master driver doesn't release sda before get_slave_ack()

[kjhhgt76]

It seems that if the last addr/data bit to sda is 0, the sda_out will keep being 0, which causes the sda in i2c_if is hold by the master driver during ack/nack period.

[dovstamler]

Do you have a pull up on the bus? The pull up should be bringing the signals back to a logic one. dov

…

On Thu, Mar 14, 2024 at 3:38 AM kjhhgt76 ***@***.***> wrote: It seems that if the last addr/data bit to sda is 0, the sda_out will keep being 0, which causes the sda in i2c_if is pulled down to 0 during ack/nack period. — Reply to this email directly, view it on GitHub <#5>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABZCJRGULA37FT22KGG5UILYYD5QNAVCNFSM6AAAAABEVGQIP6VHI2DSMVQWIX3LMV43ASLTON2WKOZSGE4DKMRRGEYTSOI> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[kjhhgt76]

In get_slave_ack() task inside i2c_master.sv,
you can see that it doesn't release the bus after sigs.drv_cb.scl_in === 1'b0 if the previous sigs.drv_cb.sda_out is 0.
And I checked that before every call of get_slave_ack() there aren't mechanism to reset sigs.drv_cb.sda_out to 1.

In i2c_if.sv, assign sda = sda_out ? 1'bz : sda_out; will drive a strong0 to sda when sda_out = 0.
In this case, during the ack/nack period the sda bus is still driven by a strong0. If the slave tries to drive it with 1, it will result in X. Even you did pullup(sda) externally.