Hi, On Thursday, 19 October 2006 10:56, Pavel Machek wrote: > Hi! > > > > > For reference, this is the diff of this function (with the above patch) > > > > against a 2.6.13 (the last one i found around here that had this > > > > working, the > > > > kernel of suse 10.0, after that it disappeared before 2.6.16): > > > > > > > > # static int prepare_processes(void) > > > > # { > > > > # int error; > > > > # > > > > # pm_prepare_console(); > > > > # - > > > > # - sys_sync(); > > > > # - > > > > # disable_nonboot_cpus(); > > > > # > > > > # if (freeze_processes()) { > > > > > > Why exacly was sys_sync removed from prepare_processes? > > > > Because it was moved to freeze_processes(). > > > > > My last power failure during suspend-to-disk caused a journal replay > > > which took a few minutes on reboot. > > > > AFAICT, sys_sync doesn't prevent journal replies from happening on failure. > > It only causes data to be written to the storage, metadata are different > > story. > > Are you sure? > > sync should sync data and metadata, AFAICT... but it is okay to sync > it to the journal, only.
If I understood correctly what the XFS people told us, they sync only as much metadata as needed to recover everything in case of a crash, so in fact after a crash they need to reply the journal to recover the rest. -- You never change things by fighting the existing reality. R. Buckminster Fuller ------------------------------------------------------------------------- Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 _______________________________________________ Suspend-devel mailing list Suspend-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/suspend-devel