3
* Copyright (c) 2019 Collabora LTD
5
* Author: Gert Wollny <gert.wollny@collabora.com>
7
* Permission is hereby granted, free of charge, to any person obtaining a
8
* copy of this software and associated documentation files (the "Software"),
9
* to deal in the Software without restriction, including without limitation
10
* on the rights to use, copy, modify, merge, publish, distribute, sub
11
* license, and/or sell copies of the Software, and to permit persons to whom
12
* the Software is furnished to do so, subject to the following conditions:
14
* The above copyright notice and this permission notice (including the next
15
* paragraph) shall be included in all copies or substantial portions of the
18
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
19
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
20
* FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
21
* THE AUTHOR(S) AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM,
22
* DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
23
* OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
24
* USE OR OTHER DEALINGS IN THE SOFTWARE.
27
#include "sfn_callstack.h"
31
CallStack::CallStack(r600_bytecode& bc):
37
CallStack::~CallStack()
41
int CallStack::push(unsigned type)
48
++m_bc.stack.push_wqm;
57
return update_max_depth(type);
60
void CallStack::pop(unsigned type)
65
assert(m_bc.stack.push >= 0);
68
--m_bc.stack.push_wqm;
69
assert(m_bc.stack.push_wqm >= 0);
73
assert(m_bc.stack.loop >= 0);
81
int CallStack::update_max_depth(unsigned type)
84
r600_stack_info& stack = m_bc.stack;
88
int entry_size = stack.entry_size;
90
elements = (stack.loop + stack.push_wqm ) * entry_size;
91
elements += stack.push;
93
switch (m_bc.chip_class) {
96
/* pre-r8xx: if any non-WQM PUSH instruction is invoked, 2 elements on
97
* the stack must be reserved to hold the current active/continue
99
if (type == FC_PUSH_VPM || stack.push > 0) {
104
/* r9xx: any stack operation on empty stack consumes 2 additional
109
/* r8xx+: 2 extra elements are not always required, but one extra
110
* element must be added for each of the following cases:
111
* 1. There is an ALU_ELSE_AFTER instruction at the point of greatest
113
* (Currently we don't use ALU_ELSE_AFTER.)
114
* 2. There are LOOP/WQM frames on the stack when any flavor of non-WQM
115
* PUSH instruction executed.
117
* NOTE: it seems we also need to reserve additional element in some
118
* other cases, e.g. when we have 4 levels of PUSH_VPM in the shader,
119
* then STACK_SIZE should be 2 instead of 1 */
120
if (type == FC_PUSH_VPM || stack.push > 0) {
131
entries = (elements + (entry_size - 1)) / entry_size;
133
if (entries > stack.max_entries)
134
stack.max_entries = entries;