ACK: [SRU][J:linux-bluefield][PATCH v1 0/1] mlxbf_gige: kernel log is flooded during heavy TX traffic

Jose Ogando jose.ogando at canonical.com
Wed Jul 2 10:42:29 UTC 2025


Looks good to me.

Acked-by: Jose Ogando <jose.ogando at canonical.com> 

On Tue, 2025-07-01 at 15:52 -0400, David Thompson wrote:
> BugLink: https://bugs.launchpad.net/bugs/2115754
> 
> SRU Justification:
> 
> [Impact]
> The GIGE driver utilizes the msglvl infrastructure to log interesting
> events during interface initialization, transmit, and receive. The
> problem is during heavy transmit traffic, the kernel log gets flooded
> with log entries such as:
> 
> mlxbf_gige MLNXBF17:00: tx_fifo_full tx_ci=0x6e51 tx_pi=0x6eca
> rx_ci=0x24e8e rx_pi=0x4f0e int_mask=0x1
> mlxbf_gige MLNXBF17:00: din_drop=0x0 rx_dma=0x4000000
> rx_fifo=0x30002000a rx_polarity=1
> mlxbf_gige MLNXBF17:00: tx_fifo_full tx_ci=0x6e52 tx_pi=0x6ece
> rx_ci=0x24e90 rx_pi=0x4f0e int_mask=0x1
> mlxbf_gige MLNXBF17:00: din_drop=0x0 rx_dma=0x4000001
> rx_fifo=0x300000000 rx_polarity=1
> 
> [Fix]
> There are two fixes here:
> 1) make sure to use rate limiting "dev_dbg()"
> 2) don't log the transmit FIFO full events, instead just count them
> 
> [Test Case]
> * Boot BF3 platform with software containing fix
> * Check the kernel log via dmesg
> * Start test that requires transmitting lots of traffic out DPU OOB
> port
> * Verify that kernel log is not flooded with above messages
> 
> [Regression Potential]
> This fix is straightforward and thus low risk
> 
> David Thompson (1):
>   UBUNTU: SAUCE: mlxbf_gige: fixes for msglvl infrastructure
> 
>  .../mellanox/mlxbf_gige/mlxbf_gige_main.c     | 20 +++++++++--------
> --
>  .../mellanox/mlxbf_gige/mlxbf_gige_tx.c       |  5 +----
>  2 files changed, 11 insertions(+), 14 deletions(-)
> 
> -- 
> 2.43.2
> 
> 




More information about the kernel-team mailing list