linux/drivers/gpu/drm/i915/i915_gpu_error.h

347 lines
7.0 KiB
C
Raw Normal View History

/*
* SPDX-License-Identifier: MIT
*
* Copyright <EFBFBD> 2008-2018 Intel Corporation
*/
#ifndef _I915_GPU_ERROR_H_
#define _I915_GPU_ERROR_H_
#include <linux/atomic.h>
#include <linux/kref.h>
#include <linux/ktime.h>
#include <linux/sched.h>
#include <drm/drm_mm.h>
#include "gt/intel_engine.h"
#include "gt/uc/intel_uc_fw.h"
#include "intel_device_info.h"
#include "i915_gem.h"
#include "i915_gem_gtt.h"
#include "i915_params.h"
#include "i915_scheduler.h"
struct drm_i915_private;
struct i915_vma_compress;
struct intel_engine_capture_vma;
struct intel_overlay_error_state;
struct intel_display_error_state;
struct i915_vma_coredump {
struct i915_vma_coredump *next;
char name[20];
u64 gtt_offset;
u64 gtt_size;
u32 gtt_page_sizes;
int num_pages;
int page_count;
int unused;
drm/i915: Replace zero-length array with flexible-array The current codebase makes use of the zero-length array language extension to the C90 standard, but the preferred mechanism to declare variable-length types such as these ones is a flexible array member[1][2], introduced in C99: struct foo { int stuff; struct boo array[]; }; By making use of the mechanism above, we will get a compiler warning in case the flexible array does not occur last in the structure, which will help us prevent some kind of undefined behavior bugs from being inadvertently introduced[3] to the codebase from now on. Also, notice that, dynamic memory allocations won't be affected by this change: "Flexible array members have incomplete type, and so the sizeof operator may not be applied. As a quirk of the original implementation of zero-length arrays, sizeof evaluates to zero."[1] sizeof(flexible-array-member) triggers a warning because flexible array members have incomplete type[1]. There are some instances of code in which the sizeof operator is being incorrectly/erroneously applied to zero-length arrays and the result is zero. Such instances may be hiding some bugs. So, this work (flexible-array member conversions) will also help to get completely rid of those sorts of issues. This issue was found with the help of Coccinelle. [1] https://gcc.gnu.org/onlinedocs/gcc/Zero-Length.html [2] https://github.com/KSPP/linux/issues/21 [3] commit 76497732932f ("cxgb3/l2t: Fix undefined behaviour") Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Link: https://patchwork.freedesktop.org/patch/msgid/20200507185408.GA14561@embeddedor
2020-05-07 18:54:08 +00:00
u32 *pages[];
};
struct i915_request_coredump {
unsigned long flags;
pid_t pid;
u32 context;
u32 seqno;
u32 head;
u32 tail;
struct i915_sched_attr sched_attr;
};
struct intel_engine_coredump {
const struct intel_engine_cs *engine;
bool simulated;
u32 reset_count;
/* position of active request inside the ring */
u32 rq_head, rq_post, rq_tail;
/* Register state */
u32 ccid;
u32 start;
u32 tail;
u32 head;
u32 ctl;
u32 mode;
u32 hws;
u32 ipeir;
u32 ipehr;
u32 esr;
u32 bbstate;
u32 instpm;
u32 instps;
u64 bbaddr;
u64 acthd;
u32 fault_reg;
u64 faddr;
u32 rc_psmi; /* sleep state */
struct intel_instdone instdone;
struct i915_gem_context_coredump {
char comm[TASK_COMM_LEN];
u64 total_runtime;
u32 avg_runtime;
pid_t pid;
int active;
int guilty;
struct i915_sched_attr sched_attr;
} context;
struct i915_vma_coredump *vma;
struct i915_request_coredump execlist[EXECLIST_MAX_PORTS];
unsigned int num_ports;
struct {
u32 gfx_mode;
union {
u64 pdp[4];
u32 pp_dir_base;
};
} vm_info;
struct intel_engine_coredump *next;
};
struct intel_gt_coredump {
const struct intel_gt *_gt;
bool awake;
bool simulated;
/* Generic register state */
u32 eir;
u32 pgtbl_er;
u32 ier;
u32 gtier[6], ngtier;
u32 derrmr;
u32 forcewake;
u32 error; /* gen6+ */
u32 err_int; /* gen7 */
u32 fault_data0; /* gen8, gen9 */
u32 fault_data1; /* gen8, gen9 */
u32 done_reg;
u32 gac_eco;
u32 gam_ecochk;
u32 gab_ctl;
u32 gfx_mode;
u32 gtt_cache;
u32 aux_err; /* gen12 */
u32 sfc_done[GEN12_SFC_DONE_MAX]; /* gen12 */
u32 gam_done; /* gen12 */
u32 nfence;
u64 fence[I915_MAX_NUM_FENCES];
struct intel_engine_coredump *engine;
struct intel_uc_coredump {
struct intel_uc_fw guc_fw;
struct intel_uc_fw huc_fw;
struct i915_vma_coredump *guc_log;
} *uc;
struct intel_gt_coredump *next;
};
struct i915_gpu_coredump {
struct kref ref;
ktime_t time;
ktime_t boottime;
ktime_t uptime;
unsigned long capture;
struct drm_i915_private *i915;
struct intel_gt_coredump *gt;
char error_msg[128];
bool simulated;
bool wakelock;
bool suspended;
int iommu;
u32 reset_count;
u32 suspend_count;
struct intel_device_info device_info;
struct intel_runtime_info runtime_info;
struct intel_driver_caps driver_caps;
struct i915_params params;
struct intel_overlay_error_state *overlay;
struct intel_display_error_state *display;
struct scatterlist *sgl, *fit;
};
struct i915_gpu_error {
/* For reset and error_state handling. */
spinlock_t lock;
/* Protected by the above dev->gpu_error.lock. */
struct i915_gpu_coredump *first_error;
atomic_t pending_fb_pin;
drm/i915: Revoke mmaps and prevent access to fence registers across reset Previously, we were able to rely on the recursive properties of struct_mutex to allow us to serialise revoking mmaps and reacquiring the FENCE registers with them being clobbered over a global device reset. I then proceeded to throw out the baby with the bath water in order to pursue a struct_mutex-less reset. Perusing LWN for alternative strategies, the dilemma on how to serialise access to a global resource on one side was answered by https://lwn.net/Articles/202847/ -- Sleepable RCU: 1 int readside(void) { 2 int idx; 3 rcu_read_lock(); 4 if (nomoresrcu) { 5 rcu_read_unlock(); 6 return -EINVAL; 7 } 8 idx = srcu_read_lock(&ss); 9 rcu_read_unlock(); 10 /* SRCU read-side critical section. */ 11 srcu_read_unlock(&ss, idx); 12 return 0; 13 } 14 15 void cleanup(void) 16 { 17 nomoresrcu = 1; 18 synchronize_rcu(); 19 synchronize_srcu(&ss); 20 cleanup_srcu_struct(&ss); 21 } No more worrying about stop_machine, just an uber-complex mutex, optimised for reads, with the overhead pushed to the rare reset path. However, we do run the risk of a deadlock as we allocate underneath the SRCU read lock, and the allocation may require a GPU reset, causing a dependency cycle via the in-flight requests. We resolve that by declaring the driver wedged and cancelling all in-flight rendering. v2: Use expedited rcu barriers to match our earlier timing characteristics. v3: Try to annotate locking contexts for sparse v4: Reduce selftest lock duration to avoid a reset deadlock with fences v5: s/srcu/reset_backoff_srcu/ v6: Remove more stale comments Testcase: igt/gem_mmap_gtt/hang Fixes: eb8d0f5af4ec ("drm/i915: Remove GPU reset dependence on struct_mutex") Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Mika Kuoppala <mika.kuoppala@intel.com> Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190208153708.20023-2-chris@chris-wilson.co.uk
2019-02-08 15:37:03 +00:00
/** Number of times the device has been reset (global) */
atomic_t reset_count;
drm/i915: Revoke mmaps and prevent access to fence registers across reset Previously, we were able to rely on the recursive properties of struct_mutex to allow us to serialise revoking mmaps and reacquiring the FENCE registers with them being clobbered over a global device reset. I then proceeded to throw out the baby with the bath water in order to pursue a struct_mutex-less reset. Perusing LWN for alternative strategies, the dilemma on how to serialise access to a global resource on one side was answered by https://lwn.net/Articles/202847/ -- Sleepable RCU: 1 int readside(void) { 2 int idx; 3 rcu_read_lock(); 4 if (nomoresrcu) { 5 rcu_read_unlock(); 6 return -EINVAL; 7 } 8 idx = srcu_read_lock(&ss); 9 rcu_read_unlock(); 10 /* SRCU read-side critical section. */ 11 srcu_read_unlock(&ss, idx); 12 return 0; 13 } 14 15 void cleanup(void) 16 { 17 nomoresrcu = 1; 18 synchronize_rcu(); 19 synchronize_srcu(&ss); 20 cleanup_srcu_struct(&ss); 21 } No more worrying about stop_machine, just an uber-complex mutex, optimised for reads, with the overhead pushed to the rare reset path. However, we do run the risk of a deadlock as we allocate underneath the SRCU read lock, and the allocation may require a GPU reset, causing a dependency cycle via the in-flight requests. We resolve that by declaring the driver wedged and cancelling all in-flight rendering. v2: Use expedited rcu barriers to match our earlier timing characteristics. v3: Try to annotate locking contexts for sparse v4: Reduce selftest lock duration to avoid a reset deadlock with fences v5: s/srcu/reset_backoff_srcu/ v6: Remove more stale comments Testcase: igt/gem_mmap_gtt/hang Fixes: eb8d0f5af4ec ("drm/i915: Remove GPU reset dependence on struct_mutex") Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Mika Kuoppala <mika.kuoppala@intel.com> Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190208153708.20023-2-chris@chris-wilson.co.uk
2019-02-08 15:37:03 +00:00
/** Number of times an engine has been reset */
atomic_t reset_engine_count[I915_NUM_ENGINES];
};
struct drm_i915_error_state_buf {
struct drm_i915_private *i915;
struct scatterlist *sgl, *cur, *end;
char *buf;
size_t bytes;
size_t size;
loff_t iter;
int err;
};
#if IS_ENABLED(CONFIG_DRM_I915_CAPTURE_ERROR)
__printf(2, 3)
void i915_error_printf(struct drm_i915_error_state_buf *e, const char *f, ...);
struct i915_gpu_coredump *i915_gpu_coredump(struct drm_i915_private *i915);
void i915_capture_error_state(struct drm_i915_private *i915);
struct i915_gpu_coredump *
i915_gpu_coredump_alloc(struct drm_i915_private *i915, gfp_t gfp);
struct intel_gt_coredump *
intel_gt_coredump_alloc(struct intel_gt *gt, gfp_t gfp);
struct intel_engine_coredump *
intel_engine_coredump_alloc(struct intel_engine_cs *engine, gfp_t gfp);
struct intel_engine_capture_vma *
intel_engine_coredump_add_request(struct intel_engine_coredump *ee,
struct i915_request *rq,
gfp_t gfp);
void intel_engine_coredump_add_vma(struct intel_engine_coredump *ee,
struct intel_engine_capture_vma *capture,
struct i915_vma_compress *compress);
struct i915_vma_compress *
i915_vma_capture_prepare(struct intel_gt_coredump *gt);
void i915_vma_capture_finish(struct intel_gt_coredump *gt,
struct i915_vma_compress *compress);
void i915_error_state_store(struct i915_gpu_coredump *error);
static inline struct i915_gpu_coredump *
i915_gpu_coredump_get(struct i915_gpu_coredump *gpu)
{
kref_get(&gpu->ref);
return gpu;
}
ssize_t
i915_gpu_coredump_copy_to_buffer(struct i915_gpu_coredump *error,
char *buf, loff_t offset, size_t count);
void __i915_gpu_coredump_free(struct kref *kref);
static inline void i915_gpu_coredump_put(struct i915_gpu_coredump *gpu)
{
if (gpu)
kref_put(&gpu->ref, __i915_gpu_coredump_free);
}
struct i915_gpu_coredump *i915_first_error_state(struct drm_i915_private *i915);
void i915_reset_error_state(struct drm_i915_private *i915);
void i915_disable_error_state(struct drm_i915_private *i915, int err);
#else
static inline void i915_capture_error_state(struct drm_i915_private *i915)
{
}
static inline struct i915_gpu_coredump *
i915_gpu_coredump_alloc(struct drm_i915_private *i915, gfp_t gfp)
{
return NULL;
}
static inline struct intel_gt_coredump *
intel_gt_coredump_alloc(struct intel_gt *gt, gfp_t gfp)
{
return NULL;
}
static inline struct intel_engine_coredump *
intel_engine_coredump_alloc(struct intel_engine_cs *engine, gfp_t gfp)
{
return NULL;
}
static inline struct intel_engine_capture_vma *
intel_engine_coredump_add_request(struct intel_engine_coredump *ee,
struct i915_request *rq,
gfp_t gfp)
{
return NULL;
}
static inline void
intel_engine_coredump_add_vma(struct intel_engine_coredump *ee,
struct intel_engine_capture_vma *capture,
struct i915_vma_compress *compress)
{
}
static inline struct i915_vma_compress *
i915_vma_capture_prepare(struct intel_gt_coredump *gt)
{
return NULL;
}
static inline void
i915_vma_capture_finish(struct intel_gt_coredump *gt,
struct i915_vma_compress *compress)
{
}
static inline void
i915_error_state_store(struct i915_gpu_coredump *error)
{
}
static inline void i915_gpu_coredump_put(struct i915_gpu_coredump *gpu)
{
}
static inline struct i915_gpu_coredump *
i915_first_error_state(struct drm_i915_private *i915)
{
return ERR_PTR(-ENODEV);
}
static inline void i915_reset_error_state(struct drm_i915_private *i915)
{
}
static inline void i915_disable_error_state(struct drm_i915_private *i915,
int err)
{
}
#endif /* IS_ENABLED(CONFIG_DRM_I915_CAPTURE_ERROR) */
#endif /* _I915_GPU_ERROR_H_ */