Dick Hollenbeck pointed out[0] that serial usage for non-consoles was dramatically affected due to the atomic console implementation. Indeed, the implementation was taking the global console_atomic_lock for _all_ 8250 ports, regardless if they were consoles or not. This series fixes that by only using the cpu-lock if the port is a console. While making those changes, I realized that the clear counter/ier-cache is not needed, so that is also removed. Dick also pointed out that the IER accessor functions were using non-typical names. These are updated and moved to inline macros. Finally, I noticed that the fsl, ingenic, and mtk variants were directly accessing IER. These are now also appropriately wrapped. In his suggestion, Dick did optimizations regarding IER caching. However, these are _not_ included because they may have some unwanted side-effects. For example, I noticed a few places in 8250 code where the value of uart_8250_port->ier is expected to be different than the hardware register. John Ogness [0] https://www.spinics.net/lists/linux-rt-users/msg21480.html John Ogness (2): serial: 8250: only atomic lock for console serial: 8250: fsl/ingenic/mtk: fix atomic console drivers/tty/serial/8250/8250.h | 65 ++++++++++++++------- drivers/tty/serial/8250/8250_core.c | 6 +- drivers/tty/serial/8250/8250_dma.c | 4 +- drivers/tty/serial/8250/8250_fsl.c | 9 +++ drivers/tty/serial/8250/8250_ingenic.c | 7 +++ drivers/tty/serial/8250/8250_mtk.c | 29 ++++++++- drivers/tty/serial/8250/8250_port.c | 81 ++++++-------------------- 7 files changed, 109 insertions(+), 92 deletions(-) -- 2.20.1
