From 34c3692e4db2cdeb3cbb24a90907e368c46fef6d Mon Sep 17 00:00:00 2001
From: Felix Kuehling <Felix.Kuehling@amd.com>
Date: Tue, 1 Nov 2022 15:02:48 -0400
Subject: drm/amdkfd: Fix error handling in criu_checkpoint
Git-commit: b91c23e099f0b65d62159da13458c5eefa76083f
Patch-mainline: v6.1-rc5
References: jsc#PED-1166 jsc#PED-1168 jsc#PED-1170 jsc#PED-1218 jsc#PED-1220 jsc#PED-1222 jsc#PED-1223 jsc#PED-1225 jsc#PED-2849
Alt-commit: a0fba3e078801b9596e61a0687d398664cdc36e6

Checkpoint BOs last. That way we don't need to close dmabuf FDs if
something else fails later. This avoids problematic access to user mode
memory in the error handling code path.

criu_checkpoint_bos has its own error handling and cleanup that does not
depend on access to user memory.

In the private data, keep BOs before the remaining objects. This is
necessary to restore things in the correct order as restoring events
depends on the events-page BO being restored first.

Fixes: be072b06c739 ("drm/amdkfd: CRIU export BOs as prime dmabuf objects")
Reported-by: Jann Horn <jannh@google.com>
CC: Rajneesh Bhardwaj <Rajneesh.Bhardwaj@amd.com>
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Reviewed-and-tested-by: Rajneesh Bhardwaj <rajneesh.bhardwaj@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Acked-by: Patrik Jakobsson <pjakobsson@suse.de>
---
 drivers/gpu/drm/amd/amdkfd/kfd_chardev.c | 34 +++++++++++-------------
 1 file changed, 15 insertions(+), 19 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
index 5feaba6a77de..6d291aa6386b 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
@@ -1950,7 +1950,7 @@ static int criu_checkpoint(struct file *filep,
 {
 	int ret;
 	uint32_t num_devices, num_bos, num_objects;
-	uint64_t priv_size, priv_offset = 0;
+	uint64_t priv_size, priv_offset = 0, bo_priv_offset;
 
 	if (!args->devices || !args->bos || !args->priv_data)
 		return -EINVAL;
@@ -1994,38 +1994,34 @@ static int criu_checkpoint(struct file *filep,
 	if (ret)
 		goto exit_unlock;
 
-	ret = criu_checkpoint_bos(p, num_bos, (uint8_t __user *)args->bos,
-			    (uint8_t __user *)args->priv_data, &priv_offset);
-	if (ret)
-		goto exit_unlock;
+	/* Leave room for BOs in the private data. They need to be restored
+	 * before events, but we checkpoint them last to simplify the error
+	 * handling.
+	 */
+	bo_priv_offset = priv_offset;
+	priv_offset += num_bos * sizeof(struct kfd_criu_bo_priv_data);
 
 	if (num_objects) {
 		ret = kfd_criu_checkpoint_queues(p, (uint8_t __user *)args->priv_data,
 						 &priv_offset);
 		if (ret)
-			goto close_bo_fds;
+			goto exit_unlock;
 
 		ret = kfd_criu_checkpoint_events(p, (uint8_t __user *)args->priv_data,
 						 &priv_offset);
 		if (ret)
-			goto close_bo_fds;
+			goto exit_unlock;
 
 		ret = kfd_criu_checkpoint_svm(p, (uint8_t __user *)args->priv_data, &priv_offset);
 		if (ret)
-			goto close_bo_fds;
+			goto exit_unlock;
 	}
 
-close_bo_fds:
-	if (ret) {
-		/* If IOCTL returns err, user assumes all FDs opened in criu_dump_bos are closed */
-		uint32_t i;
-		struct kfd_criu_bo_bucket *bo_buckets = (struct kfd_criu_bo_bucket *) args->bos;
-
-		for (i = 0; i < num_bos; i++) {
-			if (bo_buckets[i].alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_VRAM)
-				close_fd(bo_buckets[i].dmabuf_fd);
-		}
-	}
+	/* This must be the last thing in this function that can fail.
+	 * Otherwise we leak dmabuf file descriptors.
+	 */
+	ret = criu_checkpoint_bos(p, num_bos, (uint8_t __user *)args->bos,
+			   (uint8_t __user *)args->priv_data, &bo_priv_offset);
 
 exit_unlock:
 	mutex_unlock(&p->mutex);
-- 
2.38.1