User: Password:
|
|
Subscribe / Log in / New account

Deferring mtime and ctime updates

Deferring mtime and ctime updates

Posted Aug 29, 2013 3:05 UTC (Thu) by heijo (guest, #88363)
Parent article: Deferring mtime and ctime updates

Wow, how did this apparently huge issue get missed for 20 years?

Does this *really* mean that if you write a 4GB file via mmap on a 4KB page architecture, the mtime field is updated on disk a million times, or is it mitigated somehow?


(Log in to post comments)

Deferring mtime and ctime updates

Posted Aug 29, 2013 10:21 UTC (Thu) by cladisch (✭ supporter ✭, #50193) [Link]

The inode in the page cache will be updated and scheduled for an eventual write to disk a million times. How many of those writes are merged or do actually happen depends on the caching policy, and on how seriously a journaling FS takes metadata updates.


Copyright © 2018, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds