From: Dan Christian <Daniel.A.Christian@NASA.gov>
To: linux-kernel@vger.kernel.org
Subject: [PATCH] 16654 UART drops transmit characters
Date: Wed, 12 Nov 2003 14:05:01 -0800 [thread overview]
Message-ID: <200311121405.01631.Daniel.A.Christian@NASA.gov> (raw)
[-- Attachment #1: Type: text/plain, Size: 763 bytes --]
A 16654 UART (with 64 byte FIFO and hardware flow control) can drop
characters on TRANSMIT if hardware flow control is used.
No UART should ever lose data on transmit, so I consider this a major
bug. The bug only shows up if you write chunks of data >64 bytes.
Enclosed is a patch that fixes the problem. I have tested it under
2.4.20. It also applies cleanly under 2.4.22.
This has been reported before, but apparently the patch never made it
into the Linux kernel. Ed Vance (edv) at MacroLink developed this fix.
Part of it fixed some bad code for supporting the buggy Elan
interrupts. That part got fixed and is not in the attached patch.
Other than that, I just updated Ed's patch to apply cleanly against
current kernels.
-Dan Christian
[-- Attachment #2: patch_elan4 --]
[-- Type: text/x-diff, Size: 1398 bytes --]
--- linux-2.4.20-20.7/drivers/char/serial.c.orig Mon Aug 18 11:33:18 2003
+++ linux-2.4.20-20.7/drivers/char/serial.c Wed Nov 12 11:27:31 2003
@@ -797,6 +797,23 @@
}
}
+/*
+ * Returns != 0 (true) if UART is ready for more transmit data.
+ * This function contains a work-around for a silicon bug in the UARTs
+ * of the AMD Elan microcontroller. The LSR THRE bit is set late and
+ * may be missed by the interrupt routine, so an IIR THRI status is also
+ * treated as an LSR THRE status. This causes xmit data loss on 16C654
+ * UARTs (and perhaps others) so the work-around is applied only to
+ * ports detected as generic UART types 8250, 16450, 16550 and 16550A.
+ */
+static _INLINE_ int uart_transmit_ready(struct async_struct *info,
+ int status, int iir)
+{
+ return (status & UART_LSR_THRE) ||
+ ((info->state->type <= PORT_16550A) &&
+ ((iir & UART_IIR_ID) == UART_IIR_THRI));
+}
+
#ifdef CONFIG_SERIAL_SHARE_IRQ
/*
* This is the serial driver's generic interrupt routine
@@ -919,9 +936,7 @@
if (status & UART_LSR_DR)
receive_chars(info, &status, regs);
check_modem_status(info);
- if ((status & UART_LSR_THRE) ||
- /* For buggy ELAN processors */
- ((iir & UART_IIR_ID) == UART_IIR_THRI))
+ if (uart_transmit_ready(info, status, iir))
transmit_chars(info, 0);
if (pass_counter++ > RS_ISR_PASS_LIMIT) {
#if SERIAL_DEBUG_INTR
reply other threads:[~2003-11-12 22:05 UTC|newest]
Thread overview: [no followups] expand[flat|nested] mbox.gz Atom feed
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=200311121405.01631.Daniel.A.Christian@NASA.gov \
--to=daniel.a.christian@nasa.gov \
--cc=dChristian@mail.arc.nasa.gov \
--cc=linux-kernel@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
Powered by JetHome