lkml.org 
[lkml]   [2002]   [Apr]   [12]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
SubjectRE: serial driver question
Date
Guennadi Liakhovetski wrote:
> Hello all
>
> The function
> static int size_fifo(struct async_struct *info)
> {
> ...
> ends as follows:
> serial_outp(info, UART_LCR, UART_LCR_DLAB);
> serial_outp(info, UART_DLL, old_dll);
> serial_outp(info, UART_DLM, old_dlm);
>
> return count;
> }
>
> Which means, that DLAB is not re-set, and, in particular, all
> subsequent read/write operations on offsets 0 and 1 will not
> affect the data and interrupt enable registers, but the divisor
> latch register... Or is this register somehow magically restored
> elsewhere or by the hardware (say, on an interrupt)? This
> function seems to be only called for startech UARTs.

Hi Guennadi,

It is somehow magically restored elsewhere. The divisor latch bit (DLAB)
does remain set when size_fifo() returns, but the state does not persist.
Here is why the code works anyway, albeit with a trap or two for the
insufficiently paranoid ... and it can be explained without magic. :)

Notice that function size_fifo() does not save the LCR value before changing
it. Static function size_fifo() is called only from the very end of static
function autoconfig_startech_uarts(). Notice that function
autoconfig_startech_uarts() never returns with state->type set to
PORT_16550A. All code paths change state->type to something else. Static
function autoconfig_startech_uarts() is only called from the middle of
static function autoconfig(). Function autoconfig() saves the LCR value
before the testing of the alleged UART, which may include calling
autoconfig_startech_uarts(). When autoconfig_startech_uarts() returns to
autoconfig(), state->type cannot be PORT_16550A so the bodies of the next
two "if" statements that check state->type are skipped. The next line of
UART touching code that runs after size_fifo() and
autoconfig_startech_uarts() return to autoconfig(), is at about 50 lines
after the call to autoconfig_startech_uarts(), just before the check for
state->type == PORT_16450:

serial_outp(info, UART_LCR, save_lcr);

which restores autoconfig()'s saved LCR value, including the DLAB bit.

Yeah, I agree that size_fifo() should clean up after itself, but it doesn't
break anything as the code currently sits. Care to submit a one-line patch
for size_fifo() to clean up the UART state?

Best regards,
Ed Vance

----------------------------------------------------------------
Ed Vance edv@macrolink.com
Macrolink, Inc. 1500 N. Kellogg Dr Anaheim, CA 92807
----------------------------------------------------------------
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:22    [W:0.032 / U:0.880 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site