Linux kernel mirror (for testing)
git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel
os
linux
gpu: host1x: Program context stream ID on submission
Add code to do stream ID switching at the beginning of a job. The
stream ID is switched to the stream ID specified by the context
passed in the job structure.
Before switching the stream ID, an OP_DONE wait is done on the
channel's engine to ensure that there is no residual ongoing
work that might do DMA using the new stream ID.
Signed-off-by: Mikko Perttunen <mperttunen@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
authored by
Mikko Perttunen
and committed by
Thierry Reding
24862547
8aa5bcb6
+76
-4
+48
-4
drivers/gpu/host1x/hw/channel_hw.c
+48
-4
drivers/gpu/host1x/hw/channel_hw.c
···
180
180
#endif
181
181
}
182
182
183
+
static void host1x_channel_program_engine_streamid(struct host1x_job *job)
184
+
{
185
+
#if HOST1X_HW >= 6
186
+
u32 fence;
187
+
188
+
if (!job->memory_context)
189
+
return;
190
+
191
+
fence = host1x_syncpt_incr_max(job->syncpt, 1);
192
+
193
+
/* First, increment a syncpoint on OP_DONE condition.. */
194
+
195
+
host1x_cdma_push(&job->channel->cdma,
196
+
host1x_opcode_nonincr(HOST1X_UCLASS_INCR_SYNCPT, 1),
197
+
HOST1X_UCLASS_INCR_SYNCPT_INDX_F(job->syncpt->id) |
198
+
HOST1X_UCLASS_INCR_SYNCPT_COND_F(1));
199
+
200
+
/* Wait for syncpoint to increment */
201
+
202
+
host1x_cdma_push(&job->channel->cdma,
203
+
host1x_opcode_setclass(HOST1X_CLASS_HOST1X,
204
+
host1x_uclass_wait_syncpt_r(), 1),
205
+
host1x_class_host_wait_syncpt(job->syncpt->id, fence));
206
+
207
+
/*
208
+
* Now that we know the engine is idle, return to class and
209
+
* change stream ID.
210
+
*/
211
+
212
+
host1x_cdma_push(&job->channel->cdma,
213
+
host1x_opcode_setclass(job->class, 0, 0),
214
+
HOST1X_OPCODE_NOP);
215
+
216
+
host1x_cdma_push(&job->channel->cdma,
217
+
host1x_opcode_setpayload(job->memory_context->stream_id),
218
+
host1x_opcode_setstreamid(job->engine_streamid_offset / 4));
219
+
#endif
220
+
}
221
+
183
222
static int channel_submit(struct host1x_job *job)
184
223
{
185
224
struct host1x_channel *ch = job->channel;
···
275
236
if (sp->base)
276
237
synchronize_syncpt_base(job);
277
238
278
-
syncval = host1x_syncpt_incr_max(sp, user_syncpt_incrs);
279
-
280
239
host1x_hw_syncpt_assign_to_channel(host, sp, ch);
281
-
282
-
job->syncpt_end = syncval;
283
240
284
241
/* add a setclass for modules that require it */
285
242
if (job->class)
286
243
host1x_cdma_push(&ch->cdma,
287
244
host1x_opcode_setclass(job->class, 0, 0),
288
245
HOST1X_OPCODE_NOP);
246
+
247
+
/*
248
+
* Ensure engine DMA is idle and set new stream ID. May increment
249
+
* syncpt max.
250
+
*/
251
+
host1x_channel_program_engine_streamid(job);
252
+
253
+
syncval = host1x_syncpt_incr_max(sp, user_syncpt_incrs);
254
+
job->syncpt_end = syncval;
289
255
290
256
submit_gathers(job, syncval - user_syncpt_incrs);
291
257
+10
drivers/gpu/host1x/hw/host1x06_hardware.h
+10
drivers/gpu/host1x/hw/host1x06_hardware.h
···
127
127
return (6 << 28) | (offset << 16) | BIT(15) | BIT(14) | count;
128
128
}
129
129
130
+
static inline u32 host1x_opcode_setstreamid(unsigned streamid)
131
+
{
132
+
return (7 << 28) | streamid;
133
+
}
134
+
135
+
static inline u32 host1x_opcode_setpayload(unsigned payload)
136
+
{
137
+
return (9 << 28) | payload;
138
+
}
139
+
130
140
static inline u32 host1x_opcode_gather_wide(unsigned count)
131
141
{
132
142
return (12 << 28) | count;
+10
drivers/gpu/host1x/hw/host1x07_hardware.h
+10
drivers/gpu/host1x/hw/host1x07_hardware.h
···
127
127
return (6 << 28) | (offset << 16) | BIT(15) | BIT(14) | count;
128
128
}
129
129
130
+
static inline u32 host1x_opcode_setstreamid(unsigned streamid)
131
+
{
132
+
return (7 << 28) | streamid;
133
+
}
134
+
135
+
static inline u32 host1x_opcode_setpayload(unsigned payload)
136
+
{
137
+
return (9 << 28) | payload;
138
+
}
139
+
130
140
static inline u32 host1x_opcode_gather_wide(unsigned count)
131
141
{
132
142
return (12 << 28) | count;
+8
include/linux/host1x.h
+8
include/linux/host1x.h
···
327
327
328
328
/* Whether host1x-side firewall should be ran for this job or not */
329
329
bool enable_firewall;
330
+
331
+
/* Options for configuring engine data stream ID */
332
+
/* Context device to use for job */
333
+
struct host1x_memory_context *memory_context;
334
+
/* Stream ID to use if context isolation is disabled (!memory_context) */
335
+
u32 engine_fallback_streamid;
336
+
/* Engine offset to program stream ID to */
337
+
u32 engine_streamid_offset;
330
338
};
331
339
332
340
struct host1x_job *host1x_job_alloc(struct host1x_channel *ch,