lkml.org 
[lkml]   [2022]   [Jun]   [22]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
From
Subject[PATCH v2 11/13] gpu: host1x: Add MLOCK release code on Tegra234
Date
From: Mikko Perttunen <mperttunen@nvidia.com>

With the full-featured opcode sequence using MLOCKs, we need to also
unlock those MLOCKs in the event of a timeout. However, it turns out
that on Tegra186/Tegra194, by default, we don't need to do this;
furthermore, on Tegra234 it is much simpler to do; so only implement
this on Tegra234 for the time being.

Signed-off-by: Mikko Perttunen <mperttunen@nvidia.com>
---
drivers/gpu/host1x/hw/cdma_hw.c | 34 ++++++++++++++++++++++
drivers/gpu/host1x/hw/hw_host1x08_common.h | 7 +++++
2 files changed, 41 insertions(+)

diff --git a/drivers/gpu/host1x/hw/cdma_hw.c b/drivers/gpu/host1x/hw/cdma_hw.c
index e49cd5b8f735..1b65a10b9dfc 100644
--- a/drivers/gpu/host1x/hw/cdma_hw.c
+++ b/drivers/gpu/host1x/hw/cdma_hw.c
@@ -238,6 +238,37 @@ static void cdma_resume(struct host1x_cdma *cdma, u32 getptr)
cdma_timeout_restart(cdma, getptr);
}

+static void timeout_release_mlock(struct host1x_cdma *cdma)
+{
+#if HOST1X_HW >= 8
+ /* Tegra186 and Tegra194 require a more complicated MLOCK release
+ * sequence. Furthermore, those chips by default don't enforce MLOCKs,
+ * so it turns out that if we don't /actually/ need MLOCKs, we can just
+ * ignore them.
+ *
+ * As such, for now just implement this on Tegra234 where things are
+ * stricter but also easy to implement.
+ */
+ struct host1x_channel *ch = cdma_to_channel(cdma);
+ struct host1x *host1x = cdma_to_host1x(cdma);
+ u32 offset;
+
+ switch (ch->client->class) {
+ case HOST1X_CLASS_VIC:
+ offset = HOST1X_COMMON_VIC_MLOCK;
+ break;
+ case HOST1X_CLASS_NVDEC:
+ offset = HOST1X_COMMON_NVDEC_MLOCK;
+ break;
+ default:
+ WARN(1, "%s was not updated for class %u", __func__, ch->client->class);
+ return;
+ }
+
+ host1x_common_writel(host1x, 0x0, offset);
+#endif
+}
+
/*
* If this timeout fires, it indicates the current sync_queue entry has
* exceeded its TTL and the userctx should be timed out and remaining
@@ -288,6 +319,9 @@ static void cdma_timeout_handler(struct work_struct *work)
/* stop HW, resetting channel/module */
host1x_hw_cdma_freeze(host1x, cdma);

+ /* release any held MLOCK */
+ timeout_release_mlock(cdma);
+
host1x_cdma_update_sync_queue(cdma, ch->dev);
mutex_unlock(&cdma->lock);
}
diff --git a/drivers/gpu/host1x/hw/hw_host1x08_common.h b/drivers/gpu/host1x/hw/hw_host1x08_common.h
index 4df28440b86b..8e0c99150ec2 100644
--- a/drivers/gpu/host1x/hw/hw_host1x08_common.h
+++ b/drivers/gpu/host1x/hw/hw_host1x08_common.h
@@ -2,3 +2,10 @@
/*
* Copyright (c) 2022 NVIDIA Corporation.
*/
+
+#define HOST1X_COMMON_OFA_MLOCK 0x4050
+#define HOST1X_COMMON_NVJPG1_MLOCK 0x4070
+#define HOST1X_COMMON_VIC_MLOCK 0x4078
+#define HOST1X_COMMON_NVENC_MLOCK 0x407c
+#define HOST1X_COMMON_NVDEC_MLOCK 0x4080
+#define HOST1X_COMMON_NVJPG_MLOCK 0x4084
--
2.36.1
\
 
 \ /
  Last update: 2022-06-22 13:38    [W:0.130 / U:0.132 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site