gpu: nvgpu: add user API to get a syncpoint

Add new user API NVGPU_IOCTL_CHANNEL_GET_USER_SYNCPOINT which will expose
per-channel allocated syncpoint to user space
API will also return current value of the syncpoint
On supported platforms, this API will also return a RW semaphore address
(corresponding to syncpoint shim) to user space

Add new characteristics flag NVGPU_GPU_FLAGS_SUPPORT_USER_SYNCPOINT to indicate
support for this new API
Add new flag NVGPU_SUPPORT_USER_SYNCPOINT for use of core driver

Set this flag for GV11B and GP10B for now

Add a new API (*syncpt_address) in struct gk20a_channel_sync to get GPU_VA
address of a syncpoint

Add new API nvgpu_nvhost_syncpt_read_maxval() which will read and return MAX
value of syncpoint

Bug 200326065
Jira NVGPU-179

Change-Id: I9da6f17b85996f4fc6731c0bf94fca6f3181c3e0
Signed-off-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1658009
Reviewed-by: svc-mobile-coverity <svc-mobile-coverity@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Konsta Holtta <kholtta@nvidia.com>
Reviewed-by: Vijayakumar Subbu <vsubbu@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>

1 files changed, 67 insertions, 0 deletions

diff --git a/drivers/gpu/nvgpu/common/linux/ioctl_channel.c b/drivers/gpu/nvgpu/common/linux/ioctl_channel.c
index b7856ca0..ab6ac9b9 100644
--- a/drivers/gpu/nvgpu/common/linux/ioctl_channel.c
+++ b/drivers/gpu/nvgpu/common/linux/ioctl_channel.c
@@ -32,6 +32,7 @@
 #include <nvgpu/enabled.h>
 #include <nvgpu/error_notifier.h>
 #include <nvgpu/barrier.h>
+#include <nvgpu/nvhost.h>
 
 #include "gk20a/gk20a.h"
 #include "gk20a/dbg_gpu_gk20a.h"
@@ -948,6 +949,60 @@ static int nvgpu_ioctl_channel_set_preemption_mode(struct channel_gk20a *ch,
         return err;
 }
 
+static int nvgpu_ioctl_channel_get_user_syncpoint(struct channel_gk20a *ch,
+        struct nvgpu_get_user_syncpoint_args *args)
+{
+#ifdef CONFIG_TEGRA_GK20A_NVHOST
+        struct gk20a *g = ch->g;
+        int err;
+
+        if (!nvgpu_is_enabled(g, NVGPU_SUPPORT_USER_SYNCPOINT)) {
+                nvgpu_err(g, "user syncpoints not supported");
+                return -EINVAL;
+        }
+
+        if (!gk20a_platform_has_syncpoints(g)) {
+                nvgpu_err(g, "syncpoints not supported");
+                return -EINVAL;
+        }
+
+        if (g->aggressive_sync_destroy_thresh) {
+                nvgpu_err(g, "sufficient syncpoints not available");
+                return -EINVAL;
+        }
+
+        nvgpu_mutex_acquire(&ch->sync_lock);
+        if (ch->sync) {
+                nvgpu_mutex_release(&ch->sync_lock);
+        } else {
+                ch->sync = gk20a_channel_sync_create(ch);
+                if (!ch->sync) {
+                        nvgpu_mutex_release(&ch->sync_lock);
+                        return -ENOMEM;
+                }
+                nvgpu_mutex_release(&ch->sync_lock);
+
+                if (g->ops.fifo.resetup_ramfc) {
+                        err = g->ops.fifo.resetup_ramfc(ch);
+                        if (err)
+                                return err;
+                }
+        }
+
+        args->syncpoint_id = ch->sync->syncpt_id(ch->sync);
+        args->syncpoint_max = nvgpu_nvhost_syncpt_read_maxval(g->nvhost_dev,
+                                                args->syncpoint_id);
+        if (nvgpu_is_enabled(g, NVGPU_SUPPORT_SYNCPOINT_ADDRESS))
+                args->gpu_va = ch->sync->syncpt_address(ch->sync);
+        else
+                args->gpu_va = 0;
+
+        return 0;
+#else
+        return -EINVAL;
+#endif
+}
+
 long gk20a_channel_ioctl(struct file *filp,
         unsigned int cmd, unsigned long arg)
 {
@@ -1239,6 +1294,18 @@ long gk20a_channel_ioctl(struct file *filp,
                         err = -EINVAL;
                 }
                 break;
+        case NVGPU_IOCTL_CHANNEL_GET_USER_SYNCPOINT:
+                err = gk20a_busy(ch->g);
+                if (err) {
+                        dev_err(dev,
+                                "%s: failed to host gk20a for ioctl cmd: 0x%x",
+                                __func__, cmd);
+                        break;
+                }
+                err = nvgpu_ioctl_channel_get_user_syncpoint(ch,
+                      (struct nvgpu_get_user_syncpoint_args *)buf);
+                gk20a_idle(ch->g);
+                break;
         default:
                 dev_dbg(dev, "unrecognized ioctl cmd: 0x%x", cmd);
                 err = -ENOTTY;