linux_dsm_epyc7002/drivers/gpu/drm/nouveau/nvc0_graph.fuc
Ben Skeggs 0411de8548 drm/nvc0/gr: import and use our own fuc by default
The ability to use NVIDIA's fuc has been retained *temporarily* in order
to better debug any issues that may be lingering in our initial attempt
at writing this ucode.  Once I'm fairly confident we're okay, it'll be
removed.

There's a number of things not implemented by this fuc currently, but
most of it is sets of state that our context setup would not have used
anyway.  No doubt we'll find out what they're for at some point, and
implement it if required.

This has been tested on 0xc0/0xc4 thus far, and from what I could tell
it worked as well as NVIDIA's.  It's also been tested on 0xc1, but even
with NVIDIA's fuc that chipset doesn't work correctly with nouveau yet.

0xc3/0xc8/0xce should in theory be supported too, but I don't have the
hardware to check that.

There's no doubt numerous bugs to squash yet, please report any!

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
2011-06-23 15:57:38 +10:00

401 lines
8.5 KiB
Plaintext

/* fuc microcode util functions for nvc0 PGRAPH
*
* Copyright 2011 Red Hat Inc.
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
* to deal in the Software without restriction, including without limitation
* the rights to use, copy, modify, merge, publish, distribute, sublicense,
* and/or sell copies of the Software, and to permit persons to whom the
* Software is furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in
* all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
* THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
* OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
* ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
* OTHER DEALINGS IN THE SOFTWARE.
*
* Authors: Ben Skeggs
*/
define(`mmctx_data', `.b32 eval((($2 - 1) << 26) | $1)')
define(`queue_init', `.skip eval((2 * 4) + ((8 * 4) * 2))')
ifdef(`include_code', `
// Error codes
define(`E_BAD_COMMAND', 0x01)
define(`E_CMD_OVERFLOW', 0x02)
// Util macros to help with debugging ucode hangs etc
define(`T_WAIT', 0)
define(`T_MMCTX', 1)
define(`T_STRWAIT', 2)
define(`T_STRINIT', 3)
define(`T_AUTO', 4)
define(`T_CHAN', 5)
define(`T_LOAD', 6)
define(`T_SAVE', 7)
define(`T_LCHAN', 8)
define(`T_LCTXH', 9)
define(`trace_set', `
mov $r8 0x83c
shl b32 $r8 6
clear b32 $r9
bset $r9 $1
iowr I[$r8 + 0x000] $r9 // CC_SCRATCH[7]
')
define(`trace_clr', `
mov $r8 0x85c
shl b32 $r8 6
clear b32 $r9
bset $r9 $1
iowr I[$r8 + 0x000] $r9 // CC_SCRATCH[7]
')
// queue_put - add request to queue
//
// In : $r13 queue pointer
// $r14 command
// $r15 data
//
queue_put:
// make sure we have space..
ld b32 $r8 D[$r13 + 0x0] // GET
ld b32 $r9 D[$r13 + 0x4] // PUT
xor $r8 8
cmpu b32 $r8 $r9
bra ne queue_put_next
mov $r15 E_CMD_OVERFLOW
call error
ret
// store cmd/data on queue
queue_put_next:
and $r8 $r9 7
shl b32 $r8 3
add b32 $r8 $r13
add b32 $r8 8
st b32 D[$r8 + 0x0] $r14
st b32 D[$r8 + 0x4] $r15
// update PUT
add b32 $r9 1
and $r9 0xf
st b32 D[$r13 + 0x4] $r9
ret
// queue_get - fetch request from queue
//
// In : $r13 queue pointer
//
// Out: $p1 clear on success (data available)
// $r14 command
// $r15 data
//
queue_get:
bset $flags $p1
ld b32 $r8 D[$r13 + 0x0] // GET
ld b32 $r9 D[$r13 + 0x4] // PUT
cmpu b32 $r8 $r9
bra e queue_get_done
// fetch first cmd/data pair
and $r9 $r8 7
shl b32 $r9 3
add b32 $r9 $r13
add b32 $r9 8
ld b32 $r14 D[$r9 + 0x0]
ld b32 $r15 D[$r9 + 0x4]
// update GET
add b32 $r8 1
and $r8 0xf
st b32 D[$r13 + 0x0] $r8
bclr $flags $p1
queue_get_done:
ret
// nv_rd32 - read 32-bit value from nv register
//
// In : $r14 register
// Out: $r15 value
//
nv_rd32:
mov $r11 0x728
shl b32 $r11 6
mov b32 $r12 $r14
bset $r12 31 // MMIO_CTRL_PENDING
iowr I[$r11 + 0x000] $r12 // MMIO_CTRL
nv_rd32_wait:
iord $r12 I[$r11 + 0x000]
xbit $r12 $r12 31
bra ne nv_rd32_wait
mov $r10 6 // DONE_MMIO_RD
call wait_doneo
iord $r15 I[$r11 + 0x100] // MMIO_RDVAL
ret
// nv_wr32 - write 32-bit value to nv register
//
// In : $r14 register
// $r15 value
//
nv_wr32:
mov $r11 0x728
shl b32 $r11 6
iowr I[$r11 + 0x200] $r15 // MMIO_WRVAL
mov b32 $r12 $r14
bset $r12 31 // MMIO_CTRL_PENDING
bset $r12 30 // MMIO_CTRL_WRITE
iowr I[$r11 + 0x000] $r12 // MMIO_CTRL
nv_wr32_wait:
iord $r12 I[$r11 + 0x000]
xbit $r12 $r12 31
bra ne nv_wr32_wait
ret
// (re)set watchdog timer
//
// In : $r15 timeout
//
watchdog_reset:
mov $r8 0x430
shl b32 $r8 6
bset $r15 31
iowr I[$r8 + 0x000] $r15
ret
// clear watchdog timer
watchdog_clear:
mov $r8 0x430
shl b32 $r8 6
iowr I[$r8 + 0x000] $r0
ret
// wait_done{z,o} - wait on FUC_DONE bit to become clear/set
//
// In : $r10 bit to wait on
//
define(`wait_done', `
$1:
trace_set(T_WAIT);
mov $r8 0x818
shl b32 $r8 6
iowr I[$r8 + 0x000] $r10 // CC_SCRATCH[6] = wait bit
wait_done_$1:
mov $r8 0x400
shl b32 $r8 6
iord $r8 I[$r8 + 0x000] // DONE
xbit $r8 $r8 $r10
bra $2 wait_done_$1
trace_clr(T_WAIT)
ret
')
wait_done(wait_donez, ne)
wait_done(wait_doneo, e)
// mmctx_size - determine size of a mmio list transfer
//
// In : $r14 mmio list head
// $r15 mmio list tail
// Out: $r15 transfer size (in bytes)
//
mmctx_size:
clear b32 $r9
nv_mmctx_size_loop:
ld b32 $r8 D[$r14]
shr b32 $r8 26
add b32 $r8 1
shl b32 $r8 2
add b32 $r9 $r8
add b32 $r14 4
cmpu b32 $r14 $r15
bra ne nv_mmctx_size_loop
mov b32 $r15 $r9
ret
// mmctx_xfer - execute a list of mmio transfers
//
// In : $r10 flags
// bit 0: direction (0 = save, 1 = load)
// bit 1: set if first transfer
// bit 2: set if last transfer
// $r11 base
// $r12 mmio list head
// $r13 mmio list tail
// $r14 multi_stride
// $r15 multi_mask
//
mmctx_xfer:
trace_set(T_MMCTX)
mov $r8 0x710
shl b32 $r8 6
clear b32 $r9
or $r11 $r11
bra e mmctx_base_disabled
iowr I[$r8 + 0x000] $r11 // MMCTX_BASE
bset $r9 0 // BASE_EN
mmctx_base_disabled:
or $r14 $r14
bra e mmctx_multi_disabled
iowr I[$r8 + 0x200] $r14 // MMCTX_MULTI_STRIDE
iowr I[$r8 + 0x300] $r15 // MMCTX_MULTI_MASK
bset $r9 1 // MULTI_EN
mmctx_multi_disabled:
add b32 $r8 0x100
xbit $r11 $r10 0
shl b32 $r11 16 // DIR
bset $r11 12 // QLIMIT = 0x10
xbit $r14 $r10 1
shl b32 $r14 17
or $r11 $r14 // START_TRIGGER
iowr I[$r8 + 0x000] $r11 // MMCTX_CTRL
// loop over the mmio list, and send requests to the hw
mmctx_exec_loop:
// wait for space in mmctx queue
mmctx_wait_free:
iord $r14 I[$r8 + 0x000] // MMCTX_CTRL
and $r14 0x1f
bra e mmctx_wait_free
// queue up an entry
ld b32 $r14 D[$r12]
or $r14 $r9
iowr I[$r8 + 0x300] $r14
add b32 $r12 4
cmpu b32 $r12 $r13
bra ne mmctx_exec_loop
xbit $r11 $r10 2
bra ne mmctx_stop
// wait for queue to empty
mmctx_fini_wait:
iord $r11 I[$r8 + 0x000] // MMCTX_CTRL
and $r11 0x1f
cmpu b32 $r11 0x10
bra ne mmctx_fini_wait
mov $r10 2 // DONE_MMCTX
call wait_donez
bra mmctx_done
mmctx_stop:
xbit $r11 $r10 0
shl b32 $r11 16 // DIR
bset $r11 12 // QLIMIT = 0x10
bset $r11 18 // STOP_TRIGGER
iowr I[$r8 + 0x000] $r11 // MMCTX_CTRL
mmctx_stop_wait:
// wait for STOP_TRIGGER to clear
iord $r11 I[$r8 + 0x000] // MMCTX_CTRL
xbit $r11 $r11 18
bra ne mmctx_stop_wait
mmctx_done:
trace_clr(T_MMCTX)
ret
// Wait for DONE_STRAND
//
strand_wait:
push $r10
mov $r10 2
call wait_donez
pop $r10
ret
// unknown - call before issuing strand commands
//
strand_pre:
mov $r8 0x4afc
sethi $r8 0x20000
mov $r9 0xc
iowr I[$r8] $r9
call strand_wait
ret
// unknown - call after issuing strand commands
//
strand_post:
mov $r8 0x4afc
sethi $r8 0x20000
mov $r9 0xd
iowr I[$r8] $r9
call strand_wait
ret
// Selects strand set?!
//
// In: $r14 id
//
strand_set:
mov $r10 0x4ffc
sethi $r10 0x20000
sub b32 $r11 $r10 0x500
mov $r12 0xf
iowr I[$r10 + 0x000] $r12 // 0x93c = 0xf
mov $r12 0xb
iowr I[$r11 + 0x000] $r12 // 0x928 = 0xb
call strand_wait
iowr I[$r10 + 0x000] $r14 // 0x93c = <id>
mov $r12 0xa
iowr I[$r11 + 0x000] $r12 // 0x928 = 0xa
call strand_wait
ret
// Initialise strand context data
//
// In : $r15 context base
// Out: $r15 context size (in bytes)
//
// Strandset(?) 3 hardcoded currently
//
strand_ctx_init:
trace_set(T_STRINIT)
call strand_pre
mov $r14 3
call strand_set
mov $r10 0x46fc
sethi $r10 0x20000
add b32 $r11 $r10 0x400
iowr I[$r10 + 0x100] $r0 // STRAND_FIRST_GENE = 0
mov $r12 1
iowr I[$r11 + 0x000] $r12 // STRAND_CMD = LATCH_FIRST_GENE
call strand_wait
sub b32 $r12 $r0 1
iowr I[$r10 + 0x000] $r12 // STRAND_GENE_CNT = 0xffffffff
mov $r12 2
iowr I[$r11 + 0x000] $r12 // STRAND_CMD = LATCH_GENE_CNT
call strand_wait
call strand_post
// read the size of each strand, poke the context offset of
// each into STRAND_{SAVE,LOAD}_SWBASE now, no need to worry
// about it later then.
mov $r8 0x880
shl b32 $r8 6
iord $r9 I[$r8 + 0x000] // STRANDS
add b32 $r8 0x2200
shr b32 $r14 $r15 8
ctx_init_strand_loop:
iowr I[$r8 + 0x000] $r14 // STRAND_SAVE_SWBASE
iowr I[$r8 + 0x100] $r14 // STRAND_LOAD_SWBASE
iord $r10 I[$r8 + 0x200] // STRAND_SIZE
shr b32 $r10 6
add b32 $r10 1
add b32 $r14 $r10
add b32 $r8 4
sub b32 $r9 1
bra ne ctx_init_strand_loop
shl b32 $r14 8
sub b32 $r15 $r14 $r15
trace_clr(T_STRINIT)
ret
')