Thread (28 messages) 28 messages, 5 authors, 2015-12-04

Re: [PATCH v3 2/5] tty: Introduce SER_RS485_SOFTWARE read-only flag for struct serial_rs485

From: Matwey V. Kornilov <hidden>
Date: 2015-12-04 17:50:49
Also in: lkml

2015-12-03 22:45 GMT+03:00 Peter Hurley [off-list ref]:
On 12/03/2015 12:29 PM, Matwey V. Kornilov wrote:
quoted
2015-12-03 17:41 GMT+03:00 Peter Hurley [off-list ref]:
quoted
Hi Matwey,

On 12/03/2015 12:50 AM, Matwey V. Kornilov wrote:
quoted
I am working on v4, where I completely redesigned implementation. And
now I think that it is considerably better than v3.
It looks like the following:
https://github.com/matwey/linux/commits/8520_rs485_v4
But it is not ready yet, there is a bug somewhere.

In the v4, each subdriver decides separately if it needs rs485
emulation support. Then it enables it like the following:
https://github.com/matwey/linux/commit/4455e425fc045713fb921ccec695fe183f1558f0
Before calling serial8250_rs485_emul_enabled, the driver enables
interrupt on empty shift register (they are always there for omap_).
Looks good.

Are you testing with CONFIG_SERIAL_8250_DMA=n first to simplify the
debug effort? DMA adds a completely different tx path.
Many thanks for the advice. I've just found that the bug is not in my code =)
Even with pure 4.3.0 I cannot open /dev/ttyS5 more than once. It just
hangs on open() and the process is in S+ state.
Hmm, that's odd. So

$ stty -a < /dev/ttyS5

hangs if something like below is running?

$ cat > /dev/ttyS5
Nonblocking mode works, blocking mode hands on tty_port_block_til_ready
https://bugzilla.kernel.org/show_bug.cgi?id=108851
quoted
quoted
Also, before submission, please shorten the identifiers. And Greg hates
functions returning bool so just expanded serial8250_rs485_emul_enabled()
inline.
I would like to keep it as API to hide implementation details.
In other words, I don't want to inline it in omap_8250_rs485_config.
quoted
Am I allowed to use `re' instead of rs485_emul in names?
Long names and constructs tend to obscure the execution flow.
Some of the names could be reduced where the meaning is obvious:

  serial8250_rts_on_send
  serial8250_rts_after_send
  serial8250_handle_start_timer
  serial8250_handle_stop_timer

These two I would inline into their lone call site:

  serial8250_rs485_emul_startup()
  serial8250_rs485_emul_shutdown()

serial8250_rs485_emul_start_tx  => __start_tx_rs485

rs485_emul => sw485/em485/emul485/soft485 ?

Or just rs485 (except for the field name and structs so as not to confuse
it with the port->rs485)

Just my 2¢

Regards,
Peter Hurley


-- 
With best regards,
Matwey V. Kornilov.
Sternberg Astronomical Institute, Lomonosov Moscow State University, Russia
119991, Moscow, Universitetsky pr-k 13, +7 (495) 9392382
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help