cachepc-linux

Fork of AMDESE/linux with modifications for CachePC side-channel attack
git clone https://git.sinitax.com/sinitax/cachepc-linux
Log | Files | Refs | README | LICENSE | sfeed.txt

fault-codes.rst (5205B)


      1=====================
      2I2C/SMBUS Fault Codes
      3=====================
      4
      5This is a summary of the most important conventions for use of fault
      6codes in the I2C/SMBus stack.
      7
      8
      9A "Fault" is not always an "Error"
     10----------------------------------
     11Not all fault reports imply errors; "page faults" should be a familiar
     12example.  Software often retries idempotent operations after transient
     13faults.  There may be fancier recovery schemes that are appropriate in
     14some cases, such as re-initializing (and maybe resetting).  After such
     15recovery, triggered by a fault report, there is no error.
     16
     17In a similar way, sometimes a "fault" code just reports one defined
     18result for an operation ... it doesn't indicate that anything is wrong
     19at all, just that the outcome wasn't on the "golden path".
     20
     21In short, your I2C driver code may need to know these codes in order
     22to respond correctly.  Other code may need to rely on YOUR code reporting
     23the right fault code, so that it can (in turn) behave correctly.
     24
     25
     26I2C and SMBus fault codes
     27-------------------------
     28These are returned as negative numbers from most calls, with zero or
     29some positive number indicating a non-fault return.  The specific
     30numbers associated with these symbols differ between architectures,
     31though most Linux systems use <asm-generic/errno*.h> numbering.
     32
     33Note that the descriptions here are not exhaustive.  There are other
     34codes that may be returned, and other cases where these codes should
     35be returned.  However, drivers should not return other codes for these
     36cases (unless the hardware doesn't provide unique fault reports).
     37
     38Also, codes returned by adapter probe methods follow rules which are
     39specific to their host bus (such as PCI, or the platform bus).
     40
     41
     42EAGAIN
     43	Returned by I2C adapters when they lose arbitration in master
     44	transmit mode:  some other master was transmitting different
     45	data at the same time.
     46
     47	Also returned when trying to invoke an I2C operation in an
     48	atomic context, when some task is already using that I2C bus
     49	to execute some other operation.
     50
     51EBADMSG
     52	Returned by SMBus logic when an invalid Packet Error Code byte
     53	is received.  This code is a CRC covering all bytes in the
     54	transaction, and is sent before the terminating STOP.  This
     55	fault is only reported on read transactions; the SMBus slave
     56	may have a way to report PEC mismatches on writes from the
     57	host.  Note that even if PECs are in use, you should not rely
     58	on these as the only way to detect incorrect data transfers.
     59
     60EBUSY
     61	Returned by SMBus adapters when the bus was busy for longer
     62	than allowed.  This usually indicates some device (maybe the
     63	SMBus adapter) needs some fault recovery (such as resetting),
     64	or that the reset was attempted but failed.
     65
     66EINVAL
     67	This rather vague error means an invalid parameter has been
     68	detected before any I/O operation was started.  Use a more
     69	specific fault code when you can.
     70
     71EIO
     72	This rather vague error means something went wrong when
     73	performing an I/O operation.  Use a more specific fault
     74	code when you can.
     75
     76ENODEV
     77	Returned by driver probe() methods.  This is a bit more
     78	specific than ENXIO, implying the problem isn't with the
     79	address, but with the device found there.  Driver probes
     80	may verify the device returns *correct* responses, and
     81	return this as appropriate.  (The driver core will warn
     82	about probe faults other than ENXIO and ENODEV.)
     83
     84ENOMEM
     85	Returned by any component that can't allocate memory when
     86	it needs to do so.
     87
     88ENXIO
     89	Returned by I2C adapters to indicate that the address phase
     90	of a transfer didn't get an ACK.  While it might just mean
     91	an I2C device was temporarily not responding, usually it
     92	means there's nothing listening at that address.
     93
     94	Returned by driver probe() methods to indicate that they
     95	found no device to bind to.  (ENODEV may also be used.)
     96
     97EOPNOTSUPP
     98	Returned by an adapter when asked to perform an operation
     99	that it doesn't, or can't, support.
    100
    101	For example, this would be returned when an adapter that
    102	doesn't support SMBus block transfers is asked to execute
    103	one.  In that case, the driver making that request should
    104	have verified that functionality was supported before it
    105	made that block transfer request.
    106
    107	Similarly, if an I2C adapter can't execute all legal I2C
    108	messages, it should return this when asked to perform a
    109	transaction it can't.  (These limitations can't be seen in
    110	the adapter's functionality mask, since the assumption is
    111	that if an adapter supports I2C it supports all of I2C.)
    112
    113EPROTO
    114	Returned when slave does not conform to the relevant I2C
    115	or SMBus (or chip-specific) protocol specifications.  One
    116	case is when the length of an SMBus block data response
    117	(from the SMBus slave) is outside the range 1-32 bytes.
    118
    119ESHUTDOWN
    120	Returned when a transfer was requested using an adapter
    121	which is already suspended.
    122
    123ETIMEDOUT
    124	This is returned by drivers when an operation took too much
    125	time, and was aborted before it completed.
    126
    127	SMBus adapters may return it when an operation took more
    128	time than allowed by the SMBus specification; for example,
    129	when a slave stretches clocks too far.  I2C has no such
    130	timeouts, but it's normal for I2C adapters to impose some
    131	arbitrary limits (much longer than SMBus!) too.