Re: LOOKUP_SHARED is default now
Hi. On Tue, Apr 09, 2002 at 01:08:04AM -0400, Jeff Roberson wrote: This patch has seriously reduced file system deadlocks for several people. It also makes concurrent file system access much faster in certain cases. Since I have only heard good reports and no bad reports I'm going to enable it by default. If you do experience some file system deadlocks please let me know. You may revert to the previous behavior with 'options LOOKUP_EXCLUSIVE'. I will take this away after a month or so if there are no problems. I've been struggling upgrading kernel since beginning of April, and finally found I have to add options LOOKUP_EXCLUSIVE to my kernel config file. Without LOOKUP_EXCLUSIVE, - some of the processes stall in inode state, and can't be killed by any signals - shutting down(and maybe unmounting) the system results in the panic: lockmgr: upgrade exclusive lock The stalling processes are all touching files under nullfs-mounted directories, so I think nullfs code is not yet LOOKUP_SHARED-safe. If anyone is interested, I can post the backtrace and my kernel config (after upgrading the world and rebuilding the panicking kernel). Regards. To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
Re: LOOKUP_SHARED is default now
On Mon, May 06, 2002 at 03:53:27PM +0900, [EMAIL PROTECTED] wrote: I've been struggling upgrading kernel since beginning of April, and finally found I have to add options LOOKUP_EXCLUSIVE to my kernel config file. Without LOOKUP_EXCLUSIVE, - some of the processes stall in inode state, and can't be killed by any signals - shutting down(and maybe unmounting) the system results in the panic: lockmgr: upgrade exclusive lock The stalling processes are all touching files under nullfs-mounted directories, so I think nullfs code is not yet LOOKUP_SHARED-safe. If anyone is interested, I can post the backtrace and my kernel config (after upgrading the world and rebuilding the panicking kernel). If you're feeling experimental try the change in PR kern/37270. It makes nullfs partially safe w.r.t. LOOKUP_SHARED, although there are still easily reproducable deadlocks (possible interactions with the syncer and/or vnlru processes but nothing concrete; at least, not from me). Joshua To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
Re: LOOKUP_SHARED is default now
On Tue, 09 Apr 2002 15:06:19 -0400, Jeff Roberson wrote: Right, sorry. There was some minimal discussion about this on arch quite a while ago. Basically, it allows namei to return leafs locked with shared locks instead of exclusive locks when a flag is set. This not only reduces contention, but also the number of exclusive locks that are floating around in the system. A good place for this to be documented is the NAMEI OPERATION FLAGS section of the namei(9) manual page. If you're not comfortable with the mdoc(7) markup language, you can submit your documentation change in plaintext format to the -doc list, and someone will usually do the markup and commit for you. Ciao, Sheldon. To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
Re: LOOKUP_SHARED is default now
On Wed, Apr 10, 2002 at 12:00:13PM +0200, Sheldon Hearn wrote: A good place for this to be documented is the NAMEI OPERATION FLAGS section of the namei(9) manual page. I don't believe this option will exist long enought for it to need to be documented. Jeff changed the default, but left the option incase it unexpectedly caused problems for people. AFAIK, the option will totally go away RSN. To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
Re: LOOKUP_SHARED is default now
Jeff Roberson [EMAIL PROTECTED] writes: This patch has seriously reduced file system deadlocks for several people. It also makes concurrent file system access much faster in certain cases. Since I have only heard good reports and no bad reports I'm going to enable it by default. If you do experience some file system deadlocks please let me know. You may revert to the previous behavior with 'options LOOKUP_EXCLUSIVE'. I will take this away after a month or so if there are no problems. Considering that neither LOOKUP_SHARED nor LOOKUP_EXCLUSIVE is documented anywhere, could you enlighten us as to what, exactly, they do? DES -- Dag-Erling Smorgrav - [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
Re: LOOKUP_SHARED is default now
On 9 Apr 2002, Dag-Erling Smorgrav wrote: Considering that neither LOOKUP_SHARED nor LOOKUP_EXCLUSIVE is documented anywhere, could you enlighten us as to what, exactly, they do? Right, sorry. There was some minimal discussion about this on arch quite a while ago. Basically, it allows namei to return leafs locked with shared locks instead of exclusive locks when a flag is set. This not only reduces contention, but also the number of exclusive locks that are floating around in the system. Jeff To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
LOOKUP_SHARED is default now
This patch has seriously reduced file system deadlocks for several people. It also makes concurrent file system access much faster in certain cases. Since I have only heard good reports and no bad reports I'm going to enable it by default. If you do experience some file system deadlocks please let me know. You may revert to the previous behavior with 'options LOOKUP_EXCLUSIVE'. I will take this away after a month or so if there are no problems. Thanks, Jeff To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
