Skip to contentSkip to conversation

fix permission inconsistency in relation to symlinks.

PATCH Applied Michael Buesch 3 participants 9 messages Feb 5, 2005

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 fix permission inconsistency in relation to symlinks.

Hi,

This fixes a rules.d permission apply inconsistency. Here's an example of what happens without the patch:

First case without the patch:

KERNEL="hd*", GROUP="disk", MODE="0660"
KERNEL="hdc", SYMLINK="cdrecorder", MODE="0666"

mb@lfs:~$ ls -la /dev/hdc 
brw-rw----  1 root disk 22, 0 Feb  5 14:08 /dev/hdc

Second case without the patch:

KERNEL="hd*", GROUP="disk", MODE="0660"
KERNEL="hdc", MODE="0666"

mb@lfs:~$ ls -la /dev/hdc 
brw-rw-rw-  1 root disk 22, 0 Feb  5 14:08 /dev/hdc

With the patch applied:

KERNEL="hd*", GROUP="disk", MODE="0660"
KERNEL="hdc", SYMLINK="cdrecorder", MODE="0666"

mb@lfs:~$ ls -la /dev/hdc 
brw-rw-rw-  1 root disk 22, 0 Feb  5 14:08 /dev/hdc
Signature · 29 lines
Regards Michael Buesch  [ http://www.tuxsoft.de.vu ]


===== namedev.c 1.173 vs edited =====
--- 1.173/namedev.c	2004-12-21 17:41:02 +01:00
+++ edited/namedev.c	2005-02-05 14:04:44 +01:00
@@ -716,14 +716,16 @@
 	list_for_each_entry(dev, &config_device_list, node) {
 		dbg("process rule");
 		if (match_rule(udev, dev, class_dev, sysfs_device) == 0) {
-			if (dev->name[0] == '\0' && dev->symlink[0] == '\0') {
+
+			if (dev->name[0] == '\0' && dev->symlink[0] == '\0' &&
+			    dev->mode == 0000 && dev->owner[0] == '\0' && dev->group[0] == '\0') {
 				/* empty name, symlink, perms will not create any node */
-				if (dev->mode == 0000 && dev->owner[0] == '\0' && dev->group[0] == '\0') {
-					info("configured rule in '%s[%i]' applied, '%s' is ignored",
-					     dev->config_file, dev->config_line, udev->kernel_name);
-					return -1;
-				}
+				info("configured rule in '%s[%i]' applied, '%s' is ignored",
+				     dev->config_file, dev->config_line, udev->kernel_name);
+				return -1;
+			}
 
+			if (dev->name[0] == '\0') {
 				/* apply permissions only rule values */
 				if (dev->mode != 0000) {
 					udev->mode = dev->mode;

===== namedev.c 1.173 vs edited =====

namedev.c +8 -6
@@ -716,14 +716,16 @@
 	list_for_each_entry(dev, &config_device_list, node) {
 		dbg("process rule");
 		if (match_rule(udev, dev, class_dev, sysfs_device) == 0) {
-			if (dev->name[0] == '\0' && dev->symlink[0] == '\0') {
+
+			if (dev->name[0] == '\0' && dev->symlink[0] == '\0' &&
+			    dev->mode == 0000 && dev->owner[0] == '\0' && dev->group[0] == '\0') {
 				/* empty name, symlink, perms will not create any node */
-				if (dev->mode == 0000 && dev->owner[0] == '\0' && dev->group[0] == '\0') {
-					info("configured rule in '%s[%i]' applied, '%s' is ignored",
-					     dev->config_file, dev->config_line, udev->kernel_name);
-					return -1;
-				}
+				info("configured rule in '%s[%i]' applied, '%s' is ignored",
+				     dev->config_file, dev->config_line, udev->kernel_name);
+				return -1;
+			}
 
+			if (dev->name[0] == '\0') {
 				/* apply permissions only rule values */
 				if (dev->mode != 0000) {
 					udev->mode = dev->mode;
#1 lore raw

Re: [PATCH, UDEV] fix permission inconsistency in relation to

Quoted 9 lines · Michael Buesch

This fixes a rules.d permission apply inconsistency. Here's an example of what happens without the patch:

First case without the patch:

KERNEL="hd*", GROUP="disk", MODE="0660" KERNEL="hdc", SYMLINK="cdrecorder", MODE="0666" brw-rw---- 1 root disk 22, 0 Feb 5 14:08 /dev/hdc

Quoted 5 lines

Second case without the patch:

KERNEL="hd*", GROUP="disk", MODE="0660" KERNEL="hdc", MODE="0666"
brw-rw-rw- 1 root disk 22, 0 Feb 5 14:08 /dev/hdc

Both cases are correct behavior in the current model, where symlink-only rules can not set the permissions of a node. It was the same with the "old" .permissions files.

Should we change that now, so that permissions are applied from a symlink-only rule too?

Kay

-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting Tool for open source databases. Create drag-&-drop reports. Save time by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc. Download a FREE copy at http://www.intelliview.com/go/osdn_nl
_______________________________________________
Linux-hotplug-devel mailing list http://linux-hotplug.sourceforge.net
[email protected]
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

#2 lore raw
In reply to Michael Buesch
Quoted 2 lines · Kay Sievers

Should we change that now, so that permissions are applied from a symlink-only rule too?

Yes!
I think it's the sensible semantic: the second rule adds a new SYMLINK *and* sets the permissions of the node, which are two distinct operations applied by the same rule.
Unless you think it's broken, I will add the patch to the debian package of 051 (almost ready).

Signature · 2 lines
ciao,
Marco
#3 lore raw
In reply to Michael Buesch

Quoting Kay Sievers <[email protected]>:

Quoted 16 lines · Michael Buesch
Quoted 9 lines

This fixes a rules.d permission apply inconsistency. Here's an example of what happens without the patch:

First case without the patch:

KERNEL="hd*", GROUP="disk", MODE="0660" KERNEL="hdc", SYMLINK="cdrecorder", MODE="0666" brw-rw---- 1 root disk 22, 0 Feb 5 14:08 /dev/hdc

Quoted 5 lines

Second case without the patch:

KERNEL="hd*", GROUP="disk", MODE="0660" KERNEL="hdc", MODE="0666"
brw-rw-rw- 1 root disk 22, 0 Feb 5 14:08 /dev/hdc

Quoted 6 lines

Both cases are correct behavior in the current model, where symlink-only rules can not set the permissions of a node. It was the same with the "old" .permissions files.

Should we change that now, so that permissions are applied from a symlink-only rule too?

I was _very_ confused when I saw this behaviour. It took me a lot of time to figure out what's really going on there and how to workaround it.

Quoted 4 lines

Kay

Signature · 1 line
Regards Michael Buesch  [ http://www.tuxsoft.de.vu ]
#4 lore raw

Re: [PATCH, UDEV] fix permission inconsistency in relation to

In reply to Michael Buesch
Quoted 8 lines · Marco d'Itri
Quoted 2 lines · Kay Sievers

Should we change that now, so that permissions are applied from a symlink-only rule too?

Yes!
I think it's the sensible semantic: the second rule adds a new SYMLINK *and* sets the permissions of the node, which are two distinct operations applied by the same rule.

Yeah, we should do that now.

Quoted 2 lines

Unless you think it's broken, I will add the patch to the debian package of 051 (almost ready).

I think we can make it even simpler, by applying the permission unconditionally, right?

Thanks,
Kay

===== namedev.c 1.173 vs edited =====

namedev.c +26 -37
@@ -670,7 +670,6 @@ try_parent:
 		dbg("sysfs_device->path='%s'", sysfs_device->path);
 		dbg("sysfs_device->bus_id='%s'", sysfs_device->bus_id);
 	}
-
 }
 
 int namedev_name_device(struct udevice *udev, struct sysfs_class_device *class_dev)
@@ -716,32 +715,32 @@ int namedev_name_device(struct udevice *
 	list_for_each_entry(dev, &config_device_list, node) {
 		dbg("process rule");
 		if (match_rule(udev, dev, class_dev, sysfs_device) == 0) {
-			if (dev->name[0] == '\0' && dev->symlink[0] == '\0') {
-				/* empty name, symlink, perms will not create any node */
-				if (dev->mode == 0000 && dev->owner[0] == '\0' && dev->group[0] == '\0') {
-					info("configured rule in '%s[%i]' applied, '%s' is ignored",
-					     dev->config_file, dev->config_line, udev->kernel_name);
-					return -1;
-				}
-
-				/* apply permissions only rule values */
-				if (dev->mode != 0000) {
-					udev->mode = dev->mode;
-					dbg("applied mode=%#o to '%s'", udev->mode, udev->kernel_name);
-				}
-				if (dev->owner[0] != '\0') {
-					strfieldcpy(udev->owner, dev->owner);
-					apply_format(udev, udev->owner, sizeof(udev->owner), class_dev, sysfs_device);
-					dbg("applied owner='%s' to '%s'", udev->owner, udev->kernel_name);
-				}
-				if (dev->group[0] != '\0') {
-					strfieldcpy(udev->group, dev->group);
-					apply_format(udev, udev->group, sizeof(udev->group), class_dev, sysfs_device);
-					dbg("applied group='%s' to '%s'", udev->group, udev->kernel_name);
-				}
+
+			/* empty name, symlink, perms will not create any node */
+			if (dev->name[0] == '\0' && dev->symlink[0] == '\0' &&
+			    dev->mode == 0000 && dev->owner[0] == '\0' && dev->group[0] == '\0') {
+				info("configured rule in '%s[%i]' applied, '%s' is ignored",
+				     dev->config_file, dev->config_line, udev->kernel_name);
+				return -1;
+			}
+
+			/* apply permissions */
+			if (dev->mode != 0000) {
+				udev->mode = dev->mode;
+				dbg("applied mode=%#o to '%s'", udev->mode, udev->kernel_name);
+			}
+			if (dev->owner[0] != '\0') {
+				strfieldcpy(udev->owner, dev->owner);
+				apply_format(udev, udev->owner, sizeof(udev->owner), class_dev, sysfs_device);
+				dbg("applied owner='%s' to '%s'", udev->owner, udev->kernel_name);
+			}
+			if (dev->group[0] != '\0') {
+				strfieldcpy(udev->group, dev->group);
+				apply_format(udev, udev->group, sizeof(udev->group), class_dev, sysfs_device);
+				dbg("applied group='%s' to '%s'", udev->group, udev->kernel_name);
 			}
 
-			/* collect symlinks for the final matching rule */
+			/* collect symlinks for this or the final matching rule */
 			if (dev->symlink[0] != '\0') {
 				char temp[NAME_SIZE];
 
@@ -774,19 +773,9 @@ int namedev_name_device(struct udevice *
 					goto exit;
 
 				udev->partitions = dev->partitions;
-				if (dev->mode != 0000)
-					udev->mode = dev->mode;
-				if (dev->owner[0] != '\0') {
-					strfieldcpy(udev->owner, dev->owner);
-					apply_format(udev, udev->owner, sizeof(udev->owner), class_dev, sysfs_device);
-				}
-				if (dev->group[0] != '\0') {
-					strfieldcpy(udev->group, dev->group);
-					apply_format(udev, udev->group, sizeof(udev->group), class_dev, sysfs_device);
-				}
 
-				dbg("name, '%s' is going to have owner='%s', group='%s', mode = %#o",
-				    udev->name, udev->owner, udev->group, udev->mode);
+				dbg("name, '%s' is going to have owner='%s', group='%s', mode=%#o partitions=%i",
+				    udev->name, udev->owner, udev->group, udev->mode, udev->partitions);
 
 				goto exit;
 			}
#5 lore raw
In reply to Michael Buesch

Quoting Kay Sievers <[email protected]>:

Quoted 10 lines · Marco d'Itri
Quoted 8 lines · Kay Sievers
Quoted 3 lines

Should we change that now, so that permissions are applied from a symlink-only rule too?

Yes!
I think it's the sensible semantic: the second rule adds a new SYMLINK *and* sets the permissions of the node, which are two distinct operations applied by the same rule.

Quoted 7 lines

Yeah, we should do that now.

Quoted 2 lines

Unless you think it's broken, I will add the patch to the debian package of 051 (almost ready).

I think we can make it even simpler, by applying the permission unconditionally, right?

Yes, seems correct.
A first test showed that this works.

PS: Please CC: me on this topic, as I'm not subscribed to the mailing list. Thanks.

Quoted 3 lines

Thanks,
Kay

Signature · 1 line
Regards Michael Buesch  [ http://www.tuxsoft.de.vu ]
Applied
#6 lore raw

Re: [PATCH, UDEV] fix permission inconsistency in relation to

In reply to Michael Buesch
Quoted 13 lines · Michael Buesch

Quoting Kay Sievers <[email protected]>:

Quoted 12 lines · Marco d'Itri
Quoted 8 lines · Kay Sievers
Quoted 3 lines

Should we change that now, so that permissions are applied from a symlink-only rule too?

Yes!
I think it's the sensible semantic: the second rule adds a new SYMLINK *and* sets the permissions of the node, which are two distinct operations applied by the same rule.

Yeah, we should do that now.

Quoted 8 lines
Quoted 5 lines
Quoted 2 lines

Unless you think it's broken, I will add the patch to the debian package of 051 (almost ready).

I think we can make it even simpler, by applying the permission unconditionally, right?

Yes, seems correct.
A first test showed that this works.

I've applied it to my tree, thanks.

Quoted 2 lines

PS: Please CC: me on this topic, as I'm not subscribed to the mailing list. Thanks.

Sorry, you slipped it the To:.
Didn't you get message directly?

Thanks,
Kay

-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting Tool for open source databases. Create drag-&-drop reports. Save time by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc. Download a FREE copy at http://www.intelliview.com/go/osdn_nl
_______________________________________________
Linux-hotplug-devel mailing list http://linux-hotplug.sourceforge.net
[email protected]
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel

#7 lore raw
In reply to Michael Buesch

Quoting Kay Sievers <[email protected]>:

Quoted 5 lines
Quoted 2 lines

PS: Please CC: me on this topic, as I'm not subscribed to the mailing list. Thanks.

Sorry, you slipped it the To:.
Didn't you get message directly?

Oh, I got the message. The PS was for Marco d'Itri. ;) But I'm subscribed now. So the problem does not exist any longer.

Quoted 2 lines

Thanks,
Kay

Signature · 1 line
Regards Michael Buesch  [ http://www.tuxsoft.de.vu ]
#8 lore raw

Re: [PATCH, UDEV] fix permission inconsistency in relation to

In reply to Michael Buesch
Quoted 10 lines · Michael Buesch

Quoting Kay Sievers <[email protected]>:

Quoted 5 lines
Quoted 2 lines

PS: Please CC: me on this topic, as I'm not subscribed to the mailing list. Thanks.

Sorry, you slipped it the To:.
Didn't you get message directly?

Oh, I got the message. The PS was for Marco d'Itri. ;) But I'm subscribed now. So the problem does not exist any longer.

Yeah, but with _this_ list, which sometimes takes a couple of hours to deliver, you still have the same problem... So CC: is better, sure.

Kay

-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting Tool for open source databases. Create drag-&-drop reports. Save time by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc. Download a FREE copy at http://www.intelliview.com/go/osdn_nl
_______________________________________________
Linux-hotplug-devel mailing list http://linux-hotplug.sourceforge.net
[email protected]
https://lists.sourceforge.net/lists/listinfo/linux-hotplug-devel