1) finally you did what I suggested - great

2) it should, as long as the pull-ups still are at at least at 3v

from I2C Specification: http://www.nxp.com/documents/user_manual/UM10204.p​df

3.1.1 SDA and SCL signals Both SDA and SCL are bidirectional lines, connected to a positive supply voltage via a current-source or pull-up resistor (see Figure 3). When the bus is free, both lines are HIGH.

again, you need to cut the connection to see which chip holds them down, \

from I2C Specification: http://www.nxp.com/documents/user_manual/UM10204.p​df

3.1.1 SDA and SCL signals Both SDA and SCL are bidirectional lines, connected to a positive supply voltage via a current-source or pull-up resistor (see Figure 3). When the bus is free, both lines are HIGH.

again, you need to cut the connection to see which chip holds them down, my guess is that you somehow kick the external device into "wait mode"

Why not power the pull-ups from a spare output (configured for push-pull)?

that's a band-aid, something is wrong, I²C idle IS both high.

using band-aids often come back and bite you

again, @menchopez if you break the connection which chip has low o[pins?

the 'standard method of this is a "walking write"

erased cells read 0xff, so make your write a struct/array that has a 'marker byte (anything but 0xff) at the end. then read is one before the first 'entry' where the marker is 0xff a write is at the location of the first 'entry' with 0xff

the 'standard method of this is a "walking write"

erased cells read 0xff, so make your write a struct/array that has a 'marker byte (anything but oxff) at the end. then read is one before the first 'entry' where the marker is 0xff a wswrite is the location of the firts 'entry' with 0xff