netdev - Re: [net-next,PATCH] net: dsa: microchip: Do not count RX/TX Bytes and discards on KSZ87xx

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <5fa3a420-e25a-4178-b0f3-b3b7bad4665e@mailbox.org>
Date: Mon, 11 Aug 2025 22:59:09 +0200
From: Marek Vasut <marek.vasut@...lbox.org>
To: Tristram.Ha@...rochip.com
Cc: andrew@...n.ch, Arun.Ramadoss@...rochip.com, edumazet@...gle.com,
 kuba@...nel.org, pabeni@...hat.com, UNGLinuxDriver@...rochip.com,
 olteanv@...il.com, Woojung.Huh@...rochip.com, netdev@...r.kernel.org
Subject: Re: [net-next,PATCH] net: dsa: microchip: Do not count RX/TX Bytes
 and discards on KSZ87xx

On 8/11/25 10:47 PM, Tristram.Ha@...rochip.com wrote:
>> Unlike KSZ94xx, the KSZ87xx is missing the last four MIB counters at
>> address 0x20..0x23, namely rx_total, tx_total, rx_discards, tx_discards.
>>
>> Using values from rx_total / tx_total in rx_bytes / tx_bytes calculation
>> results in an underflow, because rx_total / tx_total returns values 0,
>> and the "raw->rx_total - stats->rx_packets * ETH_FCS_LEN;" calculation
>> undeflows if rx_packets / tx_packets is > 0 .
>>
>> Stop using the missing MIB counters on KSZ87xx .
>>
>> Fixes: c6101dd7ffb8 ("net: dsa: ksz9477: move get_stats64 to ksz_common.c")
>> Signed-off-by: Marek Vasut <marek.vasut@...lbox.org>
>> ---
>> Cc: Andrew Lunn <andrew@...n.ch>
>> Cc: Arun Ramadoss <arun.ramadoss@...rochip.com>
>> Cc: Eric Dumazet <edumazet@...gle.com>
>> Cc: Jakub Kicinski <kuba@...nel.org>
>> Cc: Paolo Abeni <pabeni@...hat.com>
>> Cc: Tristram Ha <tristram.ha@...rochip.com>
>> Cc: UNGLinuxDriver@...rochip.com
>> Cc: Vladimir Oltean <olteanv@...il.com>
>> Cc: Woojung Huh <woojung.huh@...rochip.com>
>> Cc: netdev@...r.kernel.org
>> ---
>>   drivers/net/dsa/microchip/ksz_common.c | 13 ++++++++-----
>>   1 file changed, 8 insertions(+), 5 deletions(-)
>>
>> diff --git a/drivers/net/dsa/microchip/ksz_common.c
>> b/drivers/net/dsa/microchip/ksz_common.c
>> index 7292bfe2f7cac..9c01526779a6d 100644
>> --- a/drivers/net/dsa/microchip/ksz_common.c
>> +++ b/drivers/net/dsa/microchip/ksz_common.c
>> @@ -2239,20 +2239,23 @@ void ksz_r_mib_stats64(struct ksz_device *dev, int port)
>>          /* HW counters are counting bytes + FCS which is not acceptable
>>           * for rtnl_link_stats64 interface
>>           */
>> -       stats->rx_bytes = raw->rx_total - stats->rx_packets * ETH_FCS_LEN;
>> -       stats->tx_bytes = raw->tx_total - stats->tx_packets * ETH_FCS_LEN;
>> -
>> +       if (!ksz_is_ksz87xx(dev)) {
>> +               stats->rx_bytes = raw->rx_total - stats->rx_packets * ETH_FCS_LEN;
>> +               stats->tx_bytes = raw->tx_total - stats->tx_packets * ETH_FCS_LEN;
>> +       }
>>          stats->rx_length_errors = raw->rx_undersize + raw->rx_fragments +
>>                  raw->rx_oversize;
>>
>>          stats->rx_crc_errors = raw->rx_crc_err;
>>          stats->rx_frame_errors = raw->rx_align_err;
>> -       stats->rx_dropped = raw->rx_discards;
>> +       if (!ksz_is_ksz87xx(dev))
>> +               stats->rx_dropped = raw->rx_discards;
>>          stats->rx_errors = stats->rx_length_errors + stats->rx_crc_errors +
>>                  stats->rx_frame_errors  + stats->rx_dropped;
>>
>>          stats->tx_window_errors = raw->tx_late_col;
>> -       stats->tx_fifo_errors = raw->tx_discards;
>> +       if (!ksz_is_ksz87xx(dev))
>> +               stats->tx_fifo_errors = raw->tx_discards;
>>          stats->tx_aborted_errors = raw->tx_exc_col;
>>          stats->tx_errors = stats->tx_window_errors + stats->tx_fifo_errors +
>>                  stats->tx_aborted_errors;
> 
> I am not sure why you have that problem.  In my KSZ8795 board running in
> kernel 6.16 those counters are working.  Using "ethtool -S lan3" or
> "ethtool -S eth0" shows rx_total and tx_total.  The "rx_discards" and
> "tx_discards" are hard to get.  In KSZ8863 the "tx_discards" counter is
> incremented when the port does not have link and a packet is sent to that
> port.  In newer switches that behavior was changed.  Occasionally you may
> get 1 or 2 at the beginning when the tail tagging function is not enabled
> yet but the MAC sends out packets.  The "rx_discards" count typically
> seldom happens, but somehow in my first bootup there are many from my
> KSZ8795 board.
> 
> Actually that many rx_discards may be a problem I need to find out.
> 
> I think you are confused about how those MIB counters are read from
> KSZ8795.  They are not using the code in ksz9477.c but in ksz8.c.

See [1] TABLE 4-26: PORT MIB COUNTER INDIRECT MEMORY OFFSETS (CONTINUED) 
, page 108 at the very end . Notice the table ends at 0x1F 
TxMultipleCollision .

See [2] TABLE 5-6: MIB COUNTERS (CONTINUED) , page 219 at the end of the 
table . Notice the table contains four extra entries , 0x80 RxByteCnt , 
0x81 TxByteCnt , 0x82 RxDropPackets , 0x83 TXDropPackets .

These entries are present on KSZ9477 and missing on KSZ8795 .

This is what this patch attempts to address.

[1] 
https://ww1.microchip.com/downloads/aemDocuments/documents/UNG/ProductDocuments/DataSheets/KSZ8795CLX-Data-Sheet-DS00002112.pdf
[2] 
https://ww1.microchip.com/downloads/aemDocuments/documents/OTH/ProductDocuments/DataSheets/KSZ9477S-Data-Sheet-DS00002392C.pdf

-- 
Best regards,
Marek Vasut