Re: [PATCH v4 0/4] i2c: designware: improve performance for transfers
From: Christian Ruppert
Date: Thu Aug 25 2016 - 12:23:57 EST
On 24.08.2016 00:18, Lucas De Marchi wrote:
> Diff from v3:
> - Fix over 80chars in one place
> - Move check for adapter being able to dynamically update TAR to be done
> on probe time rather than init as requested by Jarkko
> For the previous version, Christian had added:
> Tested-by: Christian Ruppert <christian.ruppert@xxxxxxxxxxx>
> on TB101 with Linux-4.7
Tested-by: Christian Ruppert <christian.ruppert@xxxxxxxxxxx>
Re-tested this iteration in the same configuration (TB101, Linux-4.7).
Still works perfectly.
> And Jarkko added his Acked-by: Jarkko Nikula <jarkko.nikula@xxxxxxxxxxxxxxx>
> on patches 1 and 3 (now 4).
> There's a new patch #2 as a preparatory work to move the check
> mentioned above to i2c_dw_probe().
> v3 of "i2c: designware: do not disable adapter after transfer". Differences
> - Now there's a first patch that does not depend on IC_TAR being dynamically
> enabled/disabled: it just doesn't wait for the state change when not needed.
> - We added a patch that allows detecting if HW supports the dynamic TAR updates
> - In the last patch the bits were changed as suggested by Jarkko.
> - This is tested on BayTrail and CherryTrail, both of them returning true for
> "dynamically update TAR"
> JosÃ Roberto de Souza (1):
> i2c: designware: wait for disable/enable only if necessary
> Lucas De Marchi (3):
> i2c: designware: add common functions for locking
> i2c: designware: detect when dynamic tar update is possible
> i2c: designware: do not disable adapter after transfer
> drivers/i2c/busses/i2c-designware-core.c | 162 +++++++++++++++++++++----------
> drivers/i2c/busses/i2c-designware-core.h | 1 +
> 2 files changed, 111 insertions(+), 52 deletions(-)