Once set, file->private_data remains constant. So it's safe to access
it without holding ppp_mutex.
The PPP unit fields accessed while handling PPPIOCDETACH (pf->kind and
ppp->owner) are also constant and have been set before
file->private_data got assigned. So these too can be read without
holding ppp_mutex.
Finally, ppp_release() is called only if we're the only user of the
unit.
Therefore, we can avoid locking ppp_mutex completely for handling
PPPIOCDETACH. This removes locking dependency between ppp_mutex and
rtnl_mutex, which will allow holding ppp_mutex from an rtnetlink
context.
Signed-off-by: Guillaume Nault <g.nault@xxxxxxxxxxxx>
---
drivers/net/ppp/ppp_generic.c | 21 +++++++++++++--------
1 file changed, 13 insertions(+), 8 deletions(-)
diff --git a/drivers/net/ppp/ppp_generic.c b/drivers/net/ppp/ppp_generic.c
index 7329c72..c81e257 100644
--- a/drivers/net/ppp/ppp_generic.c
+++ b/drivers/net/ppp/ppp_generic.c
@@ -592,15 +592,11 @@ static long ppp_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
cmd, arg);
}
- mutex_lock(&ppp_mutex);
-
- pf = file->private_data;
- if (!pf) {
- err = -ENOTTY;
- goto out;
- }
-
if (cmd == PPPIOCDETACH) {
+ pf = file->private_data;
+ if (!pf)
+ return -ENOTTY;
+
/*
* We have to be careful here... if the file descriptor
* has been dup'd, we could have another process in the
@@ -626,6 +622,15 @@ static long ppp_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
} else
pr_warn("PPPIOCDETACH file->f_count=%ld\n",
atomic_long_read(&file->f_count));
+
+ return err;
+ }
+
+ mutex_lock(&ppp_mutex);
+
+ pf = file->private_data;
+ if (!pf) {
+ err = -ENOTTY;
goto out;
}
--
2.8.0.rc3
--
To unsubscribe from this list: send the line "unsubscribe linux-ppp" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
