Skip to content
  1. Sep 24, 2019
    • Andreas Baierl's avatar
      WIP: lima/ppir: Introduce late abs modifier lowering · fbfe42b8
      Andreas Baierl authored 
      

Some ops can't deal with abs sources directly, so we have to lower
the abs modifier. Lower abs like sqrt(mul(x, x)).

This is what the blob does and we pass some more piglit tests.
This ppir lowering pass is executed at the very end when lowering
of all the other alu instructions is already done but before
lowering the consts.

Signed-off-by: default avatarAndreas Baierl <ichgeh@imkreisrum.de>
      fbfe42b8
    • Connor Abbott's avatar
      lima/gpir: Fix 64-bit shift in scheduler spilling · fed5b605
      Connor Abbott authored 
      

There are 64 physical registers so the shift must be 64 bits.

Reviewed-by: default avatarVasily Khoruzhick <anarsoul@gmail.com>
      fed5b605
    • Connor Abbott's avatar
      lima/gpir: Don't emit movs when translating from NIR · ef38a659
      Connor Abbott authored 
      

The scheduler doesn't expect them. To do this, I had to refactor the
registration part of gpir_node_create_dest() to be separate from
creating and inserting the node, since the last two now aren't done when
handling moves. This adds more code but creates the possibility of
automatically inserting input dependencies when inserting nodes, similar
to what's done in NIR with the use-def lists (this isn't done yet).

Reviewed-by: default avatarVasily Khoruzhick <anarsoul@gmail.com>
      ef38a659
    • Connor Abbott's avatar
      lima/gpir: Fix postlog2 fixup handling · 96c31d9a
      Connor Abbott authored 
      

We guarantee that a complex1 op is always used by postlog2 directly by
rewriting the postlog2 op to be a move when there would be a move
inserted between them. But we weren't doing this in all circumstances
where there might be a move. Move the logic to place_move() so that it
always happens. Fixes a few log tests that happened to start failing due
to changes in the register allocator leading to a different scheduling
order.

Reviewed-by: default avatarVasily Khoruzhick <anarsoul@gmail.com>
      96c31d9a
    • Connor Abbott's avatar
      lima/gpir: Use registers for values live in multiple blocks · 1cd1cce0
      Connor Abbott authored 
      

This commit adds the framework for cross-basic-block register
allocation. Like ARM's compiler, we assume that the value registers
aren't usable across branches, which means we have to use physical
registers to store any value that crosses a basic block. There are three
parts to this:

1. When translating from NIR, we rely on the NIR out-of-ssa pass to
coalesce values into registers. We insert store_reg instructions for
values used in more than one basic block, and load_reg instructions for
values not defined in the same basic block (or defined after their use,
for loops). So by the time we've translated out of NIR we've already
split things into values (which are only used in the same basic block)
and registers (which are only used in different basic blocks than where
they're defined).

2. We allocate the registers at the same time that we allocate the
values, before the final scheduler. Unlike the values, where the
assigned color is fake, we assign the actual physical index & component
to physregs at this stage. load_reg and store_reg are treated as moves
in the allocator and when creating write-after-read dependencies.

3. Finally, in the main scheduler we have to avoid overwriting existing
live physregs when spilling. First, we have to tell the scheduler which
physical registers are live at the end of each block, to avoid
overwriting those. If a register is only live at the beginning, we can
reuse it for spilling after the last original use in the final program
happens, i.e. before any original use is scheduled, but we have to be
careful to add the proper dependencies so that the spill write is
scheduled before the original reads. To handle this we repurpose
reg_link for uses to be used by the scheduler.

A few register-related things copied over from NIR or from other
drivers can be dropped.

Reviewed-by: default avatarVasily Khoruzhick <anarsoul@gmail.com>
      1cd1cce0
    • Connor Abbott's avatar
      lima/gpir: Support branch instructions · 7594ef6e
      Connor Abbott authored 
      

Because branch conditions have to be in the pass slot, there is no
unconditional branch, and realistically the pass slot has to contain a
move when branching (there's nothing it does that would be useful for
operating on booleans, so we can't use it for anything when computing
the branch condition), we put the branch instruction in the pass slot
and at codegen time turn it into a move of the branch condition. This
means that it doesn't have to be special-cased like store instructions
are in the scheduler. Because of this decision we can remove the
half-implemented BRANCH codegen slot. Finally, we (ab)use the existing
schedule_first mechanism to make sure that branches are always last in
the basic block.

Reviewed-by: default avatarVasily Khoruzhick <anarsoul@gmail.com>
      7594ef6e
    • Connor Abbott's avatar
      lima/gpir: Only try to place actual children · 2df2e081
      Connor Abbott authored 
      

When picking a node to be scheduled, we try to schedule its children as
well. But we shouldn't try to schedule nodes which only have a fake
dependency on the original node, since this isn't the point of
scheduling children at the same time and can break some expectations of
the rest of the code.

Reviewed-by: default avatarVasily Khoruzhick <anarsoul@gmail.com>
      2df2e081
    • Connor Abbott's avatar
      lima/gpir: Fix compiler warning · f989a024
      Connor Abbott authored 
      

Reviewed-by: default avatarVasily Khoruzhick <anarsoul@gmail.com>
      f989a024
    • Adam Jackson's avatar
      glx: Implement GLX_EXT_no_config_context · 0d635ccc
      Adam Jackson authored 
      This is the GLX counterpart to EGL_KHR_no_config_context. Contexts may
now be created without reference to an fbconfig, in which case it is
treated as compatible with any fbconfig (and thus any GLX drawable).

Khronos: https://github.com/KhronosGroup/OpenGL-Registry/pull/102


Reviewed-by: Kenneth Graunke's avatarKenneth Graunke <kenneth@whitecape.org>
Reviewed-by: default avatarIan Romanick <ian.d.romanick@intel.com>
      0d635ccc
    • Adam Jackson's avatar
      glx: Lift sending the MakeCurrent request to top-level code · 999c2aed
      Adam Jackson authored 
      Somewhat terrifyingly, we never sent this for direct contexts, which
means the server never knew the context/drawable bindings. To handle
this sanely, pull the request code up out of the indirect backend, and
rewrite the context switch path to call it as appropriate.  This
attempts to preserve the existing behavior of not calling unbind() on
the context if its refcount would not drop to zero.

Of course, you can't just do this indiscriminately, because this is GLX
and extant X servers have bugs and everything is terrible. To wit:

- For 1.20.x prior to 1.20.6, you can bind a direct context once, but
the second time you try to modify the context's binding you will get
GLXBadContextTag. This includes unbinding the context. And "deleting"
the context will leak memory, because it will still appear to be
current.

- For 1.19 and earlier, glXMakeCurrent(dpy, None, ctx) should be legal
for GL 3.0+ contexts, but the server will throw BadMatch.

To guard against this, we only send the request for indirect contexts
unless the server is known good, and only mention one context at a time
in such a request; if switching between contexts, we first unbind the
old, and then bind the new. Note that the second VendorRelease() version
is to catch XFree86 4.x and Xorg [67].x, which almost certainly have the
above bugs. Other servers might report different version numbers here,
but we can't do direct rendering against them, so this should be safe.

Fixes glx-make-context, glx-multi-window-single-context and
glx-query-drawable-glx_fbconfig_id-window. Sufficiently old piglit will
regress on glx-make-glxdrawable-current (throwing BadMatch), which is
fixed by mesa/piglit!116.
      999c2aed
    • Adam Jackson's avatar
      glx: Move vertex array protocol state into the indirect backend · 01e43798
      Adam Jackson authored 
      Only relevant for indirect contexts, so let's get that code out of the
common path.
      01e43798
  3. Sep 23, 2019