On Tue, 2021-08-17 at 19:51 +0200, Pavel Machek wrote:
Hi!
[ Upstream commit bd37c2888ccaa5ceb9895718f6909b247cc372e0 ]
Check return value of mlx5_fw_tracer_start(), set error path and fix return value of mlx5_fw_tracer_init() accordingly.
This is actually two fixes in one: There's cancel_work_sync() added to
Yes, the reasoning was that the patch is fixing the whole error path of the function in one-shot since we can blame it on a single commit.
the error path, but there's additional error that needs fixing.
Yes.
Could someone familiar with the code verify it after me?
Best regards, Pavel
Signed-off-by: Pavel Machek (CIP) pavel@denx.de
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.c b/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.c index 3dfcb20e97c6..857be86b4a11 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.c @@ -1007,7 +1007,7 @@ int mlx5_fw_tracer_init(struct mlx5_fw_tracer *tracer) err = mlx5_core_alloc_pd(dev, &tracer->buff.pdn); if (err) { mlx5_core_warn(dev, "FWTracer: Failed to allocate PD %d\n", err); - return err; + goto err_cancel_work; } err = mlx5_fw_tracer_create_mkey(tracer); @@ -1031,6 +1031,7 @@ int mlx5_fw_tracer_init(struct mlx5_fw_tracer *tracer) mlx5_core_destroy_mkey(dev, &tracer->buff.mkey); err_dealloc_pd: mlx5_core_dealloc_pd(dev, tracer->buff.pdn); +err_cancel_work: cancel_work_sync(&tracer->read_fw_strings_work); return err;
this is correct, do you want to submit this patch or do you want us to handle ? maybe it is better if we delayed queue_work() to after all the fragile code behind it, to reduce the error path handling ..
thanks for pointing this out.