Hello Henrik Rydberg,
This is a semi-automatic email about new static checker warnings.
The patch 4fa3a5837b84: "HID: hid-multitouch: Switch to device
groups" from May 1, 2012, leads to the following Smatch complaint:
drivers/hid/hid-multitouch.c:735 mt_probe()
error: we previously assumed 'id' could be null (see line 694)
drivers/hid/hid-multitouch.c
693
694 if (id) {
^^^^
Old check.
695 for (i = 0; mt_classes[i].name ; i++) {
696 if (id->driver_data == mt_classes[i].name) {
697 mtclass = &(mt_classes[i]);
698 break;
699 }
700 }
701 }
702
[snip]
733 mt_post_parse(td);
734
735 if (id->vendor == HID_ANY_ID && id->product == HID_ANY_ID)
^^^^^^^^^^
New dereference removed a check.
736 mt_post_parse_default_settings(td);
737
regards,
dan carpenter
Hi Dan,
This is a semi-automatic email about new static checker warnings.
The patch 4fa3a5837b84: "HID: hid-multitouch: Switch to device
groups" from May 1, 2012, leads to the following Smatch complaint:
drivers/hid/hid-multitouch.c:735 mt_probe()
error: we previously assumed 'id' could be null (see line 694)
Yes, and the same patch also removes the reason 'id' could be null.
Thanks,
Henrik
On Wed, Aug 08, 2012 at 07:28:29PM +0200, Henrik Rydberg wrote:
Hi Dan,
quoted
This is a semi-automatic email about new static checker warnings.
The patch 4fa3a5837b84: "HID: hid-multitouch: Switch to device
groups" from May 1, 2012, leads to the following Smatch complaint:
drivers/hid/hid-multitouch.c:735 mt_probe()
error: we previously assumed 'id' could be null (see line 694)
Yes, and the same patch also removes the reason 'id' could be null.
Great. Could you remove the unneeded check in mt_probe() and give
me a Reported-by: tag?
regards,
dan carpenter
On Wed, Aug 08, 2012 at 09:07:58PM +0300, Dan Carpenter wrote:
On Wed, Aug 08, 2012 at 07:28:29PM +0200, Henrik Rydberg wrote:
quoted
Hi Dan,
quoted
This is a semi-automatic email about new static checker warnings.
The patch 4fa3a5837b84: "HID: hid-multitouch: Switch to device
groups" from May 1, 2012, leads to the following Smatch complaint:
drivers/hid/hid-multitouch.c:735 mt_probe()
error: we previously assumed 'id' could be null (see line 694)
Yes, and the same patch also removes the reason 'id' could be null.
Great. Could you remove the unneeded check in mt_probe() and give
me a Reported-by: tag?
Ok, now I get it - I was pretty sure that test was not in mainline
anymore, but obviously I was wrong. Jiri, here is a patch for it, at
your convenience.
Thanks,
Henrik
---
From 9e763540dd474648544da165955e44549a3990a0 Mon Sep 17 00:00:00 2001
From: Henrik Rydberg <redacted>
Date: Wed, 8 Aug 2012 21:45:50 +0200
Subject: [PATCH] HID: hid-multitouch: Remove misleading null test
A null test was left behind during the autoloading work;
the test was introduced by 8d179a9e, but was never completely
reverted.
Reported-by: Dan Carpenter <redacted>
Signed-off-by: Henrik Rydberg <redacted>
---
drivers/hid/hid-multitouch.c | 10 ++++------
1 file changed, 4 insertions(+), 6 deletions(-)
diff --git a/drivers/hid/hid-multitouch.c b/drivers/hid/hid-multitouch.c
index 59c8b5c..f027cc9 100644
--- a/drivers/hid/hid-multitouch.c
+++ b/drivers/hid/hid-multitouch.c
@@ -691,12 +691,10 @@ static int mt_probe(struct hid_device *hdev, const struct hid_device_id *id)
struct mt_device *td;
struct mt_class *mtclass = mt_classes; /* MT_CLS_DEFAULT */
- if (id) {
- for (i = 0; mt_classes[i].name ; i++) {
- if (id->driver_data == mt_classes[i].name) {
- mtclass = &(mt_classes[i]);
- break;
- }
+ for (i = 0; mt_classes[i].name ; i++) {
+ if (id->driver_data == mt_classes[i].name) {
+ mtclass = &(mt_classes[i]);
+ break;
}
}
--
1.7.11.4
On Wed, 8 Aug 2012, Henrik Rydberg wrote:
quoted
quoted
quoted
This is a semi-automatic email about new static checker warnings.
The patch 4fa3a5837b84: "HID: hid-multitouch: Switch to device
groups" from May 1, 2012, leads to the following Smatch complaint:
drivers/hid/hid-multitouch.c:735 mt_probe()
error: we previously assumed 'id' could be null (see line 694)
Yes, and the same patch also removes the reason 'id' could be null.
Great. Could you remove the unneeded check in mt_probe() and give
me a Reported-by: tag?
Ok, now I get it - I was pretty sure that test was not in mainline
anymore, but obviously I was wrong. Jiri, here is a patch for it, at
your convenience.
Applied, thanks Dan, thanks Henrik.
--
Jiri Kosina
SUSE Labs