Re: [PATCH 1/4] net: sfp: add workaround for Realtek RTL8672 and RTL9601C chips

[Pali Rohár]

On Wednesday 30 December 2020 16:10:37 Russell King - ARM Linux admin wrote:
> On Wed, Dec 30, 2020 at 04:47:52PM +0100, Pali Rohár wrote:
> > Workaround for GPON SFP modules based on VSOL V2801F brand was added in
> > commit 0d035bed2a4a ("net: sfp: VSOL V2801F / CarlitoxxPro CPGOS03-0490
> > v2.0 workaround"). But it works only for ids explicitly added to the list.
> > As there are more rebraded VSOL V2801F modules and OEM vendors are putting
> > into vendor name random strings we cannot build workaround based on ids.
> > 
> > Moreover issue which that commit tried to workaround is generic not only to
> > VSOL based modules, but rather to all GPON modules based on Realtek RTL8672
> > and RTL9601C chips.
> > 
> > They can be found for example in following GPON modules:
> > * V-SOL V2801F
> > * C-Data FD511GX-RM0
> > * OPTON GP801R
> > * BAUDCOM BD-1234-SFM
> > * CPGOS03-0490 v2.0
> > * Ubiquiti U-Fiber Instant
> > * EXOT EGS1
> > 
> > Those Realtek chips have broken EEPROM emulator which for N-byte read
> > operation returns just one byte of EEPROM data followed by N-1 zeros.
> > 
> > So introduce a new function sfp_id_needs_byte_io() which detects SFP
> > modules with these Realtek chips which have broken EEPROM emulator based on
> > N-1 zeros and switch to 1 byte EEPROM reading operation which workaround
> > this issue.
> > 
> > This patch fixes reading EEPROM content from SFP modules based on Realtek
> > RTL8672 and RTL9601C chips.
> > 
> > Fixes: 0d035bed2a4a ("net: sfp: VSOL V2801F / CarlitoxxPro CPGOS03-0490 v2.0 workaround")
> > Co-developed-by: Russell King <rmk+kernel@xxxxxxxxxxxxxxx>
> > Signed-off-by: Russell King <rmk+kernel@xxxxxxxxxxxxxxx>
> > Signed-off-by: Pali Rohár <pali@xxxxxxxxxx>
> > ---
> >  drivers/net/phy/sfp.c | 78 ++++++++++++++++++++++++-------------------
> >  1 file changed, 44 insertions(+), 34 deletions(-)
> > 
> > diff --git a/drivers/net/phy/sfp.c b/drivers/net/phy/sfp.c
> > index 91d74c1a920a..490e78a72dd6 100644
> > --- a/drivers/net/phy/sfp.c
> > +++ b/drivers/net/phy/sfp.c
> > @@ -336,19 +336,11 @@ static int sfp_i2c_read(struct sfp *sfp, bool a2, u8 dev_addr, void *buf,
> >  			size_t len)
> >  {
> >  	struct i2c_msg msgs[2];
> > -	size_t block_size;
> > +	u8 bus_addr = a2 ? 0x51 : 0x50;
> > +	size_t block_size = sfp->i2c_block_size;
> >  	size_t this_len;
> > -	u8 bus_addr;
> >  	int ret;
> >  
> > -	if (a2) {
> > -		block_size = 16;
> > -		bus_addr = 0x51;
> > -	} else {
> > -		block_size = sfp->i2c_block_size;
> > -		bus_addr = 0x50;
> > -	}
> > -
> 
> NAK. You are undoing something that is definitely needed. The
> diagnostics must be read with sequential reads to be able to properly
> read the 16-bit values.

This change is really required for those Realtek chips. I thought that
it is obvious that from *both* addresses 0x50 and 0x51 can be read only
one byte at the same time. Reading 2 bytes (for be16 value) cannot be
really done by one i2 transfer, it must be done in two.

Therefore I had to "undo" this change as it is breaking support for all
those Realtek SFPs, including VSOL. That is why I also put Fixes tag
into commit message as it is really fixing reading for VSOL SFP from
address 0x51.

I understand that you want to read __be16 val in sfp_hwmon_read_sensor()
function via one i2c transfer to have "consistent" value, but it is
impossible on these Realtek chips.

I have already tested that sfp_hwmon_read_sensor() function see just
zero on second byte when is trying to use 2-byte read via one i2c
transfer.

When it use two i2c transfers, one for low 8bits and one for high 8bits
then reading is working.

> The rest of the patch is fine; it's a shame the entire thing has
> been spoilt by this extra addition that was not in the patch we had
> been discussing off-list.

Sorry for that. I really thought that it is obvious that this change
needs to be undone and read must always use byte transfer if we detect
these broken Realtek chips.