[RFC Patch v4 0/1] i2c statistics as sysfs attributes

[Sui Chen <suichen@google.com>]

This change renames the I2C debug counters as suggested, and moves their
processing into the i2c core:
- bus_errors
- transfers
- nacks
- recovery_successes
- recovery_failures
- timeouts

Did some brief tests with a few test programs that saves/replays I2C
trace by reading hwmon sensors. The test program and hardware run in
QEMU. The test programs are located at
https://gerrit.openbmc-project.xyz/c/openbmc/openbmc-tools/+/52527

(A normal read)

root@gsj:/tmp# cat /sys/class/i2c-adapter/i2c-1/stats/transfers
264
root@gsj:/tmp# ./i2c_bmk_bmc  0
idx=0
Processing 1 inputs
/sys/class/hwmon/hwmon0/temp1_input
(../../devices/platform/ahb/ahb:apb/f0081000.i2c/i2c-1/1-005c/hwmon/hwmon0):
0
[FindTraceEntries] t0=391.000000 t1=393.000000
Found 4 interesting I2C trace entries:
 i2c_write: i2c-1 #0 a=05c f=0000 l=1 [00]
 i2c_read: i2c-1 #1 a=05c f=0001 l=2
 i2c_reply: i2c-1 #1 a=05c f=0001 l=2 [00-00]
 i2c_result: i2c-1 n=2 ret=2
root@gsj:/tmp# cat /sys/class/i2c-adapter/i2c-1/stats/transfers
265

(Read from an inexistent address, NACK)

root@gsj:/tmp# cat /sys/class/i2c-adapter/i2c-1/stats/nacks
6
root@gsj:/tmp# cat /sys/class/i2c-adapter/i2c-1/stats/transfers
265

root@gsj:/tmp# cat i2c_trace.txt
 i2c_write: i2c-1 #0 a=0ff f=0000 l=1 [00]
 i2c_read: i2c-1 #1 a=0ff f=0001 l=2
root@gsj:/tmp# ./i2c_replay_bmc i2c_trace.txt
(program runs and finishes with 6 errors)

root@gsj:/tmp# cat /sys/class/i2c-adapter/i2c-1/stats/nacks
12
root@gsj:/tmp# cat /sys/class/i2c-adapter/i2c-1/stats/transfers
265

The program makes 2 attempts, performing 3 I2C operations at each
attempt, which translates to the 6 nack events.

We will do more testing.

Sui Chen (1):
  i2c debug counters as sysfs attributes

 drivers/i2c/i2c-core-base.c | 32 ++++++++++++-
 drivers/i2c/i2c-dev.c       | 94 +++++++++++++++++++++++++++++++++++++
 include/linux/i2c.h         | 26 ++++++++++
 3 files changed, 151 insertions(+), 1 deletion(-)

-- 
2.35.1.1094.g7c7d902a7c-goog