From: Andy Walls <awalls@md.metrocast.net>
To: linux-media@vger.kernel.org
Subject: [PATCH 02/13] lirc_zilog: Remove broken, ineffective reference counting
Date: Thu, 17 Feb 2011 20:13:50 -0500 [thread overview]
Message-ID: <1297991630.9399.18.camel@localhost> (raw)
In-Reply-To: <1297991502.9399.16.camel@localhost>
The set_use_inc() and set_use_dec() functions tried to lock
the underlying bridge driver device instance in memory by
changing the use count on the device's i2c_clients. This
worked for PCI devices (ivtv, cx18, bttv). It doesn't
work for hot-pluggable usb devices (pvrusb2 and hdpvr).
With usb device instances, the driver may get locked into
memory, but the unplugged hardware is gone.
The set_use_inc() set_use_dec() functions also tried to have
lirc_zilog change its own module refernce count, which is
racy and not guaranteed to work. The lirc_dev module does
actually perform proper module ref count manipulation on the
lirc_zilog module, so there is need for lirc_zilog to
attempt a buggy module get on itself anyway.
lirc_zilog also errantly called these functions on itself
in open() and close(), but lirc_dev did that already too.
So let's just gut the bodies of the set_use_*() functions,
and remove the extra calls to them from within lirc_zilog.
Proper reference counting of the struct IR, IR_rx, and IR_tx
objects -- to handle the case when the underlying
bttv, ivtv, cx18, hdpvr, or pvrusb2 bridge driver module or
device instance goes away -- will be added in subsequent
patches.
Signed-off-by: Andy Walls <awalls@md.metrocast.net>
---
drivers/staging/lirc/lirc_zilog.c | 32 +-------------------------------
1 files changed, 1 insertions(+), 31 deletions(-)
diff --git a/drivers/staging/lirc/lirc_zilog.c b/drivers/staging/lirc/lirc_zilog.c
index 7389b77..3a91257 100644
--- a/drivers/staging/lirc/lirc_zilog.c
+++ b/drivers/staging/lirc/lirc_zilog.c
@@ -305,34 +305,12 @@ static int lirc_thread(void *arg)
static int set_use_inc(void *data)
{
- struct IR *ir = data;
-
- if (ir->l.owner == NULL || try_module_get(ir->l.owner) == 0)
- return -ENODEV;
-
- /* lock bttv in memory while /dev/lirc is in use */
- /*
- * this is completely broken code. lirc_unregister_driver()
- * must be possible even when the device is open
- */
- if (ir->rx != NULL)
- i2c_use_client(ir->rx->c);
- if (ir->tx != NULL)
- i2c_use_client(ir->tx->c);
-
return 0;
}
static void set_use_dec(void *data)
{
- struct IR *ir = data;
-
- if (ir->rx)
- i2c_release_client(ir->rx->c);
- if (ir->tx)
- i2c_release_client(ir->tx->c);
- if (ir->l.owner != NULL)
- module_put(ir->l.owner);
+ return;
}
/* safe read of a uint32 (always network byte order) */
@@ -1098,7 +1076,6 @@ static struct IR *find_ir_device_by_minor(unsigned int minor)
static int open(struct inode *node, struct file *filep)
{
struct IR *ir;
- int ret;
unsigned int minor = MINOR(node->i_rdev);
/* find our IR struct */
@@ -1112,12 +1089,6 @@ static int open(struct inode *node, struct file *filep)
/* increment in use count */
mutex_lock(&ir->ir_lock);
++ir->open;
- ret = set_use_inc(ir);
- if (ret != 0) {
- --ir->open;
- mutex_unlock(&ir->ir_lock);
- return ret;
- }
mutex_unlock(&ir->ir_lock);
/* stash our IR struct */
@@ -1139,7 +1110,6 @@ static int close(struct inode *node, struct file *filep)
/* decrement in use count */
mutex_lock(&ir->ir_lock);
--ir->open;
- set_use_dec(ir);
mutex_unlock(&ir->ir_lock);
return 0;
--
1.7.2.1
next prev parent reply other threads:[~2011-02-18 1:13 UTC|newest]
Thread overview: 15+ messages / expand[flat|nested] mbox.gz Atom feed top
2011-02-18 1:11 [PATCH 0/13] lirc_zilog: Ref-counting and locking cleanup Andy Walls
2011-02-18 1:12 ` [PATCH 01/13] lirc_zilog: Restore checks for existence of the IR_tx object Andy Walls
2011-02-18 1:13 ` Andy Walls [this message]
2011-02-18 1:14 ` [PATCH 03/13] lirc_zilog: Convert ir_device instance array to a linked list Andy Walls
2011-02-18 1:15 ` [PATCH 04/13] lirc_zilog: Convert the instance open count to an atomic_t Andy Walls
2011-02-18 1:15 ` [PATCH 05/13] lirc_zilog: Use kernel standard methods for marking device non-seekable Andy Walls
2011-02-18 1:16 ` [PATCH 06/13] lirc_zilog: Don't acquire the rx->buf_lock in the poll() function Andy Walls
2011-02-18 1:17 ` [PATCH 07/13] lirc_zilog: Remove unneeded rx->buf_lock Andy Walls
2011-02-18 1:18 ` [PATCH 08/13] lirc_zilog: Always allocate a Rx lirc_buffer object Andy Walls
2011-02-18 1:19 ` [PATCH 09/13] lirc_zilog: Move constants from ir_probe() into the lirc_driver template Andy Walls
2011-02-18 1:20 ` [PATCH 10/13] lirc_zilog: Add ref counting of struct IR, IR_tx, and IR_rx Andy Walls
2011-02-18 1:20 ` [PATCH 11/13] lirc_zilog: Add locking of the i2c_clients when in use Andy Walls
2011-02-18 1:21 ` [PATCH 12/13] lirc_zilog: Fix somewhat confusing information messages in ir_probe() Andy Walls
2011-02-18 1:22 ` [PATCH 13/13] lirc_zilog: Update TODO list based on work completed and revised plans Andy Walls
2011-03-06 4:00 ` [PATCH 0/13] lirc_zilog: Ref-counting and locking cleanup Jarod Wilson
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=1297991630.9399.18.camel@localhost \
--to=awalls@md.metrocast.net \
--cc=linux-media@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