From: Frank Rowand <frowand.list@gmail.com>
To: unlisted-recipients:; (no To-header on input)
Cc: Stephen Boyd <sboyd@codeaurora.org>,
David Brown <davidb@codeaurora.org>,
Daniel Walker <dwalker@fifo99.com>,
Bryan Huntsman <bryanh@codeaurora.org>,
Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
Jiri Slaby <jslaby@suse.cz>,
"linux-arm-msm@vger.kernel.org" <linux-arm-msm@vger.kernel.org>,
linux-serial@vger.kernel.org,
Linux Kernel list <linux-kernel@vger.kernel.org>,
Jason Wessel <jason.wessel@windriver.com>,
kgdb-bugreport@lists.sourceforge.net
Subject: [test patch 3/4] use poll_post_exception in msm_serial
Date: Mon, 04 Aug 2014 18:06:32 -0700 [thread overview]
Message-ID: <53E02E18.60602@gmail.com> (raw)
In-Reply-To: <53E02C7C.4090206@gmail.com>
From: Frank Rowand <frank.rowand@sonymobile.com>
**********************************************************
I need help with this patch - it does not fix the issue.
**********************************************************
Use framework to allow msm_serial driver to fixup state after operating in
polled mode, before returning to interrupt mode.
I do not have complete documentation on the hardware, so this patch is a
set of shots in the dark. Each experiment can be enabled by enabling
a #define.
The issue I am trying to resolve with this patch is:
kgdb properly communicates with the dragon board, but
following the continue command, the serial driver does not get any stale
(UART_IMR_RXSTALE) interrupts until 48 characters have been read, which
triggers a high water interrupt. After the high water interrupt has been
processed, the driver resumes properly getting stale interrupts.
msm_poll_post_exception() will be called near the tail end of kgdb_cpu_enter(),
via the path invoked by:
/* Call the I/O driver's post_exception routine */
if (dbg_io_ops->post_exception)
dbg_io_ops->post_exception();
Not-signed-off-by-yet: Frank Rowand <frank.rowand@sonymobile.com>
---
drivers/tty/serial/msm_serial.c | 196 ++++++++++++++++++++++++++++++++++++++++
1 file changed, 196 insertions(+)
Index: b/drivers/tty/serial/msm_serial.c
===================================================================
--- a/drivers/tty/serial/msm_serial.c
+++ b/drivers/tty/serial/msm_serial.c
@@ -881,6 +881,201 @@ static void msm_poll_put_char(struct uar
return;
}
+
+int zzz_poll_put_char_reset;
+static void msm_poll_post_exception(struct uart_port *port)
+{
+ int misr;
+ struct msm_port *msm_port = UART_TO_MSM(port);
+
+ zzz_poll_put_char_reset++;
+
+ msm_port->old_snap_state = 0;
+
+// #define ZZZ_ALT_1
+#ifdef ZZZ_ALT_1
+ /*
+ * zzz alternate 1 -- trigger high water after continue <---- bad
+ * high water count == 48, set old_snap_state = 48
+ * then stale total snap == 49, so count == 1 <---- good
+ *
+ * better than alternate 2
+ */
+ msm_read(port, UARTDM_RX_TOTAL_SNAP);
+
+#if 0
+ misr = msm_read(port, UART_MISR);
+ if (misr & (UART_IMR_RXSTALE))
+#endif
+ msm_write(port, UART_CR_CMD_RESET_STALE_INT, UART_CR);
+ msm_write(port, 0xFFFFFF, UARTDM_DMRX);
+ msm_write(port, UART_CR_CMD_STALE_EVENT_ENABLE, UART_CR);
+#endif /* ZZZ_ALT_1 */
+
+
+// #define ZZZ_ALT_2
+#ifdef ZZZ_ALT_2
+ /*
+ * zzz alternate 2 -- trigger high water after continue <---- bad
+ * high water count == 48, set old_snap_state = 48
+ * next total snap == 310
+ * then stale total snap == 310, so count == 262 <---- bad
+ */
+ msm_read(port, UARTDM_RX_TOTAL_SNAP);
+
+ msm_write(port, UART_CR_CMD_FORCE_STALE, UART_CR);
+#endif
+
+
+// #define ZZZ_ALT_3_4
+#ifdef ZZZ_ALT_3_4
+ /*
+ * zzz alternate 3 -- never trigger after continue <---- bad
+ */
+ msm_read(port, UARTDM_RX_TOTAL_SNAP);
+
+ msm_write(port, 0xFFFFFF, UARTDM_DMRX);
+ msm_write(port, UART_CR_CMD_STALE_EVENT_ENABLE, UART_CR);
+
+ msm_write(port, UART_CR_CMD_RESET_STALE_INT, UART_CR);
+ msm_write(port, UART_CR_CMD_FORCE_STALE, UART_CR);
+
+#if 0
+ /*
+ * zzz alternate 3, #if 1 this block
+ * zzz alternate 4, #if 0 this block
+ * zzz alternate 4 -- trigger high water after continue <---- bad
+ * high water count == 48, set old_snap_state = 48
+ * then stale total snap == 49, so count == 1 <---- good
+ */
+ misr = msm_read(port, UART_MISR);
+ while (!(misr & (UART_IMR_RXSTALE))) {
+ cpu_relax();
+ misr = msm_read(port, UART_MISR);
+ }
+#endif
+
+#endif /* ZZZ_ALT_3_4 */
+
+
+// #define ZZZ_ALT_5
+#ifdef ZZZ_ALT_5
+ /*
+ * zzz alternate 5 -- trigger high water after continue <---- bad
+ * high water count == 48, set old_snap_state = 48
+ * then stale total snap == 49, so count == 1 <---- good
+ */
+ msm_read(port, UARTDM_RX_TOTAL_SNAP);
+
+#if 0
+ misr = msm_read(port, UART_MISR);
+ if (misr & (UART_IMR_RXSTALE))
+#endif
+ msm_write(port, UART_CR_CMD_RESET_STALE_INT, UART_CR);
+ msm_write(port, 0xFFFFFF, UARTDM_DMRX);
+ msm_write(port, UART_CR_CMD_STALE_EVENT_ENABLE, UART_CR);
+
+ /* zzz should not have to do this, trying anyway */
+ /* restore interrupt */
+ msm_write(port, msm_port->imr, UART_IMR);
+#endif /* ZZZ_ALT_5 */
+
+
+// #define ZZZ_ALT_6
+#ifdef ZZZ_ALT_6
+ /*
+ * zzz alternate 6 -- trigger high water after continue <---- bad
+ * high water count == 48, set old_snap_state = 48
+ * then stale total snap == 49, so count == 1 <---- good
+ */
+ msm_read(port, UARTDM_RX_TOTAL_SNAP);
+
+ msm_write(port, 0, UART_IMR); /* disable interrupt */
+
+ misr = msm_read(port, UART_MISR);
+ if (misr & (UART_IMR_RXSTALE))
+ msm_write(port, UART_CR_CMD_RESET_STALE_INT, UART_CR);
+ msm_write(port, 0xFFFFFF, UARTDM_DMRX);
+ msm_write(port, UART_CR_CMD_STALE_EVENT_ENABLE, UART_CR);
+
+ msm_write(port, UART_CR_CMD_FORCE_STALE, UART_CR);
+
+ /* restore interrupt */
+ msm_write(port, msm_port->imr, UART_IMR);
+
+#endif /* ZZZ_ALT_6 */
+
+
+#define ZZZ_ALT_7
+#ifdef ZZZ_ALT_7
+ /*
+ * Try to emulate entire interrupt driver read path....
+ *
+ * zzzzzz what is missing that is fixed by the high water irq???
+ *
+ * same result for versions A, B, C
+ * zzz alternate 7 -- trigger high water after continue <---- bad
+ * high water count == 48, set old_snap_state = 48
+ * then stale total snap == 49, so count == 1 <---- good
+ */
+
+{
+ int imr_rx_stale;
+ unsigned int sr;
+
+ /*
+ * zzz Grabbing a lock here will result in a deadlock if a breakpoint
+ * zzz is hit while the lock is held elsewhere. It would be best to
+ * zzz avoid this lock if possible.
+ *
+ * zzz It would be better to do a trylock and warn on failure to
+ * zzz acquire.
+ */
+ spin_lock(&port->lock);
+
+ misr = msm_read(port, UART_MISR);
+ imr_rx_stale = misr & UART_IMR_RXSTALE;
+
+ msm_write(port, 0, UART_IMR); /* disable interrupt */
+
+ if ((msm_read(port, UART_SR) & UART_SR_OVERRUN))
+ msm_write(port, UART_CR_CMD_RESET_ERR, UART_CR);
+
+
+ /* zzz version A, read _either_ UARTDM_RX_TOTAL_SNAP or UART_RFWR */
+ /* zzz version B, read UARTDM_RX_TOTAL_SNAP and UART_RFWR */
+ /* zzz version C, read UART_RFWR */
+ if (imr_rx_stale)
+ msm_read(port, UARTDM_RX_TOTAL_SNAP);
+ else
+ msm_read(port, UART_RFWR);
+
+ sr = msm_read(port, UART_SR);
+ /* zzz this could be: while (sr & ...) {UARTDM_RF; UART_SR;} */
+ while ((sr & UART_SR_RX_READY) != 0) {
+ msm_read(port, UARTDM_RF);
+ sr = msm_read(port, UART_SR);
+ }
+
+ if (imr_rx_stale)
+ msm_write(port, UART_CR_CMD_RESET_STALE_INT, UART_CR);
+ msm_write(port, 0xFFFFFF, UARTDM_DMRX);
+ msm_write(port, UART_CR_CMD_STALE_EVENT_ENABLE, UART_CR);
+
+ msm_write(port, msm_port->imr, UART_IMR); /* restore interrupt */
+
+ spin_unlock(&port->lock);
+}
+
+#endif /* ZZZ_ALT_7 */
+
+
+
+ return;
+}
+
+
+
#endif
static struct uart_ops msm_uart_pops = {
@@ -905,6 +1100,7 @@ static struct uart_ops msm_uart_pops = {
.poll_init = msm_poll_init,
.poll_get_char = msm_poll_get_char,
.poll_put_char = msm_poll_put_char,
+ .poll_post_exception = msm_poll_post_exception,
#endif
};
next prev parent reply other threads:[~2014-08-05 1:06 UTC|newest]
Thread overview: 11+ messages / expand[flat|nested] mbox.gz Atom feed top
2014-08-05 0:59 [request for help] tty: serial: kgdb: fix msm_serial for kgdb Frank Rowand
2014-08-05 1:03 ` [test patch 1/4] " Frank Rowand
2014-08-05 1:04 ` [test patch 2/4] add poll_post_exception framework Frank Rowand
2014-08-05 1:06 ` Frank Rowand [this message]
2014-08-05 1:08 ` [test patch 4/4] debug hackery to trace msm_serial receive interrupts Frank Rowand
2014-08-05 1:33 ` [request for help] tty: serial: kgdb: fix msm_serial for kgdb Stephen Boyd
2014-08-05 2:09 ` Frank Rowand
2014-08-05 19:22 ` Stephen Boyd
2014-08-05 23:53 ` Stephen Boyd
2014-08-06 0:55 ` Frank Rowand
2014-08-06 1:19 ` Stephen Boyd
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=53E02E18.60602@gmail.com \
--to=frowand.list@gmail.com \
--cc=bryanh@codeaurora.org \
--cc=davidb@codeaurora.org \
--cc=dwalker@fifo99.com \
--cc=gregkh@linuxfoundation.org \
--cc=jason.wessel@windriver.com \
--cc=jslaby@suse.cz \
--cc=kgdb-bugreport@lists.sourceforge.net \
--cc=linux-arm-msm@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-serial@vger.kernel.org \
--cc=sboyd@codeaurora.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;
as well as URLs for NNTP newsgroup(s).