[Intel-gfx] [PATCH] drm/i915: Fix the race between the GEM close and debugfs

Tvrtko Ursulin tvrtko.ursulin at linux.intel.com
Mon Sep 14 16:47:38 UTC 2020


On 14/09/2020 12:00, Nikunj A. Dadhania wrote:
> As we close GEM object and set file_priv to -EBADF which is protected
> by ctx->mutex, populating the GEM debugfs info is not protected
> and results in the crash shown below.
> 
> Make sure to protect the access to file_priv using ctx->mutex to avoid
> race.
> 
> BUG: unable to handle page fault for address: ffffffffffffffff
> RIP: 0010:i915_gem_object_info+0x26b/0x3eb
> Code: 89 44 24 48 48 89 44 24 40 48 89 44 24 38 48 89 44 24 30 48 89 44 24 28 48 89 44 24 20 49 8b 46 f0 48 89 44 24 20 49 8b 46 a0 <48> 8b 58 08 b9 0a 00 00 00 48 b8 aa aa aa aa aa aa aa aa 48 8d bc
> RSP: 0018:ffffac81c14cfc30 EFLAGS: 00010246
> RAX: fffffffffffffff7 RBX: ffff95094429c218 RCX: ffff95096756c740
> RDX: 0000000000000000 RSI: ffffffff919b93ee RDI: ffff95094429c218
> RBP: ffffac81c14cfd58 R08: ffff9509746fab80 R09: 0000000000000000
> R10: 0000000000000001 R11: 0000000000000000 R12: ffff9509753f8e80
> R13: ffffac81c14cfc98 R14: ffff95094429c268 R15: ffffac81c14cfc88
> FS:  00007a1bdcd52900(0000) GS:ffff950977e00000(0000) knlGS:0000000000000000
> CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> CR2: ffffffffffffffff CR3: 000000026b4e0000 CR4: 0000000000340ef0
> Call Trace:
>   seq_read+0x162/0x3ca
>   full_proxy_read+0x5b/0x8d
>   __vfs_read+0x45/0x1b9
>   vfs_read+0xc9/0x15e
>   ksys_read+0x7e/0xde
>   do_syscall_64+0x54/0x7e
>   entry_SYSCALL_64_after_hwframe+0x44/0xa9
> RIP: 0033:0x7a1bdd34cf03
> 
> Signed-off-by: Nikunj A. Dadhania <nikunj.dadhania at linux.intel.com>
> ---
>   drivers/gpu/drm/i915/i915_debugfs.c | 2 ++
>   1 file changed, 2 insertions(+)
> 
> diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c
> index 784219962193..ea469168cd44 100644
> --- a/drivers/gpu/drm/i915/i915_debugfs.c
> +++ b/drivers/gpu/drm/i915/i915_debugfs.c
> @@ -326,6 +326,7 @@ static void print_context_stats(struct seq_file *m,
>   		}
>   		i915_gem_context_unlock_engines(ctx);
>   
> +		mutex_lock(&ctx->mutex);
>   		if (!IS_ERR_OR_NULL(ctx->file_priv)) {
>   			struct file_stats stats = {
>   				.vm = rcu_access_pointer(ctx->vm),
> @@ -346,6 +347,7 @@ static void print_context_stats(struct seq_file *m,
>   
>   			print_file_stats(m, name, stats);
>   		}
> +		mutex_unlock(&ctx->mutex);
>   
>   		spin_lock(&i915->gem.contexts.lock);
>   		list_safe_reset_next(ctx, cn, link);
> 

Fix is correct, but it looked familiar and indeed I found a fix for the 
same issues back from July. Copied you on that one which now has an r-b. 
This one can have it as well but please also copy stable.

Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin at intel.com>

Regards,

Tvrtko


More information about the Intel-gfx mailing list