native.c 36.1 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
/*
 * Copyright 2014 IBM Corp.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License
 * as published by the Free Software Foundation; either version
 * 2 of the License, or (at your option) any later version.
 */

#include <linux/spinlock.h>
#include <linux/sched.h>
#include <linux/slab.h>
#include <linux/mutex.h>
#include <linux/mm.h>
#include <linux/uaccess.h>
16
#include <linux/delay.h>
17
#include <asm/synch.h>
18
#include <misc/cxl-base.h>
19 20

#include "cxl.h"
I
Ian Munsie 已提交
21
#include "trace.h"
22

23
static int afu_control(struct cxl_afu *afu, u64 command, u64 clear,
24 25
		       u64 result, u64 mask, bool enabled)
{
26
	u64 AFU_Cntl;
27
	unsigned long timeout = jiffies + (HZ * CXL_TIMEOUT);
I
Ian Munsie 已提交
28
	int rc = 0;
29 30 31 32

	spin_lock(&afu->afu_cntl_lock);
	pr_devel("AFU command starting: %llx\n", command);

I
Ian Munsie 已提交
33 34
	trace_cxl_afu_ctrl(afu, command);

35 36
	AFU_Cntl = cxl_p2n_read(afu, CXL_AFU_Cntl_An);
	cxl_p2n_write(afu, CXL_AFU_Cntl_An, (AFU_Cntl & ~clear) | command);
37 38 39 40 41

	AFU_Cntl = cxl_p2n_read(afu, CXL_AFU_Cntl_An);
	while ((AFU_Cntl & mask) != result) {
		if (time_after_eq(jiffies, timeout)) {
			dev_warn(&afu->dev, "WARNING: AFU control timed out!\n");
I
Ian Munsie 已提交
42 43
			rc = -EBUSY;
			goto out;
44
		}
45

46
		if (!cxl_ops->link_ok(afu->adapter, afu)) {
47 48 49 50 51
			afu->enabled = enabled;
			rc = -EIO;
			goto out;
		}

52
		pr_devel_ratelimited("AFU control... (0x%016llx)\n",
53 54 55
				     AFU_Cntl | command);
		cpu_relax();
		AFU_Cntl = cxl_p2n_read(afu, CXL_AFU_Cntl_An);
A
Andrew Donnellan 已提交
56
	}
57 58 59 60 61 62 63 64 65 66

	if (AFU_Cntl & CXL_AFU_Cntl_An_RA) {
		/*
		 * Workaround for a bug in the XSL used in the Mellanox CX4
		 * that fails to clear the RA bit after an AFU reset,
		 * preventing subsequent AFU resets from working.
		 */
		cxl_p2n_write(afu, CXL_AFU_Cntl_An, AFU_Cntl & ~CXL_AFU_Cntl_An_RA);
	}

67 68
	pr_devel("AFU command complete: %llx\n", command);
	afu->enabled = enabled;
I
Ian Munsie 已提交
69 70
out:
	trace_cxl_afu_ctrl_done(afu, command, rc);
71 72
	spin_unlock(&afu->afu_cntl_lock);

I
Ian Munsie 已提交
73
	return rc;
74 75 76 77 78 79
}

static int afu_enable(struct cxl_afu *afu)
{
	pr_devel("AFU enable request\n");

80
	return afu_control(afu, CXL_AFU_Cntl_An_E, 0,
81 82 83 84 85 86 87 88
			   CXL_AFU_Cntl_An_ES_Enabled,
			   CXL_AFU_Cntl_An_ES_MASK, true);
}

int cxl_afu_disable(struct cxl_afu *afu)
{
	pr_devel("AFU disable request\n");

89 90
	return afu_control(afu, 0, CXL_AFU_Cntl_An_E,
			   CXL_AFU_Cntl_An_ES_Disabled,
91 92 93 94
			   CXL_AFU_Cntl_An_ES_MASK, false);
}

/* This will disable as well as reset */
95
static int native_afu_reset(struct cxl_afu *afu)
96 97 98
{
	pr_devel("AFU reset request\n");

99
	return afu_control(afu, CXL_AFU_Cntl_An_RA, 0,
100 101 102 103 104
			   CXL_AFU_Cntl_An_RS_Complete | CXL_AFU_Cntl_An_ES_Disabled,
			   CXL_AFU_Cntl_An_RS_MASK | CXL_AFU_Cntl_An_ES_MASK,
			   false);
}

105
static int native_afu_check_and_enable(struct cxl_afu *afu)
106
{
107
	if (!cxl_ops->link_ok(afu->adapter, afu)) {
108 109 110
		WARN(1, "Refusing to enable afu while link down!\n");
		return -EIO;
	}
111 112 113 114 115 116 117 118 119 120 121 122
	if (afu->enabled)
		return 0;
	return afu_enable(afu);
}

int cxl_psl_purge(struct cxl_afu *afu)
{
	u64 PSL_CNTL = cxl_p1n_read(afu, CXL_PSL_SCNTL_An);
	u64 AFU_Cntl = cxl_p2n_read(afu, CXL_AFU_Cntl_An);
	u64 dsisr, dar;
	u64 start, end;
	unsigned long timeout = jiffies + (HZ * CXL_TIMEOUT);
I
Ian Munsie 已提交
123 124 125
	int rc = 0;

	trace_cxl_psl_ctrl(afu, CXL_PSL_SCNTL_An_Pc);
126 127 128

	pr_devel("PSL purge request\n");

129
	if (!cxl_ops->link_ok(afu->adapter, afu)) {
130 131 132 133 134
		dev_warn(&afu->dev, "PSL Purge called with link down, ignoring\n");
		rc = -EIO;
		goto out;
	}

135 136 137 138 139 140 141 142 143 144 145 146 147
	if ((AFU_Cntl & CXL_AFU_Cntl_An_ES_MASK) != CXL_AFU_Cntl_An_ES_Disabled) {
		WARN(1, "psl_purge request while AFU not disabled!\n");
		cxl_afu_disable(afu);
	}

	cxl_p1n_write(afu, CXL_PSL_SCNTL_An,
		       PSL_CNTL | CXL_PSL_SCNTL_An_Pc);
	start = local_clock();
	PSL_CNTL = cxl_p1n_read(afu, CXL_PSL_SCNTL_An);
	while ((PSL_CNTL &  CXL_PSL_SCNTL_An_Ps_MASK)
			== CXL_PSL_SCNTL_An_Ps_Pending) {
		if (time_after_eq(jiffies, timeout)) {
			dev_warn(&afu->dev, "WARNING: PSL Purge timed out!\n");
I
Ian Munsie 已提交
148 149
			rc = -EBUSY;
			goto out;
150
		}
151
		if (!cxl_ops->link_ok(afu->adapter, afu)) {
152 153 154 155
			rc = -EIO;
			goto out;
		}

156
		dsisr = cxl_p2n_read(afu, CXL_PSL_DSISR_An);
157
		pr_devel_ratelimited("PSL purging... PSL_CNTL: 0x%016llx  PSL_DSISR: 0x%016llx\n", PSL_CNTL, dsisr);
158 159
		if (dsisr & CXL_PSL_DSISR_TRANS) {
			dar = cxl_p2n_read(afu, CXL_PSL_DAR_An);
160
			dev_notice(&afu->dev, "PSL purge terminating pending translation, DSISR: 0x%016llx, DAR: 0x%016llx\n", dsisr, dar);
161 162
			cxl_p2n_write(afu, CXL_PSL_TFC_An, CXL_PSL_TFC_An_AE);
		} else if (dsisr) {
163
			dev_notice(&afu->dev, "PSL purge acknowledging pending non-translation fault, DSISR: 0x%016llx\n", dsisr);
164 165 166 167 168
			cxl_p2n_write(afu, CXL_PSL_TFC_An, CXL_PSL_TFC_An_A);
		} else {
			cpu_relax();
		}
		PSL_CNTL = cxl_p1n_read(afu, CXL_PSL_SCNTL_An);
A
Andrew Donnellan 已提交
169
	}
170 171 172 173 174
	end = local_clock();
	pr_devel("PSL purged in %lld ns\n", end - start);

	cxl_p1n_write(afu, CXL_PSL_SCNTL_An,
		       PSL_CNTL & ~CXL_PSL_SCNTL_An_Pc);
I
Ian Munsie 已提交
175 176 177
out:
	trace_cxl_psl_ctrl_done(afu, CXL_PSL_SCNTL_An_Pc, rc);
	return rc;
178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197
}

static int spa_max_procs(int spa_size)
{
	/*
	 * From the CAIA:
	 *    end_of_SPA_area = SPA_Base + ((n+4) * 128) + (( ((n*8) + 127) >> 7) * 128) + 255
	 * Most of that junk is really just an overly-complicated way of saying
	 * the last 256 bytes are __aligned(128), so it's really:
	 *    end_of_SPA_area = end_of_PSL_queue_area + __aligned(128) 255
	 * and
	 *    end_of_PSL_queue_area = SPA_Base + ((n+4) * 128) + (n*8) - 1
	 * so
	 *    sizeof(SPA) = ((n+4) * 128) + (n*8) + __aligned(128) 256
	 * Ignore the alignment (which is safe in this case as long as we are
	 * careful with our rounding) and solve for n:
	 */
	return ((spa_size / 8) - 96) / 17;
}

198
int cxl_alloc_spa(struct cxl_afu *afu)
199
{
200 201
	unsigned spa_size;

202
	/* Work out how many pages to allocate */
203
	afu->native->spa_order = -1;
204
	do {
205
		afu->native->spa_order++;
206 207 208 209 210 211 212 213 214 215
		spa_size = (1 << afu->native->spa_order) * PAGE_SIZE;

		if (spa_size > 0x100000) {
			dev_warn(&afu->dev, "num_of_processes too large for the SPA, limiting to %i (0x%x)\n",
					afu->native->spa_max_procs, afu->native->spa_size);
			afu->num_procs = afu->native->spa_max_procs;
			break;
		}

		afu->native->spa_size = spa_size;
216 217
		afu->native->spa_max_procs = spa_max_procs(afu->native->spa_size);
	} while (afu->native->spa_max_procs < afu->num_procs);
218

219 220
	if (!(afu->native->spa = (struct cxl_process_element *)
	      __get_free_pages(GFP_KERNEL | __GFP_ZERO, afu->native->spa_order))) {
221 222 223 224
		pr_err("cxl_alloc_spa: Unable to allocate scheduled process area\n");
		return -ENOMEM;
	}
	pr_devel("spa pages: %i afu->spa_max_procs: %i   afu->num_procs: %i\n",
225
		 1<<afu->native->spa_order, afu->native->spa_max_procs, afu->num_procs);
226

227 228 229 230 231 232 233
	return 0;
}

static void attach_spa(struct cxl_afu *afu)
{
	u64 spap;

234 235
	afu->native->sw_command_status = (__be64 *)((char *)afu->native->spa +
					    ((afu->native->spa_max_procs + 3) * 128));
236

237 238
	spap = virt_to_phys(afu->native->spa) & CXL_PSL_SPAP_Addr;
	spap |= ((afu->native->spa_size >> (12 - CXL_PSL_SPAP_Size_Shift)) - 1) & CXL_PSL_SPAP_Size;
239
	spap |= CXL_PSL_SPAP_V;
240 241 242
	pr_devel("cxl: SPA allocated at 0x%p. Max processes: %i, sw_command_status: 0x%p CXL_PSL_SPAP_An=0x%016llx\n",
		afu->native->spa, afu->native->spa_max_procs,
		afu->native->sw_command_status, spap);
243 244 245
	cxl_p1n_write(afu, CXL_PSL_SPAP_An, spap);
}

246
static inline void detach_spa(struct cxl_afu *afu)
247
{
248
	cxl_p1n_write(afu, CXL_PSL_SPAP_An, 0);
249 250 251 252
}

void cxl_release_spa(struct cxl_afu *afu)
{
253 254 255 256
	if (afu->native->spa) {
		free_pages((unsigned long) afu->native->spa,
			afu->native->spa_order);
		afu->native->spa = NULL;
257
	}
258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273
}

int cxl_tlb_slb_invalidate(struct cxl *adapter)
{
	unsigned long timeout = jiffies + (HZ * CXL_TIMEOUT);

	pr_devel("CXL adapter wide TLBIA & SLBIA\n");

	cxl_p1_write(adapter, CXL_PSL_AFUSEL, CXL_PSL_AFUSEL_A);

	cxl_p1_write(adapter, CXL_PSL_TLBIA, CXL_TLB_SLB_IQ_ALL);
	while (cxl_p1_read(adapter, CXL_PSL_TLBIA) & CXL_TLB_SLB_P) {
		if (time_after_eq(jiffies, timeout)) {
			dev_warn(&adapter->dev, "WARNING: CXL adapter wide TLBIA timed out!\n");
			return -EBUSY;
		}
274
		if (!cxl_ops->link_ok(adapter, NULL))
275
			return -EIO;
276 277 278 279 280 281 282 283 284
		cpu_relax();
	}

	cxl_p1_write(adapter, CXL_PSL_SLBIA, CXL_TLB_SLB_IQ_ALL);
	while (cxl_p1_read(adapter, CXL_PSL_SLBIA) & CXL_TLB_SLB_P) {
		if (time_after_eq(jiffies, timeout)) {
			dev_warn(&adapter->dev, "WARNING: CXL adapter wide SLBIA timed out!\n");
			return -EBUSY;
		}
285
		if (!cxl_ops->link_ok(adapter, NULL))
286
			return -EIO;
287 288 289 290 291
		cpu_relax();
	}
	return 0;
}

292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322
int cxl_data_cache_flush(struct cxl *adapter)
{
	u64 reg;
	unsigned long timeout = jiffies + (HZ * CXL_TIMEOUT);

	pr_devel("Flushing data cache\n");

	reg = cxl_p1_read(adapter, CXL_PSL_Control);
	reg |= CXL_PSL_Control_Fr;
	cxl_p1_write(adapter, CXL_PSL_Control, reg);

	reg = cxl_p1_read(adapter, CXL_PSL_Control);
	while ((reg & CXL_PSL_Control_Fs_MASK) != CXL_PSL_Control_Fs_Complete) {
		if (time_after_eq(jiffies, timeout)) {
			dev_warn(&adapter->dev, "WARNING: cache flush timed out!\n");
			return -EBUSY;
		}

		if (!cxl_ops->link_ok(adapter, NULL)) {
			dev_warn(&adapter->dev, "WARNING: link down when flushing cache\n");
			return -EIO;
		}
		cpu_relax();
		reg = cxl_p1_read(adapter, CXL_PSL_Control);
	}

	reg &= ~CXL_PSL_Control_Fr;
	cxl_p1_write(adapter, CXL_PSL_Control, reg);
	return 0;
}

323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348
static int cxl_write_sstp(struct cxl_afu *afu, u64 sstp0, u64 sstp1)
{
	int rc;

	/* 1. Disable SSTP by writing 0 to SSTP1[V] */
	cxl_p2n_write(afu, CXL_SSTP1_An, 0);

	/* 2. Invalidate all SLB entries */
	if ((rc = cxl_afu_slbia(afu)))
		return rc;

	/* 3. Set SSTP0_An */
	cxl_p2n_write(afu, CXL_SSTP0_An, sstp0);

	/* 4. Set SSTP1_An */
	cxl_p2n_write(afu, CXL_SSTP1_An, sstp1);

	return 0;
}

/* Using per slice version may improve performance here. (ie. SLBIA_An) */
static void slb_invalid(struct cxl_context *ctx)
{
	struct cxl *adapter = ctx->afu->adapter;
	u64 slbia;

349
	WARN_ON(!mutex_is_locked(&ctx->afu->native->spa_mutex));
350 351 352 353 354 355 356

	cxl_p1_write(adapter, CXL_PSL_LBISEL,
			((u64)be32_to_cpu(ctx->elem->common.pid) << 32) |
			be32_to_cpu(ctx->elem->lpid));
	cxl_p1_write(adapter, CXL_PSL_SLBIA, CXL_TLB_SLB_IQ_LPIDPID);

	while (1) {
357
		if (!cxl_ops->link_ok(adapter, NULL))
358
			break;
359 360 361 362 363 364 365 366 367 368 369
		slbia = cxl_p1_read(adapter, CXL_PSL_SLBIA);
		if (!(slbia & CXL_TLB_SLB_P))
			break;
		cpu_relax();
	}
}

static int do_process_element_cmd(struct cxl_context *ctx,
				  u64 cmd, u64 pe_state)
{
	u64 state;
370
	unsigned long timeout = jiffies + (HZ * CXL_TIMEOUT);
I
Ian Munsie 已提交
371 372 373
	int rc = 0;

	trace_cxl_llcmd(ctx, cmd);
374 375 376 377 378

	WARN_ON(!ctx->afu->enabled);

	ctx->elem->software_state = cpu_to_be32(pe_state);
	smp_wmb();
379
	*(ctx->afu->native->sw_command_status) = cpu_to_be64(cmd | 0 | ctx->pe);
380 381 382
	smp_mb();
	cxl_p1n_write(ctx->afu, CXL_PSL_LLCMD_An, cmd | ctx->pe);
	while (1) {
383 384
		if (time_after_eq(jiffies, timeout)) {
			dev_warn(&ctx->afu->dev, "WARNING: Process Element Command timed out!\n");
I
Ian Munsie 已提交
385 386
			rc = -EBUSY;
			goto out;
387
		}
388
		if (!cxl_ops->link_ok(ctx->afu->adapter, ctx->afu)) {
389 390 391 392
			dev_warn(&ctx->afu->dev, "WARNING: Device link down, aborting Process Element Command!\n");
			rc = -EIO;
			goto out;
		}
393
		state = be64_to_cpup(ctx->afu->native->sw_command_status);
394 395
		if (state == ~0ULL) {
			pr_err("cxl: Error adding process element to AFU\n");
I
Ian Munsie 已提交
396 397
			rc = -1;
			goto out;
398 399 400 401 402 403 404 405 406 407 408 409 410 411
		}
		if ((state & (CXL_SPA_SW_CMD_MASK | CXL_SPA_SW_STATE_MASK  | CXL_SPA_SW_LINK_MASK)) ==
		    (cmd | (cmd >> 16) | ctx->pe))
			break;
		/*
		 * The command won't finish in the PSL if there are
		 * outstanding DSIs.  Hence we need to yield here in
		 * case there are outstanding DSIs that we need to
		 * service.  Tuning possiblity: we could wait for a
		 * while before sched
		 */
		schedule();

	}
I
Ian Munsie 已提交
412 413 414
out:
	trace_cxl_llcmd_done(ctx, cmd, rc);
	return rc;
415 416 417 418 419 420
}

static int add_process_element(struct cxl_context *ctx)
{
	int rc = 0;

421
	mutex_lock(&ctx->afu->native->spa_mutex);
422 423 424 425
	pr_devel("%s Adding pe: %i started\n", __func__, ctx->pe);
	if (!(rc = do_process_element_cmd(ctx, CXL_SPA_SW_CMD_ADD, CXL_PE_SOFTWARE_STATE_V)))
		ctx->pe_inserted = true;
	pr_devel("%s Adding pe: %i finished\n", __func__, ctx->pe);
426
	mutex_unlock(&ctx->afu->native->spa_mutex);
427 428 429 430 431 432 433 434 435 436 437
	return rc;
}

static int terminate_process_element(struct cxl_context *ctx)
{
	int rc = 0;

	/* fast path terminate if it's already invalid */
	if (!(ctx->elem->software_state & cpu_to_be32(CXL_PE_SOFTWARE_STATE_V)))
		return rc;

438
	mutex_lock(&ctx->afu->native->spa_mutex);
439
	pr_devel("%s Terminate pe: %i started\n", __func__, ctx->pe);
440 441 442 443
	/* We could be asked to terminate when the hw is down. That
	 * should always succeed: it's not running if the hw has gone
	 * away and is being reset.
	 */
444
	if (cxl_ops->link_ok(ctx->afu->adapter, ctx->afu))
445 446
		rc = do_process_element_cmd(ctx, CXL_SPA_SW_CMD_TERMINATE,
					    CXL_PE_SOFTWARE_STATE_V | CXL_PE_SOFTWARE_STATE_T);
447 448
	ctx->elem->software_state = 0;	/* Remove Valid bit */
	pr_devel("%s Terminate pe: %i finished\n", __func__, ctx->pe);
449
	mutex_unlock(&ctx->afu->native->spa_mutex);
450 451 452 453 454 455 456
	return rc;
}

static int remove_process_element(struct cxl_context *ctx)
{
	int rc = 0;

457
	mutex_lock(&ctx->afu->native->spa_mutex);
458
	pr_devel("%s Remove pe: %i started\n", __func__, ctx->pe);
459 460 461 462

	/* We could be asked to remove when the hw is down. Again, if
	 * the hw is down, the PE is gone, so we succeed.
	 */
463
	if (cxl_ops->link_ok(ctx->afu->adapter, ctx->afu))
464 465 466
		rc = do_process_element_cmd(ctx, CXL_SPA_SW_CMD_REMOVE, 0);

	if (!rc)
467 468 469
		ctx->pe_inserted = false;
	slb_invalid(ctx);
	pr_devel("%s Remove pe: %i finished\n", __func__, ctx->pe);
470
	mutex_unlock(&ctx->afu->native->spa_mutex);
471 472 473 474

	return rc;
}

M
Michael Neuling 已提交
475
void cxl_assign_psn_space(struct cxl_context *ctx)
476 477 478 479 480 481
{
	if (!ctx->afu->pp_size || ctx->master) {
		ctx->psn_phys = ctx->afu->psn_phys;
		ctx->psn_size = ctx->afu->adapter->ps_size;
	} else {
		ctx->psn_phys = ctx->afu->psn_phys +
482
			(ctx->afu->native->pp_offset + ctx->afu->pp_size * ctx->pe);
483 484 485 486 487 488 489 490 491 492
		ctx->psn_size = ctx->afu->pp_size;
	}
}

static int activate_afu_directed(struct cxl_afu *afu)
{
	int rc;

	dev_info(&afu->dev, "Activating AFU directed mode\n");

493
	afu->num_procs = afu->max_procs_virtualised;
494
	if (afu->native->spa == NULL) {
495 496 497 498
		if (cxl_alloc_spa(afu))
			return -ENOMEM;
	}
	attach_spa(afu);
499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528

	cxl_p1n_write(afu, CXL_PSL_SCNTL_An, CXL_PSL_SCNTL_An_PM_AFU);
	cxl_p1n_write(afu, CXL_PSL_AMOR_An, 0xFFFFFFFFFFFFFFFFULL);
	cxl_p1n_write(afu, CXL_PSL_ID_An, CXL_PSL_ID_An_F | CXL_PSL_ID_An_L);

	afu->current_mode = CXL_MODE_DIRECTED;

	if ((rc = cxl_chardev_m_afu_add(afu)))
		return rc;

	if ((rc = cxl_sysfs_afu_m_add(afu)))
		goto err;

	if ((rc = cxl_chardev_s_afu_add(afu)))
		goto err1;

	return 0;
err1:
	cxl_sysfs_afu_m_remove(afu);
err:
	cxl_chardev_afu_remove(afu);
	return rc;
}

#ifdef CONFIG_CPU_LITTLE_ENDIAN
#define set_endian(sr) ((sr) |= CXL_PSL_SR_An_LE)
#else
#define set_endian(sr) ((sr) &= ~(CXL_PSL_SR_An_LE))
#endif

529 530 531 532
static u64 calculate_sr(struct cxl_context *ctx)
{
	u64 sr = 0;

533
	set_endian(sr);
534 535 536 537 538
	if (ctx->master)
		sr |= CXL_PSL_SR_An_MP;
	if (mfspr(SPRN_LPCR) & LPCR_TC)
		sr |= CXL_PSL_SR_An_TC;
	if (ctx->kernel) {
539 540 541
		if (!ctx->real_mode)
			sr |= CXL_PSL_SR_An_R;
		sr |= (mfmsr() & MSR_SF) | CXL_PSL_SR_An_HV;
542 543 544 545 546 547 548 549 550
	} else {
		sr |= CXL_PSL_SR_An_PR | CXL_PSL_SR_An_R;
		sr &= ~(CXL_PSL_SR_An_HV);
		if (!test_tsk_thread_flag(current, TIF_32BIT))
			sr |= CXL_PSL_SR_An_SF;
	}
	return sr;
}

551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579
static void update_ivtes_directed(struct cxl_context *ctx)
{
	bool need_update = (ctx->status == STARTED);
	int r;

	if (need_update) {
		WARN_ON(terminate_process_element(ctx));
		WARN_ON(remove_process_element(ctx));
	}

	for (r = 0; r < CXL_IRQ_RANGES; r++) {
		ctx->elem->ivte_offsets[r] = cpu_to_be16(ctx->irqs.offset[r]);
		ctx->elem->ivte_ranges[r] = cpu_to_be16(ctx->irqs.range[r]);
	}

	/*
	 * Theoretically we could use the update llcmd, instead of a
	 * terminate/remove/add (or if an atomic update was required we could
	 * do a suspend/update/resume), however it seems there might be issues
	 * with the update llcmd on some cards (including those using an XSL on
	 * an ASIC) so for now it's safest to go with the commands that are
	 * known to work. In the future if we come across a situation where the
	 * card may be performing transactions using the same PE while we are
	 * doing this update we might need to revisit this.
	 */
	if (need_update)
		WARN_ON(add_process_element(ctx));
}

580 581
static int attach_afu_directed(struct cxl_context *ctx, u64 wed, u64 amr)
{
582
	u32 pid;
583
	int result;
584

M
Michael Neuling 已提交
585
	cxl_assign_psn_space(ctx);
586 587 588 589 590 591

	ctx->elem->ctxtime = 0; /* disable */
	ctx->elem->lpid = cpu_to_be32(mfspr(SPRN_LPID));
	ctx->elem->haurp = 0; /* disable */
	ctx->elem->sdr = cpu_to_be64(mfspr(SPRN_SDR1));

592 593 594
	pid = current->pid;
	if (ctx->kernel)
		pid = 0;
595
	ctx->elem->common.tid = 0;
596 597 598
	ctx->elem->common.pid = cpu_to_be32(pid);

	ctx->elem->sr = cpu_to_be64(calculate_sr(ctx));
599 600 601 602 603 604 605 606 607 608

	ctx->elem->common.csrp = 0; /* disable */
	ctx->elem->common.aurp0 = 0; /* disable */
	ctx->elem->common.aurp1 = 0; /* disable */

	cxl_prefault(ctx, wed);

	ctx->elem->common.sstp0 = cpu_to_be64(ctx->sstp0);
	ctx->elem->common.sstp1 = cpu_to_be64(ctx->sstp1);

609 610 611 612 613 614 615 616 617
	/*
	 * Ensure we have the multiplexed PSL interrupt set up to take faults
	 * for kernel contexts that may not have allocated any AFU IRQs at all:
	 */
	if (ctx->irqs.range[0] == 0) {
		ctx->irqs.offset[0] = ctx->afu->native->psl_hwirq;
		ctx->irqs.range[0] = 1;
	}

618
	update_ivtes_directed(ctx);
619 620 621 622 623

	ctx->elem->common.amr = cpu_to_be64(amr);
	ctx->elem->common.wed = cpu_to_be64(wed);

	/* first guy needs to enable */
624
	if ((result = cxl_ops->afu_check_and_enable(ctx->afu)))
625 626
		return result;

627
	return add_process_element(ctx);
628 629 630 631 632 633 634 635 636 637 638 639
}

static int deactivate_afu_directed(struct cxl_afu *afu)
{
	dev_info(&afu->dev, "Deactivating AFU directed mode\n");

	afu->current_mode = 0;
	afu->num_procs = 0;

	cxl_sysfs_afu_m_remove(afu);
	cxl_chardev_afu_remove(afu);

640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666
	/*
	 * The CAIA section 2.2.1 indicates that the procedure for starting and
	 * stopping an AFU in AFU directed mode is AFU specific, which is not
	 * ideal since this code is generic and with one exception has no
	 * knowledge of the AFU. This is in contrast to the procedure for
	 * disabling a dedicated process AFU, which is documented to just
	 * require a reset. The architecture does indicate that both an AFU
	 * reset and an AFU disable should result in the AFU being disabled and
	 * we do both followed by a PSL purge for safety.
	 *
	 * Notably we used to have some issues with the disable sequence on PSL
	 * cards, which is why we ended up using this heavy weight procedure in
	 * the first place, however a bug was discovered that had rendered the
	 * disable operation ineffective, so it is conceivable that was the
	 * sole explanation for those difficulties. Careful regression testing
	 * is recommended if anyone attempts to remove or reorder these
	 * operations.
	 *
	 * The XSL on the Mellanox CX4 behaves a little differently from the
	 * PSL based cards and will time out an AFU reset if the AFU is still
	 * enabled. That card is special in that we do have a means to identify
	 * it from this code, so in that case we skip the reset and just use a
	 * disable/purge to avoid the timeout and corresponding noise in the
	 * kernel log.
	 */
	if (afu->adapter->native->sl_ops->needs_reset_before_disable)
		cxl_ops->afu_reset(afu);
667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695
	cxl_afu_disable(afu);
	cxl_psl_purge(afu);

	return 0;
}

static int activate_dedicated_process(struct cxl_afu *afu)
{
	dev_info(&afu->dev, "Activating dedicated process mode\n");

	cxl_p1n_write(afu, CXL_PSL_SCNTL_An, CXL_PSL_SCNTL_An_PM_Process);

	cxl_p1n_write(afu, CXL_PSL_CtxTime_An, 0); /* disable */
	cxl_p1n_write(afu, CXL_PSL_SPAP_An, 0);    /* disable */
	cxl_p1n_write(afu, CXL_PSL_AMOR_An, 0xFFFFFFFFFFFFFFFFULL);
	cxl_p1n_write(afu, CXL_PSL_LPID_An, mfspr(SPRN_LPID));
	cxl_p1n_write(afu, CXL_HAURP_An, 0);       /* disable */
	cxl_p1n_write(afu, CXL_PSL_SDR_An, mfspr(SPRN_SDR1));

	cxl_p2n_write(afu, CXL_CSRP_An, 0);        /* disable */
	cxl_p2n_write(afu, CXL_AURP0_An, 0);       /* disable */
	cxl_p2n_write(afu, CXL_AURP1_An, 0);       /* disable */

	afu->current_mode = CXL_MODE_DEDICATED;
	afu->num_procs = 1;

	return cxl_chardev_d_afu_add(afu);
}

696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711
static void update_ivtes_dedicated(struct cxl_context *ctx)
{
	struct cxl_afu *afu = ctx->afu;

	cxl_p1n_write(afu, CXL_PSL_IVTE_Offset_An,
		       (((u64)ctx->irqs.offset[0] & 0xffff) << 48) |
		       (((u64)ctx->irqs.offset[1] & 0xffff) << 32) |
		       (((u64)ctx->irqs.offset[2] & 0xffff) << 16) |
			((u64)ctx->irqs.offset[3] & 0xffff));
	cxl_p1n_write(afu, CXL_PSL_IVTE_Limit_An, (u64)
		       (((u64)ctx->irqs.range[0] & 0xffff) << 48) |
		       (((u64)ctx->irqs.range[1] & 0xffff) << 32) |
		       (((u64)ctx->irqs.range[2] & 0xffff) << 16) |
			((u64)ctx->irqs.range[3] & 0xffff));
}

712 713 714
static int attach_dedicated(struct cxl_context *ctx, u64 wed, u64 amr)
{
	struct cxl_afu *afu = ctx->afu;
715
	u64 pid;
716 717
	int rc;

718 719 720 721 722 723
	pid = (u64)current->pid << 32;
	if (ctx->kernel)
		pid = 0;
	cxl_p2n_write(afu, CXL_PSL_PID_TID_An, pid);

	cxl_p1n_write(afu, CXL_PSL_SR_An, calculate_sr(ctx));
724 725 726 727 728 729

	if ((rc = cxl_write_sstp(afu, ctx->sstp0, ctx->sstp1)))
		return rc;

	cxl_prefault(ctx, wed);

730
	update_ivtes_dedicated(ctx);
731 732 733 734

	cxl_p2n_write(afu, CXL_PSL_AMR_An, amr);

	/* master only context for dedicated */
M
Michael Neuling 已提交
735
	cxl_assign_psn_space(ctx);
736

737
	if ((rc = cxl_ops->afu_reset(afu)))
738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756
		return rc;

	cxl_p2n_write(afu, CXL_PSL_WED_An, wed);

	return afu_enable(afu);
}

static int deactivate_dedicated_process(struct cxl_afu *afu)
{
	dev_info(&afu->dev, "Deactivating dedicated process mode\n");

	afu->current_mode = 0;
	afu->num_procs = 0;

	cxl_chardev_afu_remove(afu);

	return 0;
}

757
static int native_afu_deactivate_mode(struct cxl_afu *afu, int mode)
758 759 760 761 762 763 764 765
{
	if (mode == CXL_MODE_DIRECTED)
		return deactivate_afu_directed(afu);
	if (mode == CXL_MODE_DEDICATED)
		return deactivate_dedicated_process(afu);
	return 0;
}

766
static int native_afu_activate_mode(struct cxl_afu *afu, int mode)
767 768 769 770 771 772
{
	if (!mode)
		return 0;
	if (!(mode & afu->modes_supported))
		return -EINVAL;

773
	if (!cxl_ops->link_ok(afu->adapter, afu)) {
774 775 776 777
		WARN(1, "Device link is down, refusing to activate!\n");
		return -EIO;
	}

778 779 780 781 782 783 784 785
	if (mode == CXL_MODE_DIRECTED)
		return activate_afu_directed(afu);
	if (mode == CXL_MODE_DEDICATED)
		return activate_dedicated_process(afu);

	return -EINVAL;
}

786 787
static int native_attach_process(struct cxl_context *ctx, bool kernel,
				u64 wed, u64 amr)
788
{
789
	if (!cxl_ops->link_ok(ctx->afu->adapter, ctx->afu)) {
790 791 792 793
		WARN(1, "Device link is down, refusing to attach process!\n");
		return -EIO;
	}

794 795 796 797 798 799 800 801 802 803 804 805
	ctx->kernel = kernel;
	if (ctx->afu->current_mode == CXL_MODE_DIRECTED)
		return attach_afu_directed(ctx, wed, amr);

	if (ctx->afu->current_mode == CXL_MODE_DEDICATED)
		return attach_dedicated(ctx, wed, amr);

	return -EINVAL;
}

static inline int detach_process_native_dedicated(struct cxl_context *ctx)
{
806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821
	/*
	 * The CAIA section 2.1.1 indicates that we need to do an AFU reset to
	 * stop the AFU in dedicated mode (we therefore do not make that
	 * optional like we do in the afu directed path). It does not indicate
	 * that we need to do an explicit disable (which should occur
	 * implicitly as part of the reset) or purge, but we do these as well
	 * to be on the safe side.
	 *
	 * Notably we used to have some issues with the disable sequence
	 * (before the sequence was spelled out in the architecture) which is
	 * why we were so heavy weight in the first place, however a bug was
	 * discovered that had rendered the disable operation ineffective, so
	 * it is conceivable that was the sole explanation for those
	 * difficulties. Point is, we should be careful and do some regression
	 * testing if we ever attempt to remove any part of this procedure.
	 */
822
	cxl_ops->afu_reset(ctx->afu);
823 824 825 826 827
	cxl_afu_disable(ctx->afu);
	cxl_psl_purge(ctx->afu);
	return 0;
}

828 829 830 831 832 833 834 835 836
static void native_update_ivtes(struct cxl_context *ctx)
{
	if (ctx->afu->current_mode == CXL_MODE_DIRECTED)
		return update_ivtes_directed(ctx);
	if (ctx->afu->current_mode == CXL_MODE_DEDICATED)
		return update_ivtes_dedicated(ctx);
	WARN(1, "native_update_ivtes: Bad mode\n");
}

837 838 839 840 841 842 843 844 845 846 847 848
static inline int detach_process_native_afu_directed(struct cxl_context *ctx)
{
	if (!ctx->pe_inserted)
		return 0;
	if (terminate_process_element(ctx))
		return -1;
	if (remove_process_element(ctx))
		return -1;

	return 0;
}

849
static int native_detach_process(struct cxl_context *ctx)
850
{
I
Ian Munsie 已提交
851 852
	trace_cxl_detach(ctx);

853 854 855 856 857 858
	if (ctx->afu->current_mode == CXL_MODE_DEDICATED)
		return detach_process_native_dedicated(ctx);

	return detach_process_native_afu_directed(ctx);
}

859
static int native_get_irq_info(struct cxl_afu *afu, struct cxl_irq_info *info)
860 861 862
{
	u64 pidtid;

863 864 865
	/* If the adapter has gone away, we can't get any meaningful
	 * information.
	 */
866
	if (!cxl_ops->link_ok(afu->adapter, afu))
867 868
		return -EIO;

869 870 871 872
	info->dsisr = cxl_p2n_read(afu, CXL_PSL_DSISR_An);
	info->dar = cxl_p2n_read(afu, CXL_PSL_DAR_An);
	info->dsr = cxl_p2n_read(afu, CXL_PSL_DSR_An);
	pidtid = cxl_p2n_read(afu, CXL_PSL_PID_TID_An);
873 874
	info->pid = pidtid >> 32;
	info->tid = pidtid & 0xffffffff;
875 876
	info->afu_err = cxl_p2n_read(afu, CXL_AFU_ERR_An);
	info->errstat = cxl_p2n_read(afu, CXL_PSL_ErrStat_An);
877
	info->proc_handle = 0;
878 879 880 881

	return 0;
}

882
void cxl_native_psl_irq_dump_regs(struct cxl_context *ctx)
883 884 885 886 887 888 889 890 891 892
{
	u64 fir1, fir2, fir_slice, serr, afu_debug;

	fir1 = cxl_p1_read(ctx->afu->adapter, CXL_PSL_FIR1);
	fir2 = cxl_p1_read(ctx->afu->adapter, CXL_PSL_FIR2);
	fir_slice = cxl_p1n_read(ctx->afu, CXL_PSL_FIR_SLICE_An);
	afu_debug = cxl_p1n_read(ctx->afu, CXL_AFU_DEBUG_An);

	dev_crit(&ctx->afu->dev, "PSL_FIR1: 0x%016llx\n", fir1);
	dev_crit(&ctx->afu->dev, "PSL_FIR2: 0x%016llx\n", fir2);
893 894
	if (ctx->afu->adapter->native->sl_ops->register_serr_irq) {
		serr = cxl_p1n_read(ctx->afu, CXL_PSL_SERR_An);
895
		cxl_afu_decode_psl_serr(ctx->afu, serr);
896
	}
897 898
	dev_crit(&ctx->afu->dev, "PSL_FIR_SLICE_An: 0x%016llx\n", fir_slice);
	dev_crit(&ctx->afu->dev, "CXL_PSL_AFU_DEBUG_An: 0x%016llx\n", afu_debug);
899 900 901 902 903 904 905
}

static irqreturn_t native_handle_psl_slice_error(struct cxl_context *ctx,
						u64 dsisr, u64 errstat)
{

	dev_crit(&ctx->afu->dev, "PSL ERROR STATUS: 0x%016llx\n", errstat);
906

907 908 909 910 911 912 913
	if (ctx->afu->adapter->native->sl_ops->psl_irq_dump_registers)
		ctx->afu->adapter->native->sl_ops->psl_irq_dump_registers(ctx);

	if (ctx->afu->adapter->native->sl_ops->debugfs_stop_trace) {
		dev_crit(&ctx->afu->dev, "STOPPING CXL TRACE\n");
		ctx->afu->adapter->native->sl_ops->debugfs_stop_trace(ctx->afu->adapter);
	}
914

915
	return cxl_ops->ack_irq(ctx, 0, errstat);
916 917 918 919 920 921 922 923 924 925 926 927
}

static irqreturn_t fail_psl_irq(struct cxl_afu *afu, struct cxl_irq_info *irq_info)
{
	if (irq_info->dsisr & CXL_PSL_DSISR_TRANS)
		cxl_p2n_write(afu, CXL_PSL_TFC_An, CXL_PSL_TFC_An_AE);
	else
		cxl_p2n_write(afu, CXL_PSL_TFC_An, CXL_PSL_TFC_An_A);

	return IRQ_HANDLED;
}

928
static irqreturn_t native_irq_multiplexed(int irq, void *data)
929 930 931 932
{
	struct cxl_afu *afu = data;
	struct cxl_context *ctx;
	struct cxl_irq_info irq_info;
933 934 935 936 937 938 939 940 941 942 943 944
	u64 phreg = cxl_p2n_read(afu, CXL_PSL_PEHandle_An);
	int ph, ret;

	/* check if eeh kicked in while the interrupt was in flight */
	if (unlikely(phreg == ~0ULL)) {
		dev_warn(&afu->dev,
			 "Ignoring slice interrupt(%d) due to fenced card",
			 irq);
		return IRQ_HANDLED;
	}
	/* Mask the pe-handle from register value */
	ph = phreg & 0xffff;
945
	if ((ret = native_get_irq_info(afu, &irq_info))) {
946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965
		WARN(1, "Unable to get CXL IRQ Info: %i\n", ret);
		return fail_psl_irq(afu, &irq_info);
	}

	rcu_read_lock();
	ctx = idr_find(&afu->contexts_idr, ph);
	if (ctx) {
		ret = cxl_irq(irq, ctx, &irq_info);
		rcu_read_unlock();
		return ret;
	}
	rcu_read_unlock();

	WARN(1, "Unable to demultiplex CXL PSL IRQ for PE %i DSISR %016llx DAR"
		" %016llx\n(Possible AFU HW issue - was a term/remove acked"
		" with outstanding transactions?)\n", ph, irq_info.dsisr,
		irq_info.dar);
	return fail_psl_irq(afu, &irq_info);
}

A
Andrew Donnellan 已提交
966
static void native_irq_wait(struct cxl_context *ctx)
967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994
{
	u64 dsisr;
	int timeout = 1000;
	int ph;

	/*
	 * Wait until no further interrupts are presented by the PSL
	 * for this context.
	 */
	while (timeout--) {
		ph = cxl_p2n_read(ctx->afu, CXL_PSL_PEHandle_An) & 0xffff;
		if (ph != ctx->pe)
			return;
		dsisr = cxl_p2n_read(ctx->afu, CXL_PSL_DSISR_An);
		if ((dsisr & CXL_PSL_DSISR_PENDING) == 0)
			return;
		/*
		 * We are waiting for the workqueue to process our
		 * irq, so need to let that run here.
		 */
		msleep(1);
	}

	dev_warn(&ctx->afu->dev, "WARNING: waiting on DSI for PE %i"
		 " DSISR %016llx!\n", ph, dsisr);
	return;
}

995
static irqreturn_t native_slice_irq_err(int irq, void *data)
996 997
{
	struct cxl_afu *afu = data;
998
	u64 fir_slice, errstat, serr, afu_debug, afu_error, dsisr;
999

1000 1001 1002
	/*
	 * slice err interrupt is only used with full PSL (no XSL)
	 */
1003 1004 1005 1006
	serr = cxl_p1n_read(afu, CXL_PSL_SERR_An);
	fir_slice = cxl_p1n_read(afu, CXL_PSL_FIR_SLICE_An);
	errstat = cxl_p2n_read(afu, CXL_PSL_ErrStat_An);
	afu_debug = cxl_p1n_read(afu, CXL_AFU_DEBUG_An);
1007 1008 1009
	afu_error = cxl_p2n_read(afu, CXL_AFU_ERR_An);
	dsisr = cxl_p2n_read(afu, CXL_PSL_DSISR_An);
	cxl_afu_decode_psl_serr(afu, serr);
1010 1011 1012
	dev_crit(&afu->dev, "PSL_FIR_SLICE_An: 0x%016llx\n", fir_slice);
	dev_crit(&afu->dev, "CXL_PSL_ErrStat_An: 0x%016llx\n", errstat);
	dev_crit(&afu->dev, "CXL_PSL_AFU_DEBUG_An: 0x%016llx\n", afu_debug);
1013 1014
	dev_crit(&afu->dev, "AFU_ERR_An: 0x%.16llx\n", afu_error);
	dev_crit(&afu->dev, "PSL_DSISR_An: 0x%.16llx\n", dsisr);
1015 1016 1017 1018 1019 1020

	cxl_p1n_write(afu, CXL_PSL_SERR_An, serr);

	return IRQ_HANDLED;
}

1021 1022 1023 1024 1025 1026 1027 1028 1029 1030
void cxl_native_err_irq_dump_regs(struct cxl *adapter)
{
	u64 fir1, fir2;

	fir1 = cxl_p1_read(adapter, CXL_PSL_FIR1);
	fir2 = cxl_p1_read(adapter, CXL_PSL_FIR2);

	dev_crit(&adapter->dev, "PSL_FIR1: 0x%016llx\nPSL_FIR2: 0x%016llx\n", fir1, fir2);
}

1031
static irqreturn_t native_irq_err(int irq, void *data)
1032 1033
{
	struct cxl *adapter = data;
1034
	u64 err_ivte;
1035 1036 1037 1038 1039 1040

	WARN(1, "CXL ERROR interrupt %i\n", irq);

	err_ivte = cxl_p1_read(adapter, CXL_PSL_ErrIVTE);
	dev_crit(&adapter->dev, "PSL_ErrIVTE: 0x%016llx\n", err_ivte);

1041 1042 1043 1044
	if (adapter->native->sl_ops->debugfs_stop_trace) {
		dev_crit(&adapter->dev, "STOPPING CXL TRACE\n");
		adapter->native->sl_ops->debugfs_stop_trace(adapter);
	}
1045

1046 1047
	if (adapter->native->sl_ops->err_irq_dump_registers)
		adapter->native->sl_ops->err_irq_dump_registers(adapter);
1048 1049 1050 1051

	return IRQ_HANDLED;
}

1052
int cxl_native_register_psl_err_irq(struct cxl *adapter)
1053 1054 1055 1056 1057 1058 1059 1060
{
	int rc;

	adapter->irq_name = kasprintf(GFP_KERNEL, "cxl-%s-err",
				      dev_name(&adapter->dev));
	if (!adapter->irq_name)
		return -ENOMEM;

1061
	if ((rc = cxl_register_one_irq(adapter, native_irq_err, adapter,
1062 1063
				       &adapter->native->err_hwirq,
				       &adapter->native->err_virq,
1064 1065 1066 1067 1068 1069
				       adapter->irq_name))) {
		kfree(adapter->irq_name);
		adapter->irq_name = NULL;
		return rc;
	}

1070
	cxl_p1_write(adapter, CXL_PSL_ErrIVTE, adapter->native->err_hwirq & 0xffff);
1071 1072 1073 1074

	return 0;
}

1075
void cxl_native_release_psl_err_irq(struct cxl *adapter)
1076
{
1077
	if (adapter->native->err_virq != irq_find_mapping(NULL, adapter->native->err_hwirq))
1078 1079 1080
		return;

	cxl_p1_write(adapter, CXL_PSL_ErrIVTE, 0x0000000000000000);
1081 1082
	cxl_unmap_irq(adapter->native->err_virq, adapter);
	cxl_ops->release_one_irq(adapter, adapter->native->err_hwirq);
1083 1084 1085
	kfree(adapter->irq_name);
}

1086
int cxl_native_register_serr_irq(struct cxl_afu *afu)
1087 1088 1089 1090 1091 1092 1093 1094 1095
{
	u64 serr;
	int rc;

	afu->err_irq_name = kasprintf(GFP_KERNEL, "cxl-%s-err",
				      dev_name(&afu->dev));
	if (!afu->err_irq_name)
		return -ENOMEM;

1096
	if ((rc = cxl_register_one_irq(afu->adapter, native_slice_irq_err, afu,
1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110
				       &afu->serr_hwirq,
				       &afu->serr_virq, afu->err_irq_name))) {
		kfree(afu->err_irq_name);
		afu->err_irq_name = NULL;
		return rc;
	}

	serr = cxl_p1n_read(afu, CXL_PSL_SERR_An);
	serr = (serr & 0x00ffffffffff0000ULL) | (afu->serr_hwirq & 0xffff);
	cxl_p1n_write(afu, CXL_PSL_SERR_An, serr);

	return 0;
}

1111
void cxl_native_release_serr_irq(struct cxl_afu *afu)
1112 1113 1114 1115 1116 1117
{
	if (afu->serr_virq != irq_find_mapping(NULL, afu->serr_hwirq))
		return;

	cxl_p1n_write(afu, CXL_PSL_SERR_An, 0x0000000000000000);
	cxl_unmap_irq(afu->serr_virq, afu);
1118
	cxl_ops->release_one_irq(afu->adapter, afu->serr_hwirq);
1119 1120 1121
	kfree(afu->err_irq_name);
}

1122
int cxl_native_register_psl_irq(struct cxl_afu *afu)
1123 1124 1125 1126 1127 1128 1129 1130
{
	int rc;

	afu->psl_irq_name = kasprintf(GFP_KERNEL, "cxl-%s",
				      dev_name(&afu->dev));
	if (!afu->psl_irq_name)
		return -ENOMEM;

1131 1132
	if ((rc = cxl_register_one_irq(afu->adapter, native_irq_multiplexed,
				    afu, &afu->native->psl_hwirq, &afu->native->psl_virq,
1133 1134 1135 1136 1137 1138 1139
				    afu->psl_irq_name))) {
		kfree(afu->psl_irq_name);
		afu->psl_irq_name = NULL;
	}
	return rc;
}

1140
void cxl_native_release_psl_irq(struct cxl_afu *afu)
1141
{
1142
	if (afu->native->psl_virq != irq_find_mapping(NULL, afu->native->psl_hwirq))
1143 1144
		return;

1145 1146
	cxl_unmap_irq(afu->native->psl_virq, afu);
	cxl_ops->release_one_irq(afu->adapter, afu->native->psl_hwirq);
1147 1148 1149
	kfree(afu->psl_irq_name);
}

1150 1151 1152 1153
static void recover_psl_err(struct cxl_afu *afu, u64 errstat)
{
	u64 dsisr;

1154
	pr_devel("RECOVERING FROM PSL ERROR... (0x%016llx)\n", errstat);
1155 1156 1157 1158 1159 1160 1161 1162 1163

	/* Clear PSL_DSISR[PE] */
	dsisr = cxl_p2n_read(afu, CXL_PSL_DSISR_An);
	cxl_p2n_write(afu, CXL_PSL_DSISR_An, dsisr & ~CXL_PSL_DSISR_An_PE);

	/* Write 1s to clear error status bits */
	cxl_p2n_write(afu, CXL_PSL_ErrStat_An, errstat);
}

1164
static int native_ack_irq(struct cxl_context *ctx, u64 tfc, u64 psl_reset_mask)
1165
{
I
Ian Munsie 已提交
1166
	trace_cxl_psl_irq_ack(ctx, tfc);
1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178
	if (tfc)
		cxl_p2n_write(ctx->afu, CXL_PSL_TFC_An, tfc);
	if (psl_reset_mask)
		recover_psl_err(ctx->afu, psl_reset_mask);

	return 0;
}

int cxl_check_error(struct cxl_afu *afu)
{
	return (cxl_p1n_read(afu, CXL_PSL_SCNTL_An) == ~0ULL);
}
1179

1180 1181 1182 1183 1184 1185
static bool native_support_attributes(const char *attr_name,
				      enum cxl_attrs type)
{
	return true;
}

1186
static int native_afu_cr_read64(struct cxl_afu *afu, int cr, u64 off, u64 *out)
1187
{
1188
	if (unlikely(!cxl_ops->link_ok(afu->adapter, afu)))
1189 1190 1191
		return -EIO;
	if (unlikely(off >= afu->crs_len))
		return -ERANGE;
1192
	*out = in_le64(afu->native->afu_desc_mmio + afu->crs_offset +
1193 1194
		(cr * afu->crs_len) + off);
	return 0;
1195 1196
}

1197
static int native_afu_cr_read32(struct cxl_afu *afu, int cr, u64 off, u32 *out)
1198
{
1199
	if (unlikely(!cxl_ops->link_ok(afu->adapter, afu)))
1200 1201 1202
		return -EIO;
	if (unlikely(off >= afu->crs_len))
		return -ERANGE;
1203
	*out = in_le32(afu->native->afu_desc_mmio + afu->crs_offset +
1204 1205
		(cr * afu->crs_len) + off);
	return 0;
1206 1207
}

1208
static int native_afu_cr_read16(struct cxl_afu *afu, int cr, u64 off, u16 *out)
1209 1210 1211
{
	u64 aligned_off = off & ~0x3L;
	u32 val;
1212
	int rc;
1213

1214
	rc = native_afu_cr_read32(afu, cr, aligned_off, &val);
1215 1216 1217
	if (!rc)
		*out = (val >> ((off & 0x3) * 8)) & 0xffff;
	return rc;
1218 1219
}

1220
static int native_afu_cr_read8(struct cxl_afu *afu, int cr, u64 off, u8 *out)
1221 1222 1223
{
	u64 aligned_off = off & ~0x3L;
	u32 val;
1224
	int rc;
1225

1226
	rc = native_afu_cr_read32(afu, cr, aligned_off, &val);
1227 1228 1229
	if (!rc)
		*out = (val >> ((off & 0x3) * 8)) & 0xff;
	return rc;
1230
}
1231

1232 1233
static int native_afu_cr_write32(struct cxl_afu *afu, int cr, u64 off, u32 in)
{
1234
	if (unlikely(!cxl_ops->link_ok(afu->adapter, afu)))
1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277
		return -EIO;
	if (unlikely(off >= afu->crs_len))
		return -ERANGE;
	out_le32(afu->native->afu_desc_mmio + afu->crs_offset +
		(cr * afu->crs_len) + off, in);
	return 0;
}

static int native_afu_cr_write16(struct cxl_afu *afu, int cr, u64 off, u16 in)
{
	u64 aligned_off = off & ~0x3L;
	u32 val32, mask, shift;
	int rc;

	rc = native_afu_cr_read32(afu, cr, aligned_off, &val32);
	if (rc)
		return rc;
	shift = (off & 0x3) * 8;
	WARN_ON(shift == 24);
	mask = 0xffff << shift;
	val32 = (val32 & ~mask) | (in << shift);

	rc = native_afu_cr_write32(afu, cr, aligned_off, val32);
	return rc;
}

static int native_afu_cr_write8(struct cxl_afu *afu, int cr, u64 off, u8 in)
{
	u64 aligned_off = off & ~0x3L;
	u32 val32, mask, shift;
	int rc;

	rc = native_afu_cr_read32(afu, cr, aligned_off, &val32);
	if (rc)
		return rc;
	shift = (off & 0x3) * 8;
	mask = 0xff << shift;
	val32 = (val32 & ~mask) | (in << shift);

	rc = native_afu_cr_write32(afu, cr, aligned_off, val32);
	return rc;
}

1278 1279
const struct cxl_backend_ops cxl_native_ops = {
	.module = THIS_MODULE,
1280 1281 1282 1283 1284 1285 1286
	.adapter_reset = cxl_pci_reset,
	.alloc_one_irq = cxl_pci_alloc_one_irq,
	.release_one_irq = cxl_pci_release_one_irq,
	.alloc_irq_ranges = cxl_pci_alloc_irq_ranges,
	.release_irq_ranges = cxl_pci_release_irq_ranges,
	.setup_irq = cxl_pci_setup_irq,
	.handle_psl_slice_error = native_handle_psl_slice_error,
1287
	.psl_interrupt = NULL,
1288
	.ack_irq = native_ack_irq,
1289
	.irq_wait = native_irq_wait,
1290 1291
	.attach_process = native_attach_process,
	.detach_process = native_detach_process,
1292
	.update_ivtes = native_update_ivtes,
1293
	.support_attributes = native_support_attributes,
1294
	.link_ok = cxl_adapter_link_ok,
1295 1296 1297 1298 1299 1300 1301 1302 1303 1304
	.release_afu = cxl_pci_release_afu,
	.afu_read_err_buffer = cxl_pci_afu_read_err_buffer,
	.afu_check_and_enable = native_afu_check_and_enable,
	.afu_activate_mode = native_afu_activate_mode,
	.afu_deactivate_mode = native_afu_deactivate_mode,
	.afu_reset = native_afu_reset,
	.afu_cr_read8 = native_afu_cr_read8,
	.afu_cr_read16 = native_afu_cr_read16,
	.afu_cr_read32 = native_afu_cr_read32,
	.afu_cr_read64 = native_afu_cr_read64,
1305 1306 1307 1308
	.afu_cr_write8 = native_afu_cr_write8,
	.afu_cr_write16 = native_afu_cr_write16,
	.afu_cr_write32 = native_afu_cr_write32,
	.read_adapter_vpd = cxl_pci_read_adapter_vpd,
1309
};