Files
linux-nvgpu/drivers/gpu/nvgpu/hal/mc/mc_gm20b.c
Tejal Kudav b80b2bdab8 gpu: nvgpu: Add CE interrupt handling
a. LAUNCH_ERR
    - Userspace error.
    - Triggered due to faulty launch.
    - Handle using recovery to reset CE engine and teardown the
      faulty channel.

b. An INVALID_CONFIG -
    - Triggered when LCE is mapped to floorswept PCE.
    - On iGPU, we use the default PCE 2 LCE  HW mapping.
      The default mapping can be read from NV_CE_PCE2LCE_CONFIG
      INIT value in CE refmanual.
    - NvGPU driver configures the mapping on dGPUs (currently only on
      Turing).
    - So, this interrupt can only be triggered if there is
      kernel or HW error
    - Recovery ( which is killing the context + engine reset) will
      not help resolve this error.
    - Trigger Quiesce as part of handling.

c. A MTHD_BUFFER_FAULT -
    - NvGPU driver allocates fault buffers for all TSGs or contexts,
      maps them in BAR2 VA space and writes the VA into channel
      instance block.
    - Can be triggered only due to kernel bug
    - Recovery will not help, need quiesce

d. FBUF_CRC_FAIL
    - Triggered when the CRC entry read from the method fault buffer
      does not match the computed CRC from the methods contained in
      the buffer.
    - This indicates memory corruption and is a fatal interrupt which
      at least requires the LCE to be reset before operations can
      start again, if not the entire GPU.
    - Better to quiesce on memory corruption
      CE Engine reset (via recovery) will not help.

e. FBUF_MAGIC_CHK_FAIL
    - Triggered when the MAGIC_NUM entry read from the method fault
      buf does not match NV_CE_MTHD_BUFFER_GLOBAL_HDR_MAGIC_NUM_VAL
    - This indicates memory corruption and is a fatal interrupt
    - Better to quiesce on memory corruption

f. STALLING_DEBUG
    - Only triggered with SW write for debug purposes
    - Debug interrupt, currently ignored

Move launch error handling from GP10b to GV11b HAL as -
1. LAUNCHERR_REPORT errcode METHOD_BUFFER_ACCESS_FAULT is not
   defined on Pascal
2. We do not support GP10b on dev-main ToT

JIRA NVGPU-8102

Change-Id: Idc84119bc23b5e85f3479fe62cc8720e98b627a5
Signed-off-by: Tejal Kudav <tkudav@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvgpu/+/2678893
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
2022-03-14 17:12:14 -07:00

295 lines
7.0 KiB
C

/*
* GM20B Master Control
*
* Copyright (c) 2014-2022, NVIDIA CORPORATION. All rights reserved.
*
* 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 AUTHORS OR COPYRIGHT HOLDERS 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.
*/
#include <nvgpu/timers.h>
#include <nvgpu/atomic.h>
#include <nvgpu/io.h>
#include <nvgpu/mc.h>
#include <nvgpu/cic_mon.h>
#include <nvgpu/ltc.h>
#include <nvgpu/gk20a.h>
#include <nvgpu/bug.h>
#include <nvgpu/engines.h>
#include <nvgpu/device.h>
#include <nvgpu/power_features/pg.h>
#include <nvgpu/ce.h>
#include "mc_gm20b.h"
#include <nvgpu/hw/gm20b/hw_mc_gm20b.h>
void gm20b_mc_isr_stall(struct gk20a *g)
{
u32 mc_intr_0;
u32 i;
const struct nvgpu_device *dev;
mc_intr_0 = g->ops.mc.intr_stall(g);
nvgpu_log(g, gpu_dbg_intr, "stall intr %08x", mc_intr_0);
for (i = 0U; i < g->fifo.num_engines; i++) {
dev = g->fifo.active_engines[i];
if ((mc_intr_0 & BIT32(dev->intr_id)) == 0U) {
continue;
}
/* GR Engine */
if (nvgpu_device_is_graphics(g, dev)) {
nvgpu_pg_elpg_protected_call(g,
g->ops.gr.intr.stall_isr(g));
}
/* CE Engine */
if (nvgpu_device_is_ce(g, dev)) {
nvgpu_ce_stall_isr(g, dev->inst_id, dev->pri_base);
}
}
if ((mc_intr_0 & mc_intr_pfifo_pending_f()) != 0U) {
g->ops.fifo.intr_0_isr(g);
}
#ifdef CONFIG_NVGPU_LS_PMU
if ((mc_intr_0 & mc_intr_pmu_pending_f()) != 0U) {
g->ops.pmu.pmu_isr(g);
}
#endif
if ((mc_intr_0 & mc_intr_priv_ring_pending_f()) != 0U) {
g->ops.priv_ring.isr(g);
}
if ((mc_intr_0 & mc_intr_ltc_pending_f()) != 0U) {
g->ops.mc.ltc_isr(g);
}
}
void gm20b_mc_intr_mask(struct gk20a *g)
{
nvgpu_writel(g, mc_intr_en_0_r(),
mc_intr_en_0_inta_disabled_f());
nvgpu_writel(g, mc_intr_en_1_r(),
mc_intr_en_1_inta_disabled_f());
nvgpu_writel(g, mc_intr_mask_0_r(), 0);
nvgpu_writel(g, mc_intr_mask_1_r(), 0);
}
void gm20b_mc_intr_enable(struct gk20a *g)
{
nvgpu_writel(g, mc_intr_en_1_r(),
mc_intr_en_1_inta_hardware_f());
nvgpu_writel(g, mc_intr_en_0_r(),
mc_intr_en_0_inta_hardware_f());
}
static u32 gm20b_mc_intr_pending_f(struct gk20a *g, u32 unit)
{
u32 intr_pending_f = 0;
switch (unit) {
case NVGPU_CIC_INTR_UNIT_BUS:
intr_pending_f = mc_intr_pbus_pending_f();
break;
case NVGPU_CIC_INTR_UNIT_PRIV_RING:
intr_pending_f = mc_intr_priv_ring_pending_f();
break;
case NVGPU_CIC_INTR_UNIT_FIFO:
intr_pending_f = mc_intr_pfifo_pending_f();
break;
case NVGPU_CIC_INTR_UNIT_LTC:
intr_pending_f = mc_intr_ltc_pending_f();
break;
case NVGPU_CIC_INTR_UNIT_GR:
intr_pending_f = nvgpu_gr_engine_interrupt_mask(g);
break;
case NVGPU_CIC_INTR_UNIT_PMU:
intr_pending_f = mc_intr_mask_0_pmu_enabled_f();
break;
case NVGPU_CIC_INTR_UNIT_CE:
intr_pending_f = nvgpu_ce_engine_interrupt_mask(g);
break;
default:
nvgpu_err(g, "Invalid MC interrupt unit specified !!!");
break;
}
return intr_pending_f;
}
void gm20b_mc_intr_stall_unit_config(struct gk20a *g, u32 unit, bool enable)
{
u32 unit_pending_f = gm20b_mc_intr_pending_f(g, unit);
if (enable) {
nvgpu_writel(g, mc_intr_mask_0_r(),
nvgpu_readl(g, mc_intr_mask_0_r()) |
unit_pending_f);
} else {
nvgpu_writel(g, mc_intr_mask_0_r(),
nvgpu_readl(g, mc_intr_mask_0_r()) &
~unit_pending_f);
}
}
void gm20b_mc_intr_nonstall_unit_config(struct gk20a *g, u32 unit, bool enable)
{
u32 unit_pending_f = gm20b_mc_intr_pending_f(g, unit);
if (enable) {
nvgpu_writel(g, mc_intr_mask_1_r(),
nvgpu_readl(g, mc_intr_mask_1_r()) |
unit_pending_f);
} else {
nvgpu_writel(g, mc_intr_mask_1_r(),
nvgpu_readl(g, mc_intr_mask_1_r()) &
~unit_pending_f);
}
}
void gm20b_mc_intr_stall_pause(struct gk20a *g)
{
nvgpu_writel(g, mc_intr_en_0_r(),
mc_intr_en_0_inta_disabled_f());
/* flush previous write */
(void) nvgpu_readl(g, mc_intr_en_0_r());
}
void gm20b_mc_intr_stall_resume(struct gk20a *g)
{
nvgpu_writel(g, mc_intr_en_0_r(),
mc_intr_en_0_inta_hardware_f());
/* flush previous write */
(void) nvgpu_readl(g, mc_intr_en_0_r());
}
void gm20b_mc_intr_nonstall_pause(struct gk20a *g)
{
nvgpu_writel(g, mc_intr_en_1_r(),
mc_intr_en_0_inta_disabled_f());
/* flush previous write */
(void) nvgpu_readl(g, mc_intr_en_1_r());
}
void gm20b_mc_intr_nonstall_resume(struct gk20a *g)
{
nvgpu_writel(g, mc_intr_en_1_r(),
mc_intr_en_0_inta_hardware_f());
/* flush previous write */
(void) nvgpu_readl(g, mc_intr_en_1_r());
}
u32 gm20b_mc_intr_stall(struct gk20a *g)
{
return nvgpu_readl(g, mc_intr_r(NVGPU_CIC_INTR_STALLING));
}
u32 gm20b_mc_intr_nonstall(struct gk20a *g)
{
return nvgpu_readl(g, mc_intr_r(NVGPU_CIC_INTR_NONSTALLING));
}
bool gm20b_mc_is_intr1_pending(struct gk20a *g, u32 unit, u32 mc_intr_1)
{
u32 mask;
bool is_pending;
switch (unit) {
case NVGPU_UNIT_FIFO:
mask = mc_intr_pfifo_pending_f();
break;
default:
mask = 0U;
break;
}
if (mask == 0U) {
nvgpu_err(g, "unknown unit %d", unit);
is_pending = false;
} else {
is_pending = ((mc_intr_1 & mask) != 0U);
}
return is_pending;
}
void gm20b_mc_log_pending_intrs(struct gk20a *g)
{
u32 mc_intr_0;
u32 mc_intr_1;
mc_intr_0 = g->ops.mc.intr_stall(g);
if (mc_intr_0 != 0U) {
if ((mc_intr_0 & mc_intr_priv_ring_pending_f()) != 0U) {
/* clear priv ring interrupts */
g->ops.priv_ring.isr(g);
}
mc_intr_0 = g->ops.mc.intr_stall(g);
if (mc_intr_0 != 0U) {
nvgpu_info(g, "Pending stall intr0=0x%08x", mc_intr_0);
}
}
mc_intr_1 = g->ops.mc.intr_nonstall(g);
if (mc_intr_1 != 0U) {
nvgpu_info(g, "Pending nonstall intr1=0x%08x", mc_intr_1);
}
}
void gm20b_mc_fb_reset(struct gk20a *g)
{
u32 val;
nvgpu_log_info(g, "reset gk20a fb");
val = nvgpu_readl(g, mc_elpg_enable_r());
val |= mc_elpg_enable_xbar_enabled_f()
| mc_elpg_enable_pfb_enabled_f()
| mc_elpg_enable_hub_enabled_f();
nvgpu_writel(g, mc_elpg_enable_r(), val);
}
void gm20b_mc_ltc_isr(struct gk20a *g)
{
u32 mc_intr;
u32 ltc;
mc_intr = nvgpu_readl(g, mc_intr_ltc_r());
nvgpu_log(g, gpu_dbg_intr, "mc_ltc_intr: %08x", mc_intr);
for (ltc = 0; ltc < nvgpu_ltc_get_ltc_count(g); ltc++) {
if ((mc_intr & BIT32(ltc)) == 0U) {
continue;
}
g->ops.ltc.intr.isr(g, ltc);
}
}
bool gm20b_mc_is_mmu_fault_pending(struct gk20a *g)
{
return g->ops.fifo.is_mmu_fault_pending(g);
}