gpu: nvgpu: fix double jiffies conversion

The gk20a_channel_wait() ioctl still considers a timeout value type to
be jiffies, but NVGPU_COND_WAIT_INTERRUPTIBLE takes milliseconds. Pass
the millisecond value from args directly to the cond wait.

NVGPU_NO_TIMEOUT is the max unsigned value that fits in 32 bits, so it
does not need to be tested explicitly.

Jira NVGPU-83

Change-Id: If7979c015696b2d1e0bbe9612c546410a0f41f2e
Signed-off-by: Konsta Holtta <kholtta@nvidia.com>
Reviewed-on: http://git-master/r/1505422
Reviewed-by: Automatic_Commit_Validation_User
GVS: Gerrit_Virtual_Submit
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>

1 files changed, 3 insertions, 9 deletions

diff --git a/drivers/gpu/nvgpu/common/linux/ioctl_channel.c b/drivers/gpu/nvgpu/common/linux/ioctl_channel.c
index a083cb81..08e483a3 100644
--- a/drivers/gpu/nvgpu/common/linux/ioctl_channel.c
+++ b/drivers/gpu/nvgpu/common/linux/ioctl_channel.c
@@ -401,7 +401,7 @@ int nvgpu_channel_ioctl_alloc_gpfifo(struct channel_gk20a *c,
 
 static int gk20a_channel_wait_semaphore(struct channel_gk20a *ch,
                                         ulong id, u32 offset,
-                                        u32 payload, long timeout)
+                                        u32 payload, u32 timeout)
 {
         struct dma_buf *dmabuf;
         void *data;
@@ -448,7 +448,6 @@ static int gk20a_channel_wait(struct channel_gk20a *ch,
         u64 jiffies;
         ulong id;
         u32 offset;
-        unsigned long timeout;
         int remain, ret = 0;
         u64 end;
 
@@ -457,11 +456,6 @@ static int gk20a_channel_wait(struct channel_gk20a *ch,
         if (ch->has_timedout)
                 return -ETIMEDOUT;
 
-        if (args->timeout == NVGPU_NO_TIMEOUT)
-                timeout = MAX_SCHEDULE_TIMEOUT;
-        else
-                timeout = (u32)msecs_to_jiffies(args->timeout);
-
         switch (args->type) {
         case NVGPU_WAIT_TYPE_NOTIFIER:
                 id = args->condition.notifier.dmabuf_fd;
@@ -494,7 +488,7 @@ static int gk20a_channel_wait(struct channel_gk20a *ch,
                 remain = NVGPU_COND_WAIT_INTERRUPTIBLE(
                                 &ch->notifier_wq,
                                 notif->status == 0 || ch->has_timedout,
-                                timeout);
+                                args->timeout);
 
                 if (remain == 0 && notif->status != 0) {
                         ret = -ETIMEDOUT;
@@ -521,7 +515,7 @@ notif_clean_up:
                                 args->condition.semaphore.dmabuf_fd,
                                 args->condition.semaphore.offset,
                                 args->condition.semaphore.payload,
-                                timeout);
+                                args->timeout);
 
                 break;