nv-tegra-mirror/linux-nvgpu
1
0
Fork 0
mirror of git://nv-tegra.nvidia.com/linux-nvgpu.git synced 2025-12-24 02:22:34 +03:00
Code Issues Packages Projects Releases Wiki Activity

gpu: nvgpu: Fix gk20a_sync_pt_has_signaled()

Browse Source 
Fix a reentrancy problem in gk20a_sync_pt_has_signaled()
where one thread could clear a pointer before another
thread tried to access it. A spinlock was added to the
gk20a_sync_pt struct which is used to ensure that the
underyling gk20a_sync_pt data is accessed in a sane
manner.

Bug 1604892

Change-Id: I270d89def7b986405a3167285d51ceda950c7b82
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: http://git-master/r/935909
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
GVS: Gerrit_Virtual_Submit
This commit is contained in:
Alex Waterman
2016-01-21 14:44:43 -08:00
parent 04092f74bb
commit aa74098f29
1 changed files with 22 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

25
drivers/gpu/nvgpu/gk20a/sync_gk20a.c
View File

@@ -1,7 +1,7 @@
/* /*
* GK20A Sync Framework Integration * GK20A Sync Framework Integration
* *
* Copyright (c) 2014-2015, NVIDIA CORPORATION. All rights reserved. * Copyright (c) 2014-2016, NVIDIA CORPORATION. All rights reserved.
* *
* This program is free software; you can redistribute it and/or modify it * This program is free software; you can redistribute it and/or modify it
* under the terms and conditions of the GNU General Public License, * under the terms and conditions of the GNU General Public License,
@@ -21,6 +21,7 @@
#include <linux/hrtimer.h> #include <linux/hrtimer.h>
#include <linux/module.h> #include <linux/module.h>
#include <linux/slab.h> #include <linux/slab.h>
#include <linux/spinlock.h>
#include <uapi/linux/nvgpu.h> #include <uapi/linux/nvgpu.h>
#include "../../../staging/android/sync.h" #include "../../../staging/android/sync.h"
#include "semaphore_gk20a.h" #include "semaphore_gk20a.h"
@@ -44,6 +45,12 @@ struct gk20a_sync_pt {
struct gk20a_sync_timeline *obj; struct gk20a_sync_timeline *obj;
struct sync_fence *dep; struct sync_fence *dep;
ktime_t dep_timestamp; ktime_t dep_timestamp;
/*
* A spinlock is necessary since there are times when this lock
* will be acquired in interrupt context.
*/
spinlock_t lock;
}; };
struct gk20a_sync_pt_inst { struct gk20a_sync_pt_inst {
@@ -149,7 +156,10 @@ static struct gk20a_sync_pt *gk20a_sync_pt_create_shared(
} }
} }
spin_lock_init(&shared->lock);
gk20a_semaphore_get(sema); gk20a_semaphore_get(sema);
return shared; return shared;
} }
@@ -194,14 +204,20 @@ static struct sync_pt *gk20a_sync_pt_dup_inst(struct sync_pt *sync_pt)
return &pti->pt; return &pti->pt;
} }
/*
* This function must be able to run on the same sync_pt concurrently. This
* requires a lock to protect access to the sync_pt's internal data structures
* which are modified as a side effect of calling this function.
*/
static int gk20a_sync_pt_has_signaled(struct sync_pt *sync_pt) static int gk20a_sync_pt_has_signaled(struct sync_pt *sync_pt)
{ {
struct gk20a_sync_pt *pt = to_gk20a_sync_pt(sync_pt); struct gk20a_sync_pt *pt = to_gk20a_sync_pt(sync_pt);
struct gk20a_sync_timeline *obj = pt->obj; struct gk20a_sync_timeline *obj = pt->obj;
bool signaled; bool signaled = true;
spin_lock(&pt->lock);
if (!pt->sema) if (!pt->sema)
return true; goto done;
/* Acquired == not realeased yet == active == not signaled. */ /* Acquired == not realeased yet == active == not signaled. */
signaled = !gk20a_semaphore_is_acquired(pt->sema); signaled = !gk20a_semaphore_is_acquired(pt->sema);
@@ -232,6 +248,9 @@ static int gk20a_sync_pt_has_signaled(struct sync_pt *sync_pt)
gk20a_semaphore_put(pt->sema); gk20a_semaphore_put(pt->sema);
pt->sema = NULL; pt->sema = NULL;
} }
done:
spin_unlock(&pt->lock);
return signaled; return signaled;
} }