Re: Bug#558406: gnome-keyring-daemon doesn't work on GNU/kFreeBSD
reopen 558406
thanks
Hi,
Sorry to reopen the bug, but the first patch was not the good one and
it doesn't fix the bug at all.
Attached is a patch which definitely fix this bug.
I'm also CC-ing the debian-bsd@ list. Even if the patch fix the bug,
i'm not really statisfied. I suspect a deeper problem on how socket
credentials should work on kFreeBSD.
from egg/egg-unix-credentials.c
--8<- line 100 -
#if defined(HAVE_CMSGCRED) || defined(LOCAL_CREDS)
if (cmsg.hdr.cmsg_len < CMSG_LEN (sizeof *cred) ||
cmsg.hdr.cmsg_type != SCM_CREDS) {
fprintf (stderr, "message from recvmsg() was not\n");
return -1;
}
#endif
=> always return -1
>8--
If the previous test is disabled then if fails on:
--8<- line 122 -
#elif defined(HAVE_CMSGCRED)
cred = (struct cmsgcred *) CMSG_DATA (&cmsg.hdr);
*pid = cred->cmcred_pid;
*uid = cred->cmcred_euid;
#elif defined(LOCAL_CREDS)
=> uid is always 0, it should be the user id (eg. >= 1000)
>8--
The fact that the source code builds fine and
HAVE_CMSGCRED/LOCAL_CREDS are defined makes me feel that something is
weird.
I will be glad to get debian-bsd@ folks opinions on it.
Regards,
--
Emmanuel Bouthenot
mail: kol...@{openics,debian}.orggpg: 1024D/0x414EC36E
xmpp: kol...@im.openics.org irc: kol...@{freenode,oftc}
--- gnome-keyring-2.28.1-orig/egg/egg-unix-credentials.c 2009-10-18 01:27:22.0 +0200
+++ gnome-keyring-2.28.1/egg/egg-unix-credentials.c 2009-12-07 01:13:17.0 +0100
@@ -37,6 +37,10 @@
#include
#endif
+#if defined(__FreeBSD_kernel__)
+#include
+#endif
+
int
egg_unix_credentials_read (int sock, pid_t *pid, uid_t *uid)
{
@@ -45,7 +49,7 @@
char buf;
int ret;
-#if defined(HAVE_CMSGCRED) || defined(LOCAL_CREDS)
+#if (defined(HAVE_CMSGCRED) || defined(LOCAL_CREDS)) && !defined(__FreeBSD_kernel__)
/* Prefer CMSGCRED over LOCAL_CREDS because the former provides the
* remote PID. */
#if defined(HAVE_CMSGCRED)
@@ -73,7 +77,7 @@
msg.msg_iov = &iov;
msg.msg_iovlen = 1;
-#if defined(HAVE_CMSGCRED) || defined(LOCAL_CREDS)
+#if (defined(HAVE_CMSGCRED) || defined(LOCAL_CREDS)) && !defined(__FreeBSD_kernel__)
memset (&cmsg, 0, sizeof (cmsg));
msg.msg_control = (caddr_t) &cmsg;
msg.msg_controllen = CMSG_SPACE(sizeof *cred);
@@ -97,7 +101,7 @@
return -1;
}
-#if defined(HAVE_CMSGCRED) || defined(LOCAL_CREDS)
+#if (defined(HAVE_CMSGCRED) || defined(LOCAL_CREDS)) && !defined(__FreeBSD_kernel__)
if (cmsg.hdr.cmsg_len < CMSG_LEN (sizeof *cred) ||
cmsg.hdr.cmsg_type != SCM_CREDS) {
fprintf (stderr, "message from recvmsg() was not SCM_CREDS\n");
@@ -119,6 +123,18 @@
cr_len, (int) sizeof (cr));
return -1;
}
+#elif defined(LOCAL_PEERCRED) /* GNU/kFreeBSD */
+ struct xucred cr;
+ socklen_t cr_len = sizeof(cr);
+ *pid = 0;
+
+ if (getsockopt (sock, 0, LOCAL_PEERCRED, &cr, &cr_len) == 0) {
+ *uid = cr.cr_uid;
+ } else {
+ fprintf (stderr, "failed to getsockopt() credentials, returned len %d/%d\n",
+ cr_len, (int) sizeof (cr));
+ return -1;
+ }
#elif defined(HAVE_CMSGCRED)
cred = (struct cmsgcred *) CMSG_DATA (&cmsg.hdr);
*pid = cred->cmcred_pid;
Re: devd, acpiconf and such
On Tue, Dec 15, 2009 at 03:26:39PM +0100, Werner Koch wrote: > Hi! > > Looking at the freebsd kernel event system I obviously came across the > devd. Are there any plans to include it or is the plan to port the > Linux udev system to kfreebsd? The plan is to include it, the udev file system is not something that can be ported, it is to dependent on the kernel. > Looking at devd.conf and its usage I have to say that I really like > it; in contrast to udev (or whatever the current hotplug system on > GNU/Linux is) it is easy to understand and thus I assume also easy to > debug. It is much like pf.conf in contrast to iptables. The devd > code does not look too complex and it should be straightforward to > port it to glibc. > > To support acpi suspend we also need the acpiconf tool, which is even > more trivial. > > Anyone working on it? Shall I take it up? > As far as I know, no one is working on that. Ideally this should be integrated to the freebsd-utils source package (possibly producing new binary packages), but this is probably a detail, the most important job being to port it to glibc. -- Aurelien Jarno GPG: 1024D/F1BCDB73 aurel...@aurel32.net http://www.aurel32.net -- To UNSUBSCRIBE, email to debian-bsd-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Re: installation fails
On Tue, Dec 15, 2009 at 10:24:57PM +0200, Anton Andreev wrote: > Hello, > > After the debian kfreeBSD installer reboots the system then then first boot > fails. > Does anyone has the same problem? > This is due to a bug in initramfs, that mounts a tmpfs filesystem over /dev (as on GNU/Linux), it used to be workarounded in freebsd-utils, but a bug now prevents the workaround to work. The bug is fixed in freebsd-utils from unstable, but the package still needs to migrate into testing. In the meanwhile, it is possible to workaround the problem by running "ln -s /proc/mtab /target/etc/mtab" in the d-i terminal (using alt-f2), at the end of the installation, and before rebooting. -- Aurelien Jarno GPG: 1024D/F1BCDB73 aurel...@aurel32.net http://www.aurel32.net -- To UNSUBSCRIBE, email to debian-bsd-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Re: installation fails
On Tue, 2009-12-15 at 22:24 +0200, Anton Andreev wrote: > Hello, > > After the debian kfreeBSD installer reboots the system then then first boot > fails. > Does anyone has the same problem? > Is this with a d-i image? How does it fail? Any errors? I've recently used the latest d-i image with my EeePC and its seems devfs isn't being mounted for some reason (which breaks the boot process). Not sure if this is what you're referring to. -- Joshua -- To UNSUBSCRIBE, email to debian-bsd-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Re: installation fails
On Tue, Dec 15, 2009 at 10:24:57PM +0200, Anton Andreev wrote: > Hello, Hi. > After the debian kfreeBSD installer reboots the system then then first boot > fails. > Does anyone has the same problem? Can you please briefly describe your hardware and the kind of failure? If yours is VirtualBox (virtual machine), I tested that first boot fails in VirtualBox because of a VirtualBox bug. Cheers, Luca Favatella -- To UNSUBSCRIBE, email to debian-bsd-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
installation fails
Hello, After the debian kfreeBSD installer reboots the system then then first boot fails. Does anyone has the same problem? Cheers, Anton -- To UNSUBSCRIBE, email to debian-bsd-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
devd, acpiconf and such
Hi! Looking at the freebsd kernel event system I obviously came across the devd. Are there any plans to include it or is the plan to port the Linux udev system to kfreebsd? Looking at devd.conf and its usage I have to say that I really like it; in contrast to udev (or whatever the current hotplug system on GNU/Linux is) it is easy to understand and thus I assume also easy to debug. It is much like pf.conf in contrast to iptables. The devd code does not look too complex and it should be straightforward to port it to glibc. To support acpi suspend we also need the acpiconf tool, which is even more trivial. Anyone working on it? Shall I take it up? Salam-Shalom, Werner -- Die Gedanken sind frei. Ausnahmen regelt ein Bundesgesetz. -- To UNSUBSCRIBE, email to debian-bsd-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
