Re: [Bluez-devel] rfcomm (utils)

Re: [Bluez-devel] rfcomm (utils)

[Robert Wlaschin]

In poking around in \bluez-utils-2.21\rfcomm\main.c, I noticed in
several functions (including print_dev_list) there seems to be a malloc
statement but no corresponding free.  Is this the recommended way to
handle the rfcomm_dev_list_req structures?

Thanks,
~Robert




_______________________________________________
Bluez-devel mailing list
Bluez-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-devel

Re: [Bluez-devel] rfcomm (utils)

[Marcel Holtmann]

Hi Robert,

> In poking around in \bluez-utils-2.21\rfcomm\main.c, I noticed in
> several functions (including print_dev_list) there seems to be a malloc
> statement but no corresponding free.  Is this the recommended way to
> handle the rfcomm_dev_list_req structures?

this seems to be a bug, but I would prefer you actually check the latest
version from CVS instead of an old package and be more specific of the
cases you are talking about. Otherwise we can fix them.

Regards

Marcel




_______________________________________________
Bluez-devel mailing list
Bluez-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-devel

Re: [Bluez-devel] rfcomm (utils)

[Robert Wlaschin]

Thanks for the quick response.

In reference, the current CVS available
http://bluez.cvs.sourceforge.net/bluez/ web site seems to have the same
issue.

Here's a suggested patch (using the 2.25 source)

--- bluez-utils-2.25/rfcomm/main.c.orig 2006-06-07 08:01:03.149584500
-0700
+++ bluez-utils-2.25/rfcomm/main.c.new  2006-06-07 08:02:25.915739200
-0700
@@ -132,11 +132,14 @@
 
        if (ioctl(ctl, RFCOMMGETDEVLIST, (void *) dl) < 0) {
                perror("Can't get device list");
+               free(dl);
                exit(1);
        }
 
        for (i = 0; i < dl->dev_num; i++)
                print_dev_info(di + i);
+
+       free(dl);
 }
 
 static int create_dev(int ctl, int dev, uint32_t flags, bdaddr_t
*bdaddr, int argc, char **argv)
@@ -237,12 +240,15 @@
 
        if (ioctl(ctl, RFCOMMGETDEVLIST, (void *) dl) < 0) {
                perror("Can't get device list");
+               free(dl);
                exit(1);
        }
 
        for (i = 0; i < dl->dev_num; i++)
                release_dev(ctl, (di + i)->id, 0);
 
+       free(dl);
+
        return 0;
 }


This should catch the missing deallocs.

Hope this helps.

Thanks,
Robert

Hi Robert,

> In poking around in \bluez-utils-2.21\rfcomm\main.c, I noticed in
> several functions (including print_dev_list) there seems to be a
malloc
> statement but no corresponding free.  Is this the recommended way to
> handle the rfcomm_dev_list_req structures?

this seems to be a bug, but I would prefer you actually check the latest
version from CVS instead of an old package and be more specific of the
cases you are talking about. Otherwise we can fix them.

Regards

Marcel




_______________________________________________
Bluez-devel mailing list
Bluez-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-devel

Re: [Bluez-devel] rfcomm (utils)

[Robert Wlaschin]

Marcel Holtmann <marcel <at> holtmann.org> writes:

> 
> Hi Robert,
> 
> > In poking around in \bluez-utils-2.21\rfcomm\main.c, I noticed in
> > several functions (including print_dev_list) there seems to be a malloc
> > statement but no corresponding free.  Is this the recommended way to
> > handle the rfcomm_dev_list_req structures?
> 
> this seems to be a bug, but I would prefer you actually check the latest
> version from CVS instead of an old package and be more specific of the
> cases you are talking about. Otherwise we can fix them.
> 
> Regards
> 
> Marcel
> 

Marcel,

After reading some of the other posts I came to find out that diff -u is the
preferred format.  This is using utils/rfcomm/main.c v 1.18 from the CVS 
repository.

--- bluez-utils/rfcomm/main.c.orig 2006-06-07 09:20:23.180048500 -0700
+++ bluez-utils/rfcomm/main.c.new  2006-06-07 09:21:34.899257500 -0700
@@ -132,11 +132,14 @@
 
        if (ioctl(ctl, RFCOMMGETDEVLIST, (void *) dl) < 0) {
                perror("Can't get device list");
+               free(dl);
                exit(1);
        }
 
        for (i = 0; i < dl->dev_num; i++)
                print_dev_info(di + i);
+
+       free(dl);
 }
 
 static int create_dev(int ctl, int dev, uint32_t flags, bdaddr_t *bdaddr, int
argc, char **argv)
@@ -243,12 +246,15 @@
 
        if (ioctl(ctl, RFCOMMGETDEVLIST, (void *) dl) < 0) {
                perror("Can't get device list");
+               free(dl);
                exit(1);
        }
 
        for (i = 0; i < dl->dev_num; i++)
                release_dev(ctl, (di + i)->id, 0);
 
+       free(dl);
+
        return 0;
 }

Thanks,
Robert



_______________________________________________
Bluez-devel mailing list
Bluez-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-devel

Re: [Bluez-devel] rfcomm (utils)

[Marcel Holtmann]

Hi Robert,

> After reading some of the other posts I came to find out that diff -u is the
> preferred format.  This is using utils/rfcomm/main.c v 1.18 from the CVS 
> repository.

the context diffs are not really readable by any humans. However please
attach this patch, because your mail client converted all tabs into
whitespaces.

Regards

Marcel




_______________________________________________
Bluez-devel mailing list
Bluez-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-devel