May 2005 - Darwin-dev - lists.apple.com

Re: Patch to mount_nfs to make NFSMNT_RESVPORT default
by Markus Hitter 12 May '05

12 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Am 12.05.2005 um 16:49 schrieb Craig Rodrigues: I am a bit new to Darwin, so how would I submit this patch to Apple? You have at least three choices: 3) Both. HTH, Markus [1] My own experience and what I heard from mailing lists. - - - - - - - - - - - - - - - - - - - Dipl. Ing. Markus Hitter http://www.jump-ing.de/ _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… 1) Report a bug/enhancement request at <http://developer.apple.com/bugreporter/> and wait for Apple to incorporate it into one of their next releases. Apple takes bug reports seriously, but is usually next to mute to their reporters[1], even canned/automated messages were dropped about a year ago. 2) Head over to <http://www.opendarwin.org/>, join the hackers lists and ask there. One of the goals of OpenDarwin is to get enhancements/bugfixes/experimental stuff out to the public more quickly. Making your patch compatible to darwinbuild would help a lot, probably. This email sent to site_archiver(a)lists.apple.com

1 0

Re: mutex PTHREAD_MUTEX_ERRORCHECK and pthread_cancel issues
by Matt Watson 12 May '05

12 May '05

site_archiver(a)lists.apple.com Delivered-To: Darwin-dev(a)lists.apple.com On May 12, 2005, at 8:40 AM, Stéphane Letz wrote: Is there a workaround for that? This works fine for me on Tiger: #include <assert.h> #include <pthread.h> mach_port_t machThread = pthread_mach_thread_np(fThread); thread_terminate(machThread); matt. _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… I am porting code that needs the PTHREAD_MUTEX_ERRORCHECK type of pthread_mutex. But it seems that PTHREAD_MUTEX_ERRORCHECK is not supported on Darwin. I get an EINVAL error when trying to use pthread_mutexattr_settype. int main (void) { pthread_mutexattr_t a; assert(!pthread_mutexattr_init(&a)); assert(!pthread_mutexattr_settype(&a, PTHREAD_MUTEX_ERRORCHECK)); return 0; } A second problem on the behaviour of pthread_cancel that is not correct. Is the use of explicit pthread_testcancel the way to go or can the following termination code be used safely? No, it can't. In previous versions of Mac OS X, pthread_testcancel() was the only cancellation point. Some code depended on this, so we had to preserve this behavior by default. If you define _APPLE_C_SOURCE when building your pthreads code on Tiger, you should get the new (proper) cancellation behavior. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Problems building automount
by Craig Rodrigues 12 May '05

12 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com User-agent: Mutt/1.5.9i On Thu, May 12, 2005 at 08:17:37AM -0400, Andrew Gallatin wrote: > In the meantime, you can work around this by adding the -P flag to your > automount maps. Eg: > > # cat /etc/auto.home > home -P,-nosuid sw:/home > common -P,-nosuid sw:/common > <...> Hmmm....thanks for the tip! Do you know how to pass options like -P using the "-host" option in automount? I invoke automount like this: automount -m /net -host What this allows me to do is: cd /net/[hostname]/[mntpoint] I don't have any map file. We use this a lot with the Berkeley amd under FreeBSD. I'm still going to try to push Apple to pass NFSMNT_RESVPORT to the mount() command in the automount program, because it is a one line change which is very useful. Thanks. -- Craig Rodrigues rodrigc(a)crodrigues.org _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

mutex PTHREAD_MUTEX_ERRORCHECK and pthread_cancel issues
by Stéphane Letz 12 May '05

12 May '05

site_archiver(a)lists.apple.com Delivered-To: Darwin-dev(a)lists.apple.com Hi, Is there a workaround for that? mach_port_t machThread = pthread_mach_thread_np(fThread); thread_terminate(machThread); Thanks Stephane Letz _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… I am porting code that needs the PTHREAD_MUTEX_ERRORCHECK type of pthread_mutex. But it seems that PTHREAD_MUTEX_ERRORCHECK is not supported on Darwin. I get an EINVAL error when trying to use pthread_mutexattr_settype. A second problem on the behaviour of pthread_cancel that is not correct. Is the use of explicit pthread_testcancel the way to go or can the following termination code be used safely? This email sent to site_archiver(a)lists.apple.com

1 0

Patch to mount_nfs to make NFSMNT_RESVPORT default
by Craig Rodrigues 12 May '05

12 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com User-agent: Mutt/1.5.9i Hi, Can someone comment on the attached patch? This patch makes NFSMNT_RESVPORT a default option for mount_nfs, which greatly improves interoperability with Linux and FreeBSD NFS servers. I looked at the FreeBSD source code for mount_nfs, and took the change directly from there, including the new wording on the man page. I am a bit new to Darwin, so how would I submit this patch to Apple? Thanks. -- Craig Rodrigues rodrigc(a)crodrigues.org Index: src/diskdev_cmds/mount_nfs.tproj/mount_nfs.8 =================================================================== RCS file: /Volumes/src/cvs/od/src/diskdev_cmds/mount_nfs.tproj/mount_nfs.8,v retrieving revision 1.1.1.3 diff -u -r1.1.1.3 mount_nfs.8 --- src/diskdev_cmds/mount_nfs.tproj/mount_nfs.8 13 Dec 2003 15:34:40 -0000 1.1.1.3 +++ src/diskdev_cmds/mount_nfs.tproj/mount_nfs.8 12 May 2005 14:42:40 -0000 @@ -102,11 +102,15 @@ Any attempt to perform file locking operations on this mount will return the error EOPNOTSUPP regardless of whether or not the NFS server supports NFS file locking. +.It Fl N +Do +.Em not +use a reserved socket port number (see below). .It Fl P Use a reserved socket port number. -This is useful for mounting servers that require clients to use a -reserved port number on the mistaken belief that this makes NFS -more secure. (For the rare case where the client has a trusted root account +This flag is obsolete, and only retained for compatibility reasons. +Reserved port numbers are used by default now. +(For the rare case where the client has a trusted root account but untrustworthy users and the network cables are in secure areas this does help, but for normal desktop clients this does not apply.) .It Fl R Index: src/diskdev_cmds/mount_nfs.tproj/mount_nfs.c =================================================================== RCS file: /Volumes/src/cvs/od/src/diskdev_cmds/mount_nfs.tproj/mount_nfs.c,v retrieving revision 1.1.1.2 diff -u -r1.1.1.2 mount_nfs.c --- src/diskdev_cmds/mount_nfs.tproj/mount_nfs.c 13 Dec 2003 15:34:40 -0000 1.1.1.2 +++ src/diskdev_cmds/mount_nfs.tproj/mount_nfs.c 12 May 2005 14:42:41 -0000 @@ -149,7 +149,7 @@ 0, (u_char *)0, 0, - NFSMNT_NFSV3, + NFSMNT_NFSV3 | NFSMNT_RESVPORT, NFS_WSIZE, NFS_RSIZE, NFS_READDIRSIZE, @@ -305,6 +305,9 @@ case 'l': nfsargsp->flags |= NFSMNT_RDIRPLUS; break; + case 'N': + nfsargsp->flags &= ~NFSMNT_RESVPORT; + break; #ifdef NFSKERB case 'm': (void)strncpy(realm, optarg, REALM_SZ - 1); @@ -361,7 +364,7 @@ altflags = 0; break; case 'P': - nfsargsp->flags |= NFSMNT_RESVPORT; + /* obsolete for NFSMNT_RESVPORT, now default */ break; #ifdef ISO case 'p': _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

Re: launchd checkin
by Dave Zarzycki 12 May '05

12 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com You need to set the "ServiceIPC" flag to true in your plist. davez On May 12, 2005, at 2:31 AM, Braden Thomas wrote: launch_data_alloc(LAUNCH_DATA_STRING); launch_data_set_string(msg, LAUNCH_KEY_CHECKIN); resp = launch_msg(msg) launch_data_dict_lookup(resp, LAUNCH_JOBKEY_SOCKETS); ... _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… That tells launchd that you promise to talk to it, which at the very least means that you check in. Hello everyone. I'm trying to get launchd working for a TCP daemon using the "Sockets" plist entry. Unfortunately the launch interface isn't well undocumented (the lack of the launch(3) manpage referred to by launchd.plist(5)), but from reading launch.h for the API and launchproxy.c and portmap.c for an example, I was able to determine that the correct course of events goes as follows: Unfortunately I can't get this working -- response launch_data_t ("resp") is of type LAUNCH_DATA_ERRNO. The error is always 13 (EACCESS). Apparently my daemon doesn't have permissions to get the response to LAUNCH_KEY_CHECKIN...? Anyone know why I don't have permission to receive this dictionary? Anyone know how to get this working (or an alternative method to get active FDs from within the daemon?) _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/zarzycki%40apple.com This email sent to zarzycki(a)apple.com This email sent to site_archiver(a)lists.apple.com

1 0

Re: launchd checkin
by Markus Hitter 12 May '05

12 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Am 12.05.2005 um 11:31 schrieb Braden Thomas: ... "well _un_documented" ;-) ... Cheers, Markus - - - - - - - - - - - - - - - - - - - Dipl. Ing. Markus Hitter http://www.jump-ing.de/ _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… Unfortunately the launch interface isn't well undocumented ... ... (the lack of the launch(3) manpage referred to by launchd.plist(5)), ... Just as a note, launchd's sources are open: <http://darwinsource.opendarwin.org/10.4/> This email sent to site_archiver(a)lists.apple.com

1 0

CFMessagePortSendRequest + kCFRunLoopDefaultMode + Dsitributed notification
by Stephane Sudre 12 May '05

12 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… Is there a run loop mode which can be used to prevent distributed notification from being "seen" when you're waiting for the reply of a CFMessagePortSendRequest? The distributed notification observer is being registered with CFNotificationSuspensionBehaviorDeliverImmediately. Maybe there is an existing run loop mode which could be passed to CFMessagePortSendRequest to define exceptions. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Problems building automount
by Andrew Gallatin 12 May '05

12 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Craig Rodrigues writes: > > I want to improve interoperability of the automounter > with Freebsd and Linux by enabling the NFSMNT_RESVPORT > by default with this patch: That would have saved me an hour or two of frustration when setting up my first MacOSX client. In the meantime, you can work around this by adding the -P flag to your automount maps. Eg: # cat /etc/auto.home home -P,-nosuid sw:/home common -P,-nosuid sw:/common <...> Drew _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

Re: Problems building automount
by Craig Rodrigues 12 May '05

12 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com User-agent: Mutt/1.5.9i On Wed, May 11, 2005 at 11:18:48PM -0700, Kevin Van Vechten wrote: > Sorry, URLMount is a private framework, so those headers are not > available, and there's no real workaround at this time (Bad engineer, > no cookie). You could patch around the URLMount code to get > automount compiled, but that wouldn't work as desired on Mac OS X. What a shame. It defeats the point of having the automount source code open if it cannot be compiled due to missing headers. I've filed bug 4116865 with Apple, so hopefully something will come of it. I want to improve interoperability of the automounter with Freebsd and Linux by enabling the NFSMNT_RESVPORT by default with this patch: --- Controller.m.orig Thu May 12 07:42:16 2005 +++ Controller.m Thu May 12 07:44:04 2005 @@ -529,7 +529,7 @@ args.maxgrouplist = NFS_MAXGRPS; args.readahead = NFS_DEFRAHEAD; args.fhsize = sizeof(nfs_fh); - args.flags = NFSMNT_INT | NFSMNT_TIMEO | NFSMNT_RETRANS | NFSMNT_NOLOCKS; + args.flags = NFSMNT_INT | NFSMNT_TIMEO | NFSMNT_RETRANS | NFSMNT_NOLOCKS | NFSMNT_RESVPORT; args.wsize = NFS_WSIZE; args.rsize = NFS_RSIZE; #else FreeBSD and Linux NFS servers are configured by default to reject NFS mount requests from clients that don't use reserved ports. -- Craig Rodrigues rodrigc(a)crodrigues.org _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0