From: Stephen Evanchik <evanchsa@clarkson.edu>
To: linux-kernel@vger.kernel.org
Subject: [PATCH 2.5.54] hermes: serialization fixes
Date: Fri, 3 Jan 2003 12:39:29 -0500 [thread overview]
Message-ID: <200301031239.29226.evanchsa@clarkson.edu> (raw)
The hermes MAC controller module wasn't serializing BAP seek calls. This
caused an eventual Rx/Tx failure which equates tens of thousands of errors on
the wireless interface. A simple spinlock is used to keep things in line.
Any comments are appreciated, I don't believe this is the best solution, but
it is working well. Patches can be downloaded from here:
http://www.clarkson.edu/~evanchsa/software/kernel/patches/hermes.patch-2.4.20
http://www.clarkson.edu/~evanchsa/software/kernel/patches/hermes.patch-2.5.54
--- linux-2.5.54/drivers/net/wireless/hermes.h 2003-01-01 22:21:02.000000000
-0500
+++ linux-2.5.54-devel/drivers/net/wireless/hermes.h 2003-01-03
11:38:25.000000000 -0500
@@ -288,6 +288,9 @@
u16 inten; /* Which interrupts should be enabled? */
+ /* Lock to avoid tripping over ourselves */
+ spinlock_t baplock;
+
#ifdef HERMES_DEBUG_BUFFER
struct hermes_debug_entry dbuf[HERMES_DEBUG_BUFSIZE];
unsigned long dbufp;
--- linux-2.5.54/drivers/net/wireless/hermes.c 2003-01-01 22:21:13.000000000
-0500
+++ linux-2.5.54-devel/drivers/net/wireless/hermes.c 2003-01-03
11:41:51.000000000 -0500
@@ -407,11 +407,18 @@
{
int dreg = bap ? HERMES_DATA1 : HERMES_DATA0;
int err = 0;
+ unsigned long flags;
if ( (len < 0) || (len % 2) )
return -EINVAL;
+ /* Without this, system instability occurs */
+ spin_lock_irqsave((&hw->baplock), flags);
+
err = hermes_bap_seek(hw, bap, id, offset);
+
+ spin_unlock_irqrestore((&hw->baplock), flags);
+
if (err)
goto out;
@@ -433,11 +440,17 @@
{
int dreg = bap ? HERMES_DATA1 : HERMES_DATA0;
int err = 0;
+ unsigned long flags;
if ( (len < 0) || (len % 2) )
return -EINVAL;
+ spin_lock_irqsave((&hw->baplock), flags);
+
err = hermes_bap_seek(hw, bap, id, offset);
+
+ spin_unlock_irqrestore((&hw->baplock), flags);
+
if (err)
goto out;
@@ -463,6 +476,7 @@
int dreg = bap ? HERMES_DATA1 : HERMES_DATA0;
u16 rlength, rtype;
int nwords;
+ unsigned long flags;
if ( (bufsize < 0) || (bufsize % 2) )
return -EINVAL;
@@ -471,7 +485,12 @@
if (err)
goto out;
+ spin_lock_irqsave((&hw->baplock), flags);
+
err = hermes_bap_seek(hw, bap, rid, 0);
+
+ spin_unlock_irqrestore((&hw->baplock), flags);
+
if (err)
goto out;
@@ -505,8 +524,14 @@
int dreg = bap ? HERMES_DATA1 : HERMES_DATA0;
int err = 0;
int count;
+ unsigned long flags;
+ spin_lock_irqsave((&hw->baplock), flags);
+
err = hermes_bap_seek(hw, bap, rid, 0);
+
+ spin_unlock_irqrestore((&hw->baplock), flags);
+
if (err)
goto out;
next reply other threads:[~2003-01-03 17:34 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2003-01-03 17:39 Stephen Evanchik [this message]
2003-01-03 17:47 ` Benjamin LaHaise
2003-01-03 17:56 ` Stephen Evanchik
2003-01-06 17:40 ` Alexander Hoogerhuis
2003-01-06 18:36 ` Stephen Evanchik
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=200301031239.29226.evanchsa@clarkson.edu \
--to=evanchsa@clarkson.edu \
--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
all inboxes | Powered by JetHome®