 -07:00
dabd056bf3
Move mmfs ioctls into the DRM. Untested.
2008-04-29 13:32:52 -07:00
cb33133ef3
i915: fix off by one in VGA save/restore of AR & CR regs
...
Turns out it's important to save/restore AR14 in particular.
2008-04-29 12:39:38 -07:00
Keith Packard
Eric Anholt
Dave Airlie
Ben Skeggs
Jesse Barnes