Re: [PATCH] i2c-mpc: generate START condition after STOP caused by read i2c_msg

[Esben Haabendal <esbenhaabendal-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>]

Hi

Your patch (and the small addition to mpc_i2c_start) does not work for me.

[   35.765803] mpc_xfer()
[   35.785480] writeccr 0
[   35.785505] writeccr 80
[   35.785523] mpc_xfer: 1 bytes to 2c:W - 1 of 2 messages
[   35.798817] mpc_write addr=2c len=1 restart=0
[   35.815327] writeccr f0
[   35.815503] I2C: SR=a2
[   35.818675] I2C: SR=a6
[   35.821450] mpc_xfer: 1 bytes to 2c:R - 2 of 2 messages
[   35.827119] mpc_read addr=2c len=1 restart=1
[   35.837463] writeccr f4
[   35.837641] I2C: SR=a6
[   35.840011] writeccr e8
[   35.840133] I2C: SR=a3
[   35.843596] writeccr 80
[   35.843632] mpc_xfer()
[   35.855068] writeccr 0
[   35.855093] writeccr 80
[   35.855111] mpc_xfer: 1 bytes to 2c:W - 1 of 2 messages
[   35.865346] mpc_write addr=2c len=1 restart=0
[   35.870109] writeccr f0
[   35.870272] I2C: SR=a2
[   35.873372] I2C: SR=a6
[   35.875757] mpc_xfer: 1 bytes to 2c:R - 2 of 2 messages
[   35.881606] mpc_read addr=2c len=1 restart=1
[   35.886290] writeccr f4
[   35.886463] I2C: SR=a6
[   35.889425] writeccr e8
[   35.889575] I2C: SR=a7
[   35.891944] writeccr 80
[   35.961177] mpc_xfer()
[   35.972517] writeccr 0
[   35.972541] writeccr 80
[   35.972559] mpc_xfer: 1 bytes to 4e:W - 1 of 20 messages
[   35.982628] mpc_write addr=4e len=1 restart=0
[   35.987389] writeccr f0
[   35.987424] I2C: SR=b3
[   35.990386] I2C: MAL
[   35.992971] i2c_wait(address) error: -5
[   35.997215] writeccr 80
[   35.997241] Error: i2c_transfer failed: -5

I have now spent a few hours trying a lot of different paths to fix
this approach, but I simply cannot find a way to get i2c read to work
without a trailing STOP condition with this controller.

Is there a problem in applying my patch, as it should be "clean"
improvement over the current implementation, which uses STOP
condition, but does not work.

Until Isomeone finds a way to get it to work without STOP, we will
have a working driver, which I assume is what we want.

Best regards,
Esben Haabendal

-- 
Esben Haabendal, Senior Software Consultant
DoréDevelopment ApS, Ved Stranden 1, 9560 Hadsund, DK-Denmark
Phone: +45 51 92 53 93, E-mail: eha-SIcX2qgNSybd/EuGEV170n9LOBIZ5rWg@public.gmane.org
WWW: http://www.doredevelopment.dk