mbox series

[v3,0/6] mtd: rawnand: atmel: Convert the driver to exec_op()

Message ID 20200518161754.302415-1-boris.brezillon@collabora.com
Headers show
Series mtd: rawnand: atmel: Convert the driver to exec_op() | expand

Message

Boris Brezillon May 18, 2020, 4:17 p.m. UTC 
Hello,

This v3 addresses a few problems reported by Tudor (one of the fix has
been sent separately [1]). There's still a perf regression on sama5
(200KB/s drop in the write path), but it seems to be related to a core
change (when switching to exec_op(), the page_prog_end() does a
read_status() which was not present in cmdfunc()) and the fact that
the NFC is bad at handling simple commands.

What's probably more worrisome is the difference between the NFC and
the !NFC case (not introduced by this series). I suspect it was to do
with HSMC specific timings, or maybe the NFC simply behaves poorly
compared to direct bus accesses. In any case, this issue is orthogonal
to the exec_op() conversion, and can, IMHO, be addressed separately.

Regards,

Boris

[1]https://patchwork.ozlabs.org/project/linux-mtd/patch/20200518155237.297549-1-boris.brezillon@collabora.com/

Boris Brezillon (6):
  mtd: rawnand: atmel: Enable the NFC controller at probe time
  mtd: rawnand: atmel: Drop redundant nand_read_page_op()
  mtd: rawnand: atmel: Use nand_{write,read}_data_op()
  mtd: rawnand: atmel: Use nand_prog_page_end_op()
  mtd: rawnand: atmel: Convert the driver to exec_op()
  mtd: rawnand: atmel: Get rid of the legacy interface implementation

 drivers/mtd/nand/raw/atmel/nand-controller.c | 470 +++++++++++--------
 1 file changed, 264 insertions(+), 206 deletions(-)