Previously, the lock would get released on the first close by the X Server
(during AIGLX setup), and the Radeon driver would then hang in initialization
due to unexpected failure in DRM calls that required the lock to be held.
Based on a patch by Kostik Belousov.
The data is now in kernel space, copied in/out as appropriate according to the
This results in DRM_COPY_{TO,FROM}_USER going away, and error paths to deal
with those failures. This also means that XFree86 4.2.0 support for i810 DRM
is lost.
As a fallout, replace filp storage with file_priv storage for "unique
identifier of a client" all over the DRM. There is a 1:1 mapping, so this
should be a noop. This could be a minor performance improvement, as everything
on Linux dereferenced filp to get file_priv anyway, while only the mmap ioctls
went the other direction.
This was used to make all ioctl handlers return -errno on linux and errno on
*BSD. Instead, just return -errno in shared code, and flip sign on return from
shared code to *BSD code.
The current version didn't build on BSD, where the new functionality isn't used
yet anyway. Whoever changes that will hopefully be able to make the OSes share
this file as well.
Actually make the existing ioctls for adding and removing drawables do
something useful, and add another ioctl for the X server to update drawable
information. The only kind of drawable information tracked so far is cliprects.
drm_mtrr_{add,del} for handling the MTRR setup. Still has a LOR issue
with DRM_VERIFYAREA_READ/DRM_COPY_FROM_USER_UNCHECKED in savage_bci.c
-- this won't work with the fine-grained locking in use, and just doing
a single copyin to a temporary will probably work fine. Also note that
the module leaks approximately 4 kb on unload.
up a good bit, I think. Also, remove the agp_uninit() function which
has lain around as a noop for years now. The FreeBSD DRM is now all
compiling, with the exception of via. One known sleeping-with-lock-held
issue remains.
radeon_cp.c to use a drm_local_map_t-type mapping (drm_core_ioremap
rather than drm_ioremap), which contains private device mapping
information on BSD. I also changed the ati_pcigart interface to use
"void *" for pointers to kva rather than "unsigned long". While PCIGART
support appears to be broken on FreeBSD currently, I think this is not
new, and BusType PCI remains working on my r100 in Linux.
firstopen, by making drm_addmap require the drm device lock to be held.
Also, make matching of kernel maps match linux by requiring shm matches
to have the contains_lock flag set if the offset doesn't match.
- Comment out the "is this mapping/bufs in allocated AGP" bits in BSD
because they break mga (which uses AGP allocation that doesn't track
entries). It's not a security issue when we still have the related
ioctls marked root-only.
- Apply some power-of-two alignment restrictions to hopefully avoid some
panicing in bad cases of drm_pci_alloc() on FreeBSD.
- Add verbosity to some error handling that I found useful while debugging.
allocate the resource RF_ACTIVE, pull out the appropriate value, and
return it. However, allocating large framebuffers RF_ACTIVE would run
the system out of KVA, and this also left open the possibility of the
resource getting moved after getting the offset. Instead, when either
of these are called, allocate the resource if it isn't allocated
already (non-RF_ACTIVE) and store it in the DRM device, to be cleaned
up on lastclose.
me to match other drivers and avoid ifdeffing. The linux via_drv.c will
be moved from shared-core to linux-core soon by repocopy.
Submitted by: Jake Burkholder <jake@FreeBSD.org> Tested by: unichrome
understandable: preinit -> load postinit -> (removed) presetup ->
firstopen postsetup -> (removed) open_helper -> open prerelease ->
preclose free_filp_priv -> postclose pretakedown -> lastclose
postcleanup -> unload release -> reclaim_buffers_locked version ->
(removed)
postinit and version were replaced with generic code in the Linux DRM
(drivers now set their version numbers and description in the driver
structure, like on BSD). postsetup wasn't used at all. Fixes the savage
hooks for initializing and tearing down mappings at the right times.
Testing involved at least starting X, running glxgears, killing
glxgears, exiting X, and repeating.
Tested on: FreeBSD (g200, g400, r200, r128) Linux (r200, savage4)
driver's preinit routine, and by using DRM_COPY_TO_USER_IOCTL when
copying out to an ioctl's data pointer. Pulled from the latest version
of my drm-hook-rename.diff and only compile-tested after that.
This patch adds serveral new ioctls and a new query to get_param query to
support PCI MGA cards.
Two ioctls were added to implement interrupt based waiting. With this
change, the client-side driver no longer needs to map the primary DMA
region or the MMIO region. Previously, end-of-frame waiting was done by
busy waiting in the client-side driver until one of the MMIO registers
(the current DMA pointer) matched a pointer to the end of primary DMA
space. By using interrupts, the busy waiting and the extra mappings are
removed.
A third ioctl was added to bootstrap DMA. This ioctl, which is used by the
X-server, moves a *LOT* of code from the X-server into the kernel. This
allows the kernel to do whatever needs to be done to setup DMA buffers.
The entire process and the locations of the buffers are hidden from
user-mode.
Additionally, a get_param query was added to differentiate between G4x0
cards and G550 cards. A gap was left in the numbering sequence so that,
if needed, G450 cards could be distinguished from G400 cards. According
to Ville Syrjälä, the G4x0 cards and the G550 cards handle
anisotropic filtering differently. This seems the most compatible way
to let the client-side driver know which card it's own. Doing this very
small change now eliminates the need to bump the DRM minor version
twice.
http://marc.theaimsgroup.com/?l=dri-devel&m=106625815319773&w=2
A number of ioctl handlers in linux-core were also modified so that they
could be called in-kernel. In these cases, the in-kernel callable
version kept the existing name (e.g., drm_agp_acquire) and the ioctl
handler added _ioctl to the name (e.g., drm_agp_acquire_ioctl).
This patch also replaces the drm_agp_do_release function with
drm_agp_release. drm_agp_release (drm_core_agp_release in the previous
patch) is very similar to drm_agp_do_release, and I saw no reason to
have both.
This commit *breaks the build* on BSD. Eric said that he would make the
required updates to the BSD side soon.
Xorg bug: 3259 Reviewed by: Eric Anholt
There were two problems. First, the 'warp' and 'primary' pointers weren't
cleared, so mga_do_cleanup_dma, which gets called multiple times, would
try to ioremapfree them multiple times. This resulted in the new error
messages to syslog. The second problem was the, since the dev_private
structure isn't reallocated and cleaned out in mga_do_init_dma, when
the server is reloaded idle-waits would wait for impossible values.
I have given this patch some more riggorous testing. This includes:
- Load module, start server, run GL app, stop server, unload module.
- Load module, start server, run GL app, stop server, unload module, reload
module, restart server, run GL app.
- Load module, start server, run GL app, stop server, restart server, run
GL app, stop server, unload module.
In all three cases, everything worked as expected. Please let me know if
there are any further regressions with this patch.
Xorg bug: 3408 Reported by: Chris Rankin
that a device absolutely is, absolutely is not, or may or may not be
AGP. Modify the i915 DRM to use this to force all i9x5 devices to be
"AGP" (even the PCI-e devices).
Reported by: Lukas Hejtmanek
is now allocated (and partially filled in) by the new
mga_driver_preinit function.
This allows the driver to detect the type of card (i.e., G200 class vs.
G400 class) on its own. The chipset value passed to mga_dma_init is now
ignored. This same technique is used by the radeon DRM.
As a result of this, mga_driver_pretakedown was converted to
mga_driver_postcleanup. This routine gets called in some other places
than might be expected, and it sets the dev_private pointer to NULL.
That little gem took over an hour to track down. :(
platform-specific drm_device_is_agp function. Added implementation of
this function the the Linux-specific portion of the MGA driver to
detect PCI G450 cards. Added code to the Linux-specific portion of the
generic DRM layer to not initialize AGP infrastructure if the card is
not AGP (this matches what already existed in BSD).
Bumped the driver date and the driver patch-level for MGA.
This mostly fixes bugzilla #3248. The BSD side still needs an
implementation of mga_driver_device_is_agp.
with IOMMUs and such. There is one usage of the forbidden vtophys()
left in drm_scatter.c which will be fixed up soon. This required a KPI
change for drm_pci_alloc/free() to return/use a drm_dma_handle_t that
keeps track of os-specific bits, rather than just passing around the
vaddr/busaddr/size.
Submitted by: Tonnerre Lombard (partially) Tested on: FreeBSD: Rage128
AGP/PCI Linux: Savage4 AGP/PCI
privileges on Radeon hardware. Essentially, a malicious program could
submit a packet containing an offset (possibly in main memory) to be
rendered from/to, while a separate thread switched that offset in
userspace rapidly between a valid value and an invalid one.
radeon_check_and_fixup_offset() would pull the offset in from user
space, check it, and spit it back out to user space to be copied in
later by the emit code. It would sometimes catch the bad value, but
sometimes the malicious program could modify it after the check and get
an invalid offset rendered from/to.
Fix this by allocating a temporary buffer and copying the data in at once.
While here, make the cliprects stuff not do the VERIFYAREA_READ and
COPY_FROM_USER_UNCHECKED gymnastics, avoiding a lock order reversal on
FreeBSD. Performance impact is negligible -- no difference on r200 to
~1% improvement on rv200 in quake3 tests (P4 1Ghz, demofour at
1024x768, n=4 or 5).
FreeBSD. Add drm_get_resource_{start|len} so linux-specific stuff
doesn't need to be in shared code.
- Fix mach64 build by using __DECONST to work around passing a const
pointer to useracc, which is unfortunately not marked const.
- Get rid of a lot of maplist code by not having dev->maplist be a pointer,
and by sticking the link entries directly in drm_local_map_t rather
than having a separate structure for the linked list.
- Factor out map uninit and removal into its own routine, rather than
duplicating in both drm_takedown() and drm_rmmap().
- Hook up more driver functions, and correct FreeBSD-specific bits of
radeon_cp.c, making radeon work.
- Baby steps towards using bus_space as we should.
ioctls with dev_lock, which is a major step toward being able to remove
Giant. Covers some new pieces (dev->unique*) in the core, and avoids
one call down into system internals with the drm lock held, which is
usually bad (FreeBSD LOR #23, #27).
code. Remove the "drv" from sisdrv, as it's unnecessary. Use the
drm_pci functions in i915 instead of per-os implementations of the
same. Avoid whitespace within fields in drm_pciids.txt (one of the r300
definitions), since it breaks the bsd pciids script. Tested on sis,
mga, r128. i915 needs more work.
done by Martin Lexa (martin at martinlexa dot cz). Now that we've got
porting for all three major BSDs (and the fourth being very similar to
FreeBSD), move the mostly-duplication drm_os_* files into drmP.h.
Remove some cruft from linux heritage and from pieces of the DRM that
have since been removed.
Note that things are still not quite working for even FreeBSD, but these
are first steps at cleanup, and just a WIP checkpoint.
martinlexa dot cz). Now that we've got porting for all three major BSDs
(and the fourth being very similar to FreeBSD), move the
mostly-duplication drm_os_* files into drmP.h. Remove some cruft from
linux heritage and from pieces of the DRM that have since been removed.
Note that things are still not quite working for even FreeBSD, but these
are first steps at cleanup, and just a WIP checkpoint.
DRM_IOCTL_VIA_DMA_INIT DRM_IOCTL_VIA_CMDBUFFER DRM_IOCTL_VIA_FLUSH
The first ioctl sets up an area in AGP memory that will be used as the ring
buffer. The second ioctl copies a command buffer from user space memory
to the ring buffer. The third ioctl waits for engine idle until it
returns.
The motivation for this patch is to avoid the wait for engine idle call
before each buffer flush in the current DRI driver. With this patch,
the DRI driver can continue to flush its buffer as long as there is
free space in the ring buffer.
This patch adds an additional copy operation on the command buffer. This
buffer copying is necessary to support multiple DRI clients rendering
simultaneously. Otherwise, more CPU time will be spent in the busy loop
waiting for engine idle between DRI context switch. Even in the single
client case, the tradeoff is reasonable in comparision to the kernel
call to check for free buffer space for the client to render directly
to the ring buffer.
dev->lock.hw_lock is already set. This fixes the case of two X Servers
running on the same head on different VTs with interface 1.1, by making
the 2nd head fail to inizialize like before.
a busid that doesn't correspond to the device the DRM is attached to.
This is a breaking of backwards-compatibility only for the
multiple-DRI-head case with X Servers that don't use interface 1.1.
- Move irq_busid to drm_irq.h and make it only return the IRQ for the
current device. Retains compatibility with previous X Servers, cleans
up unnecessary code. This means no irq_busid on !__HAVE_IRQ, but can be
changed if necessary.
- Bump interface version to 1.2. This version when set signifies that the
control ioctl should ignore the irq number passed in and enable the
interrupt handler for the attached device. Otherwise it errors out when
the passed-in irq is not equal to the device's.
- Store the highest version the interface has been set to in the device.
- Fix a recursion on DRM_LOCK in irq_uninstall on FreeBSD. This leaves
irq_uninstall being done without the lock in some cases, but it was
racey anyways.
the 2D driver initializes MC_FB_LOCATION and related registers sanely
the DRM deduces the layout from these registers
clients use the new SETPARAM ioctl to tell the DRM where they think the
framebuffer is located in the card's address space
the DRM uses all this information to check client state and fix it up if
necessary
This is a prerequisite for things like direct rendering with IGP chips and
video capturing.
in bufs_info sysctl handler. dev->dma and dev->dma_lock existence are
protected by DRM_LOCK(). Fixes panic on sysctl hw.dri when the device
is uninitialied (when you aren't in X).
server or client to notify the DRM that it expects a certain version of
the device dependent or device independent interface. If the major
doesn't match or minor is too large, EINVAL is returned. A major of -1
means that the requestor doesn't care about that portion of the
interface. The ioctl returns the actual versions in the same struct.
- Introduce DRM DI interface version 1.1. If the server requests version
1.1, then the DRM sets the unique itself according to the busid of the
device it probed, which may then be accessed as normal using getunique.
- Request version 1.1 in libdrm's drmOpenByBusID, allowing the X Server to
request based on a BusID. Introduce a wrapper for DRM_IOCTL_SET_VERSION
and bump libdrm minor version.
- Pass the busid in DRIScreenInit if libdrm can handle both a busid and
name. This allows drmOpenByBusID to be used to find the DRM instead of
just the driver name, which allows us in the future to tie a DRM more
strongly to the device it probed to. Introduce a function
DRICreatePCIBusID which creates a busid in the form pci:oooo:bb:dd.f
similar to linux's pci_name() function. This matches the format used by
the DRM in version 1.1. libdrm knows how to match both this format and
the old PCI🅱️d:f format.
- Use the new DRICreatePCIBusID function in the *_dri.c to request the new,
more exact busid format.
get the drm_file_t * based on the filp passed in ioctl handlers.
- Use this macro on BSD for simplification and improve its error reporting.
Make failure to find the drm_file_t * print as an error, not debug.
This failure may be part of the problem with KDE.
- Make debug and error print macros include the pid on BSD.
data into temporary variables with the lock held then outputting to
sysctls with the lock released. Rearranged a little extra code to aid
this. Note that drm_memory_debug.h hasn't had this fix applied, but I
consider that code to be just about dead anyway.
used by root (the X Server) which are not locked. However, it should
deal with lost-IRQ issues on -current which I think people have been
experiencing but I am unable to reproduce (though I understand why they
would occur, because of a bug of mine). Note that most of the locking
(DRM_LOCK()/UNLOCK()) is all covered by Giant still, so it doesn't
matter yet.
- Remove locking on FreeBSD-stable and NetBSD. These are covered by the
fact that there is no reentrancy of the kernel except by interrupts,
which are locked using spldrm()/splx() instead.
from __HAVE_DMA. This will be useful for adding vblank sync support to
sis and tdfx. Rename dma_service to irq_handler, which is more
accurately what it is.
- Fix the #if _HAVE_DMA_IRQ in radeon, r128, mga, i810, i830, gamma to have
the right number of underscores. This may have been a problem in the
case that the server died without doing its DRM_IOCTL_CONTROL to
uninit.
just a single instance. Moved the PCI ID lists from <card>_drv.c in BSD
to <card>.h. The PCI ID lists include a driver private field, which may
be used by drivers for chip family or other information. Based on work
by jonsmirl.
- Make tdfx_drv.c and tdfx.h match other drivers.
- Fixed up linking of sis shared files.
Tested with Radeon and SiS on Linux and FreeBSD, including a Linux setup
with
2 SiS cards in a machine, but only one head being used (with DRI)
Change some nearby memset()s to bzero()s or to calloc allocations to
take advantage of M_ZERO). Reverse some error tests to reduce high
levels of indentation. Move the sg_cleanup() call out of the maplist
loop in DRM(takedown)-- I can't see any need for it to be inside.
Jung-uk Kim
Eric Anholt
vehemens
Arthur Huillet
Michel Dänzer
Alan Hourihane
Eric Anholt
Dave Airlie
Ian Romanick
Erdi Chen
Michel Daenzer